<<

Electrochemical real-time spectrometry: A novel tool for time-resolved characterization of the products of electrochemical reactions

Elektrochemische Realzeit-Massenspektrometrie: Eine neuartige Methode zur zeitaufgelösten Charakterisierung der Produkte elektrochemischer Reaktionen

Der Technischen Fakultät der Friedrich-Alexander-Universität Erlangen-Nürnberg

zur

Erlangung des Doktorgrades Dr.-Ingenieur

vorgelegt von

Peyman Khanipour Mehrin aus Shiraz, Iran

Als Dissertation genehmigt von der Technischen Fakultät der Friedrich-Alexander-Universität Erlangen-Nürnberg Tag der mündlichen Prüfung: 17.11.2020

Vorsitzender des Promotionsorgans: Prof. Dr.-Ing. habil. Andreas Paul Fröba

Gutachter: Prof. Dr. Karl J.J. Mayrhofer Prof. Dr. Frank-Michael Matysik I

Acknowledgements

This study is done in the team of the electrocatalysis unit at Helmholtz- Institut Erlangen-Nürnberg (HI ERN) with the financial support of Forschungszentrum Jülich. I would like to express my deep gratitude to Prof. Dr. Karl J. J. Mayrhofer for accepting me as a Ph.D. student and also for all his encouragement, supports, and freedoms during my study. I’m grateful to Prof. Dr. Frank-Michael Matysik for kindly accepting to act as a second reviewer and also for the time he has invested in reading this thesis. This piece of work is enabled by collaboration with from different expertise. I would like to express my appreciation to Dr. Sandra Haschke from FAU for providing shape-controlled high surface area which I used for performing oxidation of primary alcohols and also the characterization of the provided material SEM, EDX, and XRD. Mr. Mario Löffler from HI ERN for obtaining the XPS data and his remarkable knowledge with the interpretation of the spectra on -based electrodes for the CO 2 electroreduction reaction. Mr. Fabian Waidhas from FAU for the execution of the infrared for the IPA oxidation reaction. Dr. Florian D. Speck from HI-ERN for performing the dissolution studies of platinum and platinum-ruthenium during IPA oxidation with ICP-MS. Mr. Iosif Mangoufis-Giasin from HI ERN for carrying out the experiment with the rotating disk during IPA oxidation. I would like to appreciate my very motivated master student, Mr. Felix Haase who has supported me at the beginning with technical developments of SFC-EC-RTMS and performing the electrochemical measurements. I would like to extend my thanks to Dr. Jan- Philipp Grote who has trained me during my first weeks in Max-Planck-Institut für Eisenforschung GmbH, with his technical and scientific expertise. My thanks go to Dr. Andreas Bösmann from Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) for the introduction of the DART mass and his help at the beginning phase. I’m very grateful to all my good colleagues and friends, especially, Mr. Mario Löffler, Mr. Andreas M. Reichert, Dr. Alexander Eberle, Dr. Markus D. Pohl, Dr. Susanne Spörler, Mrs. Ricarda Kloth, Mr. Iosif Mangoufis-Giasin, Dr. Dmitry Vasilyev, and Mr. Felix Derleth with the excellent provided working and your scientific inputs. I express special appreciation to Mrs. Anja Kraus and Dr. Sabine Lösel for supporting me in the initial phase of entering the country with the organization process. All other individuals, whom I did not mention by name I thank you all. I’m grateful to the HI ERN, particularly to the electrocatalysis unit for their support and the common, cheerful activities. These years here without a doubt were the most rewarding time of my . My sincere thanks go to my inspiring supervisor Dr. Ioannis Katsounaros, a passionate excellent who was very patient with me throughout these years; for his continuous supports, trust, and for sharing his deep knowledge of with me, and for fruitful discussions we had especially in the evenings. I wish him all the best for his scientific career. Finally, my special thanks go to my parents without their support, impulses, and advice throughout this endeavor I was not standing here. I would like also to express sincere gratitude to my younger brother for his supports of my parents while I was studying abroad.

II

Abstract

Electrochemical synthesis is an attractive approach to exploit the excess of renewable . However, the fluctuations occurring in the provision of renewable energy induces dynamic reaction conditions, which to rapid changes in the and consequently impose transient events. The study of the catalytic activity and selectivity of the electrochemical reactions under dynamic conditions demands the analysis of the reaction products with techniques that can operate under operando conditions with sufficient time resolution to enable capturing transients in product formation. However, the temporal resolution of the previously developed methods for product analysis is of the order of several minutes, while the existing real-time methods can only determine a limited number of products (e.g. only gaseous and volatile products) or require conditions that make electrochemical experiments difficult (e.g. low ). Electrochemical real-time (EC-RTMS) is a novel and potent methodology was developed during this thesis for the characterization of the complete series of products during complex electrosynthesis reactions or power-to-X conversions with excellent time and potential resolution. This sophisticated is a hyphenation of two inlet modified mass and it can cover a large range of products by combining two mechanisms. Moreover, it is capable of characterizing products independent of their pressure with a very high tolerance to the presence of salts. Here, EC-RTMS is coupled to a modified scanning flow cell (SFC) where the sampling of the products happens at the electrode-electrolyte interface. The potential of the methodology is demonstrated for the characterization of products during different reactions. Electroreduction of CO 2 on copper is the most representative example of a complex reaction where the power of the technique for the determination of a dozen different gaseous and liquid products in the presence of excessive amounts of salt is demonstrated. An enhanced formation of several C 2+ products over C 1 products with lower onset potential for anodized copper in comparison to pristine copper is demonstrated with potential step or sweep experiments. It is hypothesized that the effect of anodization is related to the structural changes caused by surface oxidation followed by reduction. The role of as the intermediate of the reaction is remarkable. The capability of the EC- RTMS for the characterization of products of electrooxidation of , , and 1- propanol on the surface of platinum during the potentiodynamic experiment is demonstrated. The mechanism of the ionization in liquid analysis when the reactant is ionizable and the extent of contribution of liquid products in the analysis are discussed in detail and interpretation of the mass spectra is provided. A fundamental study combining EC-RTMS and in situ vibrational spectroscopy reveals that the conversion of isopropanol (IPA) to (ACE) is selective in the operational potential range of the direct alcohol and only traces of CO 2 are formed at high potentials. However, the accumulation of adsorbed ACE on the electrode surface to a gradual electrode poisoning. Moreover, product analysis in real time reveals that alloying platinum with inactive ruthenium reduces the for the oxidation of IPA to ACE and improves the tolerance towards the poisoning. However, the investigation in real time, shows that stability of ruthenium is significantly lower compared to platinum which is detrimental for the long-lasting efficiency of the IPA fuel cell. III

Zusammenfassung

Elektrochemische Synthesereaktionen sind ein interessanter Ansatz, um den Überschuss an erneuerbarer Energie auszunutzen. Die Schwankungen in der Zurverfügungstellung von erneuerbarer Energie bewirken jedoch dynamische Reaktionsbedingungen. Diese führen zu schnell veränderlichen Elektrodenpotentialen und in Folge dessen zu Übergangseffekten. Die Untersuchung der katalytischen Aktivität und Selektivität elektrochemischer Reaktionen unter dynamischen Bedingungen erfordert die Analyse von Reaktionsprodukten mit Methoden, die während dem Betrieb mit ausreichender Zeitauflösung funktionieren, um Übergänge in Produktentstehungsraten aufzunehmen. Die Zeitauflösung von bisher entwickelten Methoden ist jedoch in der Größenordnung von mehreren Minuten und bestehende Realzeitmethoden sind auf bestimmte Produkte (z.B. lediglich Gase und volatile Produkte) beschränkt, oder erfordern Bedingungen, die elektrochemische Experimente erschweren (z.B. niedrige Elektrolytkonzentrationen). Die elektrochemische Realzeitmassenspektrometrie (electrochemical real-time mass spectrometry, EC-RTMS) ist eine neue und vielseitige Methode, die im Rahmen dieser Arbeit entwickelt wurde, um die komplette Serie von Reaktionsprodukten während komplexer Elektrosynthesereaktionen, oder Power-to-X Umwandlungen mit herausragender Zeit- und Potentialauflösung charakterisieren zu können. Diese hochentwickelte Methode besteht aus der Kopplung von zwei Massenspektrometern, deren Einlasssysteme modifiziert wurden und erlaubt es, einen weiten Bereich an Produkten durch die Kombination von zwei Ionisationsmechanismen abzudecken. Darüber hinaus kann die Methode flüssige Produkte unabhängig von ihrem Dampfdruck und mit hoher Toleranz gegenüber der Anwesenheit von Salz charakterisieren. In der vorliegenden Arbeit wird EC-RTMS an eine modifizierte Rasterflusszelle (scanning flow cell, SFC) gekoppelt, in der die Probennahme der Produkte an der Elektroden-Elektrolyt-Grenzfläche stattfindet. Das Potential der Methode zur Charakterisierung von Produkten wird am Beispiel von verschiedenen Reaktionen aufgezeigt. Die Elektroreduktion von CO 2 auf Kupfer ist das am meisten repräsentative Beispiel einer komplexen Reaktion anhand dessen die Stärke der Methode zur Bestimmung eines Dutzends an verschiedenen gasförmigen und flüssigen Produkten in Anwesenheit hoher Mengen an Salz demonstriert wird. Eine erhöhte

Bildungsrate von mehreren C 2+ Produkten gegenüber C 1 Produkten auf anodisiertem Kupfer verglichen mit reinem Kupfer wird mittels Potentialhalte- und -rampenexperimenten in der Methode gezeigt. Der Effekt tritt begleitet von einem niedrigeren Onsetpotential auf. Es wird vermutet, dass der Effekt des Anodisierens in Verbindung steht mit strukturellen Veränderungen, die durch Oberflächenoxidation und –reduktion hervorgerufen werden. Die Rolle von Acetaldehyd als Zwischenprodukt ist bemerkenswert. Des weiteren wird die Fähigkeit der Charakterisierung von Produkten aus der Elektrooxidation von Methanol, Ethanol und 1-Propanol auf der Oberfläche von Platin während potentiodynamischer Experimente durch EC-RTMS gezeigt. Der Reaktionsmechanismus der Ionisation zur Analyse von Flüssigkeiten wird für den Fall diskutiert, dass der Reaktant ionisierbar ist. Außerdem wird der Einfluss von Flüssigprodukten auf die Gasanalyse detailliert adressiert. Eine Interpretation der Massenspektren wird gezeigt. Eine fundamentale Studie aus einer Kombination von EC-RTMS und in situ Schwingungsspektroskopie zeigt, dass die IV

Umwandlung von Isopropanol (IPA) zu Aceton (ACE) auf der Elektrodenoberfläche zu einer zunehmenden Vergiftung der Elektrode führt. Außerdem zeigt eine Produktanalyse in Realzeit, dass Legieren von Platin mit inaktivem Ruthenium die Überspannung der Oxidation von IPA zu ACE erniedrigt und die Toleranz gegenüber der Vergiftung erhöht. Eine Korrosionsuntersuchung in Realzeit allerdings zeigt, dass die Stabilität von Ruthenium signifikant niedriger ist als für Platin, was nachteilig für die Langzeiteffizienz der IPA Brennstoffzelle ist.

V

Table of content

Acknowledgements ...... I Abstract ...... II Zusammenfassung ...... III Table of content ...... V List of Abbreviations and Acronyms ...... IX 1 Motivation ...... 1 1.1 Transition towards a fossil fuel-free energy landscape ...... 1 1.2 Intermittency of renewable energy generation: mismatch to the consumption ...... 1 1.3 Storage of renewable electricity ...... 2 1.4 Power to X ...... 5 1.4.1 Electrofuels: grid-scale storage of electricity ...... 5 1.4.2 A green alternative for the production of chemicals ...... 6 2 Time-resolved techniques in electrochemical processes ...... 7 2.1 Introduction ...... 7 2.2 Study of the mechanism of electrocatalysis: product/intermediate analysis ...... 7 2.3 Categorization of different analytical techniques for characterization of the products/intermediates ...... 8 2.4 Pro and cons of time-resolved analysis of reaction species over classical methods ...... 9 3 Scope and outline of the thesis ...... 11 4 Introduction to mass spectrometry and its application in electrochemistry ...... 12 4.1 Introduction to mass spectrometry ...... 12 4.2 sources ...... 13 4.2.1 Hard ionization mass spectrometry: impact ...... 13 4.2.2 Soft ionization mass spectrometry: ...... 14 4.2.3 mass spectrometry ...... 15 4.2.4 DART ...... 15 4.2.5 Mechanism of ionization in DART ...... 16 4.3 Mass analyzers and detectors ...... 20 4.3.1 and SEM detector ...... 20 4.3.2 Time of flight (TOF) mass analyzer and microchannel plate detector (MCP) ...... 21 4.3.3 Resolving power of mass analyzers ...... 23 4.4 Analysis of gaseous products in electrochemistry ...... 23 VI

4.4.1 The background: Membrane inlet mass spectrometry (MIMS) ...... 23 4.4.2 Differential electrochemical mass spectrometry (DEMS) ...... 24 4.5 Analysis of liquid products in electrochemistry ...... 25 4.5.1 General challenges of the liquid analysis with EI-QMS ...... 25 4.5.2 Selected-ion flow-tube quadrupole mass spectrometry (SIFT-QMS) vs. EI-QMS 25 4.5.2 EC-SIFT-MS vs. EC-RTMS ...... 27 4.5.3 EC-ESI-MS ...... 27 4.5.4 EC-DESI-MS ...... 28 4.5.5 Ion suppression ...... 28 5 Materials and methods ...... 30 5.1 SFC-EC-RTMS ...... 30 5.1.1 Fabrication of SFC ...... 30 5.1.2 Operational parameters of SFC ...... 30 5.1.3 Gas analysis in EC-RTMS ...... 31 5.1.4 Liquid analysis in EC-RTMS ...... 32

5.2 CO 2 electroreduction on the surface of pristine and anodized copper ...... 33 5.2.1 Electrochemical measurements ...... 33 5.2.2 Product analysis with EC-RTMS ...... 34 5.2.3 X-ray photoelectron spectroscopy (XPS) ...... 34 5.3 Electrooxidation of saturated C1-C3 primary alcohols on Pt-ALD ...... 35 5.3.1 EC-RTMS ...... 35 5.3.2 Pt-ALD preparation ...... 36 5.4 Isopropanol oxidation on Pt and PtRu ...... 37 5.4.1 Electrochemical measurements ...... 37 5.4.2 Product analysis with EC-RTMS ...... 38 5.5.3 Dissolution with inductively coupled mass spectrometry (ICP-MS) ...... 38 5.4.4 Electrochemical infrared reflection absorption spectroscopy (EC-IRRAS) ...... 38

5.4.5 Pt xRu y/C nanoparticles preparation ...... 38 5.4.6 Quantification of products during steady-state oxidation of IPA ...... 39 6 Electrochemical real time mass spectrometry (EC-RTMS) ...... 41 6.1 Introduction ...... 41 6.2 Scanning flow cell ...... 41 6.3 SFC-EC-RTMS ...... 42 6.4 Gas analysis in EC-RTMS ...... 43 VII

6.5 Liquid analysis in EC-RTMS ...... 45 6.5.1 The nebulizer-spray chamber unit ...... 45 6.5.2 Peltier controller for the spray chamber ...... 47 6.5.3 Salt trap unit ...... 48 6.5.4 Reaction chamber ...... 48 6.5.5 The parameters of inlet- ion source assembly in DART mass spectrometer ...... 49 6.5.6 Tuning parameters of the DART-TOF-MS ...... 51 6.5.7 Ionization mechanism of EC-RTMS during aqueous electrochemistry ...... 53 6.6 Unique specifications of EC-RTMS ...... 55

7 A comparative time-resolved product analysis during CO 2 electroreduction on pristine and anodized copper ...... 57 7.1 Introduction ...... 57 7.2 Results and discussion ...... 58 7.2.1 Double step chronoamperometry ...... 58 7.2.2 Sweep ...... 59 7.2.3 Detection of MeOH ...... 62 7.2.4 Characterization of the electrodes ...... 62 7.2.5 Surface area measurement ...... 64 7.3 Conclusions ...... 65 8 Real time product characterization during electrooxidation of saturated C1-C3 primary alcohols on the surface of platinum ...... 66 8.1 Introduction ...... 66 8.2 Results and discussion ...... 66 8.2.1 Characterization of Pt-ALD ...... 66 8.2.2 Electrochemical protocol ...... 68 8.2.3 MeOH electrooxidation ...... 68 8.2.4 EtOH electrooxidation ...... 69 8.2.5 1-PrOH electrooxidation ...... 74 8.3 Conclusions ...... 75 9 Study of the mechanism of electrooxidation of isopropanol with time-resolved analytics on Pt and bimetallic Pt 1-xRu x ...... 77 9.1 Introduction ...... 77 9.2 Isopropanol electrooxidation on Pt ...... 79 9.2.1 Sweep experiment with EC-RTMS ...... 79 9.2.2 Examination of adsorbates with EC-IRRAS ...... 81 VIII

9.2.3 Electrode cleaning protocol ...... 82 9.2.4 The kinetics of sorption during IPA oxidation on Pt ...... 82

9.3 IPA electrooxidation on Pt/C and bimetallic Pt xRu (1-x) /C nanoparticles ...... 84 9.3.1 Electrode cleaning protocol ...... 84 9.3.2 Sweep experiment with EC-RTMS ...... 85

9.3.3 Origin of CO 2 during IPA oxidation ...... 88 9.3.4 IPA oxidation during step experiment ...... 89 9.3.5 Open-circuit potential (OCP) measurement ...... 90 9.3.6 Dissolution of PtRu bimetallic alloys during sweep voltammetry ...... 91 9.3.7 Dissolution by changing the upper potential limit (UPL) ...... 92 9.3.8 Impact of IPA on dissolution ...... 93 9.3.9 Accelerated stress test (AST) ...... 94 9.3.10 Performance of the catalyst after AST ...... 95 9.3.12 Quantified product analysis during steady-state oxidation of IPA ...... 96 9.3.11 Detection of ACE during IPA oxidation on NPs with EC-RTMS ...... 97 9.4 Conclusions ...... 99 10 Summary and outlook ...... 101 Bibliography ...... 103 Appendix ...... 126

IX

List of Abbreviations and Acronyms

1-PrOH 1-Propanol ACE Acetone

ACE ads Adsorbed acetone ALD Atomic layer deposition AN-Cu Anodized copper APCI Atmospheric pressure chemical ionization AST Accelerated stress test ATR Attenuated total reflection CE Counter electrode CI Chemical ionization CID Collision-induced

CO 2RR CO 2 reduction reaction CO ads Adsorbed CTR Charge transfer reaction Cu Copper CV DAFC Direct alcohol fuel cell DART Direct analysis in real time DEMS Differential electrochemical mass spectrometry DESI Desorption DMF 2,5-dimethylfuran EC-IRRAS Electrochemical infrared reflection absorption spectroscopy EC-RTMS Electrochemical real time mass spectrometry EDX Energy-dispersive X-ray EI ESI Electrospray ionization EtOH Ethanol FTIR Fourier-transform FWHM Full width at half maximum HS-GC-MS Headspace gas -mass spectrometry

Hupd underpotential deposition IC Ion chromatography ICP-MS Inductively coupled plasma mass spectrometry IE Ionization energy IPA Isopropanol IPAOR Isopropanol oxidation reaction j Current density LOD Limit of detection LOHC Liquid organic hydrogen carrier LSV Linear sweep voltammetry m/z Mass-to-charge ratio X

MCP Microchannel plate ME Metastable energy MeOH Methanol MIMS Membrane inlet mass spectrometry MS Mass spectrometry MSCV Mass spectrometry cyclic voltammogram MSLSV Mass spectrometry linear sweep voltammogram NMR Nuclear magnetic resonance NP OCP Open circuit potential OCV Open circuit PA Proton affinity P-Cu Polycrystalline copper PEM-FC Proton-exchange membrane fuel cell Pumped hydro storage Pt Platinum PTR Proton transfer reaction PV Photovoltaic QMS Quadrupole mass spectrometry RDE RE Reference electrode RF Radio frequency RHE Reversible hydrogen electrode RP Resolving power Ru Ruthenium S/N Signal to noise ratio SEM Secondary electron detector SFC Scanning flow cell SIFT Selected ion flow tube SIM Selected ion monitoring TMEM Transient microenvironment mechanism TOF-MS Time of flight mass spectrometry UPL Upper potential limit VOCs Volatile organic compounds XPS X-ray photoelectron spectroscopy XRD X-Ray diffraction 1

1 Motivation

1.1 Transition towards a fossil fuel-free energy landscape The outcome of current global energy production in different sectors is the disastrous impact on the environment. For instance, one-third of worldwide greenhouse emissions is the consequence of the generation of fossil fuel-based electricity, while only 3% of the electricity (15% of European grid supply) originates from renewables 1. This urge to develop new global-scale sustainable technologies to eliminate fossil fuels in the energy sector, in the foreseeing future 2. For example, the concept of transition of energy (Energiewende) as the world energy outlook is about modernizing energy systems by migrating massively from fossil energy-predominant economy towards a renewable and sustainable society in all different sectors including mobility, , and in or domestic. For instance, targets to increase the renewable share up to 60% along with greenhouse gas emissions mitigation down to 80-95% till 2050 in comparison to 1990 3. The necessity of the energy transition is not only because of the catastrophic impact of anthropogenic greenhouse gases on the earth, namely global warming 4. The fossil fuel scarcity and the expectation that the global energy demand will increase due to population and economic growth, calls for a sustainable energy 5. The latter can be elucidated better by the exemplary Peak oil theory which is a bell-shaped curve of crude oil production versus time. This theory tells us that the extraction rate of the most representative fossil fuel in the near future reaches a maximum. More expensive and scarcer hydrocarbon deposits are co- incentive to environmental concerns to rescue our planet 6, 7 .

1.2 Intermittency of renewable energy generation: mismatch to the consumption Most types of renewable energies (, wind turbines, hydroelectric, and tidal) yield electricity. Figure 1.1 illustrates the energy consumption vs. generation from renewable energy for Germany in 2018, evidently, the imbalance between the supply and demand is substantial. This mismatch is a big obstacle for the direct incorporation of renewable resources into the electricity network. In detail, in a high share renewable electricity grid, the network is not stable and reliable, and it can lead either to wasting of electricity due to overgeneration or to a blackout resulting from the shortfall of electricity. Besides, because of the present dynamic climate change, the prediction of the consumption and generation trends is rather challenging 8. A stable, reliable, and uninterrupted (in respect of voltage and frequency) grid system based on renewable electricity needs efficient (buffer) in a large quantity with very high flexibility. In other words, the surplus renewable electricity can be stored and used in the time/location of demand. To address the importance of the latter, utility-scale storage of renewable energy became one of the top 10 emerging technologies in 2019 9. 2

Figure 1.1 Energy consumption vs. power generation from renewable resources in Germany 2018 10 .

1.3 Storage of renewable electricity The intermittency challenge and the necessity to develop efficient storage technologies will become increasingly significant, the more renewable electricity is incorporated in the energy sector. An ideal electricity storage technology should address the seven following features: (i) large capacity, (ii) long lifetime of storage, (iii) high gravimetric and volumetric , (iv) fast responding time, (v) the capability of shipping the surplus of energy, (vi) low cycle loss, and, ( vii ) being affordable (low cost of investment and operation). (i) First and foremost, the demand of the society for the energy storage system could be extended to grid-scale calls for a large-capacity storage system to save immense amounts of electricity without limitation. Moreover, (ii) the long lifetime of storage means that storage technology is capable to store energy for a very long time without losing its content (ideally unlimited time). It should be mentioned that there are many developed technologies for storing electricity in a short time scale i.e. on a daily basis, but the seasonal storage is a challenge 11 . Figure 1.2 represents, the capacity and time scale of existing technologies for storing electricity. For instance, a capacitor is capable to store a negligible amount of energy for a very short time with applications in electronic circuits. Batteries, in the middle of the Y-axis in the graph, can deliver more energy on a longer time scale which is enough to power not only small electric devices but also low-range vehicles, however, its time and size scale still is not sufficient for grid-scale storage. Saving electricity in chemical bonds can fulfill the above mentioned two of the criteria of an idealistic storage system very well. The size of storage of chemicals in practice is widespread, since the infrastructure of storage, transfer, handle safe chemicals exists, is well developed, and is very cheap to extend (sometimes an ordinary tank is sufficient to store liquid fuel). The time of keeping a stable chemical under appropriate conditions can be virtually unlimited (the most tangible example is the natural resources of fossil fuels that last over millions of years). A competitive technology to chemical storage is the pumped hydro storage (PHS), which is a large-scale (time and size) storage of electricity 3 by transferring from a lower altitude reservoir to a higher elevation. By reversing the process, the electricity can be regenerated back by a hydroelectric turbine. Regardless of the type (natural occurring or man-made reservoir), the technology is very limited by geographical constraints (e.g. hilly regions) 12, 13 .

Figure 1.2 Comparison of different electrical energy storage technologies in terms of time and capacity (adapted from 14 ).

(iii) Gravimetric and volumetric energy density mean how much energy can be stored in a unit of weight and volume, respectively. The latter is the most decisive factor for the implementation of renewable energy in the transportation sector or for transferring energy from one location to another. Figure 1.3 shows the weight and volume of three different fuel/ tank systems namely, conventional diesel, compressed hydrogen-fuel cell, and battery-driven technologies for a 500 km range vehicle 15 .

Figure 1.3 Energy equivalency in terms of weight and volume of three different fuel and tank systems for a 500 km range vehicle (reprinted with permission from 15 ). 4

As a matter of fact, for the same range, the battery-driven car requires roughly 20 and 15 times higher weight and volume of fuel system respectively compared to a diesel car. Due to this limitation, it is very hard to extend such technology from low range vehicles to heavy- duty trucks or aircraft. In comparison, the stored energy in traditional fuels e.g. jet fuel is immense that can propel an aircraft. Therefore , synthetic fuels from electrochemical processes (electrofuels) not only enable the storage of renewable electricity for the transportation sector in high energy density, but also offer a promising pathway of keeping the existing mobility infrastructure. (iv) The storage technology should be fast-responding , which means it should be capable of delivering or storing enough energy in a very short time at the moment of high consumption or generation, respectively, without restriction. The response time can be clarified better by the refueling time in the transportation terminology. Table 1.1 shows some more features of different vehicles, namely, the refueling frequency and duration. Lithium-ion battery vehicle, for instance, has the charging duration of a couple of with refueling frequency of 3 days compared to gasoline or hydrogen fuel cell vehicles that filling a car takes only a couple of minutes and it lasts for weeks.

Table 1.1 Selected properties of the different driving concepts (adapted from 16 ). Property/type of Lithium-ion Fuel cell Gasoline vehicle Plug-in hybrid vehicle battery vehicle vehicle Energy content 445 kWh 200 + 10 kWh 24 kWh 140 kWh (tank) Volume (tank) 50 liters 25 + 50 liters 90 to 170 liters 120 to 180 liters 150 to 250 kg 4 + 80 kg Weight (tank) 37 kg 20 + 100 kg (cell + system) (fuel + system) Range > 700 km 50 + 600 km < 150 km ~ 400 km Every day + Every 3 days full, Every 1 to 2 Refueling frequency Every 2 weeks every 2 weeks 30% every day weeks 3 minutes+ Refueling duration 3 minutes 0.5 to 8 hours 3 minutes 2 hours

(v) The capability to renewable energy to make a geographical balance between locations with high generation and consumption rates is a big bonus for a storage system since the production of most renewables is spatiotemporally limited 17 . For instance, solar energy as the most prevailing, abundant renewable can be exploited with photovoltaic (PV) panels, nevertheless, PV efficiency is highly dependent on solar radiation intensity. As a matter of fact, the African continent is blessed with solar energy and the solar park in deserted areas can supply the European countries or perhaps the whole world with clean energy. A case study shows that harnessing only 1% of whole deserted areas in Africa is sufficient to supply the whole world population with green solar energy by considering the current efficiency of PV panels 18, 19 . High-voltage (HVDC) lines from north African countries to Europe is a scenario to transmit electricity over long distances 20 . Aside from the huge financing it still suffers from the imbalance between supply and consumption, the transportation of electricity to the rest with this technology is not feasible, while carriers can. Hydrogen from (e-Hydrogen) can be a competent candidate 5 for transferring energy from one location to another, for instance in the form of , , or liquid organic hydrogen carriers (LOHC). (vi) The cycle efficiency of a storage system is the ratio between output to input energy. For instance, batteries can reach up to 90% efficiency whereas, PHS up to 85% 12 . On the other hand, electrolyzer and fuel cell are the most representative devices to transform the electrical to chemical energy and back, respectively. For instance, hydrogen fuel cells can reach 40%-60%, efficiencies, nevertheless, by cogeneration of heat and electricity, this can increase up to 85-90%, while the highest reported efficiency for the water electrolyzer is 80%, the roundtrip efficiency or overall energy recovery a cycle based on hydrogen fuel cell/ water electrolyzer is only up to 50%. From this aspect, still, the chemical storage of electricity is behind the batteries, therefore, chemical storage is suited more to long term storage and larger quantities of energy 21 . Last but not least, (vii) the cost of operation and investments to develop a new energy system and apply it to different sectors should be affordable and realistic not only to modern countries i.e. developing countries contribute largely to greenhouse gas emissions. As an example, the electrification access rate of some countries like Nepal is 15.4% compared to Singapore with a 100% access rate 22 . In conclusion, the storage of electricity in chemicals for energy purposes (electrofuels or chemical energy carriers) offer the use of the current infrastructure of fossil fuels, which would cut down investment costs and is the most efficient, since nearly all of the prerequisites of an ideal system are fulfilled.

1.4 Power to X Power to X is a smart way to harvest the surplus of renewable energy by converting the electricity into other forms of energy, mainly to chemical energy 23-25 . This field is currently growing rapidly to produce highly efficient fuels or value-added chemicals 26-28 . Such technologies can help the transition of industrial production of chemicals away from fossil fuels, thereby decreasing the worldwide anthropogenic emissions of greenhouse gases by one-seventh 1.

1.4.1 Electrofuels: grid-scale storage of electricity Renewable electricity can convert the chemical compounds with a lower thermodynamic energy state to a higher state (electrofuels) through electrocatalytic processes. The raw material for such conversions in the best scenario should be selected from cheap and 28 29 26 abundant chemicals such as water, CO 2 , furanic compounds . In detail, direct 30 CO 2 electroreduction to fuels or indirect by pairing the produced syngas from co- electrolysis of CO 2 to CO and water to H 2 with Fischer-Tropsch, can generate liquid hydrocarbons 31 . Producing synthetic fuel from biomass feedstock is another hot topic e.g. by converting 5-hydroxymethylfurfural (5-HMF) which is the product of dehydration of certain sugars to 2,5-dimethylfuran (DMF) 32 . DMF has superior fuel properties compared to gasoline 33 . The produced e-fuel can be used directly in a conventional way as fossil fuels and can be implemented easily in different energy sectors since the current infrastructure (like pipelines, tanks, oil tankers) is already developed. The latter enables the delocalized electricity storage by using as mentioned infrastructure and be transported to the point of consumption. Plus, the 6 stand-alone energy system is sometimes inevitable for remote locations e.g. hydrogen can be a good candidate to have a fully isolated eco-village 34 . The alternative is of course to convert back in case of specific fuels like hydrogen or alcohol to electricity in times of necessity via a fuel cell.

1.4.2 A green alternative for the production of chemicals So far efficient storing of renewable electricity in chemical bonds is discussed. Here, the target is more about the role of electrochemistry in the synthesis of valuable products in the chemical industry. Although electrochemistry is widely applied in inorganic electrochemical processes like the chloralkali to produce and hydroxide or Hall-Heroult for production of elemental aluminum on an industrial scale, the application of electrochemistry in organic synthesis over the years remained limited 35 . The first report on electrifying the organic synthesis was dated back to the 19 th century, with the pioneering research of Faraday on the electrolysis of 36 . Later on, this was followed by Kolbe who introduced a new process to make hydrocarbons from organic acids (first C-C coupling report in electrochemistry) 37 . Recently, a plethora of research is done to make simple to complex by electrosynthesis 24 . Electrosynthesis is beneficial compared to conventional chemical processes since (i) it substitutes dangerous reagents with the inherently safe pump of renewable or (ii) by that reactive, short-lifetime species or intermediates as reagents can be generated which can offer new pathways of synthesis (indirect electrosynthesis) by coupling electrochemistry to organic 27 . For these two reasons, (iii) the step of separation of reagent is omitted 24, 38 since (iv) the rate of reaction can be easily controlled by variation of the current. (v) Electrosynthesis is an alternative to the traditional chemical processes based on high pressure and temperature with (close-) ambient processes 39 . However, the main limitation is to find a pair of organic and supporting electrolyte to provide enough conductivity that reaction can be carried out with reasonable current density and low resistance of the cell as well as a suitable reaction medium 38 . (vi) Most of the chemicals like ammonia, hydrogen peroxide, methanol are manufactured in centralized restricted areas on large scales and are distributed worldwide 40 . The latter is mostly because of high temperature, pressure, and risk of danger of such processes (e.g. ~500°C, 150–300 bar for ammonia synthesis) 41 . Decentralized production means to shift the production to the point of consumption. For example, H 2O2 is produced almost exclusively from the anthraquinone process and is transported in a high (up to 70% wt% which is explosive and contaminated with stabilizer) to decrease its shipping cost. However, for most applications concentration of ˂9 wt% (for water treatment, only 0.1 wt%) is sufficient and the end-user has to dilute it again. Alternatively,

H2O2 can be produced in an appropriate concentration (up to ~2%) onsite through electrocatalytic processes in a safer and cost-efficient way 40, 42 . Therefore, synthesis with electrochemistry offers the possibility of production of chemicals onsite, where it is needed in the desired amount. 7

2 Time-resolved techniques in electrochemical processes

2.1 Introduction In the direction of the evolution of modern energy technologies for storage and conversion as well as the sustainable synthesis of valuable products, electrocatalysis has a key role. The main objective of this chapter is to discuss shortly why the development of advanced time-resolved methodologies is notably important to capture the transient occurring events under dynamic conditions and consequently elucidate some aspects of the electrocatalytic reaction mechanisms which are not accessible by classical techniques. Electrocatalysis includes processes where the happens at the surface of a heterogeneous catalyst. According to thermodynamics, the surface of the catalyst at a certain comes to equilibrium in and thin layer composition. Applying voltage (alternation of chemical potential) to an induces perturbation at the electrode-electrolyte interface and results in a dynamic zone where the environment is changing towards another (quasi-) equilibrium state. Time-resolved measurements are “movies” at the atomic level which provide direct information on how the various properties of interface evolve versus time 43 . However, tracking properties of the catalytic process at two states of the equilibrium (before and after the reaction) is not necessarily the same as during the reaction. Thus, a classical low time resolution technique at the best case provides information only about the average of these events over a certain period. The time-domain of the processes on the surface can vary a lot and the importance of time-resolved experiments is highlighted when these transformations are fast. Operando measurements are the studies of different properties of a process during the operation by employing in-situ or real time techniques. A myriad of operando techniques are developed and invented and each of them elucidates one/several specific properties of the (electro)catalytic reactions. These methodologies can be categorized based on the provided information. For instance, a correlation between catalytic structure or composition to activity/selectivity changes of the electrode can be acquired by various in-situ X-ray techniques like X-ray photoelectron spectroscopy (XPS) 44 . Assessment of the active sites can be established by e.g. in-situ Raman 45 or electrochemical scanning tunneling (EC-STM) 46 . In-situ transmission electron microscopy (TEM) or scanning electron microscopy (SEM) gives insights to morphological evolution or structural changes (reversible/ irreversible) over time such as degradation processes or aging effect down to nanometer-size regime 47 . The transient dissolution of metallic electrodes is revealed by the aid of time-resolved Inductively coupled plasma mass spectrometry (ICP-MS) 48 . One of the most important categories of techniques belongs to those to study the intermediates and reaction products which are discussed further below.

2.2 Study of the mechanism of electrocatalysis: product/intermediate analysis The kinetics of a reaction on an electrocatalyst in the simplest case comprises of the adsorption of reactant, electron transfer to the adsorbed intermediate, (in some cases dissociation or association of the intermediate), and desorption of product. However, in reality, this image can become very complex (e.g. Figure 2.1 demonstrates the proposed 8

49 mechanism of the reduction of CO 2 on copper with several parallel pathways) . The development of an efficient electrocatalyst systematically and rationally needs to study the kinetics of sorption which most of the time controls the overall process. The latter requires to get insight into the full series of products/intermediates on various surfaces and by that correlate the activity/ selectivity to catalytic sites. Such information is valuable in understanding the mechanism of the reaction as well as for engineering the structure to a more efficient, robust, selective, and active catalyst 50 . For such assessments, the coupling of electrochemistry to analytical techniques is demanding.

2.3 Categorization of different analytical techniques for characterization of the products/intermediates The kinetics of sorption can be examined by monitoring the temporal evolution of the adsorbed species directly on the electrode and the composition of the thin layer by employing different techniques such as in-situ IR. The latter has to be complemented by analyzing desorbed products. According to Figure 2.1, an electrochemical reaction can form various gaseous and liquid desorbed products. The distribution and rate of formation of such products are highly dependent on the potential as well as on the electrode material 51 . This correlation can be acquired with the classical approach namely, by characterizing the products during steady-state electrolysis (potentiostatic or galvanostatic). The product analysis can be done Figure 2.1 Proposed reaction pathways for with off-line, at-line, on-line techniques. The electroreduction of CO 2 on the surface of classification of analytical techniques is based on copper to different intermediates/products (reprinted with permission from 49 . how automated the analytical technique is. Copyright (2020) American Chemical The off-line analysis normally involves Society. the manual collection of an aliquot of the electrolyte or a certain volume of headspace that contains the liquid and gaseous products respectively with a sampling bag from the , and transfer the sample to the analytical technique for post-analysis. The main drawbacks of offline techniques are that they are time-consuming and less reproducible due to human error. To automatize the process of collection, reduce the workload, and eliminate human error, at-line or online techniques are favored. The at-line analysis which is mostly for liquid products involves the automatic collection of the products of the reactor via a fraction collector periodically and the analysis is done via an instrument that is equipped with an autosampler and sits in the vicinity. The liquid products typically are analyzed by collecting the sample either automatically by fraction collector or by the user and be analyzed with different offline or at line techniques e.g. nuclear magnetic resonance (NMR) and with ion, liquid, or 51-56 . 9

On-line techniques are fully automated in terms of sample collection and analysis methods. This includes the coupling of the electrochemical cell directly to the analytical technique with a product transfer line. The most known example is the determination of gaseous products with on-line gas chromatography 51 . However, these techniques are not necessarily real time. In other words, the frequency or time resolution of the analysis is far from the time scope of the kinetics of the process in electrochemistry. Liquid products normally are very difficult to be analyzed with online techniques, however, when the sampling from the interface happens the concentration of the product by sampling it locally from the surface can be increased to be detectable with employed analytical instrument 57 . However, when a voltammetry experiment with high time resolution is desired, due to the different time scales of voltammetry and chromatography, a liquid fraction collector to collect the sample for analysis is needed and in that way, very volatile or unstable compounds can be missed again, since the transferring the electrolyte to an appropriate container for analysis is not a closed system 58 . Real-time methods are online techniques that the time resolution of the analysis meets the time length of the electrochemical process which may happen in the range of sub-seconds to seconds. In situ techniques are named after the methods that the measurement happens locally at the interface while the process is taking place without any alternation triggered by analysis mechanism and mostly belongs to the interaction of noninvasive light e.g. in-situ spectroscopy for characterization of products or intermediates 59 (ex-situ characterization mostly describes the characterization of the surface once the event has already occurred). Mass spectrometry as an invasive technique is categorized in real time techniques where the measurement cannot happen at the interface without influencing the surface event because of the process of ionization, therefore sampling cannot happen exactly at the interface rather the generated products have to be withdrawn from the electrode to the ion source. These techniques are capable of continuously monitoring products or tracking the evolution of adsorbed species on the surface. Real time/In situ methods are rapid, with minimized errors and costs, they provide full control of the system and they are very desirable not only because they are faster but also because of their capability of elucidating the underlying mechanism of surface processes.

2.4 Pro and cons of time-resolved analysis of reaction species over classical methods The advantages of the steady-state electrolysis approach are that they offer quantitative information about product distribution, which can be used for calculation kinetic parameters like production rates (partial current) or faradaic efficiencies (selectivity). Note that, quantification of products, under dynamic conditions, is not trivial with real time or in situ techniques. Nevertheless, breakthrough studies show this capability 60, 61 . Moreover, they enable the examination of the long-term performance of the catalyst under steady-state conditions. However, there are several limitations of the classical approach in comparison to time- resolved techniques: (i) Although the calculation from theory can give hints about active, high-performance catalysts for a certain reaction, screening the materials is the fastest approach to find electrocatalytic active material for a certain reaction under real conditions. The fact is that, 10 during the steady-state experiment, only one material-electrolyte-potential combination is addressed per experiment, thus screening the materials to find a selective electrode for a new reaction under various conditions is very time-consuming, while the process of determination of selectivity/activity as a function of the potential or the electrode material is accelerated by employing time- or potential-resolved product analysis. (ii) The temporal resolution of the classical techniques is in the order of minutes (the low time resolution of the analytical technique is also for analysis, e.g. separation with chromatography) therefore, transient processes at the dynamic electrode which may occur within seconds or sub-seconds cannot be captured. (iii) To assess the kinetics of the reaction especially in the case of liquid (dissolved) products, steady-state investigations have to be performed long enough to collect the adequate product, so their concentration in the sample will be well above the of the analytical instrument and by that a lower the error of the measurement is ensured. Therefore this approach is also called often long-term electrolysis. In most cases, long term electrolysis in combination with low time resolution analytics gives valuable information about the kinetics of most reactions. However, when the kinetics of product desorption is very slow and the catalyst gets poisoned quickly (e.g. during the oxidation of many organic molecules), this approach may be inappropriate. Techniques that provide information about the transient formation of products (e.g. real time mass spectrometry) are very sensitive, therefore no accumulation of product is necessary. (iv) The formation of short-lived intermediates 62 or very volatile products may be lost by employing low time-resolved product analysis techniques 55 . The latter can be eliminated by simultaneous acquisition of the generated products via a transfer line and a close system. Real time measurement of the hydrogen peroxide during evolution is an example that is done with a rotating ring disk electrode (RRDE) because the hydrogen peroxide is quite unstable and therefore real time measurements via electrochemical sensor become important. (v) Interference of the generated products in the electrochemistry due to the cross over or exchange of the species between the counter and working electrode in the standard bulk two-compartment cell with typical membranes e.g. Nafion® can change the real product distribution and leads to misinterpretation of the mechanism of the reaction. This issue can be resolved by employing a flow cell combined with the fast analysis of the products. In conclusion, the time-resolved experiments should be complemented with steady- state electrolysis to take advantage of both approaches.

11

3 Scope and outline of the thesis

In the continuation of developing novel hyphenated analytical techniques in electrocatalytic studies, the scope of this thesis is to introduce electrochemical real time mass spectrometry (EC-RTMS) as a novel technique for the determination of the liquid and gaseous products during electrochemistry. The potential of the technique is showcased during some exemplary electrocatalytic processes. This is done by interpretation of the recorded data of these model reactions and the assignment of the mass signals to each product. This thesis is organized as follows: The objective of Chapter 4 is to give an introduction to mass spectrometry by providing fundamentals and basic information on the instrumental components of the EC- RTMS. The latter comes along with a short literature survey of the previous efforts for real time product analysis during electrochemistry. The experimental part including materials and methodologies is given in Chapter 5 where the components of the EC-RTMS, as well as the experimental conditions of electrochemical investigation for each reaction, are provided. Chapter 6 is devoted to technical details of the EC-RTMS, where the configuration of the inlet systems for both gas and liquid analysis are presented. This is followed by explaining the role of each instrumental parameter on the signal of the mass spectrometer. Chapter 7 is a proof of concept and demonstrates the distribution of products during the CO 2 reduction reaction on the surface of copper in the presence of a non-volatile electrolyte. Understanding the change in selectivity of the reaction when the surface of copper gets oxidized and reduced is a subject of extensive research in the past with classical techniques and here is aimed to be revealed in real time. Surface characterization as well as surface area measurements on two electrodes in combination with time-resolved product analysis give more insight into the correlation between surface structure/composition to activity and selectivity of the reaction. Applicability of the technique in Chapter 8 is verified for another class of reactions with an ionizable liquid reactant where electrooxidation of primary alcohols on the surface of platinum is model studied. The possibility of ionization or adequate sensitivity of the technique is examined for the detection of all previously reported products in real time. Particularly, the detection of non-volatile liquid products e.g. organic acids is enabled which were not reported during time-resolved experiments previously. Further investigation to examine the contribution of liquid reaction products in the gas analysis is provided which is stated before as one of the main sophistication of a class of real time techniques based on mass spectrometry. The target of Chapter 9 is to evaluate the activity and selectively of electrooxidation of isopropanol (IPA) to acetone (ACE) on platinum. The latter is necessary because IPA can be oxidized to ACE in a fuel cell and ACE can later be converted catalytically in another reactor back to IPA and this concept can be implemented in a carbon- free energy scenario. A fundamental study by employing real time product analysis with EC- RTMS and in-situ infrared spectroscopy gives insight into the relationship between the potential of the platinum planar electrode to products/adsorbed species distributions. The operando investigations of product analysis and dissolution of the electrocatalytic material on

Pt/C and Pt (1-x) Ru x/C (the state of the art of catalyst for electrooxidation of alcohols) are carried out to obtain information on the activity, selectivity, and stability of the reaction since the nanoparticles can be used in the membrane electrode assembly of an IPA-based fuel cell. 12

4 Introduction to mass spectrometry and its application in electrochemistry

4.1 Introduction to mass spectrometry The immediate, real-time determination of the whole series of products during electrochemical reactions at the -liquid interface has been a grand challenge for electrochemists 63 . Still, intense research is going on to develop appropriate techniques that combine: (a) high time resolution which enables tracking the transient formation of products or short-lifetime species, (b) high sensitivity to trace products with low formation rate, (c) versatility to detect products in different phases with various properties including gases, or , and (d) robustness and resilience towards high concentration of salts or corrosive . Mass spectrometry (MS) is one of the most powerful, widespread tools to characterize a broad range of species (, ions, or molecules) based on their weight and structure. Moreover, it is a highly sensitive and fast-responding technique. Because of its capabilities, other (e.g. chromatographic or electrochemical) methods can be empowered when combined with mass spectrometry. For example, it can be applied for real-time, operando characterization of products during electrochemical reactions under dynamic conditions. Generally, mass spectrometry techniques have a temporal resolution in the range of milliseconds to seconds 65 . This is longer than spectroscopic techniques, because of the lower speed of ions compared to photons and the time delay introduced by using an inlet system in mass spectrometry. Nevertheless, the temporal resolution of mass spectrometry is still sufficient to monitor intermediate or unstable reaction products 66 . On the other side, mass spectrometry offers some important advantages compared to spectroscopy, such as versatility and universality to identify a broad range of compounds. Moreover, the spectral analysis for complex matrices is simpler than in spectroscopy, where the interpretation of Figure 4.1 Block diagram of different spectra and the assignment of bands to components of a mass spectrometer, equipped specific compounds is very challenging. with (a) vacuum-based or (b) ambient ion source 64 A mass spectrometer, as illustrated in (adapted with permission from ). Figure 4.1, is composed typically of the following key components: inlet system, ionization source, mass analyzer, detector, and a signal processing system. The mass analyzer, the detector, and in most cases the ion source, have to be positioned in a vacuum housing, which is generated by a series of vacuum pumps. This is first to protect high pressure- or O 2- sensitive MS elements, but also to control the ion chemistry inside the MS, by minimizing the collision of generated ions with other species through the long mean free path. While in most cases the ion source is also preferred to operate at low pressures, some methods enable the 13 ionization at ambient pressures. Therefore, mass spectrometry techniques can be classified into two main categories based on the operating pressure of their ion source, namely, ambient and vacuum-based (see Figure 4.1). The inlet has to bring the species of interest to the ionization region without interrupting the vacuum or the intended ionization process. The ionization source is afterwards responsible for charging the transported species, which are then separated in the mass analyzer and eventually detected by an appropriate detector . Finally, the signal processor integrates the analog ion current for different mass-to-charge ratios (m/z) and generates the mass , which is the attribution of the relative abundance of ions or the signal of the mass spectrometer to m/z values.

4.2 Ion sources 4.2.1 Hard ionization mass spectrometry: electron impact The electron impact (EI) is the most important vacuum-based ion source, introduced in 1918 by A.J. Dempster 67 . The mechanism of the ionization is based on the interaction of a cascade of electrons with gas-phase molecules of the analyte in the ionization chamber (Figure 4.2). When an electron from the cascade collides with an analyte (M), one electron is expelled from the latter and results in the molecular ion (M +●) (see reaction 4.1).

M + e - → M +● + 2e - ( 4.1)

The electrons in the EI source are typically accelerated to 70 eV. This energy is much higher than the first ionization energy of most organic compounds (typically below 10 eV). Hence, the excess energy turns into kinetic energy, resulting in an unstable molecular ion which may undergo extensive fragmentation (bond rupture) and yield charged and non- charged species (S+ and N● respectively) as in (reaction 4.2).

M+● → S + + N● ( 4.2)

Diminishing the fragmentation is possible by reducing the energy of the electron stream down to the threshold of the ionization energy of the analyte. However, this can lead to considerable loss of sensitivity due to the fact that the ionization cross-section in this energy is much lower compared to the case of 70 eV 68 .

Figure 4.2 Scheme of EI ion source with the lens system (adapted with permission from 64 ).

Due to severe fragmentation at 70 eV, the EI ion source is known as a hard ionization source. Fragmentation results in a for a single compound that contains several 14 different mass-to-charge ratios (m/z), other than that of the molecular ion. The fragmentation pattern can be used to obtain information about the structure of the compound (functional groups). The EI spectrum for 1-Pentanol is shown as an example in Figure 4.3. The intensity of the molecular ion peak (M +●) at m/z=88 is very low compared to the base peak at m/z=44, + which corresponds to CH 2CHOH fragment. This means that the major part of the formed M+● dissociates to this fragment. The mass spectrum also shows other fragments, such as m/z= 31, which corresponds to the cleavage of the butyl group.

Figure 4.3 EI spectrum of 1-Pentanol (reprinted with permission from 64 ).

4.2.2 Soft ionization mass spectrometry: chemical ionization The concept of soft ionization was first introduced with chemical ionization (CI) as the vacuum-based, soft alternative of EI 64, 69 . In CI, a reagent gas in excess gets first ionized in the ionization region, by electron ionization under pressure controlled conditions (~1 torr). The resulting cloud of reagent ions, subsequently, ionizes the analyte molecules with very low kinetic energy, via ion-molecule reaction, and with minor fragmentation. This results in simpler spectra of complex mixtures compared to EI, however, the soft ionization is not capable to ionize many compounds due to low ionization energy. is one of the most commonly used reagent gases in CI. The prevailing + mechanism of ionization, in that case, involves predominantly the generation of CH 4 and + CH 3 primary ions by the bombardment of methane with high energy electrons, followed by + + the formation of secondary ions of CH 5 (reaction 4.3) or C2H5 (reaction 4.4) from reactions between the primary ions and methane gas.

+ + CH 4 + CH 4 → CH 5 + CH 3 ( 4.3) + + CH 3 + CH 4 → C 2H5 + H 2 ( 4.4)

Afterward, secondary ions ionize the analyte molecule (M) by proton transfer with + + CH 5 (reaction 4.5) and C2H5 (reaction 4.6) or hydride transfer (reaction 4.7). The mechanism of ionization depends on the functional groups and the structure of the analyte molecules.

+ + CH 5 + M → MH + CH 4 ( 4.5) + + C2H5 + M → MH + C2H4 ( 4.6) + + C2H5 + MH → M + C 2H6 ( 4.7) 15

Other gases that are often used as reagents in CI are , ammonia, or isobutane, and depending on the reagent, the ion chemistry, and consequently, the resulting mass spectrum can be different.

4.2.3 Ambient ionization mass spectrometry Even though the spectral analysis of complex reaction mixtures would be simpler for CI compared to EI, they are both still vacuum-based ionization sources; therefore, developing an interface that is permeable to non-volatile liquid products would be very challenging. Ambient ionization mass spectrometry offers a convenient solution to ionize compounds with very low vapor pressure or thermally labile analytes. The ionization mechanism has to be soft to prevent the ionization of atmospheric species, which would otherwise interfere with the ionization of the analytes of interest, and due to the relatively high abundance of the air components in the ion source, this leads to ionization suppression of the analytes 70 . From a mechanistic standpoint, the two main categories in ambient-based ionization techniques are atmospheric pressure chemical ionization (APCI) and electrospray ionization (ESI). Discussion about ESI ion source is presented later in section 4.5.3. APCI as the atmospheric version of the CI is developed to analyze the polar to less polar compounds in the liquid phase in the low mass range 71 . In APCI, the liquid is converted to a spray and, afterward, a electrode ionizes the analytes in the spray phase.

4.2.4 DART ion source Direct analysis in real time (DART) is one of the most known versions of APCI. It was invented by Robert Cody in 2003 and was commercialized in 2005 72 . DART is a revolutionary ionization method in for rapid and cheap characterization of samples e.g. in forensic science, pharmaceutics, quality control, or the food industry. It is capable to ionize a wide range of compounds in gaseous, liquid, and solid forms in different shapes, without in their native state. DART is an open-source, where ionization occurs at the region between the ion source and the inlet of the MS, giving qualitative information on the sample composition. Figure 4.4 shows the basic configuration of the DART ion source which consists of two main chambers. Initially, an inert gas (typically He or N 2) enters the DC region where a high voltage between the needle electrode and a grounded perforated counter electrode generates a cold plasma that consists of ions, electrons, and excimer species. Some of these species have long enough lifetime (metastables) that can survive to travel towards the grid electrode, while a heater is used to aid the molecules to be desorbed from their matrix. In the end, the grid electrode neutralizes the electrons and negatively charged species (in positive mode) or the positively charged species (in negative mode) from the Penning ionization region, to avoid recombination of charged species 73 . Lastly, in this mode, the positive bias of the grid electrode in DART neutralizes the electrons or anions, and positive ions of water or protonated molecules are repelled into the reaction chamber and later to the mass spectrometer.

16

Figure 4.4 Direct analysis in real time (DART) ion source for the analysis of samples in the open air (adapted from 74 ).

DART is capable of ionizing compounds with a below 1 KDa. An advantage of DART compared to the other ion sources like ESI is the simpler spectrum of mixtures because it does not cause the formation of any multiply charged species 75 and adducts of alkali . Moreover, it is very robust towards ion suppression effects, which will be discussed later in detail 71, 76 .

4.2.5 Mechanism of ionization in DART The initial step of ionization in DART in both positive and negative modes is the Penning ionization (reaction 4.8).

M*+S→ M + S +● + e - ( 4.8)

The Penning effect is based on the interaction of an excited-state neutral in the gas phase (M *) with a neutral molecule (S) that yields the corresponding positive ion of the molecule (S+●), the ground state of the atom (M), and one electron. The prerequisite for the above-mentioned reaction is that the internal energy of M * should be higher than the ionization energy (IE) of the molecule S. Helium is the most efficient gas for Penning ionization in DART. During the glow discharge of He gas, many higher energy state species are yielded, including the metastable atom of He (23S) with the internal or metastable energy (ME) of 19.8 eV 77 and a lifetime of roughly 8000 s 78 . Ions and electrons in glow discharge recombine and only long lifetime species such as He(2 3S) survive to the grid electrode.

4.2.5.1 Positive mode Since DART is an open-air source, the mechanism of the ionization of the ambient air is firstly discussed. Among the main species present in the air, nitrogen (N 2), oxygen (O 2), 3 and water (H 2O) all can be ionized directly with He(2 S) through the Penning ionization mechanism as is shown in (reaction 4.9).

3 +● 1 - He(2 S) + M → M + He(1 S) + e ME He(2 3S) ˃ IE M ( 4.9)

The above reaction is feasible for N 2, O 2, and H 2O, since the ionization energies of

= 15.58 eV, IE O2 = 12.07 eV and IE H2O= 12.62 eV are much lower than the = IE N2 ME ( ) +● +● , +● 79, 80 19.8 eV. The products of the reactions are N 2 , O2 and H2O respectively . The 17 proposed mechanism of the Penning ionization of air species is explained in detail in Figure 4.5. Among all, reaction 2 is one of the fastest, opposite to the clustering reactions 3-5 which are the slowest. As it is illustrated, the formation of hydronium ion as a primary ion can be triggered by N 2 (e.g. reaction cluster of 6→7→8→2) or O 2 (e.g. reaction cluster 10→18→2).

Figure 4.5 Ion chemistry of ambient air in positive mode of DART (reprinted with permission from 80 ).

+● The H 2O formed from different pathways can react further with another molecule of water to yield protonated water (reaction 2 in Figure 4.5). This process can be repeated, eventually, to form different protonated water clusters (reactions 3, 4 and, 5 in Figure 4.5). The ionization of the analyte molecules (M) in the positive mode of DART mainly happens via the proton transfer reaction (PTR) mechanism. The latter yield the protonated molecule (reaction 4.10) or protonated (reaction 4.11) depending on the instrumental conditions. Humidity, the temperature of the ion source, and the orifice 1 voltage (described in chapter 6) are the most important parameters that determine which form of the protonated molecule or the n in protonated water prevails as the primary ion.

+ + [(H 2O) nH] + M → [M + H] + nH 2O ( 4.10) + + [(H 2O) nH] + M → [M (H 2O) nH] ( 4.11)

Note that, charge transfer reaction (CTR) can happen in addition to PTR as a side +● 81 + reaction (reaction 4.12) with O2 or NO in the ion source especially under dry conditions i.e. when water loading is very low and air species are dominant. The mechanism of formation of such primary ions is predicted in Figure 4.6. NO + as primary hydride transfer like aldehydes 82 .

+● +● O2 + M → M + O 2 IE of M˂12.07 eV ( 4.12) 18

Figure 4.6. Mass spectrum background of helium DART-MS in the open air in positive mode (adapted with permission from 73 ).

4.2.5.2 Thermodynamics of proton transfer reaction Thermodynamics of PTR can be expressed according to the of the (reaction 4.13) reaction which is the transfer of the proton from the primary ion (PH +) to the analyte molecule (M) in the gas phase:

PH + (g) + M (g) → P (g) + MH + (g) ( 4.13) M (g) + H + (g) → MH + (g) ( 4.14) P (g) + H + (g) → PH + (g) ( 4.15)

Based on Hess’s law, the Gibbs free energy of the PTR is equal to the difference of the Gibbs free energies of reactions of 4.14 and 4.15.

ΔG0 4.13 = ΔG0 4.14 - ΔG0 4.15 ( 4.16) T ( ) T ( ) T ( )

The Gibbs free energy of the PTR is related to its according to equation 4.17. [MH+][P] ΔG0 (4.13)= -RTln K where K = ( 4.17) T eq eq [M][PH +]

The affinity of molecules to accept a proton is known as basicity. Gas-phase basicity (B) and proton affinity (PA) are the negative change of the Gibbs free energy and of the enthalpy of PTR respectively. Therefore, the ΔG = ΔH TΔS relationship can be rewritten as - in equation 4.18.

0 0 0 0 -ΔHT= -ΔGT - TΔST=PA=B - TΔST ( 4.18)

The PA of a compound has always a positive value since the proton acceptance reaction is exothermic. As the entropy does not change significantly during the PTR, the PA is a good descriptor for the feasibility of the PTR. For example, if the PA of the analyte (M) is only 10 kJ mol -1 higher than that of the primary molecule, the Gibbs free energies would also -1 differ by 10 kJ mol , which would translate to an equilibrium constant of K eq ≈ 60 for the 19

PTR at the temperature of 298 K. This additionally means that small differences in proton affinity can determine whether the reaction equilibrium is shifted towards the protonated analyte or not 68 .

4.2.5.3 Kinetics of proton transfer reaction Under normal operation conditions of DART, the PTR occurs mostly with protonated water (n=1 in reaction 4.10) or protonated water dimer (n=2 in reaction 4.10). PTR is a second-order elementary reaction with respect to the analyte and primary ion. In a simple case, when protonated water (or water dimer) and M are the only analyte present and proton donor, respectively in the ionization region, the kinetics of the reaction is:

d MH + d H O+ =- 3 =k H O+ M ( 4.19) dt dt p 3

Where [M] is the concentration of liquid product in the gas phase before ionization, + [H 3O ] is the concentration of the primary ion, and k p is the proton transfer reaction rate + constant. Provided that [M] ˃˃ [H 3O ] which is even valid in the low concentration range of M in the gas phase, then it can be concluded that:

+ + -kp M t [MH ] = H3O (1-e ) ( 4.20) 0

+ Where t is the time of PTR inside the reaction chamber, [H 3O ]0 is the initial + concentration of the primary ion, while [MH ]t is the concentration of the protonated analyte + + after time t. When [MH ] << [H 3O ]0, equation 4.20 can be simplified using the Taylor expansion to equation 4.21 68 .

+ + [MH ] = kp H3O M t ( 4.21) 0

4.2.5.4 Negative mode High energy electrons are one of the products of the Penning ionization (4.8) as the first step of DART ionization. The energy of Penning electrons (e *) depends on the energy difference between the and . Later, the high energy Penning electrons are ME ∗ IE thermalized by collision with other species and can be captured by atmospheric oxygen to -● 76 form O 2 (reaction 4.22) .

- -● O2 + e → O2 ( 4.22)

Based on the functional group and polarity of the target analyte, the mechanism of the -● ionization of the analyte from O 2 can be proton abstraction (reaction 4.23), charge exchange (reaction 4.24), or oxygen attachment (reaction 4.25).

-● - ● O2 + M → [M – H] + OOH ( 4.23) -● -● O2 + M → M + O 2 ( 4.24) -● -●* -● * O2 + M → [M + O 2] + G → [M + O 2] + G ( 4.25) 20

The background spectrum of helium DART in the negative mode in the open air is shown in Figure 4.7. The formation of some ions is predicted and discussion about the mechanism is out of the scope of this thesis.

Figure 4.7 Mass spectrum background of helium DART-MS in the open air in negative mode (adapted with permission from 73 ).

4.3 Mass analyzers and detectors 4.3.1 Quadrupole mass analyzer and SEM detector The quadrupole mass analyzer is one of the most common, cheap, and robust mass analyzers. It consists of four cylindrical rods in a parallel arrangement and each rod is electrically connected with its diagonal rod (see Figure 4.8 (a)). A potential of +(U+V cos (ωt)) is applied to the one pair and -(U+V cos (ωt)) to the other pair of rods, where U is a constant potential and V cos (ωt) is a superimposed, variable radiofrequency voltage with an amplitude V and frequency ω. As accelerated ions from the ion source region enter the mass analyzer in the traverse direction, they start to oscillate. Depending on the values of U, V, and ω, only species with a specific range of m/z ratio can pass through the analyzer with a stable trajectory towards the detector, while the other ions strike the charged rods and become neutral. The scanning of the m/z ratio is done normally by changing the applied voltage 83 . Τwo types of detectors, the , and the secondary (SEM) are typically used in EI-QMS. The choice of the two detectors is based on the desired sensitivity and linear range. The faraday cup is a rugged detector that offers fairly low sensitivity but a higher upper linearity limit, hence it is suitable for the detection of analytes in high concentration. As shown in Figure 4.8 (c). the detector consists of a rectangular box or cylinder with a small orifice that is intentionally designed to capture and trap the incoming ions. When the beam of ions collides with the metallic surface, a current for detection is generated. The SEM is more sensitive than the faraday cup but with a lower upper linearity limit, therefore it is used for the detection of trace gas concentrations. The operation principle of the SEM detector is illustrated in Figure 4.8 (a) and (b). First, the stream of ions (here cations) hits the high voltage conversion dynode (electrode with opposite polarity), which releases electrons. These electrons are accelerated towards the continuous dynode (here horn- shape ) where several secondary electrons are induced. This process is repeated multiple times and eventually, a cascade of electrons (10 6 to 10 8) is liberated to generate a significant current 74 . This amplification is responsible for the increased sensitivity of this detector but also for its saturation with a large number of ions. 21

Figure 4.8 Schematic of (a) EI-QMS (b) secondary electron multiplier detector (c) Faraday cup (adapted with permission from 64, 74 ).

4.3.2 Time of flight (TOF) mass analyzer and microchannel plate detector (MCP) TOF is a non-scanning, broad mass range (without upper limit), high resolving power mass analyzer, with rapid acquisition rate (ideal for capturing the transient formation ions from analytes during e.g. electrochemistry) compared to the scanning quadrupole counterpart. The mechanism of separation of ions in TOF is based on the different times that accelerated ions of different mass need to travel in a field-free drift, evacuated tube. In order to make the continuous ion stream compatible with intermittent TOF analysis, a pulser plate in an orthogonal position has to be employed (see Figure 4.9 (a)). The latter means that the sampling or the injection of the ion stream into the flight tube happens perpendicular to the trajectory vector of the continuous ion stream. Separation of the packet of ions in TOF happens very fast (about 16,000 injections of ions per second in JEOL AccuTOF TM ). For a pulser applied voltage V, an ion with charge z gains equal to zeV, which is transformed into kinetic energy. Since the velocity ( ) of travelling ions is equal to the length of the flight tube (L) divided into the time of flight (t), the following equation 4.26 can be deduced.

1 1 L 2 zeV= mυ2= m( ) ( 4.26) 2 2 t

By rearranging the latter, equation 4.27 results.

L m t=( ) ( 4.27) 2eV z √

When L and V are kept constant, the time of flight is equal to the square root of the m/z value. Ideally, ions receive the same kinetic energy from the pulser. However, as it is illustrated in Figure 4.9 (a), ions that are closer to the pulser plate gain more energy than distant ions, which limits the resolving power of the TOF analyzer and causes divergence in the drift trajectory.

22

Figure 4.9 Schematic representation of (a) Reflectron time of flight mass analyzer (b) cross-section of microchannel plate and electron multiplication from an incident ion within a single channel (c) assembly of dual MCP and electronics of the detector in JEOL AcuuTOF (adapted with permission from 64, 84 ).

One approach to increase the resolving power is to use reflectron in the flight tube. Reflectron is an ion mirror system that corrects for time-of-flight divergency for ions with the same mass but with a different kinetic energy. This happens by retarding more the ions with higher kinetic energies. In contrast to quadrupole, the stream of ions separated from the TOF analyzer is not focused but scattered over a large area. Additionally, TOF is a fast mass analyzer that demands a fast-responding coupled detector. These two requirements can not be covered with one single long channel in an SEM detector. Instead, short-time pulses of ions can be tracked with an MCP detector which is a two-dimension, large-area plate consisting of an array of short-length channels. As Figure 4.9 (b) illustrates, MCP consists of many parallel separate cylindrical channels with approximately 0.6 mm thickness and 10 µm inner diameter and a distance of 12 µm between the center of two channels. With two-ply (dual) MCP of JEOL AccuTOF TM , each striking ion can generate 10 6 cascades of electrons. In the end, electrons are captured by an electrode resulting in an electric signal. The analysis involves the integration of these very fast pulses over a fraction of the second to have a sufficient S/N ratio. The MCP, however, has some disadvantages, such as limited dynamic range because of saturation as well as sensitivity to air and humidity (a supply of dry nitrogen during venting mass spectrometer is necessary).

23

4.3.3 Resolving power of mass analyzers Resolving power (RP) is reciprocal of mass resolution (R), that they are sometimes in mass spectrometry incorrectly interchanged. RP is the power of the MS to distinguish between two adjacent peaks with distance Δm. There are two approaches to determine the RP according to equation 4.28, namely, (a) valley definition and (b) peak width definition 74 .

1 m m RP= = a or (b) ( 4.28) R ∆m ( ) m1 2 Based on the valley definition, m is the mass of interest and the peak separation (Δm) is the distance between the two adjacent peaks with equal intensity when the overlap of two peaks (valley magnitude) is 10% of the total height. Since finding two equally magnitude peaks with specific peak separation conditions is very hard, valley definition is seldom used. The peak width definition is more straightforward as RP can be calculated with only one clearly separated peak and full width at half maximum (FWHM). Figure 4.10 shows the RP calculation based on two definitions. The RP value according to valley definitions is roughly half for the same peak separation in comparison to the FHWM definition (compare the (a) and (b) in Figure 4.10) 74 .

Figure 4.10 Comparison of peak separation based on (a) valley definition for a mass spectrometer with RP=500 and full width at half maximum definition when (b) RP=1040 and (c) RP=500 (reprinted from 74 ).

The RP is not the same for different m/z values, therefore for reporting the RP values, the mass has to be mentioned. Quadrupole mass analyzers, for instance, have the same line width (Δm) throughout the whole mass range which means, the separation of the mass 28 and 29 is the same as mas 500 and 501. Therefore, according to both definitions, the RP of quadrupole increases with the mass, but still, it is far behind the higher resolution TOF mass analyzer.

4.4 Analysis of gaseous products in electrochemistry 4.4.1 The background: Membrane inlet mass spectrometry (MIMS) EI-QMS is the most known mass spectrometry technique that has been coupled to electrochemical processes for the detection of products in real time. Since the EI is a vacuum- based ion source, its combination with classical aqueous electrochemistry needs the 24 development of an inlet system that acts as an interface and withstands the pressure difference. Additionally, it should separate the products in an efficient way from the electrolyte phase without permitting a large amount of liquid to pass through to the mass spectrometer in order to keep the vacuum of the MS in the operating range. Membrane inlet (introduction) mass spectrometry (MIMS) was developed in 1963 to be able originally to analyze the gases and dissolved species in the liquid phase 85 with a semipermeable nonporous membrane as the interface. The mass transport or steady-state flux (J) of the species through the membrane is dependent on the surface area (A) as well as the thickness of the (l) of the membrane, the diffusion coefficient of the analyte through the membrane (D), the partition coefficient of the analyte between the membrane and the liquid phase (K), and the concentration of the analyte in the liquid phase (C s) as in equation 4.29.

ADKC J s ( 4.29) ∝ l

The technique was used extensively for monitoring the analytes during the chemical, environmental or biological processes with a very low limit of detection (LOD), however, the response time of MIMS or t 10-90% (the required time that signal rises from 10 to 90% of the steady-state flux) for a certain analyte is dependent to the diffusion constant and thickness of the membrane (equation 4.30) 86, 87 .

l2 t ( 4.30) 10-90% ∝ D

Therefore, a drawback of MIMS is that the response time is directly dependent on the diffusion of the analyte molecules. Small gaseous molecules typically have a high diffusion coefficient, so the time response is relatively short (in the seconds range) and therefore they can be well analyzed with MIMS. This is not the case, however, for volatile organic compounds (VOCs) which diffuse more slowly, resulting in long response times.

4.4.2 Differential electrochemical mass spectrometry (DEMS) Due to the poor time resolution of MIMS for the analysis of VOCs, DEMS was introduced as an alternative version in 1971 88 . The time response of DEMS is improved by utilizing a thin nonwetting porous membrane (typically Teflon PTFE) with a small pore size (20 nm), which is mechanically supported with a metal or glass frit 89 . The porous structure enhances the mass transport of the analyte to the mass spectrometer. Even though the porous membrane is hydrophobic, it passes through a lot of water (in the case of aqueous electrochemistry), so the area of the membrane has to be limited to keep the pressure of EI- QMS in the operating range below 1×10 -5 torr. However, by restriction of surface area, the amount of products in mass spectrometer would not be sufficient. An alternative solution is to use a large area of the membrane (to have a large exposed area of membrane in contact with electrolyte) with a differential in the inlet system in T- configuration to remove extra gas including water and analyte species. Since DEMS was designed to follow the rate of formation of products during electrochemistry, the term “differential” in DEMS is to distinguish it from integrating techniques 63 . In comparison 25 to MIMS, DEMS has a higher limit of detection (LOD). Therefore, to gain high sensitivity, DEMS typically involves the sputter-deposition of the catalyst (ca. 0.1 μm) directly onto a porous hydrophobic Teflon membrane, so the probing is done at the interface where the product concentration is higher. Additionally, this configuration keeps the response time short (˂0.1 s) for both gases and VOCs. However, only sputter-deposited electrodes can be addressed in the classical configuration. To enable investigations on various electrode types and shapes, several inlets and cell modifications of DEMS have been presented over the last 40 years [71-75]. For instance, online electrochemical mass spectrometry (OLEMS), is a modified version of DEMS that makes the analysis of gaseous and volatile products especially with single crystal in hanging meniscus configuration possible. This is done by bringing a small capillary probe, with a membrane end, very close to the surface of the working electrode, without differential pumping by sacrificing the time resolution compared to original DEMS design 90, 91 . The diffusion of liquid products through the hydrophobic membrane highly depends on their vapor pressure, therefore products with low vapor pressure (e.g. organic acids) cannot permeate 92 . In both MIMS and DEMS approaches, ion current of a certain mass during electrochemical techniques like cyclic voltammetry can be acquired and the graph of ion current for certain m/z vs potential, so-called mass spectrometry cyclic voltammetry (MSCV), can be obtained 63 .

4.5 Analysis of liquid products in electrochemistry 4.5.1 General challenges of the liquid analysis with EI-QMS Although EI-QMS in a robust, cheap, and versatile instrument for real time characterization, when used for product analysis in electrocatalysis it is limited to gas phase or thermally stable VOCs only (non-polar to mid polar molecules in low mass range), due limitation of the interface and mechanism of ionization and therefore, analysis of non-volatile products are not possible to be detected directly. To address this limitation, an approach is to periodically collect electrolyte aliquots from the electrode vicinity and post-analyze with chromatography techniques, which however is detrimental for time resolution 93 . Moreover, EI is a hard ion source that causes extensive fragmentation. While this is helpful to determine the structure of molecules when the MS is connected to separation techniques, like GC, when coupled to an electrochemical cell a complex mixture of analytes is fed to the instrument and the interpretation of spectra can become very challenging or even impossible. The latter is discussed further in detail as below.

4.5.2 Selected-ion flow-tube quadrupole mass spectrometry (SIFT-QMS) vs. EI-QMS The principle of ionization in ionization with SIFT-MS is pretty much similar to + +, + DART-MS, where the primary ions, including H 3O , NO and O 2 are produced by discharging the moist or dry air in a plasma source. Later, a quadrupole mass filter + selects only one specific primary ion like H 3O and it delivers them to a reaction chamber, + where the ionization of the analyte molecules takes place (in the case of H 3O the mechanism is PTR). Eventually, the generated ions are separated with another quadrupole mass analyzer and eventually detected with an SEM detector 68 . 26

Figure 4.11 shows a comparison between EI-QMS and SIFT-QMS as hard and soft ionization sources respectively for the analysis of the same mixture of VOCs. It is apparent that assigning a mass to a certain compound in the case of SIFT is quite straightforward, while EI yields a very sophisticated spectrum. The latter is because, for most of the , there is a multi contribution of different compounds in a single mass. To analyze this mixture of compounds with EI, it is necessary to perform deconvolution, namely to subtract the contribution of each compound to different masses 94 . This, however, requires prior knowledge of the compounds that are present in the matrix. Finally, the fragmentation caused by the EI source can be detrimental for sensitivity, as the compound breaks into many fragments.

Figure 4.11 Comparison of mass spectra of a mixture of VOCs by (a) EI-QMS as hard ionization + 95 source and (b) SIFT-QMS as soft ionization source in H 3O mode (adapted from ).

The general drawback of both EI-QMS and SIFT-QMS is the limited resolving power of the quadrupole mass analyzer. This makes the discrimination of compounds with very close m/z values (for example CO, m/z=27.9949 and N2, m/z=28.0062) very difficult, as the 27 quadrupole can in most cases separate the m/z= 27 from m/z= 28. For such separation, a high- resolution mass analyzer like EI-TOF-MS would be required.

4.5.2 EC-SIFT-MS vs. EC-RTMS Recently breakthrough studies employed the selected ion flow tube mass spectrometry 96, 97 (SIFT-QMS) to detect the multiple liquid and gaseous products in real time of CO 2RR . Aside from the power of the technique, SIFT is equipped with a quadrupole mass analyzer which has limited resolving power and determination of a mixture of unknown liquid products that have higher molecular weight compared to gaseous products imposes a limitation the time resolution. The latter is because that quadrupole is a scanning mass analyzer and the required time of non-targeted analysis (the whole spectrum) can be as long as 90 s, while in selected ion monitoring (SIM) mode can reach only down to a couple of seconds, depending on the number of intended analyzing masses. TOF is superior from the time resolution aspect that nontargeted spectra can be acquired in the ms range. Another weak point of SIFT-MS compared to EC-RTMS is that detection of most permanent gases like hydrogen, oxygen, CO,

CO 2, etc. is not feasible due to the relatively soft ionization mechanism. Last but not least, the main disadvantage of SIFT-MS compared to DART-MS in EC- RTMS is that in general has a higher detection limit. For example, by considering the developed inlet system for DART-MS, the dilution of the liquid stream is roughly 10 3 times, since the typical flow rates of the liquid stream and nebulizing gas in the nebulizer are 0.5 mL min -1 and 0.5 L min -1 respectively (it will be discussed later in chapter 6). By taking to account that LOD of analytes in the liquid phase can reach down to 1-10 ppb and a 1000 times dilution factor of the liquid phase in the nebulizer, the sensitivity of 1-10 pptv= 1-10 part per trillion by volume in the gas phase is achievable. The reported sensitivity of SIFT-MS in the gas phase is however in the range of a few ppbv to a few hundred of pptv 98, 99 . The high sensitivity of the technique is very crucial to analyze the products directly on the electrode.

4.5.3 EC-ESI-MS The first attempt to analyze the nonvolatile electrochemical products in real time was introduced in 1986 by coupling electrochemistry to ionization mass spectrometry with a fair time resolution 100, 101 . Shortly after the invention of APCI or ESI sources, due to superiority, the application of the thermospray ion source diminished 102 . ESI as the successor of the thermospray is one of the most employed ion sources for analysis of liquid products during electrochemistry which was firstly coupled in 1985 103, 104 . The mechanism in ESI involves nebulizing a stream of liquid with a metallic nebulizer capillary (mostly stainless ) along with a coaxial flow of nitrogen as spray gas. The voltage of the capillary is maintained in the kV range with respect to the mass spectrometer. Due to the , the spray droplets get charges. As they get smaller due to the evaporation of the solvent, the charge density increases until the rayleigh limit where the surface tension of the liquid droplet cannot maintain the charge anymore and the coulombic explosion happens and gas-phase ions produce. Based on the explained mechanism, there are several inherent restrictions of coupling ESI to electrochemistry. Since the ionization is based on the electrophoretic process, spraying pure water is problematic due to high surface tension and the possibility of electric discharge 28 in ESI which causes the damage of components of the mass spectrometer 105 . The latter represents the incapability of the source for detection of liquid products in water e.g. during

CO 2 electroreduction. Thus, using an organic solvent like methanol is necessary to decrease the surface tension and increase the volatility of the liquid droplet. Another issue of such coupling arises from a high concentration of the supporting electrolyte during electrochemistry. Acidic or alkaline cause drastic corrosion of the metallic parts of ion source 106 , while nonvolatile salts like phosphate buffer can block the ion source in addition to ion suppression issues that will be discussed later. Moreover, due to the high voltage of ESI, undesirable electrochemical reaction can be induced and this effect was first identified by anodic corrosion of electrode emitter capillary in positive mode by detection of metallic ions in the mass spectrometer 107 . Water as the main constituent of liquid stream in ESI emitter can get oxidized and reduced in positive and negative mode respectively that cause a change of pH, especially in a nonbuffered solution up to 4 units 108 . When an oxidation process happens at the capillary emitter, a reduction process should happen at the mass spectrometer counter electrode side and vice versa. By controlling the current between these two electrodes, amperometric measurements are possible to be done in the ESI ion source (in source electrochemistry) although, the solution air interface resistance limits the faradaic current 108 . To perform potential controlled measurements with an electrochemical cell that is coupled to ESI, however, the potential of the WE can not be controlled without influencing the potential of ESI emitter capillary 104 . Due to this fact coupling of an electrochemical cell to ESI-MS is challenging since the high voltage of ion source can cause backward current to an electrochemical cell, therefore electrochemical cell has to be decoupled to have proper potential control, either by a long transfer line or inserting a ground point between EC and ESI 108, 109 .

4.5.4 EC-DESI-MS Desorption electrospray ionization (DESI) is an ambient mass spectrometry technique is introduced in 2004 as the modified version of the ESI 110 . It is developed to analyze the analytes without sample preparation similar to the DART ion source. Shortly after invention, its coupling was demonstrated for the characterization of the liquid products during electrochemistry 66, 111 . Some of the mentioned problems of coupling the ESI source with electrochemistry is less critical with DESI since the generation of the primary ions is decoupled from the ionization of the analytes, for instance, in source redox reactions in DESI compared to ESI is not significant 112 .

4.5.5 Ion suppression Ion suppression is a well-known phenomenon in mass spectrometry which means that the ionization efficiency of the analyte of interest decreases due to the competition with other species including ions or easily ionizable molecules that are present in the matrix during the process of ionization. In comparison to the DEMS or MIMS, where the gaseous and volatile analytes are separated by a membrane from the matrix especially the ions, in ambient mass spectrometry, the considerable presence of ions or ionizable reactants (from the electrolyte) can quench the ionization process in the ion source. To circumvent the latter a separation technique can be used prior to MS in the cost of time resolution of the technique 113, 114 . 29

DART is less prone towards ion suppression compared to other ambient ion sources (DART has less than 11%, APCI between 20% to 90%, and ESI ion source between 26 to 80%) 115 . The advantage of the high salt tolerance of DART is exhibited by coupling the ion source to the capillary since it needs a high concentration of buffer solution for its operation 116 . Moreover, sometimes the mobile phase containing the nonvolatile buffer during the separation with liquid chromatography is unavoidable, which leads to quenching of other ionization sources like ESI 117, 118 . DART is also successful to tolerate complex matrices like biological or crude oil samples without the need for sample pretreatment 119, 120 . However, all the ionization sources including DART have a limit of tolerance. For example, when highly alkaline and acidic electrolytes cause the depletion of the primary ions in positive and negative modes of DART, respectively. In other words, based on the mechanism of ionization in DART, PTR in the presence of an excess of hydroxide ions or proton abstraction in the presence of abundant hydronium ion can not take place. 30

5 Materials and methods

5.1 SFC-EC-RTMS 5.1.1 Fabrication of SFC A computerized numerical control (CNC) machine (Impression CAM 4-02, vhf camfacture AG, Germany) was employed to construct the scanning flow cell (SFC). This machine is originally developed to make delicate pieces for dental and jewelry applications with a maximum dimension of 120 x 90 x 70 mm in the μm range of precision. The cell was designed first in Rhinoceros 3D software and subsequently transferred to the Rhinocam module to program the drilling and milling steps. The cell for the CO 2 reduction reaction (CO 2RR) measurement was made from a block of polycarbonate (PC). Despite PC is a fully transparent and good processable polymer, in contact with a high concentration of organic substances in the electrolyte (e.g. in alcohol oxidation reaction), it starts to leach and decompose which leads to contaminated electrochemical measurements. The best alternative for organic contained electrolyte is the Teflon fluorinated propylene (FEP) polymer, which is semi-transparent and very good chemical resistant material. Transparency of the cell material is helpful to visualize bubble formation which can interfere with the measurements, during reactions with excessive gas formation such as the CO 2RR. After finishing the machining process, the cell was polished with fine sandpaper, it was cleaned in isopropanol, and was boiled several times in water. A methyl deactivated fused silica capillary with the size of (ID 530 µm, OD 680 µm) with the support of a metal tube is implemented later in the cell as a product analysis channel. The position of the capillary was fixed by placing the bottom of the cell on a flat surface; then the capillary was inserted in the cell until it was touching the surface, and finally, an epoxy glue was applied. After drying the glue, the position of the tip of the capillary and the bottom of the SFC cell were identical. The tightness between the SFC and the working electrode was ensured with a 100-200 μm silicone ring. A homemade silicon gasket (was made by mold casting the two- component SF45 two-component glue (Silikonfabrik, Germany)) was applied to the bottom of the cell for sealing the interface between the working electrode and the cell. The area of the elliptical opening of SFC was determined by the M125 C stereoscope (Leica, Wetzlar, Germany) at 12.5 mm 2.

5.1.2 Operational parameters of SFC All electrochemical measurements together with real time product analysis were performed with SFC. The setup comprises different components including a , positioning system, force sensor, mass flow controllers, and a pump which all were controlled with an in-house- developed LabVIEW software (National Instruments, Austin, USA). Electrochemical measurements were done with a Reference 600 potentiostat-galvanostat (Gamry, Pennsylvania, USA). To facilitate the positioning of the SFC over the catalyst location, which is especially important in the case of dropcasted nanoparticle films, the electrode was placed on an XYZ-translational stage (Physik Instrumente GmbH, Germany). The SFC was fixed on a force sensor (KD45, ME-Meßsysteme GmbH, Germany) to control the distance between the working electrode and the SFC. The flow rate of the reaction or purging gas was controlled via an EL-FLOW® Prestige mass flow controller (Bronkhorst, 31

Netherlands). A Reglo ICC peristaltic pump with 4 independent controllable channels (Ismatec, Germany) propelled different liquid streams with certain flow rates. The pump had to be calibrated to maintain the electrolyte in the purging vessel at the same level (see Figure 6.2) which is important for the stability of the meniscus in SFC. The calibration had to be done regularly since the tubing stiffness changes over time. Moreover, changing the electrolyte with various viscosities or alternation of the transfer line may alter the backpressure and changes in flow rate. All the tubing for the flow cell setup were selected from chemical resistance materials like PFA or Viton ® and prevent contamination of electrolyte. The potential of the working electrode was measured in reference to the Ag/AgCl/ 3 M KCl electrode (BASi®, Indiana, USA). The reference electrode was calibrated before measurement versus the reversible hydrogen electrode (RHE) and all the potentials throughout the thesis are expressed versus the RHE scale. The conversion of the measured potential to the RHE scale was done always by employing equation 5.1.

ERHE = E Ag/AgCl + E Ag/AgCl vs. SHE + 0.059 × pH ( 5.1)

where E RHE is the calculated potential versus RHE, EAg/AgCl is the measured potential during the experiment versus the Ag/AgCl electrode, and EAg/AgCl vs. SHE is the potential difference between the Ag/AgCl reference electrode and saturated hydrogen solution (SHE). Note that, the potential of the Ag/AgCl electrode may drift over time, therefore the electrode has to be calibrated regularly by measuring the open circuit potential between the Ag/AgCl electrode a platinum in hydrogen saturated 0.1 M HClO 4 solution, and by taking into account the potential shift caused by the pH of the latter solution in comparison to the standard conditions (pH=0). The electrolyte resistance in all measurements was determined by electrochemical impedance spectroscopy (EIS) and compensated with positive feedback to 90% of the measured value. The remaining uncompensated resistance was always less than 10 .

5.1.3 Gas analysis in EC-RTMS The gaseous products were detected with a MAX300-LG membrane inlet mass spectrometer (Extrel, Pennsylvania, USA). The original inlet of the mass was replaced by the homemade degasser. The backbone of the degasser is a 1/8” stainless steel T-Piece (Swagelok, USA). Then a sandwich structure of two pieces of 2 cm methyl deactivated fused silica capillary (ID 530 µm, OD 680 µm) were inserted from two sides of a 2 cm piece of Teflon® AF 2400 non-porous membrane tubing with 0.024” ID and 0.032” OD (Biogeneral, California, USA). Since the glass capillary is rigid and Teflon AF tubing is shapeable, the glass capillaries were pushed towards the Teflon tubing gently until the short end of the capillary was encircled by Teflon AF tubing and the junction were tight. Eventually, the assembly of tubings was positioned in the center of the metal T-piece. Two reducing ferrules made of 15% Graphite 85% Vespel with ID 0.8 mm (Trajan Scientific, Australia) were employed to make the assembly vacuum-tight inside the T-piece. A 1/16” stainless steel (Swagelok, USA) tubing was used as a transfer line to connect the degasser to the inlet of the mass spectrometer. The EI-QMS is equipped with yttrium oxide coated iridium filament as the ionization source, with a mass range of 2 to 250 amu. The energy of the EI source was set 32 to 70 eV for all the product analysis measurements. The temperature of the ionizer heater is set to 200 °C during all experiments to avoid liquid droplet formation inside the EI ion source. The instrument is equipped with two detectors, namely, the Faraday cup and electron multiplier. For the experiments performed in this thesis, the electron multiplier (M×1) was used except for the detection of CO 2 during methanol electrooxidation, where the Faraday detector (F×256) was used due to its high formation rate. During operation, the equilibrium pressure of EI-QMS was between 1x10 -6 to 8x10 -6 torr depending on the reaction, for instance, the pressure of the mass spectrometer was higher during CO 2RR since the gas permeability factor for CO 2 is very high (refer to the table in Figure 6.3). Note that the filament and MCP detector in EI-QMS should be turned on only when the pressure of the mass spectrometer is below 10 -5 torrs.

5.1.4 Liquid analysis in EC-RTMS After the degasser, the depleted electrolyte from gaseous products was directed via pump to a MicroFlow PFA-ST Nebulizer (Elemental Scientific, USA) which creates the mist with the aid of a nebulizing gas. The flow rates of both nebulizing gas and helium for ionization source were controlled with two EL-FLOW® Prestige mass flow controllers (Bronkhorst, Netherlands). The nebulizer was screwed tight in a glass baffled cyclonic spray chamber (for NexION 300/350, Perkin Elmer, USA) to sort the droplets of the spray and to fine and big. The larger condensed liquid drops were removed from the bottom of the spray chamber with a REGLO digital peristaltic pump (ISMATEC©, Germany). The outlet of the spray chamber which contains fine droplets was directed to the ionization region of the DART mass spectrometer through a heated Hastelloy line (HPS solutions GmbH, Germany) by positive pressure. The spray chamber transfer line should be resistive towards the harsh corrosive jet of electrolyte spray as well as it must be thermally conductive to be heatable. Joule heating is the principle of a heated transfer line where a nickel-chromium coated resistance wire is wrapped around the metal transfer line in a spiral form and the temperature of the line was controlled by setting current and potential with a power supply and a thermocouple in contact with metal used for temperature measurement. The transfer line should be designed in a way, that no condensation happens, and for safety, it should not let any liquid droplet to the inlet of the MS, since liquid, quench the vacuum of the mass spectrometer secondly, it corrodes the parts. The temperature of the spray chamber was controlled by a PC3 Peltier heating-cooling unit (Elemental Scientific Inc. Omaha, USA). The ionization of DART occurs inside a modified stainless steel 6 mm union cross piece (Swagelok, Ohio, USA). The modification of the cross piece is involved in cutting it in one direction from two sides symmetrically to have a thickness of 18 mm. The piece was then pressed in between the ceramic part of the DART ionization source and the metallic orifice of the mass spectrometer. Then the screw of the DART ion source was tightened sandwiched to ensure a fixed distance between the ion source and orifice of the MS during the measurements. Moreover, avoiding the ionizable species from surrounding enters the closed ionization region. The inner diameter of the cut cross piece was large enough to encircle the ceramic cap outlet of DART and the metallic orifice of the MS due to their conical shape and the distance and the ruler of the DART ion source show was showing a value of 12 mm in sandwiched structure. 33

The residual of the spray stream was released to an extraction hood for safety reasons. The DART ionization source is regulated by a DART controller (IonSense Inc., USA). The ionized liquid products are analyzed with the JMS-T100LP AccuTOF TM LC-plus 4G mass spectrometer (JEOL, Tokyo, Japan) is equipped with a time-of-flight mass analyzer that can achieve resolving power of ≥ 10,000 based on measurement of protonated reserpine molecule at m/z 609 according to FWHM definition. The ion source was always operated with helium (Air Liquide, Grade 4.7). The recording interval frequency was one spectrum per second.

5.1.4.1 Calibration of DART-TOF-MS The calibration of the instrument has to be done regularly by using PEG 600 as an external calibration material. The procedure involves first removing the reaction chamber (cross piece) and doing all the process of calibration in the open air. Then the temperature of the ion source has to be set 500-550°C to be enough to meltdown and evaporate the PEG 600 polymer. A glass capillary has to dip into the waxy polymer to pull out enough and then by exposing the capillary for short time to the stream of a hot jet of He, by melting down the polymer, the droplet of the liquid has to be in contact to He stream to be ionized. The process has to be repeated for negative mode.

5.2 CO 2 electroreduction on the surface of pristine and anodized copper 5.2.1 Electrochemical measurements All the experiments were done on a copper foil 25 mm×25 mm with a thickness of 1 mm (Goodfellow, London, England, 99.99+%). The exposed area of the copper to the electrolyte was confined by silicon sealing of the SFC. Before each experiment, the copper plate was polished with SiC sandpaper with 5000 grit and finished with 1 µm diamond paste (Saint-Gobain Daimantwerkzeug GmbH, Germany) to remove as much as possible oxide remnants and get a reproducible roughness. In the end, the surface was washed several times with isopropanol and eventually with ultrapure water and dried under a flow of compressed clean air. Since copper forms a natural oxide immediately as it is in contact with atmospheric oxygen and moisture, therefore, the polished clean surface was reduced electrochemically by reparative reductive LSVs. The electrolyte was prepared by dissolving 0.1 M KHCO 3 (Merck EMSURE®) in ultrapure water (Elga PureLab Plus system 18 MX, TOC < 3 ppb, Celle, Germany). The counter electrode was a bored-through carbon rod (Goodfellow, London, England) stuck through a hole at the outlet channel. Prior to the experiment, the electrolyte was saturated with CO 2 (Air Liquide, Grade 4.8) by bubbling the electrolyte for 30 min and throughout all the experiments the flow was maintained. The pH value of the saturated electrolyte was 6.8. The electrochemistry in combination with product analysis was performed with SFC, while the surface area measurement was conducted in a Teflon cell with a copper wire (1 mm diameter Mateck, Jülich, Germany) soaked in the electrolyte. The surface area was limited by wrapping a Teflon band on the top part of the wire.

34

5.2.2 Product analysis with EC-RTMS The gaseous products were Table 5.1 Optimized instrumental parameters for the determined by EI-QMS whereas determination of liquid products during the electroreduction liquid products were identified of CO 2 on copper. with DART-TOF-MS. For the Polarity of instrument Positive Negative detection of all liquid products, the Orifice 1 (V) 50 -30 LSV experiment was repeated in Ring Lens (V) 10 -5 the negative mode of DART-TOF- Orifice 2 (V) 5 -5 MS since it was not possible to Ion Guide RF (V) 50 150 switch between two modes Type of nebulizing gas O2 O2 automatically and periodically Temperature of ion source (°C) 250 250 during the measurement. In Grid voltage from ion source (V) 350 350 particular, switching between Flow rate of Nebulizing gas (L 0.3 1.0 modes takes roughly 3 min (1 min min -1) for changing the polarity of the Flow rate of He gas (L min -1) 3.0 2.5 instrument, and 2 min to deliver a stable background for all recorded Flow rate of electrolyte to the -1 0.4 0.4 ion currents). mass spectrometer (mL min ) Flow rate of electrolyte to the Since the supporting 0.5 0.5 mass spectrometer (mL min -1) electrolyte contains a high concentration of salt (0.1 M KHCO 3), the spray transfer line was heated to 100°C to remove the large portion of salt from the stream of the sprayed electrolyte to prevent any clogging of the TOF-MS orifice. Table 5.1 shows the mass spectrometer parameters for the detection of the liquid products in two polarities of the instrument. Table 5.2 displays the list of detected products during potentiostatic and potentiodynamic experiments of electroreduction of CO 2 on different copper surfaces with EC-RTMS in real time. Twelve different gaseous and liquid products are reported here, which reveals that the technique is capable of detecting almost all products reported in the literature from steady-state electrolysis studies 51 . In detail, the table shows the MS methods which were employed for the determination of each product as well as attributed m/z values together with associated ion . Measurement of the CO in the presence of saturated electrolyte with CO 2 was challenging because of three facts: firstly, Teflon AF membrane in the degasser is very permeable for CO 2 (refer to the table in Figure 6.3), secondly, the energy of the ionization of EI source is 70 eV, which is far more than the first ionization energy of CO 2 (13.77 eV) which causes remarkable contribution of CO 2 in m/z= 28, and last but not least, leakage of atmospheric nitrogen to the ion source vacuum chamber is unavoidable and can contribute to the m/z=28.

5.2.3 X-ray photoelectron spectroscopy (XPS) XPS measurements were carried out with a Quantera II spectrometer (Physical Electronics Inc., Chanhassen, USA) under ultra-high vacuum conditions with base pressure below 1.1 × 10 -8 mbar. The photoelectron spectrum of the samples was recorded using a monochromated Al K α X-rays (1486.6 eV) source at 15 kV and 50 W. The incident beam was focused on the 200 × 200 µm 2 spot area of the sample. During the measurement, the surface was kept neutralized by Ar + ion and electron bombardment. The pass energy of the analyzer 35

was set at 280 eV (survey), 140 eV (C 1s), and 69 eV (Cu 2p 3/2 , Cu L 3M4,5 M4,5 ). The signal in the survey and detailed elemental spectra were 3 and 12 sweeps, respectively.

Table 5.2 Assignment of the m/z values to products during electroreduction of CO 2 and the structure of the respective ion with corresponding mass spectrometry methods. Adapted with permission from 121 . Mass range/ Product Method Ion structure center [a] + Ethanol DART-TOF-MS, PM 47.0 – 47.1 [CH 3CH 2OH + H] + Acetaldehyde DART-TOF-MS, PM 45.0 – 45.1 [CH 3CHO + H] + 1-propanol DART-TOF-MS, PM 43.0 – 43.1 [CH 3(CH2) 2OH–H2O + H] + Propionaldehyde DART-TOF-MS, PM 59.0 – 59.1 [CH 3CH 2CHO + H] + Acetone DART-TOF-MS, PM 59.0 – 59.1 [CH 3COCH 3 + H] + Allyl alcohol DART-TOF-MS, PM 41.0 – 41.1 [CH 2CHCH 2OH–H2O + H] + Methanol DART-TOF-MS, PM 33.0 – 33.1 [CH 3OH + H] Formate DART-TOF-MS, NM [b ] 45.0 – 45.3 HCOO - - Acetate DART-TOF-MS, NM 59.0 – 59.3 CH 3COO [c] - Glyoxal DART-TOF-MS, NM 75.0 – 75.1 [CHOCH(OH) 2 - H] + Hydrogen EI-QMS 2 H2⦁ + Methane EI-QMS 15 CH 3 + Ethylene EI-QMS 26 C2H2 [a] PM=Positive mode; [b] NM=Negative mode; [c] The ion structure was deduced from a standard solution containing a known amount of glyoxal in the electrolyte.

All binding energy was corrected with respect to the adventitious C1s core-peak which appeared at 284.8 eV. A piece of sputter cleaned copper foil was used as an internal calibration of the instrument as well as the Cu 0 standard sample. Other standard samples including Cu 2O (≥99.99 %, Sigma-Aldrich), CuO (99.999 %, Sigma-Aldrich), and CuCO Cu(OH) (>95 wt%, Acros Organics BVBA) were unsealed in an Ar filled glove box 3∙ 2 with a load lock chamber. Sample preparation and transfer have been performed under inert. The were stuck to double-sided tape and clipped to the sample holder. The prepared sample was transferred via a protective box under the inert Ar atmosphere to the UHV chamber with minimized exposure to air and changes of the surface. The recorded data were treated and evaluated with the CasaXPS software (Casa Software Ltd.).

5.3 Electrooxidation of saturated C1-C3 primary alcohols on Pt-ALD 5.3.1 EC-RTMS The electrochemical measurements in combination with product characterization in real time were carried out with the SFC-EC-RTMS setup. The electrolyte reservoir in SFC before each measurement was deaerated with argon for 30 min and a continuous flow of argon was maintained throughout all experiments. The electrolyte was bubbling constantly with argon to keep the fresh electrolyte constantly deaerated. The counter electrode was a platinum wire (Mateck, Jülich, Germany) which was positioned at the waste channel of the ® SFC. All the electrolytes were prepared by dissolving 0.1 M HClO 4 (Suprapur , Merck) as 36 supporting electrolyte plus 0.2 M of respective alcohol (EMSURE ®, Merck) in ultrapure water (Elga PureLab Plus, 18.2 MΩ cm, TOC < 3 ppb, Celle, Germany). The optimized setting for DART-MS and flow Table 5.3 The instrumental conditions of SFC-EC-RTMS. rates are shown in Table 5.3. Polarity of instrument Positive mode Details on the interpretation of Orifice 1 (V) [a] 50,60,30 mass signals with EC-RTMS and Ring Lens (V) 10 the assignment to products are Orifice 2 (V) 5 outlined in Table 5.4. Ion Guide RF (V) 150 Type of nebulizing gas Ar 1 5.3.2 Pt-ALD preparation Temperature of ion source (°C) 300 The ALD precursor Grid voltage of DART ion source (V) 350 trimethyl(methylcyclopentadienyl Flow rate of Nebulizing gas (L min -1) 0.3 )platinum(IV) or MeCpPtMe3 -1 was purchased from STREM Flow rate of He gas (L min ) 4.0 while other chemical reagents for Temperature of the Peltier unit (°C) 30 preparation of the Pt-ALD electrode were bought from Temperature of the spray transfer line 40 Sigma-Aldrich, Alfa Aesar, (°C) ABCR, or VWR suppliers. Water Flow rate of electrolyte to the mass -1 0.4 for preparation of electrolyte was spectrometer (mL min ) ultra-purified with a Millipore Flow rate of electrolyte to the SFC cell -1 0.5 Direct-Q system (Merck). (mL min ) [a] Aluminum plates (99.99%) and Si The voltage of the orifice 1 was set to 50, 60, 30 V for the experiments with MeOH, EtOH, and 1-PrOH, respectively. (100) wafers covered with an oxide layer were supplied by Smart Membranes and Silicon Materials Inc., respectively. The nanoporous Pt-ALD electrodes were prepared according to previously published reports 122, 123 . At the first step, a porous anodic aluminum oxide with ca. 18 μm length and 380 nm diameter (6 h of the second anodization step) was fabricated as the template. One side of pore extremities was closed with a several micrometer thick electrical nickel contact. ALD process occurred at 220 °C with the reagent of MeCpPtMe 3 (it was kept in a heated stainless steel at 50 °C) and ozone. Ozone was produced from oxygen fed 803N generator (BMT Messtechnik, Berlin, Germany).

Two consecutive microcycles consisting each of a 0.5 s with MeCpPtMe 3 pulse and 40 s of exposure time were performed before the chamber was purged with nitrogen for 90 s. Ozone was introduced in a single pulse of 0.5 s, whereas exposure and purge durations were 40 and 90 s. 240 macrocycles of the ALD sequence were used in order to obtain approximately 13 nm of platinum. Later, the atomic layer of platinum was deposited on the template in a commercial Gemstar-6 XT ALD reactor (Arradiance, Sudbury, USA) under N 2 carrier gas. For the determination of the platinum layer thickness, silicon wafers coated with indium tin oxide were added to the reaction chamber and subsequently characterized with a

1 The fabrication of Pt-ALD materials was carried out by Dr. Sandra Haschke from the chair of Chemistry of Thin Film Materials at FAU.

37

SENpro spectroscopic ellipsometer (SENTECH Instruments GmbH, Berlin, Germany) equipped with a tungsten halogen lamp. 50 data points were recorded for wavelengths between 380 and 1050 nm under an angle of incidence of 70°. The data orientation Ɵ and ellipticity ε were then fitted with a fixed optical model for platinum. In a final step, the nanoporous platinum samples were laser-cut with a LaserPro Spirit (GCC LS Laser into circular pieces and glued with the nickel contact on small copper plates using double-sided conductive copper foil. The sample area exposed to the electrolyte was confined by a chemically resistant and electrically insulating polyamide (Kapton) adhesive tape featuring a laser-cut circular window of 2.0 mm diameter (geometrical sample area = 0.031 cm 2).

Table 5.4 Assignment of the detected products to the corresponding ion during the electrooxidation of primary alcohols on the surface of Pt-ALD. Adapted with permission from 124 .

Mass center Compound Detection method Ion Structure /range

Formic acid/ Methyl DART-TOF, PM [b] 61.0-61.1 [HCOOCH + H] + formate [a] 3

+ Acetaldehyde DART-TOF, PM 45.0-45.1 [CH 3COH + H] + Acetic acid DART-TOF, PM 61.0-61.1 [CH 3COOH + H] + Ethyl acetate DART-TOF, PM 89.0-89.1 [CH 3COOCH 2CH 3 + H] + Propionaldehyde DART-TOF, PM 59.0-59.1 [CH 3CH 2CHO + H] + Propionic acid DART-TOF, PM 75.0-75.1 [CH 3CH 2COOH + H] + Propyl Propionate DART-TOF, PM 117.0-117.2 [CH 3CH 2COO(CH 2)2CH 3 + H] + EI-QMS 44 CO 2⦁ 2+ Carbon dioxide EI-QMS 22 CO 2 + Methane EI-QMS 15 CH 3 + EI-QMS 30 CH 3CH 3⦁ [a] Formic acid was detected indirectly by measuring methyl formate; [b] PM=Positive mode

5.4 Isopropanol oxidation on Pt and PtRu 5.4.1 Electrochemical measurements The OCP and were performed in a Teflon two- compartment cell. The working electrode of the latter was mounted on a rotating disk electrode (RDE, Pine Instruments) system with a 0.196 cm 2 platinum and ruthenium disk electrodes insert (Mateck GmbH, Jülich, Germany). The working electrode for the investigation of NPs was prepared by drop-casting a thin film of platinum or platinum- ruthenium nanoparticles on a glassy carbon substrate (the preparation conditions are explained further below). The electrolyte before each measurement was deaerated with argon at least for 30 min and a continuous flow of argon was maintained throughout all experiments. Each measurement was done always with a fresh electrolyte to assure the initial concentration of IPA has not deviated by stripping with argon gas. The counter electrode was a platinum wire (Mateck GmbH, Jülich, Germany) for all measurements except for corrosion studies with ICP-MS which was a graphite rod. The reference electrode for all measurements was an Ag/AgCl/3 M KCl (BASi®, Indiana, USA).

The electrolytes were prepared by dissolving 0.1 M HClO 4 (Suprapur®, Merck) as supporting 38 electrolyte plus the appropriate concentration of IPA (Suprapur®, Merck) in ultrapure water (Merck, Milli-IQ 7000, 18.2 MΩ cm, TOC < 3 ppb).

5.4.2 Product analysis with EC-RTMS

CO 2 as the gaseous product is determined with EI-QMS in m/z= 44, with correlated ion structure of CO + ACE as the exclusive liquid product is determined with DART-TOF- 2⦁ . MS. In the case of IPA oxidation on Pt-polycrystalline, ACE was detected in the range of + m/z=59-59.2 which corresponds to the ion structure of [CH 3COCH 3 + H] . Due to the overload of the detector of the instrument and deviation linearity range in the study of IPA on the Pt/C and Pt (1-x) Ruy x/C, m/z=60-60.1 was selected as the selected range for detection of ACE in this case (detailed discussion in 9.3.11). The ion structure is the same as above with 13 13 12 + the difference that one of the carbons is C with the structure of [ C C2H6O + H] . The instrumental conditions were the same as in Table 5.3 except for the voltage of the orifice 1 which was 10 V.

5.5.3 Dissolution with inductively coupled plasma mass spectrometry (ICP-MS) Dissolved metallic species, namely, 195 Pt and 102 Ru were determined by coupling the downstream of the SFC with an ICP-MS equipped with a quadrupole mass analyzer (NexION 350X, Perkin Elmer). A solution of internal standards of 187 Re (10 μg L−1) and 103 Rh (100 μg L−1) was mixed with the stream of the sample and was analyzed simultaneously. The amount of dissolved metals was quantified by the daily calibration of the instrument. The standard solution of platinum and ruthenium (0, 0.5, 1, and 5 μg L−1) was prepared from mixing and dilution of concentrated solutions (Merck Centripur ®, 1000 mg L−1). the sampling intervals and dwell time of the mass analyzer for analysis of the four m/z was yielding a time resolution of ca. 1.5 s. During all the corrosion studies a lower concentration of IPA of 0.05 M dissolved (instead of 0.2 M IPA) to avoid carbon buildup in the plasma of ICP-MS. The onset potential for dissolution is determined when the ion current was higher than three times the standard deviation of the background signal.

5.4.4 Electrochemical infrared reflection absorption spectroscopy (EC-IRRAS) The composition of the thin layer of the Pt (111) single crystal electrode was characterized by an IR spectrometer (Bruker Vertex 80v) with evacuated optics and liquid nitrogen cooled MCT detector. The electrochemical measurements were performed with a homemade cell and RHE reference electrode. The counter electrode was a wire. A detailed description of the technique is reported before 125 . The background spectra were recorded at 0.05 V RHE with 256 scans per spectrum before each measurement. The potential- dependent spectra in the potential range between 0.05 and 1.1 V RHE were acquired with 128 scans per spectrum yielding an acquisition time of 57 s and a resolution of 2 cm -1.

5.4.5 Pt xRu y/C nanoparticles preparation The catalyst was immobilized on a glassy carbon substrate, by dropcasting of a catalyst ink in accordance with the following procedure. The metallic compositions of different commercially purchased Pt/C and Pt (1-x) Ru x/C nanoparticles (Tanaka Kikinzoku, Japan) are shown in Table 5.5. All the NPs were used as received. Since ruthenium is an inactive catalyst for IPAOR under acidic conditions, and platinum is the active material the 39 amount of weighted nanoparticles powders was adjusted based on the same quantity of -2 platinum in the composition (25 μg Pt cm ). The catalysts were all commercial NPs, with a nominal average size in the order of 4.5 nm and the specifications for the catalyst used are provided in Table 5.5. To ensure the homogeneity of the suspension during ink preparation, each experiment was repeated at least three times. The electrochemical experiment started as soon as the printed spot got dried. The dispersion of nanoparticles in ultrapure water was carried out by the aid of horn Sonifier (SFX150, Branson, USA) for 10 min with the amplitude of 25% in pulse mode (4 s on, 2 s off). A thin film of NPs was deposited by dropcasting of the homogeneous suspension on a glassy carbon plate (HTW Hochtemperatur-Werkstoffe GmbH, Germany) followed by -2 drying in air, resulting in a final catalyst loading of 25 μgPt cm . The GC plate was polished before each deposition with diamond paste with first with D3 followed by D0.25 (Saint- Gobain Diamantwerkzeug GmbH).

Table 5.5 Specification and physical properties of the Pt xRu y/C and Pt/C nanoparticle. Adapted with permission from 126 . Copyright (2020) American Chemical Society. Catalyst notation Pt Pt 0.5 Ru 0.5 Pt 0.4 Ru 0.6 Pt 0.33 Ru 0.67 Catalyst name TEC10E40E TEC66E50 TEC61E54 TEC62E58 Nominal atomic ratio (Pt:Ru) 1:0 1:1 1:1.5 1:2 Pt Content (wt.%) 36.8 32.8 29.9 27.7 Ru Content (wt.%) 0 16.9 23.2 28.8 Metal Content (wt.%) 36.8 49.7 53.1 56.5 Metal Surface Area (CO adsorption) 184.6 146.1 139.1 129.1 (m 2 g -1 metal) Catalyst Surface Area (N BET) (m 2 2 373.8 375.8 356.2 308.4 g-1 catalyst) FWHM (220 by XRD) (deg.) N/A 3.370 2.966 2.60 Peak Angle (220 by XRD) (deg.) N/A 68.82 68.83 69.06 Particle size by XRD (nm) 2.0 3.17 3.61 4.12

5.4.6 Quantification of products during steady-state oxidation of IPA The steady-state electrolysis experiment was done in a two-compartment, home-made three-electrode Teflon cell. The working electrode (WE)/ reference electrode (RE) compartment was divided from the counter electrode (CE) compartment by a bipolar membrane (Fumatech BWT GmbH, FBM). The WE compartment was operated under hydrodynamic conditions via a Reglo ICC peristaltic pump (Ismatec, Germany) to ensure enough mass transport to the surface of WE, while the electrolyte at the CE compartment was kept stagnant. The substrate of WE was a plate of glassy carbon (HTW Hochtemperatur-Werkstoffe GmbH, Germany) that was confined with a circular-cut chemically inert Kapton tape (Plano 2 GmbH) to have an area of 1 cm . A 100 µL of homogenous suspension the Pt 0.5 Ru 0.5 /C -2 (loading: 25 μgPt cm ) with the same preparation procedure of section 5.4.5 was dropcasted on the confined area. A platinum plate (MaTecK GmbH, 99.99 %) was used as the CE, and the electrolyte before the experiment was purged with argon flow for 30 min and it was maintained throughout the electrolysis experiment. ACE was detected with a Perkin Elmer Clarus 580 gas chromatograph (GC) equipped with a headspace autosampler (Perkin Elmer TurboMatrix 40) unit to evaporate the aqueous electrolyte containing the ACE and a 40 quadrupole mass spectrometer (Perkin Elmer Clarus SQ8T) was employed for detection and identification of separated ACE from GC column.

The argon stream that may contain CO 2 from the exhaust of the WE was passed through an alkaline solution (0.5 M KOH) to convert it all to sodium carbonate. At the end of the experiment, the accumulated carbonate in the alkaline solution was injected into an ICS-

5000 ion chromatograph (Thermo Fischer Scientific Inc.) to find the total amount of CO 2 that was formed during electrochemistry.

41

6 Electrochemical real time mass spectrometry (EC-RTMS)

6.1 Introduction EC-RTMS is a hyphenated apparatus based on the integration of two stand-alone mass spectrometers for the determination of the gaseous and liquid phase products with EI- QMS and DART-TOF-MS, respectively. Throughout this thesis, all the real time product characterization during electrochemistry was done by coupling the technique with scanning flow cell (SFC). By combining the high spatial resolution of material characterization of SFC with a high temporal resolution of the analysis of products of EC-RTMS, SFC-EC-RTMS is a powerful tool for high throughput screening of activity/ selectivity of different electrocatalytic materials for a certain reaction during potential/ current controlled experiments. The latter gives valuable information about the complex reaction mechanism on different surfaces.

6.2 Scanning flow cell SFC was introduced firstly for the high-throughput degradation test of electrocatalytic materials. The latter was done by coupling the downstream of the flow cell to an ICP-MS for the quantitative analysis of dissolved metallic species in the electrolyte originating from the corrosion of the electrodes 48, 127 . Later on, SFC was modified to be coupled to EI-QMS for the determination of gaseous and volatile products during electrocatalytic processes.

Figure 6.1 The detailed scheme of the modified version SFC used for coupling electrochemistry to EC-RTMS (a) and a close look at the interface of electrode-electrolyte (b).

This was performed by employing a Teflon porous membrane vacuum-sealed at the end of a capillary close to the surface of the working electrode for separation of the gas phase products from the stream of electrolyte 128 . Figure 6.1 demonstrates the design of the SFC which was coupled to EC-RTMS (a) as well as a zoomed view of the interface of electrode- electrolyte (b). By having a closer look, the design of the cell consists of 4 channels in total. Two channels of 3 mm in a V-shape configuration (60° angle), are the inlet and the outlet which provide a path for the flow of electrolyte from the left side to the right, respectively. An extra channel with an angle at the left side of the cell is used for the side insertion of the reference electrode (RE) with the principle of luggin capillary. To establish contact between RE and 42 working electrode (WE), the channel was filled by withdrawing the same electrolyte as used during the experiment via a , and this electrolyte was kept stagnant during electrochemistry. The sampling of both gaseous and liquid products happens at the electrochemical interface by means of a methyl deactivated glass capillary at the perpendicular position with respect to WE. The capillary is brought controllably very close to the working electrode surface to improve the time resolution of the product determination as well as higher sensitivity of detection. The latter is due to the higher local concentration of the species and less extent of dilution and diffusion of products to the electrolyte. Particularly, the distance between the end tip of the analysis channel and the working electrode is controlled by a homemade silicon sealing (150-200 µm of thickness) as well as a force sensor that was adjusted to 500 mN in contact mode. The distance is crucial to be controlled to have a reproducible collection efficiency and mass transport in the vicinity of the electrode. In this design, the counter electrode was positioned always at the waste channel to avoid interference of the counter electrode on the product distribution of the working electrode.

6.3 SFC-EC-RTMS The main components of the SFC-EC-RTMS are shown in Figure 6.2. A peristaltic pump with 4 independent adjustable channels controls the flowrate of different liquid streams throughout the setup. The first channel of the pump is used to feed the purple container (purge vessel) with a fresh electrolyte from another reservoir (it is not shown in Figure 6.2). The purge vessel is used to control first of all stability of the meniscus of the SFC (A). The latter is done by maintaining the level of the electrolyte in the vessel the same as the bottom of the SFC by the rule of hydrostatic pressure. The second reason is to purge the electrolyte with reactant gas (e.g. with CO 2 in case of CO 2RR) or deaerate the electrolyte (e.g. with Ar in case of alcohol oxidation). Then two channels of the pump are used to withdraw electrolyte (vacuum mode) continuously from the purge vessel to delivers the electrolyte plus reactant to the flow cell. One of them is devoted to withdraw a portion of products by an extraction capillary to EC-RTMS (MS channel). The third channel is employed to control the flow rate of the waste channel of the SFC at the counter electrode side. The sum of the MS channel and waste channel should be exactly equal to the flow rate of the purge channel. The latter is to ensure the same level of electrolyte in the purge vessel overtime during the experiment, thus, the peristaltic pump should be calibrated regularly. The fourth channel of the pump is employed for draining the liquid waste of the spray chamber which will be discussed later. A fraction of the electrolyte which contains the liquid and gaseous products is withdrawn continuously from the vicinity of the working electrode via a capillary to a homemade degasser. To shorten gas analysis time, the degasser has to be placed right after the SFC. The gaseous products are extracted by the aid of the vacuum of EI-QMS from the electrolyte phase by diffusion mechanism through a nonporous Teflon AF tubing in the degasser. Afterward, the degassed electrolyte flows further to the nebulizer where a high flow rate of a nebulizing gas breaks the liquid stream into a spray with a broad range of droplet sizes. The nebulizer is housed in a baffled spray chamber (E) where the spray is sorted based on the size of droplets. The bigger droplets are sunk to the bottom part and removed by a channel of the pump to the waste gray bottle. Therefore, only a portion of fine spray by positive pressure is pushed towards the inlet of the DART-TOF-MS for ionization and 43 analysis. At the last stage, all the collected data, by taking to account the delay time between electrochemistry and analysis, are synchronized with Labview-based software and plotted vs time or potential. Calculating the delay time is typically done by performing a step measurement at a potential where at least a liquid and a gas product are formed. The time difference between electrochemistry and ion current increase in mass spectrometer is the delay time.

Figure 6.2 A detailed scheme of the coupling of the EC-RTMS with the (Α) scanning flow cell (SFC). The electrolyte-containing products are sampled from the proximity of the working electrode via an extraction capillary in SFC. Separation of gaseous products from the electrolyte is done in a (B) home-built degasser equipped via a hydrophobic membrane. (C) Determination of gaseous products is done in EI-QMS. The degassed enters a (D) nebulizer to convert the electrolyte containing liquid products to a mist. In a spray chamber (E) bigger droplets are sunk to the bottom to waste while the finer portion of the mist is pushed by a continuous flow of nebulizing gas to the top for analysis. The liquid products in the mist are ionized and determined in (F) DART ion source and TOF-MS respectively. Adapted with permission from 121 .

6.4 Gas analysis in EC-RTMS The separation of gaseous-liquid products in EC-RTMS happens outside of the electrochemical cell via a specially designed homemade degasser equipped with a nonporous membrane tubing (Teflon AF-2400) with the principle of MIMS. Teflon AF is a class of amorphous, semipermeable fluoropolymers and it is a copolymer of 2,2-bis(Trifluoromethyl)- 4,5-difluoro-1,3-dioxole (PDD) and tetrafluoroethylene (TFE). Teflon AF-1600 and Teflon AF-2400 are two commercial copolymers with glass transition of 160 and 240°C, which contain 65 and 87 mol% of PDD respectively (m in molecular structure in Figure 6.3). Based on the table in Figure 6.3, Teflon AF-2400 has significantly higher permeability (almost three orders of magnitude) compared to nonpermeable PTFE for a few selected gaseous species 129 .

44

Gas Teflon AF-2400 PTFE CO 2 280,000 1200 O2 99,000 420 He 270,000

N2 49,000 980 Ethylene 35,000 140 Methane 34,000

Ethane 18,000

H O (g) 410,000 2

Figure 6.3 Structure of Teflon AF membrane and, table of gas permeability of Teflon AF-2400 compared to a PTFE membrane in centiBarrer (CB) 129 .

Construction of the degasser involves the vacuum tightening of the membrane tubing in the middle of the metallic housing of the T-Piece (see Figure 6.4). The stream of the electrolyte carries the products from the electrochemical cell to the right side of the degasser and it flows to the left. The top connector of the T-Piece is hooked to the EI-QMS, where the pressure difference between the two sides of the permeable tubing (ambient pressure at the electrolyte side to 10 -5-10 -6 torr at the MS side) causes extraction of gaseous products from the electrolyte via molecular diffusion mechanism through the hydrophobic membrane. The degassed electrolyte flows from the left side of the degasser further to the inlet of the liquid analysis mass spectrometer which will be described in the next section.

Figure 6.4 Configuration of the home-made degasser as the inlet of EI-QMS, equipped with Teflon AF membrane.

Due to the low diffusion coefficient of the VOCs through the membrane, and the limited retention time of the electrolyte in the degasser (controlled by the electrolyte flow rate), only the gaseous products have enough time to pass through the membrane with acceptable time resolution. The latter is beneficial for two reasons: first, there is minimal loss of VOCs in the gas analysis MS; second, the spectra at the EI-QMS become easily interpretable since VOCs do not have major contributions in the spectra (check Figure 4.11). The superior sensitivity of Teflon AF for gaseous products originates from the very low 45 permeability factor of water compared to other nonporous membranes like PDMS or porous Teflon in DMES 130 that a relatively higher surface area of the membrane can be used. As in MIMS, the flux of gaseous products can be expressed by (4.29). The flux of the gas (J) for

Teflon AF is dependent on the pressure difference ΔP, permeability factor (P A), and the thickness of the membrane (l) expressed by (6.1) 131 .

ΔP J=P ( 6.1) A l

This means that the higher pressure difference (determined by the tightness of the degasser as well as the vacuum of the mass spectrometer) can lead to higher flux and better sensitivity along with higher time resolution of the analysis because of faster mass transport.

6.5 Liquid analysis in EC-RTMS As it is discussed before, DART in its original configuration is an open-air ion source that gives pure qualitative information about the analytes. By constructing and assembling an appropriate custom-made inlet system, the (semi-) quantitative analysis of the analytes in a continuous fashion is enabled. The stability of the mass signal without drift is essential for comparing the activity and selectivity of electrocatalytic materials for a certain reaction. The design of the inlet system substantially influences not only the stability of the signal but also the signal to noise ratio (S/N) and consequently limit of detection (LOD) for the analysis of the liquid products. First and foremost, the signal of MS is directly dependent on the number of free molecules of analyte (desolvated from the matrix phase) available at the ion source region for the ionization process. Moreover, the ionization process should happen under laminar conditions, which means that any kind of turbulence, like the formation of big droplets, temperature or pressure gradients which cause spikes or drifts in the background of the signal should be strongly avoided. Additionally, controlling different parameters of the inlet system during measurements assures reproducibility. The designed inlet ion source assembly of DART-TOF-MS consists of the following 4 main parts that are discussed further, namely, nebulizer-spray chamber, Peltier temperature controller for the spray chamber, heated transfer line, and the reaction chamber.

6.5.1 The nebulizer-spray chamber unit The primary step of ionization of dissolved molecules in the liquid phase matrix with DART is the desolvation of species from the liquid phase. To carry these species continuously into the ionization source region regardless of the vapor pressure, one possibility is to deliver the liquid stream directly via a capillary at a very low flow rate in contact with the hot jet stream of the ion source gas. This simple design is used to couple the DART ion source with the HPLC 115, 117, 132 . Since the desolvation process of large droplets with the hot jet of the ion source is not efficient, the alternative way is to use a nebulizer. A nebulizer is an apparatus that breaks apart the continuous stream of the liquid stream into smaller droplets (aerosol) by different mechanisms e.g. in a pneumatic nebulizer by a nebulizing gas. By converting the big droplets into smaller in the nebulizer the surface area of exposed molecules and subsequently the collision frequency of molecules and ionizing gas in the ionization source increases. The 46 latter leads to the higher efficiency of the ionization in addition to the laminar non-disturbing ionization process. This approach is used to couple electrophoresis 116 , surface plasmon resonance 133 , or analysis of with DART in real time 120 . One of the most widespread nebulizers is the concentric tube pneumatic nebulizer (see Figure 6.5 (a)), where a high-speed gas in the shell compartment and low pressure of the liquid stream in the central capillary creates a mist of the liquid based on the Venturi effect. Although concentric nebulizers offer good sensitivity and stability of aerosol jet, the capillary for liquid is very prone to clogging by large particles like from dissolved catalyst nanoparticle investigations and should be cleaned from time to time.

Figure 6.5 (a) Concentric tube pneumatic nebulizer (adapted with permission from 64 ) (b) Nebulizer- spray chamber assembly.

Formation of the bigger droplet of liquid in ionization source can lead to an unsystematic/ systematic noise and spikes which results in an unstable background of the mass spectrometer. This is because the bigger droplets prevent the steady ionization and block the He stream to carry the ionized species to the MS inlet. The latter is because nebulizers make a wide range of droplet sizes and the finer droplets are only desired for the ionization process. Therefore, sorting out or filtering those larger droplets to the waste using a spray chamber can improve the LOD significantly. The size, design, and material of the spray chamber are decisive factors in yielding spray quality. The cyclonic design of the spray chamber usually offers outstanding sensitivity and precision compared to other versions like Scott or Barrel spray chamber, since, the cyclonic design has the advantages of having faster washout, less carryover, and more efficient in eliminating larger droplets from the aerosol stream by centrifugal force 134 . Baffled cyclonic spray chambers (with a central spray transfer tube) are superior from different aspects compared to unbaffled version such as finer spray due to better removal of bigger droplets that lead to laminar ionization conditions, reduce the solvent load and more efficient desorption process, better enrichment factor, plus faster wash time and diminished broadening effect 135 . The process of spray generation in a nebulizer/spray chamber assembly is illustrated in Figure 6.5 (b), the produced mist in a different range of droplet size collides with the baffle in the center, and only a portion of fine mist is sent to the inlet of the mass spectrometer by positive pressure and the rest of the bigger droplets accumulate in the drainage and either is removed by the aid of gravity or more efficiently by a peristaltic pump. 47

The performance of the developed inlet system was examined first by continuous analysis of the mixture of methanol (MeOH) and ethanol (EtOH) in water. The experiment was done by feeding the inlet of the DART-MS with alternatively pure water for 2 min followed by a solution of EtOH/ MeOH mixture with a certain concentration for 2 min. Note that in each step the concentration of each EtOH/ MeOH mixture increased stepwise (Starting from 200 to 400, 600, and 800 ppb in liquid). Figure 6.6 shows the ion current versus time of overlaid two mass range channels of DART-MS for methanol (m/z= 33-33.1) and methanol (m/z= 47-47.1). An ideal (mass signal vs. time) should yield a rectangular profile (straight-line increase and decrease) in the product signal. However, the signal broadening is inevitable, since all the components of the inlet system contribute to the peak broadening. The latter is mostly due to the dead volume of the inlet system components and the retention of the products especially, polar compounds by surface interaction. Controlling and understanding the associated parameters can diminish the broadening effect. For instance, the most significant share of the broadening effect comes from the spray chamber especially the dirty interior surface of the spray chamber promotes the formation of sticking droplets and consequently the memory effect. To reduce the latter the surface has to be cleaned on a regular Figure 6.6 Successive addition of 200 ppb from a concentrated MeOH+EtOH basis or the volume of the spray chamber has to mixture in pure water alternated by be as low as possible. Additionally, the length of switching to pure water to bring the signal the transfer line for liquid and spray has to be as back to the background. short as possible and the material of the lines has to have as lowest as possible interaction with the analyte in the solution especially with polar compounds (e.g. the liquid transfer line is the best to be out of PFA, FEP or PEEK to minimize the dispersion of the species). The standard solution test can represent the efficiency of the inlet system compared to real conditions. The inner diameter of pump tubing for the MS channel should be kept as small as possible for a shorter delay time. During the electrochemical reaction very well where a plug of products in the liquid stream is nebulized and a time-dependent population of products in vapor results. Ideally, the flow rate should be constant, however, due to the mechanism of function of the peristaltic pump, the rollers make a periodic pulse in flow rate. The latter influences the delivery of the electrolyte and reactants to the surface as well as products to mass spectrometers consequently systematic noise in both electrochemistry and signal of MS.

6.5.2 Peltier temperature controller for the spray chamber Since kinetics and thermodynamics of ionization are temperature dependent, the temperature of spray has to be independent of the surrounding at a constant value. The latter is necessary for long term signal stability and reproducibility of the measurements. Additionally, 48 temperature controls the equilibrium of the distribution of the species, including analytes and solvent between the spray phase and vapor phase, and by that, some compounds are more and some less influenced by the temperature according to the vapor pressure-temperature relationship. Generally speaking, a higher temperature is beneficial for the process of desolvation later that happens in the ionization region, as well as assist the vaporization of low volatile analytes from the liquid matrix. The temperature should not be too high, however, because it increases the solvent loading to the mass spectrometer, and it promotes the clustering of molecules with solvent molecules which affects e.g. protonation in the positive mode of DART.

6.5.3 Salt trap unit In order to perform electrochemistry, the presence of supporting electrolyte is mandatory. The latter can be an acidic, alkaline, or saline solution. For the reactions like electroreduction of CO 2, using a buffer salt to set the pH in the mild range is inevitable. However, the salts, especially nonvolatile ones, are harmful to the inlet of the mass spectrometers, particularly for ambient mass spectrometry since it causes clogging of the orifices or capillaries. In addition to that, the ion transfer units, including high voltage lenses, get covered over time with a layer of salt and due to the increment of the resistance that hinders the proper functioning of the mass spectrometers lenses and ion transfer unit. Therefore, the development of a desalination system is crucial. The most important role of the heated transfer is to trap the salts from the stream of spray. By increasing the temperature of the transfer line above 100 °C, the salt can be captured and precipitated on the inner wall of the heated metallic tubing.

6.5.4 Reaction chamber In all designs of the inlet system that have been developed before for continuous liquid analysis in DART-MS (for instance coupling with liquid chromatography), the ionization conditions are not well controlled due to the fact that the ionization process happens in open- air 115, 117 . By confining the ion source region by the aid of a closed interface primarily the ionization region is protected from the disruption of surrounding ambient species e.g. VOCs, and subsequently, ion chemistry inside gets controlled. Additionally, any fluctuation or turbulence caused by the surrounding is hugely avoided. Furthermore, this approach protects the user from the dangerous corrosive toxic spray of the electrolyte. According to Figure 6.7, the reaction chamber which is a modified metal cross-piece is sandwiched between the orifice of the MS and DART. This piece brings the stream of ionization gas (metastable helium) from the ion source and the fine mist from the spray chamber transfer line in an orthogonal way into contact with each other. The geometry of the reaction chamber determines the distance of the inlet of the mass spectrometer which is essential for reproducible ion chemistry. Besides, the inner diameter of two diagonal channels determines the directional flow velocity of both gases in the joint area. The latter is important for the dwell time of interaction of two streams together time in equation (4.21). The concentration of the analyte in the gas phase or [M g] is proportional to its concentration in the liquid phase [M L] and by considering a constant of ke (partitioning constant of analyte between liquid and spray phase) it can leads to the equation of [M g]= ke[M L] and by that equation 6.2 can be concluded. 49

+ + [MH ] = H3O ( 6.2) KK 0Mt

To be able to compare the performance for a product (M l), it is important that the time of the ionization reaction or the dwell time of the product molecule in the reaction chamber of the mass spectrometer is controlled in a constant value. Note that, the concentration of primary ion (here protonated water) which is proportional to the humidity and temperature of the ion source as well as the degree of fragmentation is constant by constructing a housing.

Figure 6.7 Reaction chamber from (a) outside (b) inside, the blue, green, orange, red, green particles are primary ions from ionization source, (ionized) reactants, (ionized) liquid product molecules, gaseous product molecules.

For instance, the smaller inner diameter of the spray channel with the same flow rate of the spray leads to more deflection of the He stream from the straight line towards the inlet of the mass spectrometer and preventing generated ions from reaching the MS. After the interaction of the spray stream and ionizer gas, the rest of the spray is transferred to the mist out (waste). The cross piece is from stainless steel controls the distance between the ionization source and inlet of the mass spectrometer. The latter is done by pressing the piece between ion source and inlet and securing the in the middle by fastening the screw of the DART positioning rail. The ionization in the reaction chamber happens under laminar conditions which leads to low noise and consequently to a higher S/N. It controls the reproducibly the distance between the DART ion source and orifice of the mass spectrometer controls noticeably the ion chemistry, sensitivity, and background of the spectrum 81, 133 .

6.5.5 The parameters of inlet- ion source assembly in DART mass spectrometer 6.5.5.1 Type of nebulizing gas The type of nebulizing gas influences the ion chemistry in the reaction chamber, for instance, as it is explained before oxygen can promote the CTR of water molecules in positive mode. During the ionization in aqueous electrochemistry, water is abundant in the ion source region, penning ionization reaction happens mostly with water but the formation of hydronium water can be boosted, when especially when oxygen is used as spray gas 81 . Since during aqueous electrochemistry, the inlet system generates very high humidity from the + + electrolyte, the extent of CTR from NO or O 2 is very minimal, whereas proton transfer is the predominant mechanism. On the other hand, the physical properties of nebulizing gases like viscosity and thermal conductivity determine the interaction with the stream of He and 50 accordingly, ionization efficiency. In negative mode the presence of oxygen is necessary namely, the nebulizing gas should be at least air or more efficiently, pure oxygen.

6.5.5.2 Temperature of the ion source As the first step of atmospheric ionization of liquid samples involves desorption of the sample (because the liquid phase shell does not allow the molecules to be ionized). DART ionization source is equipped with a gas heating unit (up to 550 °C) that facilitates the process of desorption of the molecules of analyte from the matrix by desolvation of the fine spray droplets. The higher temperature assists desorption of less volatile molecules, however, it may lead to decomposition of the molecules therefore it should be set according to experiment 136 .

Evaporation of the fine droplet (r is the radius at time t) of mist with an initial diameter of r 0 is described by equation 6.3.

M 2 r3=r3-6DσP ( ) t ( 6.3) 0 s ρRT

Where D, σ, P s, ρ, M, R, and T are the diffusion coefficient of vapor, surface tension, saturated vapor pressure, density, molecular weight, , and the temperature respectively 137 . Additionally, the temperature of the ion source controls both the thermodynamics and kinetics of ionization. It can also affect the fluid dynamics of helium, for instance, its viscosity and subsequently its interaction with the stream of the sample spray.

6.5.5.3 Flow rate of He The flow rate of He determines the extent and time of interaction of the ionizer gas (He(2 3S)) with the stream of spray. The flow rate of He should be high enough that gives enough time to all the process of desorption (desolvation), ionization happens. However, when it is too low it can be deflected or blown off by spray stream and therefore cannot carry the ions to the inlet of the mass spectrometer. On the other hand, very high flow rates cause less efficient blending of two-stream, in addition to a shorter time of interaction and increasing the turbulence inside the reaction chamber. All these mean, that, the flow rate of He is a dependent parameter to the flow rate of the spray (nebulizing gas) and it should be set accordingly. Besides, the flow rate of He influences the quality of the glow discharge region as well as the population of metastables, which means it should be high enough to provide enough He(2 3S) to the ion source region for penning ionization. Last but not least, Moreover, it acts as well as a heating gas to desolvate the analyte molecules from the spray phase and it determines the real temperature of the ion source. The typical flow rate in DART can vary between 2-6 L min -1 138 . For instance, based on equation 6.2, the time of PTR influences the signal intensity of the protonated molecules in the positive mode of DART.

6.5.6.4 Flow rate of nebulizing gas The flow rate of the spray (nebulizing gas) should be enough to produce a stable jet of spray and transport the fine droplets to the ionization source region by positive pressure. This parameter, same as the flow rate of He controls the dwell or interaction time of the molecules in the ion source region. Alternation of the flow rate of nebulizing gas has many impacts, for instance, the higher flow rate of gas results in smaller droplets which cause better dispersion 51 of liquid in the spray phase and consequently more efficient ionization due to better desolvation. Additionally, an adequate flow rate of nebulizing gas decreases the delay time in the spray transfer line along with the spray chamber due to more efficient washing, and that this diminishes the memory effect or broadening effect. However, a very high flow may deflect the stream of He, together with more dilution of the analyte as well as a higher ratio of the transported liquid to MS inlet instead of being sunk by gravity to the waste of spray chamber and increases the humidity and ion chemistry.

6.5.5.5 Flow rate of liquid Changing the flow rate of liquid (electrolyte) has several outcomes, namely, the mass transport in the vicinity of the working electrode (WE) in e.g. SFC as well as products from the WE (collection efficiency or extent of dilution of products), the delay time of signal between mass spectrometry and electrochemistry and of course, the quality of the spray by changing the ratio of liquid flow rate/ nebulizing gas flow rate. As it is explained before, in SFC the mass spectrometry channel withdraws the electrolyte plus products from the surface of the working electrode via a pump. A higher flow rate of liquid causes higher mass transport, but the very high value increases the chance that the unreacted reactants are sampled, in the case of kinetically controlled some reactions, increasing the flow of reactant just causes more dilution of the products. Therefore, the flow rate of liquid should be a compromise of enough mass transport (of the educt and product to and from the electrode, respectively) and dilution of the products to the mass spectrometer. The delay time and broadening the signal of the mass spectrometers (especially in the case of liquid products) are worse when the flow rate is lower because of the retention of the products on the surface of the transfer line and sprays chamber. The quality of spray is very dependent on the ratio of the nebulizing gas and liquid flow rate. By having the same flow rate of the gas, higher liquid uptake leads to worse quality of mist and bigger droplet size, respectively, less extraction of the liquid products in the spray phase and lower signal. On the other hand, a higher flow rate of liquid leads to larger loading of liquid in the ionization region (higher humidity in reaction chamber humidity in case of aqueous electrochemistry). This is because the higher flow rate of liquid requires a higher flow rate of nebulizing gas which leads to, and ion chemistry including the population of the primary ions (clustering, the effective temperature of ion source) changes 139 .

6.5.6 Tuning parameters of the DART-TOF-MS Adjusting parameters of the mass spectrometer can hugely influence the amount of the given kinetics energy to the ions (degree of fragmentation) as well as the sensitivity of the instrument by having an impact on the transfer of the target ions to the mass analyzer region and later on to the detector. Figure 6.8 (a) shows a closer look at the ion transfer interface of the AccuTOF TM mass spectrometer, while Figure 6.8 (b) displays a detailed scheme of DART-TOF-MS. As can be observed the ion guide has intentionally misaligned to a curvy configuration to keep the high vacuum region clean as well as to trap as possible the neutral species between Orifice 1 and Orifice 2. The latter causes better ion transport and a higher sensitivity, since ions are influenced by electric field but the neural species are not.

52

6.5.6.1 Orifice 1 voltage The orifice 1 voltage is the potential difference between orifice 1 and orifice 2 (see Figure 6.8 (a)) in JEOL AccuTOF TM atmospheric pressure interface and it controls the degree of fragmentation by collision-induced dissociation (CID) mechanism. The latter means that the ionized molecules from the ion source are accelerated due to the electric field across the region between orifice 1 and orifice 2 and because of relatively high pressure (300 Pa), the collision between analyte ions and other species including neutral species that are present in the region leads to more fragmentation and harder ionization condition. The orifice 1 voltage has to be a compromise between the highest ionization probability and the lowest extent fragmentation or clustering. More fragmentation happens when the voltage is high while lower ionization probability and clustering are favored at lower potential ranges. In another word, the softer condition of ionization is always desired since the spectra are simpler, but it should not be too soft where clustering of the molecules of the analyte to the same (when the concentration of the analyte is high) and clustering to the solvent (when the concentration of the analyte is low) happens. Note that the distribution of primary ions is also influenced largely by this voltage.

Figure 6.8 Scheme of (a) DART ion source plus AccuTOF TM atmospheric pressure interface (b) internal configuration of DART-TOF-MS (adapted from JEOL Ltd).

6.5.6.2 Other parameters Ring lens voltage is the voltage difference between the orifice 1 and ring lens, while the orifice 2 voltage is between ion guide and orifice 2. These should be enough to attract the ions from the CID region (orifice 1) and guide them towards the ion guide since the efficient transfer is necessary to have sufficient detection sensitivity. Ion guide RF voltage, high-frequency voltage (high electric field) is applied to the ion guide to converge the generated ions in the center. The role of the truncated version of quadrupole is to pass the ions from the ion source (low vacuum) to the mass analyzer (high vacuum). Besides, it helps to omit more neutral molecules from the ion stream, along with relaxing ions after gaining kinetic energy especially from the CID region and equilibrates the kinetic energy over the whole mass range. This design is helpful especially to remove the precipitates of the dirty or corrosive electrolyte vapor to protect the high vacuum region that might occur. In addition, it prefilters the ions to the mass analyzer, the lower value of the voltage filters out the higher m/z values and let pass lower m/z values, in the opposite, the higher value is necessary to permit the higher m/z ions. Sampling intervals in TOF-MS determine the resolving power of the instrument: smaller intervals (0.25 to 2 ns) result in higher resolving power. 53

6.5.7 Ionization mechanism of EC-RTMS during aqueous electrochemistry The first step of ionization in DART is the desorption or desolvation of the analytes from the matrix. The transient microenvironment mechanism (TMEM) addresses the role of the matrix in the ionization mechanism of the positive mode of DART 140 . According to the mechanism, a liquid droplet in the spray phase makes a shield that prevents the analyte from doing direct ionization via the Penning mechanism. For example, during aqueous electrochemistry when the reactant of the electrochemical reaction is not ionizable, and water is the dominant ionizable molecule in the ion source region, first water molecules get ionized, and next, analytes get protonated with the PTR mechanism. This means initially water gets ionized and protonated water and protonated water dimer with proton affinity values of 690 ± 4 kJ mol −1 and 808±6 kJ mol −1 141 respectively are mainly formed primary ions under normal conditions of DART. As it is explained before, the equilibrium constant of PTR and the extent of ionization of the analyte of interest are highly dependent on the difference between the proton affinity of the primary ion and the analyte. This difference determines the LOD of determination of the analyte, which means that analytes with relatively low proton affinity like formaldehyde (713 kJ mol -1) or glyoxal (675-691 kJ mol -1) 142 have very high LOD since their proton affinity values are close to proton affinity of the single water molecule and are far below the water dimer. This means only single protonated water as the primary ion is able to protonate such molecules and water clusters as proton donors have very high PA and are incapable of efficient ionization. Therefore, to improve the LOD for such compounds the humidity should be lowered, the temperature of ionization should be higher and the ionization condition in CID of DART should harder to increase the concentration of the single water molecule. Note that, high humidity has two impacts, namely, it promotes the backward reaction which is deprotonation, and, it boosts the clustering of water 143, 144 . On the other hand, when the reactant molecules are ionizable, the ionization mechanism as well as the mass spectrum based on TMEM can be completely different. The extent of this influence is the factor of three parameters, namely, concentration, ionization energy, and proton affinity of the reactant. When the IE reactant = 19.8 eV, the ˂ME () reactant can do penning ionization directly with He(2 3S) especially when the ionization energy is lower compared to other main species that are present in the ion source like water.

Figure 6.9 displays the DART spectra when the electrolyte for CO 2 reduction reaction (a) and MeOH oxidation reaction (b) is sprayed at open circuit potential (OCP) when no reduction or oxidation products are expected. In detail, the spectrum in Figure 6.9 (a) is recorded at the steady-state condition when the inlet of the mass spectrometer was fed with the solution of 0.1 M KHCO 3 saturated with CO 2 and O 2 as the nebulizing gas (the stable and steady spectrum was recorded after 30 min of operation). By comparing the latter with the spectrum in Figure 4.6 very large similarity is witnessed. Dominant ions originating from water indicates that the PTR from protonated water is the prevailing ionization reaction, + + although the existence of O 2 and NO can lead to CTR as the side reaction. On the other hand, by looking at the spectrum of Figure 6.9 (b) which is acquired after spraying a solution of 0.1 M HClO 4+0.2 M MeOH with Ar nebulizing gas, the spectrum is completely different from the ones in Figure 6.9 (a) or Figure 4.6. Penning ionization of solution containing a considerable amount of MeOH yields species like protonated MeOH (see reactions 6.4 and 54

+ 6.5), protonated MeOH cluster (reaction 6.6), methyl attached ions [CH 3(CH 3OH 2)] without the interference of water molecule as well as cluster of water and MeOH primary ions + [(CH 3OH) n(H 2O) mH] are formed as primary ions. This is mostly because MeOH has a lower IE compared to water (10.84 ± 0.01 eV versus 12.621 ± 0.002 eV) and its penning ionization is more efficient. 3 +● 1 - He(2 S) + CH 3OH → CH 3OH + He(1 S) + e (6.4) +● + ● CH 3OH + CH 3OH → CH 3OH 2 + CH 2OH ( 6.5) + + CH 3OH 2 + n CH 3OH → [(CH 3OH)nH] ( 6.6)

Figure 6.9 The background spectra of DART-TOF-MS during the electrochemical reaction in positive mode by continuous feed of electrolyte (a) CO 2 electroreduction with 0.1 M KHCO 3 in pure water as electrolyte, O 2 as nebulizing gas (b) electrooxidation of MeOH electrooxidation with 0.1 M HClO 4 + 0.2 M MeOH in water and Ar is the nebulizing gas (instrumental conditions are discussed in chapter 5).

By having a closer look at the spectrum of Figure 6.9 (b), the water species peaks like the protonated water dimer are suppressed to a large extent. On the other hand, when PA of the reactant is higher than the solvent, for instance, MeOH (PA=754.3 kJ mol -1) compared to water (PA=690 ± 4 kJ mol -1), depletion of water primary ions in favor of MeOH primary ions takes place. The distribution of different primary ions depends on the thermodynamics and kinetics of ionization inside the reaction chamber, which can be influenced e.g. by orifice 1 or the temperature of ionization. Finally, a higher concentration of the reactant has a larger impact on ion chemistry, which means, for the determination of products in the presence of easily ionizable or proton affine reactants it is favorable to use a lower concentration of reactant as possible to keep the sensitivity of the instrument high. The latter is mostly because proton transfer of the reaction product molecules mostly happens with methanol derived primary ions instead of water as in reactions 6.7 and 6.8.

+ + [(CH 3OH)nH] + M → [M + H] + n CH 3OH ( 6.7) + + [(CH 3OH)nH] + M → [M(CH 3OH)nH] ( 6.8)

In the negative mode of DART-TOF-MS when the reaction chamber of the inlet system restricts the contribution of the air species in the ionization process, it is necessary to 55 use oxygen as nebulizing gas. In this thesis, the negative mode is primarily used for the detection of organic acids by proton abstraction mechanism in the electrolyte in pH neutral or alkaline.

6.6 Unique specifications of EC-RTMS EC-RTMS has unique features compared to previously developed techniques for product analysis in real time as following: (i) The whole series of products in both liquid and gas phase can be characterized in real time since it employes two different ionization mechanisms to profit from them to cover the detection of a wider range of products. (ii) EC- RTMS is a very sensitive technique and in some cases, LOD of sub ppb can be achieved which means that very low partial currents below µA for a specific product can be achieved. (iii) Ion suppression in liquid analysis is limited. (iv) The analysis of liquid-phase products is not limited by the vapor pressure, i.e. organic acids and even ionic liquids with almost zero vapor pressure can be determined.

Figure 6.10 Comparison between blank and spiked aqueous solutions with commercial ionic liquids of - - [EMIM][Tf 2N ] and [BMIM][Tf 2N ].

For instance, Figure 6.10 displays the capability of the DART-TOF-MS for the determination of some selected ionic liquids in water by switching from pure water as the background to the solution of 10 ppm of two ionic liquids of 1-Ethyl-3-methylimidazolium - bis(trifluoromethylsulfonyl)imide ([EMIM][Tf 2N ]) and 1-Butyl-3-methylimidazolium - bis(trifluoromethylsulfonyl)imide [BMIM][Tf 2N ]. Considering that ionic liquids are comprised of ions and ion pairs, the detection of ionic liquids in DART involves separate - detection of cations (EMIM or BMIM) and anion (Tf 2N ) in positive and negative modes, respectively. (v) The difficulty in the interpretation of spectra from an unknown mixture of products that are formed simultaneously during electrochemical reactions at the surface of electrode scales with the number of the products. Therefore, another benefit of EC-RTMS is that it offers relatively clean and easily interpretable spectra. The latter is because, firstly, the membrane in gas analysis MS is not permeable to the liquid products, secondly the liquid analysis mass spectrometer is equipped with a relatively soft ionization source i.e. DART 73 . (vi) EC-RTMS as a product analysis tool is entirely independent of the electrochemical cell 56 and coupling of the technique is involved only by connecting it to the effluent of any types of electrochemical cells with various shapes and types of electrodes to the inlet of the degasser.

57

7 A comparative time-resolved product analysis during CO 2 electroreduction on pristine and anodized copper

7.1 Introduction

Since the industrial revolution, a huge accumulated CO 2 emission in the atmosphere has disrupted the natural carbon cycle and causes acidification of oceans as well as a serious threat to the global climate pattern. Closing the anthropogenic carbon cycle by recycling CO 2 and using it as a cheap energy carrier could serve as a temporary solution until the realization 145 of a carbon-free energy society . CO 2 electroreduction reaction (CO 2RR) is one of the most known examples of electrosynthesis of valuable feedstocks and synthetic fuels by the mixture of CO 2 and water on an electrocatalyst by the power of (renewable) electricity under (close) ambient conditions 146 . Thermodynamics predicts that the standard potential of conversion of

CO 2 in the aqueous electrolyte to some major products like carbon monoxide (E°= -0.10 VRHE ), formic acid (E°= -0.12 V RHE ), ethanol (E°= 0.09 V RHE ), methane (E°= 0.17 V RHE ), or ethene (E°= 0.08 V RHE ) should happen at potential very close the to hydrogen evolution 30 reaction (HER) (E°=0.00 V RHE ) . However, CO 2RR compared to competitive HER is remarkably hindered kinetically, since CO 2 is among the most inert molecules in chemistry. can decrease this energy barrier, however, product distribution and activity 147 of CO 2RR are very surface dependent . Among all reported materials, copper (Cu) is the only pure metal that is able to convert

CO 2 to hydrocarbon and multicarbon oxygenates. However, a process based on copper is 148 economically unfavorable because of (i) high overpotential of CO 2RR (>0.7 V RHE ) and by that (ii) waste of electricity due to excessive H 2 evolution as a parasitic reaction which can be produced much cheaper with platinum under acidic conditions 149, 150 (iii) formation of 151 multiple products, inferior selectivity, and activity to added value C 2+ products . During the last decade, many studies have confirmed that treating copper surface results in very better efficiency. One of the first reports by W. Kanan et al. showed that thermal oxidation and 148 reduction of copper resulting in lower overpotential of CO 2RR towards C 2+ products . Alternatively, the copper surface can be oxidized and reduced by the alternation of potential which results in anodized copper (AN-Cu), and similar improvement in product distribution is witnessed 152 .

Detection of the gaseous and volatile products of the CO 2RR in real time is reported with DEMS, however, as it is discussed in chapter 5, the membrane of the inlet restricts detection of the non-volatile products 94 . Another recent study demonstrates the capability of

SIFT-QMS in the detection of the products during CO2RR on copper in real time, however, there are some associated disadvantages, i.e, detection of some products like hydrogen and organic acids are not shown 96 . This chapter is the first demonstration of the performance of the EC-RTMS for the detection of all the reported products of CO 2RR on the polycrystalline copper (P-Cu) during potentiodynamic and potentiostatic experiments. The performance of AN-Cu is determined and compared.

58

7.2 Results and discussion The presented study in this chapter is originated to a large extent from a published manuscript 121 with permission from the publisher. I am the first author of the publication and my contribution is all the as shown results except the characterization of copper-based electrodes with X-ray photoelectron spectroscopy (XPS) which was done by Mr. Mario Löffler, a Ph.D. student from the Electrocatalysis unit of HI ERN.

7.2.1 Double step chronoamperometry Figure 7.1 displays the evolution of the mass spectrometry signals of different gaseous and liquid products (only positive mode in liquid analysis) during a double-potential step chronoamperometric experiment vs. time. The experiment was done in 0.1

M KHCO 3 supporting electrolyte, which was saturated with CO 2 (pH= 6.8) on the polished and reduced copper electrode. The electrochemical program involved 24 repetitions of potential switch between -0.2 V RHE (120 s) where no faradaic process occurs and -1.1 V RHE (60 s) where CO 2 electroreduction takes place. Two in-situ anodization steps were introduced at ca. Figure 7.1 Simultaneously synchronized ion currents for a) 1600 s and 2800s where the hydrogen, b) methane, c) methanol, d) ethylene, e) acetaldehyde, f) potential step alteration ethanol, g) allyl alcohol, h) propionaldehyde/ acetone, i) propanol protocol was interrupted by as products of electroreduction of CO 2 on the surface of poly- copper. The Q and TOF (+) are standing for the mass spectrometry a positive potential for a methods that are employed for the determination of products. The certain time as indicated on electrochemical protocol is involving periodic chronoamperometric top of the figure. By looking pulses alternatively between -0.2 V RHE (120 s) and -1.1 V RHE (60 s) at the mass spectrum signals in 0.1 M KHCO 3 saturated with CO 2 as electrolyte (pH= 6.8). Two anodization potentials are applied at 1600 s and 2800 s. The for different products before conditions of the anodization are indicated on the graph. Adapted imposing the first with permission from 121 . anodization step, the alternation of potentials gives reproducible intensities for most of the 59 products, except for methane and ethylene, where the peak intensities slightly increase. Figure 7.3 illustrates the closeup of the methanol (MeOH) signal during the periodic cycles, which shows the same pattern as the other major products, although the formation rate of MeOH on copper is very low and the signal is close to the noise of the mass spectrometer. At ca. 1600 s, a mild and short anodic pulse (+0.8 V RHE for 15 s) was applied, and after this anodization step, a very slight increment in signal intensities for ethylene and ethanol is observed during the CO 2RR steps. By applying a higher anodization potential at ca. 2800 s for a longer period (+1.2 V for 60 s), a considerable increase in the peak intensities of almost all mass signals is witnessed in the following CO 2RR steps, except for methane where it decreases and acetaldehyde where it does not significantly change. This shows that the electrochemical oxidation and reduction of copper generates a surface that favors the formation of C2+ products. The same formation rate of acetaldehyde after the second anodization step is likely associated with the fact that it is the intermediate or the precursor of other reduction products and as it is formed at the electrode surface it gets reduced further. By continuing the pulses at the generated surface, it seems that the anodization effect on enhancing the formation of C2+ products is retained within the time scale of the experiment, which could imply that the arrangement of surface atoms that favors C-C coupling is irreversible. However, the stability of the enhanced ratio of C 2+ /C 1 of the test needs to be proved with the long term electrolysis experiment.

7.2.2 Sweep voltammetry The pulse measurement guided the design of another experiment to understand better the impact of anodization with a potential resolution. Figure 7.2 displays the evolution of the products of CO 2RR during cyclic voltammetry in the potential range from 0.0 to -1.2 V RHE 51, 153, 154 which is the reported range in that P-Cu reduces CO2 to various C 1-C3 products . The experiments were done with the scan rate of 10 mV s-1 on P-Cu (blue curves) and AN-Cu (red curves) and the acquisition of ion currents different products was done simultaneously. The anodization was done in between the two cyclic voltammetry experiments, so both voltammograms were recorded on the same catalyst location. Hydrogen as a side-product (c), methane (d), and ethylene (h) are detected as gaseous products with EI-QMS. Methanol (e), acetaldehyde (h), ethanol (i), allyl alcohol (l), propionaldehyde/acetone (m), and propanol (n) are detected in the positive mode, whereas formic acid (f), acetic acid (j) and glyoxal (k) are detected in the negative mode of DART-TOF-MS. Based on the measured delay time between electrochemistry and mass spectrometers, the mass signals were synchronized and the time scale was converted to a potential scale. This treatment results in mass spectrometry-cyclic voltammograms (MSCVs), where the ion currents are expressed as a function of potential. The examination of freshly polished copper (P-Cu) begins with the electroreduction of the naturally formed oxide on the surface. In the negative-going scan, hydrogen begins to form when the potential is more negative than -0.5 V RHE. As the applied potential sweeps more negatively to ca. -1.0 V RHE , methane and ethylene initiate to form. The onset potential of ethylene is slightly more negative compared to what is reported in the literature. As it is discussed before, one of the drawbacks of the nonporous Teflon AF is the different response times to various gaseous products. Ethylene detection here is delayed due to the low diffusion coefficient of the ethylene in the membrane. Moreover, during the positive-going scan, the signal of ethylene compared to hydrogen shows a significant tailing for the same reason. To 60 diminish the issue, the sweep experiment could have been performed with lower scan rates (˂10 mV s -1). However, the latter was not giving accurate results due to the formation of excessive hydrogen gas at the electrode-electrolyte interface and accumulated bubbles in the the channel of the SFC which was disturbing the mass transport. The onset of acetaldehyde, allyl alcohol, propanol, and propionaldehyde or acetone formation were at ca. -0.9 V RHE, while for ethanol is ca. -1.0 V RHE . By comparing the ion currents for allyl alcohol, propanol, and propionaldehyde/acetone very large similarity is noticed in the shape of their profiles. This may indicate the fact that these products share the same precursor e.g. previously reported hydroxyacetone 51 which was not detected under the reaction conditions. The methanol signal was changed slightly when it is inspected carefully and it will be discussed later. Ion currents for formic acid, acetic acid, and glyoxal were recorded by repeating the sweep experiment with the negative mode of the DART- Figure 7.2 Electroreduction of CO 2 on the surface of pristine and TOF-MS, as the detection of anodized copper during cyclic voltammetry with the scan rate of -1 products in both positive and 10 mV s in 0.1 M KHCO 3 saturated with CO 2 (pH= 6.8). The potential (a) and current densities (b) and simultaneously negative modes at once is not recorded ion currents from different mass spectrometry methods possible. The onset for formic for c) hydrogen; d) methane; e) methanol;f) formic acid; g) acid is determined ethylene h) acetaldehyde; i) ethanol; j) acetate; k) glyoxal; l) allyl alcohol; m) propionaldehyde or acetone; n) propanol as approximately at -0.75 V RHE . products of the reaction are shown vs time. On the right side of Acetic acid could not be the panel, the MS methods are displayed Q: denotes for detected on P-Cu since it was determination of gaseous products with EI-QMS and TOF(+) below the detection limit. and TOF (-) indicate the determined liquid products with the Glyoxal was detected in a slight positive and negative modes of the DART-TOF-MS respectively. Adapted with permission from 121 . amount but the exact onset potential determination was not possible. A general remark is that the reported onset potentials here are highly dependent on the sensitivity of the mass spectrometers (e.g. proton 61 affinity of the product in positive mode of DART-TOF-MS). Additionally, electrochemistry and product analysis are done in a flow-mode, which means that the amount of the formed product is influencing the determined onset potential. Therefore, the reported onset for some of the products could deviate from what was reported before with long term electrolysis. After completion of the sweep experiment on P-Cu, the in-cell anodization began, by applying an oxidative potential step of +1.4 V RHE for 10 minutes. The formed Cu-oxide species on the resulting surface (AN-Cu) were subsequently reduced cathodically during the linear sweep voltammetry from 0 to -1.2 V RHE , where CO 2 reduction products were detected with the same protocol. A dramatic change in ion currents of different products is observed. Particularly, except for acetaldehyde and glyoxal which are known as intermediate products of the reaction, AN-Cu exhibits higher activity and selectivity towards C 2+ products (ethylene, ethanol, allyl alcohol, propanol, and propionaldehyde or acetone) in comparison to P-Cu whereas C1 products (methane and formic acid) were suppressed. The most significant changes in hydrocarbon production are a pronounced drop in the activity of methane (>2 times) and an increase in the production of ethylene (>5 times). This fact underlines again the ability of the oxidized electrode to catalyze the C–C bond formation. A remarkable behavior is the onset potential of acetaldehyde which is shifted significantly positive in comparison to the AN-Cu (ca. 150 mV) which implies the enhanced kinetics of formation. However, the amount of detected acetaldehyde did not change considerably. It should be noted that the mass transport in the proximity of the working electrode is very important since it may lead to different results as acetaldehyde as the intermediate of the reaction reabsorbed and reduced further on the surface before leaving the catalyst bed (the kinetics of formation and reduction) 155 . This finding indicates that acetaldehyde is the precursor of reduced products, like ethanol. Moreover, the onset potential for the formation of glyoxal is shifted negatively (exact numbers because of low S/N is difficult) and the formation rate did not change significantly (the same as for acetaldehyde).

The formation of other liquid C 2+ products prevailed over the production of formic acid and MeOH (see Figure 7.3 b). Same as for P-Cu, the ion currents for allyl alcohol, propanol, and propionaldehyde/acetone show almost the same mass spectrometer profile which may indicate that they are formed through a common pathway. Another aspect of the experiment is that the geometric current density and amount of hydrogen were promoted after anodization, which may be partially attributed to the increased roughness factor which will be discussed later in detail in section 7.2.5. The change in selectivity could be explained by a local pH effect, i.e. by the increase of the interfacial pH due to the rapid formation of hydroxide ions and the limited mass transport rates, which eventually induces CO dimerization 156 . The latter is in line with another investigation which shows that the ethylene pathway is favored in alkaline electrolyte 157 . Recent findings indicated that strong alkalization of the electrolyte on copper leads to 70% faradaic efficiency of ethylene production in a gas diffusion layer sandwich assembly where the CO 2 can react at the interface without being converted to carbonates 28 . In conclusion, the observed effect can be a combination of the change in morphology and electronic effect. It is known from many reports the CO 2RR to C 2+ is boosted when the adsorption of CO and its coverage is high, which leads to the coupling C-C bond. It is believed that grain boundaries are responsible for this effect, some believe that this can be attributed to the residual oxide especially Cu 2O. The same study shows that the selectivity of 62

148 C2+ products declines over time in favor of the production of C 1 products . These changes in the performance of the electrocatalyst are comparable to those observed with thermally- produced copper oxide. In contrast to classical approaches which use analytical methods with temporal resolution in the order of minutes, EC-RTMS is capable of acquiring the relationship between the product distribution and the potential during one sweep experiment as is shown in Figure 7.2.

7.2.3 Detection of MeOH Figure 7.3 displays the magnified MeOH signal monitored versus time during the (a) steps and (b) sweep experiments that are shown in Figure 7.1 and Figure 7.2. A slight MeOH increment is witnessed when applying the potential of -1.1 V RHE compared to -0.2 V RHE , i.e. when CO 2RR occurs and it stops periodically, as described in 7.2.1. Based on the fact that DART-TOF-MS is very sensitive for the detection of MeOH (see Figure 6.6), we can conclude Figure 7.3 Magnified ion current for MeOH during the (a) that the formation rate of MeOH is periodic potential pulses experiment between -0.2 VRHE very minimal, which is in good (120 s) and -1.1 V RHE (60 s). (b) The CV in the potential agreement with the low partial range of 0 to -1.2 V RHE before (blue) and after (red) the anodization. Adapted with permission from 121 . current densities for MeOH formation reported in the literature for the copper electrode 51 . The minimal amount of MeOH is witnessed again during cyclic voltammetry experiments on P-Cu (blue curve) during

CO 2RR by the poor S/N ratio in the potential range of 0.0 to -1.2 V RHE . Even this trace signal of MeOH is disappearing as the copper surface gets anodized (red curve) which confirms the finding that the anodization is detrimental for C 1 products.

7.2.4 Characterization of the electrodes 1

To evaluate the origin of the enhanced activity and selectivity to C 2+ products after anodization, ex-situ XPS measurements were conducted to characterize the composition on three electrodes. The XPS spectra of pristine (blue curve), AN-Cu before (red curve), and after the scan in the CO 2RR potential region (green curve) in the corresponding core level Cu 2p 3/2 are shown in Figure 7.4 (a). On the blue curve (P-Cu) the binding energies that are 0 assignable to the presence of both Cu 2O and Cu are superimposed and they cannot be deconvoluted because of the same biding energies which can be resolved in Auger spectroscopy. By looking at the red curve, first of all, the latte peak was considerably shifted positively which is correlated to higher copper oxidation states e.g. CuO, or CuCO 3⋅Cu(OH) 2. The presence of Cu 2+ is confirmed by a satellite peak at binding energies in the range of 940-

1 The XPS measurements were conducted by Mr. Mario Löffler from the Electrocatalysis unit of HI-ERN. 63

945 eV of the spectrum. On the other hand, the green curve displays that the effect of oxidation was reversed after the subsequent reductive cyclic voltammetry scan in the CO 2RR potential region 158 . In detail, the satellite peak disappeared after the CV, and the characteristic 0 shifts blue to the same binding energy of Cu or Cu 2O, which suggests the completion of the reduction.

Figure 7.4 (a) Ex-situ XPS measurements of the Cu 2p 3/2 core peak region of different copper surfaces pristine (blue curve), after anodization (red curve), anodization followed by a reductive in the potential region of 0.0 to -1.2 V RHE . Horizontal black lines indicate the binding energies of standard samples of Cu, Cu 2O, CuO and CuCO 3·Cu(OH) 2 (b) Ex-situ measurement of copper Auger LMM XPS spectra of pristine copper (blue curve), anodized copper (red curve), and anodized copper followed by a reductive LSV in the potential region of 0-1.2 V RHE (green curve). The fittings of the assigned to measured standard samples and are indicated by different colors. Adapted with permission from 121 .

Figure 7.4 (b) corresponds to the copper Auger spectroscopy Cu L 3M4,5 M4,5 for all examined electrodes pristine (blue curve), AN-Cu before (red curve), and after the scan in the

CO 2RR region (green curve). All spectra consist of a broad peak in the kinetic energy range of 905-924 eV. The quantitative calculation of the surface composition was performed by curve fitting of the Auger spectra and is shown in Table 7.1. The results represent that P-Cu is composed mainly of Cu and Cu 2O, and indicates that air-induced oxidation of copper is not avoidable even by fast transfer to the UHV chamber. The anodized electrode contains a very small amount of Cu 2O and the main contribution, is from CuCO 3⋅Cu(OH) 2 while no CuO was detected. CuCO 3⋅Cu(OH) 2 is formed mostly because of oxidation in the presence of CO 2 at 159 the interface in alkaline conditions . CuCO 3⋅Cu(OH) 2 or is an insoluble salt that passivates the surface of copper and prevents further corrosion to Cu 2+ species and it prevents of reduction of CO 2 unless the layer is completely reduced due to limitation of charge transfer which was revealed with an in-situ XPS study 158 .

After performing the sweep experiment, the main fraction of CuCO 3⋅Cu(OH) 2 is reduced in a couple of seconds to Cu or Cu 2O after exposing the sample to reductive potentials compared to those used by Velasco-Vélez et al 158 . The reduced surface after anodization contains mostly Cu and Cu 2O, but in different relative amounts compared to P-Cu 64

namely, it contains a higher amount of Cu 2O. These results also imply an incomplete reduction of the initial Cu 2O in the course of the CO 2RR. It may also confirm the hypothesis of surface enrichment with Cu (I) over the course of anodization-reduction.

Table 7.1 Quantitative analysis of the chemical composition of the examined electrodes acquired from Cu L 3M4.5 M4.5 Auger spectra. The numbers were calculated by curve fitting of the standard reference samples and presented in atomic percent. Deviations from 100% arise from rounding errors. Adapted with permission from 121 . 0 Sample/composition Cu [at%] Cu 2O [at%] CuO [at%] CuCO 3.Cu(OH) 2 [at%] Pristine 27 60 7 7 Anodized 1 8 0 90 Anodized+CV 15 70 7 8

To examine the role of anodization carefully, in-situ surface characterization techniques to determine the relationship between structural composition to activity/ selectivity under operando conditions is necessary.

7.2.5 Surface area measurement By following the ion current of hydrogen during the sweep experiment, HER activity is enhanced on AN-Cu compared to P-Cu. This might indicate the increase in the surface area during the process of anodization. To further explore that, the electrode surface area for the P- Cu and the AN-Cu was determined by recording cyclic voltammograms at various scan rates in a potential region where only non-faradaic processes take place. Before recording the CVs for the area determination, the oxide formed during anodization was first reduced by a potential sweep in the oxide reduction region from 0.0 to - -1 1.2 V RHE with the scan rate of 10 mV s . The charging current was then plotted as a function of the scan rate and the slope of the diagram yields the double layer capacitance (see Figure 7.5). Based on the capacitance for an atomically flat electrode (29 μF cm -2), the roughness factors (real-to-geometric area) for the pristine and the anodized electrode are 1.6 and 6.4 respectively, i.e. the area of copper increases by ca. 4 times because of the anodization. What is seen during the sweep experiment by comparing An-Cu and P-Cu the enhanced to C 2+ products over C 1 products is a consequence of two effects. The change in the higher formation rate of C 2+ products is not merely an electronic effect and might be partially correlated to the increment in the roughness factor. The surface roughness measurement may account for a part of higher activity enhancement of some products but it does not account for lower activity of C 1 products or the lower overpotential for the formation of C2+ products. Further studies in combination with quantification of the products are necessary to understand the origin of the increment and regulate the surface morphology/ composition to achieve the highest selectivity and activity to C 2+ products.

65

Figure 7.5 Determination of active surface area of pristine and anodized copper. (a-b) CVs with scan rates of 10-100 mV s -1 for both electrodes; (c) capacitive current plotted versus the scan rate for both electrodes, based on the CVs in a-b.

7.3 Conclusions The first demonstration of EC-RTMS by the analysis of a complex mixture of products during the CO 2RR on P-Cu and AN-Cu is provided. The highlights are (i) the detection of some products is done for the first time in real time, especially non-volatile products such as formic acid or glyoxal during potential resolved experiments; (ii) the anodization of copper changed the product distribution and selectivity of the reaction in favor of C2+ products over C 1; (iii) the only exception of C 2+ products were acetaldehyde and glyoxal that their production rates were almost the same in both cases of P-Cu and AN-Cu, but the of formation were decreased on anodized copper. This behavior was attributed to the fact that acetaldehyde and glyoxal are the intermediate or precursor of other reduced products like EtOH and as it is formed on the interface it gets reduced further. (iv) The XPS results showed that the composition of the AN-Cu after performing the reductive sweep experiment was indeed a bit enriched with Cu 2O but it still largely resembles the P-Cu, which indicates that the presence of oxide is not responsible for the drastic change in selectivity and rather structural changes should be considered; (v) active surface area measurement revealed that the surface area of AN-Cu is 4 times higher compared to P-Cu, which may explain the increase in the formation rate of some products like H 2 but cannot support the decline in the formation of C 1 products or lower onset potential of C 2+ products.

66

8 Real time product characterization during electrooxidation of saturated C1-C3 primary alcohols on the surface of platinum

8.1 Introduction Direct alcohol fuel cells (DAFCs) are promising alternatives to batteries for portable and low power energy generation applications. The good solubility of short-chain alcohols in water, convenience for storage and transport, as well as their higher volumetric density than molecular hydrogen, make them suitable for such applications. The energy density of methanol (MeOH) as the most electrochemically active alcohol is 6 KWh/kg which is in the range of gasoline (10-11 kWh/kg) 160 . Ethanol (EtOH) or 1-propanol (1-PrOH) are attractive fuels due to higher C/H ratio over C/O 161 , less toxicity and, lower crossover rate through the DAFC membrane, compared to MeOH. From the thermodynamic aspect, DAFCs are comparable to internal engines or hydrogen fuel cells. However, electrooxidation of alcohols is kinetically sluggish that leads to low-performance fuel cells compared to hydrogen PEM-FC. For all these alcohols, complete oxidation of alcohol to CO 2 is desirable to maximize the energy output of the fuel cell. The latter does not take place in reality and the partial oxidation of alcohols leads to toxic liquid products, e.g. acetaldehyde or formaldehyde during EtOH and MeOH oxidation respectively 162, 163 . Platinum (Pt) is one the most extensively studied catalysts for the oxidation of a vast variety of organic molecules including alcohols 164-166 .

This chapter deals with the investigation of electrooxidation of three primary C 1-C3 alcohols, namely, MeOH, EtOH, and 1-PrOH during potential sweep experiments on the surface of platinum as a model catalyst in acidic conditions. Herein, high-surface-area nanoporous platinum was fabricated by the atomic layer deposition (ALD) technique. Pt-ALD was employed for product characterization to increase the absolute amount of the formed products at the interface and thereby have a higher mass spectrometry signal for all products in EC-RTMS.

8.2 Results and discussion The results of this chapter have been published previously 124 and it is presented here with permission from the publisher. I am the first author of the published manuscript and my contribution is the product analysis while the fabrication and characterization of Pt-ALD were carried out by Dr. Sandra Haschke from the chair of Chemistry of Thin Film Materials at FAU.

8.2.1 Characterization of Pt-ALD 1 The surface morphology of the as-prepared Pt-ALD was determined by scanning electron microscopy (SEM). Figure 8.1 (a) and (b) illustrate the SEM images of nanoporous well-defined structure from two views, i.e. top and cross-section. The regular 3D

1 The measurements were performed by Dr. Sandra Haschke from the chair of Chemistry of Thin Film Materials at FAU. 67 nanonetwork structure confirms the high surface area of the electrode. As it is explained before in section 5.3.2, platinum is coated on the template of anodized Al 2O3 highly ordered array of pores (diameter D ≈ 380 nm and length L ≈ 18 μm and the specific surface area is accordingly ca. 3.06 cm 2) with a uniform thin layer of approximately 13 nm of platinum. Figure 8.1 (c) confirmed the presence of platinum with energy-dispersive X-ray (EDX) analysis, in addition to the expected elements Al, O, and Ni from the template and electrical contact.

Figure 8.1 Characterization of the fabricated Pt-ALD electrode before electrochemistry. SEM micrographs from (a) top view and (b) cross-section; (c) EDX and (d) XRD spectra were recorded from the top of the sample. The XRD peak positions are assigned to crystalline platinum, Al from the frame, and Ni of the electrical contact. Adapted with permission from 124 .

The chemical composition of the ALD is studied by X-Ray Diffraction (XRD) too. The diffraction pattern in Figure 8.1 (d) shows the distinct peaks at 39.6 °, 46.2 °, and 67.4 ° which are attributed to crystalline and metallic platinum based on the Crystallographic Open Database (COD 1011107), to Al (COD 9008460) from the Al frame, and to Ni (COD 2100649) of the electrical contact.

68

8.2.2 Electrochemical protocol During oxidation of primary alcohols on the surface of platinum at room temperature, slow desorption of reaction intermediates from the surface causes accumulation of adsorbates and eventually deactivation of the catalyst. The nature and coverage of adsorbed species are potential-dependent. For instance, adsorbed carbon monoxide (CO ads) during oxidation of MeOH, EtOH, and 1-PrOH is dominant, while the formation of poisoning CH x,ad species can take place at low potentials 167-170 . To examine all the alcohols, the electrochemical protocol -1 of a slow scan CV from 0.0 to +1.2 V RHE with a scan rate of 1 mV s is executed. Prior to each CV, a cleaning procedure was conducted with fast cycles to remove adsorbates from the platinum surface. This ensures the reactivation of the platinum after performing a slow scan. To have comparable results and to check the reproducibility, the slow CV was repeated at least three times. All the reported voltammograms for all alcohols are recorded after fast cycles based on the protocol that is showcased for the electrooxidation of EtOH combined with product analysis in section 8.2.4.3.

8.2.3 MeOH electrooxidation The oxidation of MeOH on Pt- ALD is investigated during CV with the scan rate of 1 mV s -1 in an Ar- saturated

0.1 M HClO 4 + 0.2 MeOH electrolyte. Figure 8.2 shows the cyclic voltammogram (black curve) as well as the simultaneously recorded ion currents which are also known as mass spectrometric cyclic voltammograms (MSCV) during reaction for different products. CO 2 (red curve) and methyl formate (blue curve) are detected as gaseous and liquid products respectively. An older study of MeOH oxidation under steady-state conditions in the acidic medium with quantified product analysis showed that the products are CO 2 (significant), formaldehyde (moderate), and formic acid (negligible) in HClO 4 Figure 8.2 Electrooxidation of MeOH on the surface electrolyte 171 . During MeOH oxidation, of platinum polycrystalline during sweep major CO formation takes place directly voltammetry (a) CV and ion currents for (b) CO 2 as 2 gaseous product and (c) Formic acid as the liquid from CO ads, while formaldehyde and product in 0.2 M MeOH + 0.1 M HClO 4 with a scan formic acid are produced from a parallel rate of 1 mV s -1. The arrows indicate the direction of pathway that may produce CO 2 as well the potential sweep. Adapted with permission from 124 172 . Here, formic acid was detected . indirectly as methyl formate, whereas formaldehyde was not determined which will be discussed later in this section in detail. Therefore, the I MF at m/z= 61-61.1 was attributed to the formic acid formation during electrochemistry. 69

By following the current density and the profile ion current for CO 2 (I CO2 ) it can be recognized that they have quite the same form, which may confirm the fact that CO 2 is the main product of the reaction. In the positive-going scan, oxidation of MeOH on platinum begins to increase at the potential of ca. +0.55 V RHE where I CO2 and I FA start to raise too until -2 the current density reaches the maximum of 75 mA cm at potential ca. +0.9 V RHE . Above this potential, due to the formation of platinum oxide, the current density, I CO2, and I FA decline. On the negative-going scan, the reaction is hindered, until the platinum oxide gets reduced (below ca. +1.0 V RHE ) where the current density and ion currents increase again. Formic acid and formaldehyde could not be detected under the conditions of DART- TOF-MS in the positive mode. As it is discussed before in Chapters 4 and 6, the mechanism of ionization in the DART ion source is based on PTR. This means that only products with higher proton affinity (PA) than primary ions or proton donor can be detected. During MeOH oxidation, in addition to water, there is an excessive amount of MeOH in the developed inlet system of the DART ion source, which can be easily ionized compared to water, by penning ionization mechanism. Hence, the prevailing primary ions here are not water ions, but rather MeOH species (see Figure 6.9) including protonated methanol with PA= 754.3 kJ mol -1 173 . Among the liquid products of the MeOH oxidation, formic acid (PA= 742.0 kJ mol -1), methyl formate (PA= 782.5 kJ mol -1) 174 , and formaldehyde (PA= 711±0.4 kJ mol -1) 173 , only methyl formate has higher PA than MeOH and it can be detected under our conditions. It should be noted that the measurement with negative mode under strongly acidic conditions is not possible on account of suppression of the primary ions in negative mode with an excessive amount of hydronium ion (pH=1). Therefore, formic acid, unlike neutral conditions during

CO 2RR could not be directly detected in this mode. Formic acid was detected indirectly in the form of methyl formate, which originates from the homogenous esterification reaction of formic acid as the product of the electrochemistry and the excess of reactant, namely, MeOH. The homogenous reaction was confirmed, by feeding the inlet of DART-TOF-MS first with electrolyte (0.2 M MeOH+0.1 M

HClO 4) to acquire a stable background in m/z = 61.0-61.1 under the same instrumental conditions, then the electrolyte was spiked with formic acid and an increment was observed in this mass range which corresponds to the formation of methyl formate. In this experiment, no change in m/z = 47.0-47.1 which would correlate to protonated formic acid in the absence of

MeOH was observed because of the lower PA of formic acid compared to MeOH (PA Formic -1 -1 174 acid = 742.0 kJ mol compared to PA MeOH = 754.3 kJ mol ) . Similarly, previous studies of the oxidation of MeOH with DEMS have assigned the formation of methyl formate (volatile) to the formation of formic acid (non-volatile) during electrochemistry 60, 175, 176 . Direct detection of formaldehyde with real time or in situ techniques is not reported and is very complicated 171 . The most representative example of the detection of formaldehyde is with electrospray ionization mass spectrometry combined with online derivatization 177 . It can be determined during the long term electrolysis with low sensitivity refractive index detector, only when considerable amounts are accumulated in electrolyte 178 .

8.2.4 EtOH electrooxidation 8.2.4.1 Sweep experiment The electrooxidation of EtOH on Pt-ALD was studied during a CV in an Ar- saturated -1 0.1 M HClO 4 + 0.2 M EtOH electrolyte with a scan rate of 1 mV s . Figure 8.3 shows the 70

cyclic voltammogram in the potential range of 0.0 to +1.2 V RHE (black curve) as well as simultaneously recorded ion currents for CO 2 (blue and red curves for two different m/z), methane (green) as gaseous products, as well as acetaldehyde (olive green), acetic acid (purple), and ethyl acetate (orange) as liquid products. Oxidation of EtOH yields not only CO 2 but also partial oxidation products like acetaldehyde and acetic acid 179 . Ethyl acetate is produced through the homogenous reaction between electrochemically formed acetic acid and excess of EtOH that is present in the electrolyte as reactant 180, 181 . Methane was detected only at the lower potential region as the reduction product in both forward and backward scans. In both scan directions of the CV, the oxidation results in broad asymmetric curves with several maxima. Specifically, in the positive-going scan, the oxidation begins at ca. +0.5 V RHE where the current density as well as ion currents for CO 2, acetaldehyde, acetic acid/ethyl acetate increase. The current density continues to increase until +0.9 V RHE where it reaches a distinct peak because the surface gets oxidized. Later, the current starts to increase again from 1.05 V RHE until the upper potential limit (UPL) equal to 1.2

VRHE where the j reaches the highest value at ca. 46 mA cm -2. By inspecting closer the profiles of the ion current for oxidation products in positive and negative-going scans, the formation rate of CO 2 is higher in the forward scan, opposite to acetic acid which shows a higher production rate in the backward scan. The acetaldehyde signal does not change significantly by Figure 8.3 Product analysis during electrooxidation of EtOH on the surface of Pt-ALD during sweep comparing the positive- and negative- voltammetry (a) CV and simultaneously recorded going scans. Quantitative product analysis MSCVs for carbon dioxide (b,c) and methane (d) as during EtOH oxidation on platinum from gaseous products; acetaldehyde (e), acetic acid (f) and ethyl acetate (g) as liquid products. Electrolyte: previous studies revealed that the current -1 0.2 M EtOH + 0.1 M HClO 4. Scan rate: 1 mV s . efficiency of CO 2 is very low up to 2% The arrows indicate the direction of the potential and the partial oxidation to acetaldehyde sweep. Adapted with permission from 124 . 71

182, 183 or acetic acid is favored over the oxidation to CO 2 . The structure of the adsorbates during electrooxidation has been studied previously extensively 179, 184, 185 . Iwasita et al. reported that the amount of the formed CO 2 is independent of the concentration of EtOH and it is produced through the oxidation of CO ads. Moreover, in the same study, the stripping oxidation of EtOH revealed that CO 2 is the only desorbed product that originates from the oxidation of ethanol-derived adsorbates 179 .

Ion currents for CO 2 and acetic acid in the forward scan decrease as the platinum gets oxidized at ca. +0.8 V RHE . Moreover, the acetaldehyde signal still increases until +0.9 V RHE , which indicates that the oxidation of EtOH to acetaldehyde continues even though the surface of the electrode is partially oxidized. Note that acetaldehyde is the predominant reaction product at higher concentrations of ethanol which may indicate that the OH adsorption is not involved in the oxidation of EtOH to acetaldehyde while water concentration is decisive for 182 the formation of CO 2 . By looking at the signal of acetic acid/ethyl acetate in the forward scan, the second oxidation peak starts again at 1.0 V RHE.

In the negative-going scan, j drops until ca. +1.0 V RHE where it starts to increase to two ill-separated oxidation peaks. Acetaldehyde and acetic acid form at the same onset potential of +1.0 VRHE while CO 2 ion current starts later at +0.9 V RHE . The higher formation rate of acetic acid here than in the forward scan may be explained by the fact that oxidation of the adsorbates to CO 2 happens in the forward scan which renders the surface of platinum-free for bulk oxidation of EtOH to acetic acid. The formation of acetic acid increases sharply with a second oxidation peak only when the surface is partially cleaned from adsorbates at ca. +0.7

(peak potential ca. 0.6 V RHE ). It is suggested that acetaldehyde is the precursor of acetic acid 163 by reacting directly with a surface oxygenated species such as OH ads . Such a behavior is not seen in the positive-going scan since the surface is strongly inhibited with adsorbates and the pathway to acetic acid is favored only when these adsorbates are oxidized to CO 2. The current density and ion currents for all oxidation products decline further until ca. +0.4 V RHE where the oxidation stops.

Methane is the only detected reduction product that forms in the H upd (hydrogen underpotential deposition) region in both forward and backward scans. The signal of methane starts rising in the positive-going scan, with the onset potential of 0.25 V RHE until 0.4 V RHE where the baseline is almost reached. In the negative-going scan, the signal increases only below 0.20 V RHE until the lower potential limit which indicates the slow kinetics of the reaction 179 . The mechanism of methane formation is studied before and it is correlated to the reduction of the ethanol derived species that have formed when the surface of platinum is 168, 179, exposed to higher potentials in the presence of EtOH in the electrolyte (e.g. +0.4 V RHE ) 186 . A reductive stripping experiment revealed that adsorbed *CH 3 or *CH 3-containing (e.g. 179 *CH 3CO) adsorbates are attributed to the production of methane . Note that, the formation of methane in the positive-going scan is attributed mostly to formerly generated EtOH derived adsorbates during fast cleaning cycles. Although this reaction is taking place in the flow cell, the formation of ethyl acetate can happen in the ion source too, where ion attachment reaction of dehydrated protonated + ethanol with the corresponding ion structure of [CH3CH 2OH + H - H2O] or simply + [CH 3CH 2] to the acetic acid molecule occurs. The analogy of the I AA and I EA curves certifies two hypotheses about the origin of ethyl acetate. Here, in contrast to MeOH oxidation that the direct detection of formic acid was not possible, the enabled direct detection of acetic acid 72 here by PTR is attributed to the higher PA of acetic acid (783.7 kJ mol -1) compared to the two different primary ions, namely, water (691.0 kJ mol-1) and EtOH (776.4 kJ mol -1) 174 .

8.2.4.2 Contribution of liquid products in gas analysis As it is discussed before, the gas analysis in EC-RTMS is based on the Teflon-AF membrane where the separation of gaseous molecules from the liquid stream takes place. The contribution of liquid products in the gas analysis mass spectrometer (EI-QMS) with this membrane is expected to be small due to the low diffusion coefficient of the liquid analytes. The permeability of the membrane here is examined by the extent of contribution of acetaldehyde in gas analysis during the EtOH oxidation. Based on the reference spectrum from the NIST library, in EI as a hard ion source, the vapor of acetaldehyde produces various fragment ions, including m/z=29, 44 and, 15 81 . By comparing the mass spectrum of acetaldehyde from DART-TOF-MS at m/z = 45.0-45.1, no analogy to none of the ion currents of m/z = 15, 29, and 44 in EI-QMS is observed (particularly, Figure 8.4 The comparison of the ion currents profile m/z=29 is the base peak for acetaldehyde, of (a) m/z = 45.0-45.1 correlated to acetaldehyde in but no meaningful signal enhancement in DART-TOF-MS with mass fragments (b) m/z = 44, (c) m/z = 29 and (d) m/z = 15 associated with the background). This indicates that the acetaldehyde in EI-QMS during electrooxidation of acetaldehyde in the time domain of the ethanol in a cyclic voltammetry with scan rate of 1 -1 experiment does not contribute to these mV s on Pt-ALD. The experiment is done to masses (see Figure 8.4). examine the permeability of the employed membrane in gas analysis and the extent of contribution of a The participation of acetaldehyde liquid product in gas analysis. The direction of the in DEMS which employs the porous sweep experiment is shown by arrows. Electrolyte: Teflon membrane in m/z=44 is very 0.2 M EtOH + 0.1 M HClO 4. Adapted with 124 significant not only because it is the major permission from . product of EtOH oxidation but also because of its relative abundance which is well above

80% of m/z=29. Therefore m/z=44 (base peak of CO 2 EI-spectrum) cannot be correlated entirely to CO 2. To circumvent this problem, detection of CO 2 in DEMS studies of EtOH 2+ oxidation is followed by m/z=22 (CO 2 ) which has an inferior sensitivity (relative 181 abundance of only 2.8% of m/z=44) in the cost of the exclusive fragment ion of CO 2 . The same happens at m/z=15 which is the mass of choice for the detection of methane. In this case, no exclusive alternative mass is available, but the analysis is not complicated since methane formation occurs at potentials where EtOH oxidation to acetaldehyde does not 73 happen 181, 187 . Here we can conclude that employing Teflon-AF resolves these problems by offering simple and clean spectra. The latter is additionally supported by the resemblance of the shape of two MSCVs m/z=44 and 22 in Figure 8.3. Similarly, ethyl acetate could contribute to the m/z = 15 181 ; again the comparison between the m/z = 89.0-89.1 in DART-TOF-MS and m/z = 15 in EI-QMS indicates that the membrane is not permeable to ethyl acetate either. Therefore, the m/z = 15 and 44 were attributed to methane and carbon dioxide, respectively. In conclusion, the high permeability of the membrane used in EC-RTMS for permanent gases and the low permeability for VOCs, makes the interpretation of EI-QMS spectra simpler, as fragments of VOCs have little contribution to the recorded masses.

8.2.4.3 Experimental protocol As it is described before, to ensure the reproducibility of slow scans, an oxidative/reductive removal protocol by repetitive fast cycles was conducted prior to each slow-scan experiment (fast scans were repeated until a steady voltammogram was obtained). This ensures an adsorbant lean surface. The cleaning mechanism can be explained Figure 8.5 Electrochemical procedure employed for by the displacement of all blocking the study of electrooxidation of alcohols exemplified adsorbed species with strongly for the oxidation of ethanol on Pt-ALD. CVs adsorbing platinum oxide which is (potential in black, current in red) vs. time (a), and MSCVs for CO 2 (b,c), methane (d), acetaldehyde (e), reduced back to platinum at lower acetic acid (f), and ethyl acetate (g). The potentials. The fast cycles comprised ten experimental protocol consisted of 10 fast cycles cycles from 0 to +1.05 V RHE with a scan followed by one slow scan, repeated three times in-a- -1 row. Scan rate for slow and fast scans: 1 and rate of 100 mV s . Then a potential step -1 100 mV s . Electrolyte: 0.2 M EtOH+ 0.1 M HClO 4. at 0 VRHE was applied for 3 minutes to Adapted with permission from 124 . allow the ion currents of the mass spectrometers for all products, reach the background level. Finally, a slow CV was performed -1 from 0.0 VRHE to +1.2 V RHE with a scan rate of 1 mV s . The functionality of the protocol is 74 demonstrated in Figure 8.5 during the oxidation of EtOH and the formation of the gaseous and liquid products was monitored versus time. The shape of all ion currents for different masses, as well as current density during the slow scans, shows the effectiveness of the protocol. Consecutive slow scans to show the reproducible response during the slow scan.

8.2.5 1-PrOH electrooxidation Electrooxidation of 1-PrOH is studied on the surface of Pt-ALD during a CV with the scan rate of 1 mV s -1. The experiment was done in an Ar- saturated electrolyte of 0.2 M 1-PrOH

+ 0.1 M HClO 4. The cyclic voltammogram (black) and ion currents for carbon dioxide in two mass to charge ratios, namely m/z = 44 (red) and m/z = 22 (dark blue), methane (green), and ethane (olive) as gaseous products, as well as propionaldehyde (purple), propionic acid (orange) and propyl propionate (light blue) as liquid products are presented in Figure 8.6. The detected oxidation products of the reaction are propionaldehyde, propionic acid, and CO 2 which is in agreement with previous investigations 188, 189 . Propyl propionate is formed during the reaction of propionic acid and the excess of 1- PrOH in the electrolyte. Ethane and methane are detected as the reduction products which is consistent with the literature 188 . The CV shows again multiple superimposed oxidation peaks. In the positive-going scan, the oxidation begins at ca. 0.5 V RHE , demonstrated by a broad current density profile, and ion current for all oxidation products increases. Generally, 1-PrOH with a longer chain (3-carbons) has a lower activity Figure 8.6 Combined cyclic voltammetry and compared to EtOH or MeOH 190 . The latter is product analysis during electrooxidation of 1- evidenced by the maximum current density of PrOH on Pt-ALD. CV (a) and simultaneously -2 recorded MSCVs for carbon dioxide (b,c), ca. 40 mA cm which appears at the upper methane (d), and ethane (e) as gaseous potential limit of the experiment (+1.2 V RHE ). products; propionaldehyde (e), propionic acid The profile of the ion current of (f), and propyl propionate (g) as liquid products. Electrolyte: 0.2 M 1-PrOH+ 0.1 M propionaldehyde is very similar to one from -1 HClO 4. Scan rate: 1mV s . The arrows acetaldehyde during EtOH oxidation with the indicate the direction of the potential sweep. difference that during the backward scan the Adapted with permission from 124 . 75 formation rate of propionaldehyde is slightly higher than the forward scan. This indicates the formation of aldehyde as the intermediate of oxidation of alcohols is more inhibited in the forward scan by 1-PrOH derived adsorbates compared to EtOH derived adsorbates. Again, similar to EtOH oxidation, the resemblance of ion current profiles m/z = 44 with the m/z = 22 indicates that m/z = 44 is the unique fragment of CO 2 under our conditions. The MSCV of CO 2 shows that the formation rate is higher at the forward scan, which again implies that the oxidative removal of the 1-PrOH derived adsorbates on the surface is more pronounced in the positive-going scan. Nevertheless, the peak position for both scan directions is the same at ca. +0.8 V RHE . Same to EtOH oxidation, stripping experiments revealed that CO 2 is formed as the platinum surface is exposed to 1-PrOH as early as at ca. +0.35 V RHE . Oxidative removal of adsorbates from the surface is the reason for the higher formation rate of propionaldehyde and propionic acid in the negative-going scan 188, 191 .

In the forward scan, the ion current for propionaldehyde (I PAD ) increases in the range 0.60-0.9 V RHE until it reaches a constant level, while in the negative scan declines a bit due to hindered kinetics on the platinum surface current drops as the platinum gets oxidized. the maximum formation is maximized at +0.8 V RHE . By reversing the sweep direction, the current increased along with two peaks of propionaldehyde and propionic acid. During the negative potential scan, j drops as potential ramped negative. A second oxidation peak with the onset potential of 1.0 V RHE when the signals for CO 2, propionaldehyde, and propionic acid increase. The same position with a maximum current density lower than half compared to the first peak at 0.85 V RHE . Only minimal amounts of methane and ethane are detected at lower potentials. The mechanism of formation of these two is the hydrogenation of 1-PrOH derived adsorbates on 188 the surface in the lower potential regimes (˂+0.4 VRHE in H upd region) . Ethane is formed in both directions of scans. Specifically, in the positive-going scan, ethane is formed due to the reduction of adsorbates that are partially retained on the surface from the preceding fast scan cleaning (see Figure 8.5). Based on the NIST library 81 , in the EI ion source, ethane breaks into many fragments including one with m/z = 15, which means ethane has a contribution to the signal of methane. The latter is witnessed by the slight increase at ca. +0.3 V RHE and analogy of ICH4 and IC2H6 in the positive-going scan. The formation of methane is confirmed with the significant increase (cannot be interpreted as the fragment of ethane) in the negative- going scan with the onset potential of +0.1 V RHE and it is in good agreement with the previous report 188 . The formation of propyl propionate takes place in both SFC as well as in the DART reaction chamber, similar to ethyl acetate during EtOH oxidation. From the analogy of the ion currents between propionic acid/propyl propionate and acetic acid/ethyl acetate, it can be concluded that the reaction of the electrochemically formed ester from organic acid is fast. Therefore, indirect detection of formic acid during MeOH oxidation as methyl formate is correct.

8.3 Conclusions EC-RTMS was employed to investigate the electrooxidation of MeOH, EtOH, and 1- PrOH on high surface area platinum prepared by the ALD technique. Some features and characteristics of the EC-RTMS, as well as valuable information on the relationship between 76 potential and product distribution, are presented. (i) This is the first report on the direct detection of organic acids as the nonvolatile reaction products of oxidation of alcohols in real time. (ii) The mechanism of the reaction including the role of proton affinity in the liquid analysis is clarified under operation when the reactant is ionizable and interpretation of the mass spectra and simple assignment of the m/z to the products were presented and discussed. (iii) The extent of contribution of liquid products in the gas analysis was proved that is minimal which is desirable for simple assignment of gaseous products in a hard ionization source mass spectrometer. (iv) The analysis of products revealed that oxidation of all studied primary alcohols results in corresponding aldehyde and organic acid as well as carbon dioxide as the end oxidation product. (v) Additionally, in all cases, the formation of the corresponding ester was detected which was attributed to the homogenous reaction between the alcohol as the reactant and the organic acid as the electrochemical oxidation product that takes place in both flow cell and ion source of liquid analysis. (vi) In the case of oxidation of MeOH, formaldehyde was not detected while the detection of formic acid was performed indirectly by following the formation of methyl formate. The difficulty in the determination of formaldehyde and formic acid directly was attributed to the high detection limit in the presence of the reactant MeOH in the ion source and to the mechanism of the ionization. (vii)

Except for MeOH oxidation, where the oxidation yields a significant amount of CO 2, a minimal amount of CO 2 was detected in the case of EtOH and 1-PrOH oxidation.

77

9 Study of the mechanism of electrooxidation of isopropanol

with time-resolved analytics on Pt and bimetallic Pt 1-xRu x

9.1 Introduction The is the most sustainable, decarbonized approach to store renewable electricity with the feasibility of grid-scale storage. This energy scenario was firstly introduced by John O’M. Bockris that hydrogen could serve as the primary energy carrier in different energy sectors of a society 192 . The gravimetric energy content of hydrogen is the highest among all the compounds (120 MJ Kg -1) which is ca. 4 times higher than 1 L of gasoline (31.67 MJ) 193 , however, the volumetric density (e.g. essential parameter in the transportation sector) of hydrogen gas under ambient conditions is very low and is one-third of methane 7, 194 . To enhance the volumetric energy density, hydrogen can be either liquified at temperatures close to absolute zero (boiling point: 21.15 K 195 ) or compressed at very high pressure (up to 700 bar) 196 . However, storage, distribution, and transportation of the pressurized/liquid hydrogen need special facilities plus being undesirable from safety or economical aspects. Hydrogen carriers are safe alternatives to store hydrogen reversibly or irreversibly close to ambient conditions. They can be a dense form (e.g. liquid or solid) of stored hydrogen which molecular hydrogen bounded by physisorption i.e. by condensing it into e.g. microporous materials like carbon nanotubes 197 or metal-organic frameworks 198 . The other approach is to chemically bind it in chemical compounds e.g. ammonia 199 , methanol, or formic acid, however, they decompose after releasing hydrogen. Liquid organic hydrogen carrier (LOHC) is an innovative way to store hydrogen in chemical bonds under room temperature and ambient pressure reversibly in the liquid phase 200 . Figure 9.1 represents the molecular structures of dibenzyltoluene (H0-DBT)/perhydro- dibenzyltoluene (H18-DBT) as the state of the art of hydrogen lean/rich pair of a LOHC. The structure of H0-DBT consists of a chain with 3 aromatic rings which gives high hydrogen capacity equal to 6.2 wt % to the molecule. Hydrogenation of the H0-DBT with molecular hydrogen results in H18-DBT which is a diesel-like liquid and can be stored or transported with hydrocarbon fuel infrastructure (tanks or pipelines) to the point of consumption where it is dehydrogenated in a dehydrogenation reactor onsite to H0-DBT and molecular hydrogen. The generated hydrogen can be used successively in a fuel cell for the generation of electricity. The cycles of hydrogenation and dehydrogenation can be repeated many times without the formation of a considerable amount of undesired side products. The process of hydrogenation H0-DBT is exothermic, while the dehydrogenation of H18-DBT is 201 endothermic with the heat of 65 K per mol of H 2 . This makes the latter approach less attractive especially in the transport sector on the account of the requirement of a constant supply of heat during the dehydrogenation step (~300°C), additionally the necessity of purification of produced hydrogen before introduction to the fuel cell. When the released hydrogen is used in a proton-exchange membrane fuel cell (PEM-FC), the efficiency for storing electricity and regeneration to electricity is less than 38%. Recently, a new concept based on LOHC is introduced, where H18-DBT LOHC is coupled to a PEM-FC based on isopropanol (IPA). According to Figure 9.1, the IPA-PEM-FC generates electricity, and ACE as the product of the IPA electrooxidation reaction (IPAOR) is 78 formed. Then, ACE is sent to a successive transfer hydrogenation reactor (THR), where it is catalytically hydrogenated by H18-DBT to IPA. The regenerated original fuel i.e. IPA is recycled back to the IPA-PEM-FC for the generation of electricity. This cycle continues until all H18-DBT is converted to its dehydrogenated form (H0-DBT). Then, H0-DBT can be hydrogenated later (e.g. with molecular hydrogen) to H18-DBT. If IPA is selectively oxidized to ACE and reduced back, it can serve as a rechargeable fuel by LOHC. This cycle is carbon- neutral which fits the climate policy due to the absence of a carbon footprint and it can boost the energy recovery over 50% 202 .

Figure 9.1 A scheme of the closed sequence of isopropanol proton-exchange membrane fuel cell (IPA- PEM-FC) coupled with transfer hydrogenation reactor (THR) for the hydrogenation of acetone (ACE). IPA-PEM-FC generates electricity and ACE, while in THR, ACE is reduced reversibly by reaction with hydrogen-rich LOHC (H18-DBT) to IPA as the original fuel for the generation of more electricity back in IPA-PEMFC as well as hydrogen lean LOHC (H0-DBT). Reproduced with permission from 202 with permission from The Royal Society of Chemistry.

As it is mentioned, the objective here is to make a reversible fuel cell, which means that, IPAOR should produce ACE selectively without the formation of dissociation side- products like CO 2 in the operational potential range IPA-PEM-FC to have a closed system. This is different from the target of electrooxidation of small chain primary alcohols (e.g. MeOH or EtOH) in DAFCs where decomposition of alcohol to the end oxidation product

(CO 2) is desirable in order to reach the highest power of the fuel cell. To optimize the efficiency of the overall energy system the performance of the IPA- PEM-FC has to be improved 202 . The latter means, that the understanding of important aspects of the reaction, including activity, selectivity, and durability of the IPAOR as well as the stability of the electrochemical interface has the utmost importance and is the target of this chapter. The latter is done by investigation of the reaction with various analytical techniques under acidic conditions (which is relevant to PEM-FC) on the Pt and PtRu surfaces. Platinum is selected because it is the only reported active catalyst for IPAOR under acidic although under alkaline conditions, platinum has lower activity 58, 203-205 and other catalysts like pure palladium 58, 205, 206 and palladium-based alloys 207-209 , gold 204, 210 , rhodium 211 are capable of oxidizing IPA.

79

9.2 Isopropanol electrooxidation on Pt The presented results in this section (9.3) are a part of the published manuscript 126 and my contribution was time-resolved product analysis with EC-RTMS. The presented measurement of EC-IRRAS was carried out by Mr. Fabian Waidhas, a Ph.D. student of the Chair of Interface Research and Catalysis from the department of chemistry, FAU. The powerful combination of EC-RTMS and electrochemical infrared reflection absorption spectroscopy (EC-IRRAS) is employed to study the IPAOR mechanism on platinum planar electrodes. IR spectra with p-polarized light reflect the physicochemical properties of the electrochemical interface i.e. intermediates and adsorbed species at the double layer while real-time mass spectrometry and s-polarized IR spectra display desorbed products of the reaction. The presented result is to a large extent a part of a collaborative publication 126 .

9.2.1 Sweep experiment with EC-RTMS Pt is the only reported pure metal with a satisfactory activity in the acidic medium for IPAOR. Previous investigations of the reaction on platinum polycrystalline (Pt-poly) revealed that the reaction has only two products. ACE as the bulk product is attributed to oxidation of the IPA, while a negligible formation of CO 2 is assigned to electrooxidation of the strongly adsorbed species on the surface of platinum 212 . Despite the high activity and selectivity of IPA to ACE on platinum, the performance of the electrode drops significantly over time. The latter is evidenced by a sweep voltammetry experiment in Figure 9.2. Time evolution of potential, current density (j) as well as synchronized ion current for ACE

(I ACE ) from DART-TOF-MS are shown in the three panels. The electrochemical protocol comprises 4 successive CVs in Figure 9.2 Electrooxidation of IPA on polycrystalline the potential range of from +0.1 to +1.0 platinum during 4 consecutive cyclic voltammetry VRHE . The measurement was performed experiment. The temporal evolvement of potential at 0.2 M IPA + 0.1 M HClO 4 electrolyte (blue) current density (red) and simultaneously with a scan rate of 10 mV s -1. Within recorded mass signal for ACE (green) are shown. The experiment was done with a scan rate of 10 mV s -1 in one CV, two oxidation peaks in forward 0.2 M IPA+ 0.1 M HClO 4 electrolyte. Adapted with and backward scans are observed. permission from 126 . Copyright (2020) American During the positive-going scan, the Chemical Society. 80

oxidation begins with the onset potential of ca. +0.3 VRHE. The maximum conversion rate is reached at +0.75 V RHE , where due to the formation of platinum oxide j begins to decline despite the supply of enough IPA to the surface in the flow cell. At the negative-going scan, oxidation starts at the onset potential of +0.85 V RHE as soon as the formed platinum oxide commences to reduce which renders platinum sites free for the oxidation of IPA. The higher oxidation current in the backward scan is attributed to the concomitant removal of oxide species and carbonaceous adsorbed species, so in a short time, an excess of platinum sites becomes available to carry out the reaction. These observations match with previous studies 212 .

The profile of I ACE resembles that of j, which indicates the selective production of ACE from IPA oxidation. Comparing the consecutively performed cycles, both j and I ACE decrease with the number of cycles, but neither the onset potentials nor the position of the peaks shifts. Such suppression is attributed to the progressive poisoning of the electrode with an adsorbed species which accumulates over time and introduces merely spatial restrictions to the reactant IPA on the platinum. The cycle-to- cycle decay in both j and I ACE indicates that the upper potential limit (UPL) is not sufficient to remove the adsorbates from the surface. Figure 9.3 represents a comparison of CVs and MSCVs of sweep experiments at low (+1.0

VRHE with solid curves) and high (+1.5 V RHE with dashed curves) Figure 9.3 Cyclic voltammetry experiment during the IPA UPLs. By expanding the UPL to electrooxidation on polycrystalline platinum (a) CVs and higher potential a new oxidation (b) simultaneously recorded MSCVs for ACE. The experiment is done with a scan rate of 10 mV s -1 in 0.2 M peak is observed. In the positive- IPA in 0.1 M HClO 4 electrolyte with two different upper going scan, in addition to the lower potential limits of +1.0 V RHE (solid curves) and +1.5 V RHE potential oxidation, a second peak at (dashed curves). Adapted with permission from 126 . Copyright (2020) American Chemical Society. around +1.3 V RHE is observed which is attributed to the oxidation of IPA to ACE on the partially oxidized/ cleaned surface concluded based on the perfect match between the profile of j and IACE . The higher peak, however, disappears by the formation of more oxide which is evidenced in the backward scan where the j and I ACE reached the baseline due to the complete block of the surface by formed oxide. Moreover, the onset of the oxidation of IPA to ACE in the negative-going scan is 81

delayed (onset potential is ca. 0.8 V RHE ), due to sluggish kinetics of formed massive amount of oxide. Interestingly, the maximum current in the negative-going scan is significantly higher which is a sign that the surface auto-inhibition by formed ACE ads, and it is difficult to oxidize in lower potentials and only are removed when the surface is exposed to very high potentials. It should be noted that the scan rate of the experiment is very decisive about the shape of the CV or MSCV, apart from the different charging currents, an slow scan rate of 1 mV/s enhances the deactivation of the catalyst and provokes a faster poisoning by the produced ACE.

9.2.2 Examination of adsorbates with EC-IRRAS 1 To resolve the nature of the adsorbed and desorbed species of IPAOR on platinum, EC-IRRAS measurements were carried out. Activity of different platinum single crystal orientation varies by order of Pt(111) > Pt(100) > Pt(110) 167 . Despite this difference, the reaction obeys a similar reaction mechanism through intermediates to products 213 . Figure 9.4 (a) displays the attenuated total reflection (ATR) spectra of the IPA (reactant) the ACE (product) which are used for the assignment of bands during the IPAOR with EC-IRRAS. Figure 9.4 (b,c) and (d,e) show the infrared spectra with p- and s-polarisation on a Pt(111) respectively. The experiments were done in the region from 0.05 V RHE to 1.1 VRHE , with the reference spectrum at 0.05 V RHE . Negative bands correspond to produced species, whereas positive bands associate with consumed species.

Starting at 0.3 V RHE and increasing up to -1 0.7 V RHE , negative bands at 1240 cm -1 Figure 9.4 Investigation of IPA oxidation with EC- (vasym C-C), 1372 cm (the δ sym CH 3/ -1 -1 IRRAS on Pt(111). (a) ATR reference spectra of IPA νasym C-C), 1426 cm /1447 cm (δ (blue curve) and ACE (green curve). (b,c) The -1 CH 3) and 1702 cm (v C=O) are relationship between potential and spectra with p- and s- assigned to the formation of ACE at polarized lights and (c,e) magnified spectrum in the double layer, coinciding with the ATR carbonyl region. The reference spectra were recorded in all measurements at 0.05 V RHE . Electrolyte: 0.2 M IPA+ reference spectrum of ACE. Positive 126 0.1 M HClO 4. Reprinted with permission from . -1 bands appearing at 1129 cm (v C-O), Copyright (2020) American Chemical Society.

1 The measurements were carried out by Mr. Fabian Waidhas, a Ph.D. student of Chair of Interface Research and Catalysis from Department of Chemistry at FAU. 82

-1 -1 -1 1167 cm (v C-C), 1308 cm (δ O-H) and 1468 cm (δ asym CH3/ δ sym CH3) indicate the -1 consumption of IPA. At 0.8 V RHE a band at 2344 cm appears corresponding to the minimal amount of CO 2 formation. Comparing the spectra in s- and p-polarisation, only the carbonyl vibration shifts blue for the p-polarisation by 6 cm -1 as displayed in Figure 9.4 (c) and (e). The ACE bands are more intense for p-polarisation than for s-polarisation, which shows the presence of ACE as an adsorbed species and in the bulk solution. The EC-IRRAS measurements provide unambiguous evidence that the oxidation of IPA on platinum is associated with the increase of ACE coverage on the electrode surface. Therefore, it is proposed here that the origin of the platinum deactivation during the IPA oxidation is the slow desorption of ACE ads species. This aspect is going to be discussed below in more detail. In contrast to primary alcohols that they dissociate through adsorbed CO, in situ FTIRS reveals a negligible cleavage of C-C bond during oxidation of IPA as secondary 169, 212, alcohol on the platinum and ACE ads as an intermediate is identified as a blocking agent 214 .

9.2.3 Electrode cleaning protocol In order to remove the ascorbates from the surface and do not pass the history of the electrode from one measurement to another, a surface cleaning protocol was executed before each experiment with EC-RTMS. This would ensure a reproducible surface coverage. The protocol comprises applying a very positive potential of +1.67 VRHE for 1 min to displace all the blocking carbonaceous adsorbed species by strongly adsorbing oxide of platinum. The latter was followed by a reduction of formed platinum oxide by switching potential to +0.05 for 2 min at H upd region.

9.2.4 The kinetics of sorption during IPA oxidation on Pt The nature of adsorbates on the surface is highly dependent on the applied potential 212 and at each potential, a quasi-equilibrium coverage of different adsorbates is established. The kinetic studies on the oxidation of saturated alcohols on platinum reveal that the number of α- H (the hydrogen atoms bonded to the carbon that contains –OH bond) is the determiner for the mechanism of oxidation 50 . In detail, at higher potentials, the first step of oxidation of . primary alcohols (RCH 2OH) involves dissociative chemisorption to R-. C-OH by occupying two adsorption sites per molecule 215, 216 . The difference between primary and secondary alcohols (R 1R2CHOH) is the absence of the chemisorbed species which is attributed to the single α-H. The mechanism of IPAOR is displayed accordingly on the surface of platinum in Figure 9.5.

Figure 9.5 The suggested mechanism of IPA oxidation on the platinum surface to ACE as the main bulk product of the reaction 212 .

83

Figure 9.6 addresses the correlation between adsorbate coverage and potential during IPAOR on the platinum surface. In Figure 9.6 (a) the proposed electrochemical protocol is displayed. After cleaning the surface (green zone) at +1.67 V RHE for 1 min, in the next interval, the impact of a variable potential of E 1 (0.1, 0.2,…,1.4 and, 1.5 V RHE applied for 300 s) is studied on the E 2= 0.6 V RHE for 300 s where the conversion rate of IPA to ACE is maximized with reference to the sweep experiment. Figure 9.6 (c) shows the absolute amount of produced ACE which is acquired by the integration of I ACE versus time (∫IACE dt) during the applied potential of E1 (red curve) and E 2=0.6V RHE (blue curve) versus E 1. The process of integration is exemplified in Figure 9.6 (b) when E1=0.8 V RHE . In order to achieve the amount of ACE at E 2=0.6 V RHE , when the electrode was modified previously by potential E1, the experiment once was done by performing both steps in a sequence without any cleaning process in between. Later the electrode was cleaned by removing the adsorbates which were formed from the previous experiment (middle green zone). Eventually, the experiment was repeated only at E1 and by Figure 9.6 Investigation of kinetics of sorption knowing the contribution of E 1 (red shaded during oxidation of IPA on platinum during step experiment (a) potential program as a function of area), the amount of ACE at E 2 was time comprises of three zones (i) cleaning surface in subtracted from the experiment of two green (an oxidative step of +1.67 V RHE for 60 s steps to acquire the blue shaded area. A followed by a reductive potential +0.05 V RHE for closer look at both integration curves at E 2 120 s), (ii) a variable potential of E 1 in red (0.1, 0.2, ... 1.4 and, 1.5 V for 300 s), (iii) a constant and E 1, the strong dependence of the ACE RHE oxidative step in blue (0.6 V RHE for 300 s) (b) an signal to the potential of E 1 is witnessed. exemplary cut of simultaneously recorded ion The red curve is indeed almost the same as current of ACE as the liquid product of the reaction the MSCV of ACE in Figure 9.6, which when E 1 = +0.8 VRHE (c) integrated ion currents as indicates that the results in the sweep the function of the potential E 1. The integrated amounts for E 1 (red curve) and E 2 (blue curve) are experiment are close to steady-state shown in points. Electrolyte: 0.2 M IPA+ 0.1 M 126 conditions. The valley trend of the blue HClO 4. Adapted with permission from . curve on the other hand at E 2=+0.6 V RHE is Copyright (2020) American Chemical Society. associated with the previously formed adsorbates when the electrode was exposed to E 1. A 84 higher value of the signal is translated to lower coverage that blocks the surface for the conversion of IPA to ACE at 0.6 V RHE . By following the ∫IACE dt at E 2 when E 1≥0.3 VRHE , the production of ACE is inhibited entirely when E 1 is between 0.3 to 0.6 V RHE . This means that adsorbate coverage in the middle potential range reaches its maximum. The signal begins to increase at higher potentials as

E1≥0.7 V RHE . According to the results of EC-IRRAS, ACE ads is the main predicted adsorbent when E1≥0.3 VRHE . This means that at higher potentials, the desorption of ACE is facilitated (lower coverage). At lower potential ranges, however, when E1= 0.2 VRHE , the suppressed signal indicates that another adsorbant other than ACE ads is involved since neither ACE is detected in the bulk of electrolyte with EC-RTMS nor on the surface of the electrode with EC-IRRAS. This can be hypothesized by the surface blocking during the dissociative 212 adsorption of IPA at potentials around E 1=+0.2 V RHE , as proposed before by Pastor et al. . The conclusion so far is that that IPA is oxidized selectively to ACE on platinum, but the reaction rate is restricted by desorption of the produced ACE or in other words, the kinetics of the IPAOR on platinum is controlled by the desorption of ACE ads at room temperature.

9.3 IPA electrooxidation on Pt/C and bimetallic Pt xRu (1-x) /C nanoparticles The presented results in this section (9.3) are adapted from a published manuscript 217 with permission from the publisher. I’m the first author of the manuscript and my contribution here is time-resolved product analysis with EC-RTMS as well as OCP, and steady-state electrolysis experiments. The dissolution study and RDE measurement were performed by Dr. Florian Speck, and Mr. Iosif Mangoufis-Giasin, respectively both former or current Ph.D. students from the Electrocatalysis unit of HI-ERN. PtRu is the state of the art electrocatalyst for the oxidation of MeOH and EtOH under acidic conditions. Compared to pure Pt, PtRu is more robust towards poisoning due to higher CO tolerance by lowering the oxidation potentials of water on ruthenium and adsorption of

OH. The latter assists the process of oxidation of CO ads as the intermediate of the reaction to 172, 218, 219 CO 2 and thereby hinders the CO-poisoning of the platinum catalyst . Here experiments show that the addition of ruthenium decreases the onset potential of IPA oxidation as well as promotes the desorption of the ACE ads and diminishes the poisoning effect of it. The objective of this section is to correlate the activity, selectivity, and stability on the composition of Pt xR(1-x) /C (x=0.5, 0.6, 0.67) and Pt/C nanoparticles during IPAOR and thus to find the optimum composition from different perspectives. The latter was addressed by analyzing products with SFC-EC-RTMS (ACE and CO 2) and the stability of nanoparticles (platinum and ruthenium) with SFC-ICP-MS. The electrocatalytic properties of the NPs are examined by the different potentiostatic and potentiodynamic experiments in the acidic medium which is equivalent to conditions of the IPA-PEM-FC.

9.3.1 Electrode cleaning protocol The previous cleaning protocol of positive potentials that was employed for cleaning of the platinum planar electrode is avoided here for the PtRu nanoparticles. This is mostly because, ruthenium is less noble than platinum, i.e. the stability potential window of ruthenium is narrower and the dissolution rate of ruthenium is much higher than platinum (see section 9.3.6 for more details). Thus, this would deliver a surface with a different metallic 85 composition than the pristine state. Alternatively, the poisoning species can be removed from the surface by previously reported cathodic reduction procedure 212 .

To verify that, Figure 9.7 displays an exemplary protocol on Pt-poly in which potential, current density, and ion current for ACE versus time are shown. The protocol includes a potential of 0.02 V RHE (for 10 min) to remove the ACE ads which is produced from an oxidation step (0.62

VRHE for 2 min). Note that at 0.02 V RHE since no oxidation product is expected, therefore the signal is decreasing to baseline. A decay of j and I ACE with time is witnessed due to the poisoning effect due to the accumulation of ACE ads . Due to the signal delay of the mass spectrometer, the intensity of the ACE signal reaches the baseline signal Figure 9.7 Exemplified cleaning protocol of electrode intensity only after a few seconds of the after performing IPA oxidation reaction on the platinum polycrystalline surface in 0.2 M IPA+ 0.1 completion of the oxidation step of IPA to M HClO 4. Potential (black), current density (red), ACE. By repeating this protocol four and signal of ACE in the mass spectrometer (blue) times, the same pattern for j and I ACE is are plotted vs. time. The potential was alternately observed which verifies the success of the switched from +0.62 V RHE for 2 min to oxidize IPA to ACE followed by +0.02 V RHE for 10 min to cleaning procedure and therefore it is remove adsorbates from the surface formed during used for all further investigations. the oxidation step. The protocol was repeated 4 times to show reproducibility. Reprinted with permission 217 9.3.2 Sweep experiment with EC- from . Copyright (2020) American Chemical Society. RTMS 1 By looking at the particle size that is reported in Table 5.5, the difference in the surface area is estimated to be minimal and changes in the performance of the catalyst can be attributed mostly to the impact of the composition of nanoalloys rather than to the impact of size. It can be hypothesized that two oxidation peaks on all PtRu can be correlated to two different pathways for oxidation of IPA. The appearance of the first early peak in the PtRu catalyst is observed only in the case of IPA as secondary alcohol and it is not reported in the case of MeOH or EtOH electrooxidation for example. This is also in line with the higher power density in the fuel cell of IPA-PEM-FC compared to fuel cells based on primary alcohols 220 . The enhanced activity is attributed to the PtRu ensembles or co-catalytic behavior of ruthenium although ruthenium

1 The RDE measurements were conducted by Mr. Iosif Mangoufis-Giasin from Electrocatalysis unit of HI-ERN. 86 is the completely inactive catalyst (see Figure 9.9). The appearance of two oxidation peaks on

Pt xRu 1-x/C can be explained by two mechanisms of IPA oxidation. Again, a very good analogy between j and I ACE indicates the reaction is selective to ACE. By looking at LSVs and MSLSVs for ACE, all the Pt (1-x) Ru x/C catalysts exhibit superior activity compared to Pt/C due to the appearance of a new oxidation peak at lower potentials with onset ca. 300 mV earlier. This means that the presence of ruthenium offers a lower energy path for IPA to ACE while platinum shows only one single peak at the higher potential range with almost the same peak position for all PtRu ratios (a very slight shift in peak position by the addition of ruthenium is observed). IPA activation at lower potentials may be attributed to the beneficial effect of the participation of partially covered adsorbed OH on ruthenium. The first pathway which results in the first peak, the only

ACE, and no CO 2 is identified in the sensitivity range of EI-QMS. Therefore, it can be concluded that IPA gets oxidized through the

ACE ads and it stops at the desorbed ACE as the end-oxidation product.

By looking at j and I ACE it can be concluded that the onset potential and maximum current of the first oxidation peak on all PtRu ratios are almost the same. This means this peak requires a certain amount of ruthenium and Pt 0.5 Ru 0.5 /C is the border for this peak because the addition of more ruthenium does not enhance the I ACE . The conversion of IPA to Figure 9.8 Electrooxidation of IPA on different nanoalloys ACE begins from ca. 0.1 V RHE and it of Pt (1-x) Ru x/C (x=0, 0.5, 0.6, 0.67) and Pt/C during linear declines at 0.2 V RHE , as it is sweep voltammetry and simultaneously recorded ion indicated by the decrease in both j currents from mass spectrometry for ACE (I ACE ) and carbon dioxide (I CO2 ) as the function of potential. Scan and I ACE . As the electrolyte flow -1 rate: 5 mV s . Electrolyte: 0.1 M HClO 4 + 0.2 M IPA . ensures a constant supply of IPA to Reprinted with permission from 217 . Copyright (2020) the surface, the activity decline is American Chemical Society. probably due to a change in the electrode state, e.g. to the early oxidation of the ruthenium which takes place at these potentials. This is detrimental to the performance of the IPA-PEM- FC because the first oxidation peak is mainly in the operating potential range of the fuel cell 221, 222 . The decreased current continues with a steady value until the second peak which is attributed to the second pathway oxidation of IPA on the platinum surface starts. This peak again involves both platinum and ruthenium and it proceeds through ACE ads . The main part of 87

ACE ads is desorbed as ACE as the bulk product and traces of strongly

ACE ads on the surface poisons the surface and it ends up to CO 2. By comparing j and I ACE of all nanocatalysts, Pt 0.5 Ru 0.5 /C has slightly more activity than Pt/C, however, the addition of more ruthenium causes a significant drop in the activity to ACE. There are two explanations for this behavior: first, it is evident that the second peak is mostly related to Pt-Pt sites. However, the role of ruthenium is more than a dilution of these sites since ruthenium is an Figure 9.9 Linear sweep voltammograms executed on inactive catalyst and causes less platinum and ruthenium polycrystalline disks under exposure of platinum to IPA which hydrodynamic conditions in the electrolyte containing (solid curves) 0.2 M of IPA and in the blank electrolyte will be discussed later in detail in -1 (dashed curves). Scan rate: 5 mV s . Rotation speeds of section 9.3.9. The onset potential of the RDE: 900 rpm. Supporting electrolyte: 0.1 M HClO 4. 217 the second on the Pt/C is +0.33 V RHE , Reprinted with permission from . Copyright (2020) American Chemical Society. which is determined when the I ACE exceeds three times the standard deviation of the background signal of the mass spectrometer.

This is almost the same for PtRu (+0.36 V RHE for Pt 0.5 Ru 0.5 /C). Note that the determination of the exact onset potential for the second peak on PtRu is trivial since two oxidation peaks are not deconvoluted. By the addition of more ruthenium, a slightly negative shift in the position of the second oxidation peak, as well as a decline in the j and I ACE at the peak are witnessed. By looking at the MSLSVs for CO 2, the onset potential for the formation of CO 2 for all nanoparticles are more or less the same. The I CO2 starts rising at 0.65 V RHE and keeps to increase up to the UPL of the experiment. The profile and intensity of I CO2 in cases of Pt 0.5 Ru 0.5 /C and Pt/C are almost similar while a higher ratio of ruthenium leads to a lower amount of CO 2 i.e. for Pt 0.4 Ru 0.6 /C and Pt 0.33 Ru 0.67 /C. The latter reflects that a high ratio of ruthenium can reduce the adsorption energy of the intermediates (e.g. ACE Ads) that may get oxidized to CO 2. The absolute amount of formed CO 2 for all catalysts is very low which is witnessed by lower S/N of the I CO2 as well as the unlikeness of LSV and I CO2 . As it is mentioned before, measurements with EC-RTMS here are semi-quantitative, so there is no correlation between IACE and I CO2 since the two signals are recorded from two different instruments. In conclusion, Pt 0.5 Ru 0.5 /C shows the highest activity among all examined catalysts towards ACE. These results information tells that the performance of the IPA-PEM- FC is enhanced by the addition of ruthenium. LSV during IPAOR on platinum and ruthenium polished polycrystalline disks are performed at the scan rate of 5 mV s -1 under well-defined mass transport with the RDE electrode at 900 rpm. Figure 9.9 displays the superimposed voltammogram obtained with 0.2

M IPA (solid curves) and without IPA (dashed curves) in 0.1 M HClO 4 supporting electrolyte. By comparing the black curves platinum exhibits significant activity compared to the background current. This is witnessed by the oxidation peak with the onset at ca. +0.35 V RHE 88

-2 and the maximum current density ca. 14 mA cm at +0.75 V RHE . However, ruthenium shows no feature oxidation peak at more positive potentials regardless of the presence of IPA (red curves). Therefore it can be concluded that pure ruthenium is an inactive catalyst for IPAOR in the studied potential range and since ruthenium is known as an OH donor, its bimetallic effect on platinum surface sites can be associated with the lower energy pathway for the dissociative adsorption of IPA 183 .

9.3.3 Origin of CO 2 during IPA oxidation For a sustainable IPA fuel cell operation, the oxidation of IPA must be selective to

ACE without the formation of CO 2 as the decomposition product which cannot be thereafter hydrogenated with LOHC. In order to get more insight into the source of CO 2, the sweep experiment was carried out in supporting electrolyte on Pt/C and

Pt 0.5 Ru 0.5 /C in the absence of IPA with the same protocol as in Figure 9.8 and

MSLSVs for CO 2 were recorded in parallel. The recorded results are compared to results when the electrolyte contains IPA and all are illustrated in Figure 9.10. In blank voltammograms, on both surfaces, the

CO 2 signal starts to rise at roughly the Figure 9.10 Synchronized recorded ion currents for CO 2 same onset potential. However, the during linear sweep voltammetry on Pt/C (black) and Pt 0.5 Ru 0.5 /C (red). The experiments were carried out in amount of formed CO is distinctly 2 supporting electrolyte of 0.1 M HClO 4 in the absence lower than when the catalyst is (solid curves) and in presence of 0.2 M IPA (dotted -1 exposed to IPA. The minimal amount curves). Scan rate: 5 mv s . Reprinted with permission from 217 . Copyright (2020) American Chemical Society. of CO 2 in supporting electrolyte can be associated with the corrosion of carbon or oxidation of the trace impurities of organic species that are present in the flow cell which may get oxidized to CO 2. The conclusion can be drawn that the difference between the solid curves and the dashed curve is correlated to the formation of CO 2 from IPAOR which is negligible. A kinetic study by time-resolved FT-IR revealed that the oxidation of IPA to CO 2 is a relatively sluggish reaction compared to IPA oxidation to ACE 213, 223 .

89

9.3.4 IPA oxidation during step experiment The chronoamperometry at constant potential provides insight mainly on the rate formation of desorbed ACE as the product of the reaction which is a good indicator to understand the kinetics of deactivation of the catalyst due to saturation of the surface with the adsorbate under steady-state conditions. Figure 9.11 displays the potential program during the chronoamperometry experiment that comprises of variable oxidative potential steps (0.1, 0.2, ….0.8,

0.9, and 1.0 V RHE ) to oxidize IPA to ACE for 5 min. The partially poisoned surface is regenerated by an intermittent step at 0

VRHE for 2 min between each oxidative step to electroreduce the ACE ads e.g. to propane as well as allow the IACE to decrease to the background. Upon applying each oxidative step, the j and I ACE increase instantaneously to the highest value but Figure 9.11 Chronoamperometry experiment during electrooxidation of IPA with time-resolved product then they gradually decrease with time analysis. Potential, current density, and ion current of and after 5 min only a part of the initial ACE are plotted vs. time. The potential protocol formation rate is maintained. A part of comprises of oxidation of IPA from +0.1 to +1.0 VRHE the decline in j is attributed to the for 5 min followed by a reductive potential 0 V RHE for 2 min to clean the surface of the electrode. Electrolyte: vanished charging current, however, the 0.1 M HClO 4 + 0.2 M IPA. Reprinted with permission decay in I ACE is merely due to the from 217 . Copyright (2020) American Chemical deactivation of catalysts due to the slow Society. kinetics of desorption of ACE ads . Such results can be followed only with time-resolved experiments since the time resolution of the low-resolution analytical techniques cannot capture such a profile of product formation rate. The results of the experiment are consistent with the sweep experiment that all PtRu NPs demonstrate lower onset potential compared to Pt/C i.e. all PtRu catalysts start to produce ACE from 0.1 V RHE while on platinum this value is 0.4 V RHE . All the catalysts show deactivation due to the accumulation of the ACE ads . However, by looking closer platinum compared to PtRu undergoes a steeper decline in both j and I ACE in the mid-range potentials. This tells that the presence of ruthenium facilitates desorption of ACE ads upon oxidation since the same amount of platinum is present at the surface of PtRu compared to the Pt/C case. At higher potentials (˃0.8 V RHE ), the decline is more drastic for all catalysts which can be attributed to the formed oxide at the surface, and IPAOR is hindered on the surface of metal oxide.

90

9.3.5 Open-circuit potential (OCP) measurement Figure 9.12 displays the OCP measurement for all four catalysts in an argon-saturated

0.1 M HClO 4 + 0.2 M IPA, in the absence or presence of 0.2 M of ACE. Each point represents the average potential of triplicate repeated experiments always with a fresh catalyst dropcasted film and electrolyte (the standard deviations are presented by error bars). The OCP value was obtained only after 30 min when the potential was stable and the deviation was < 0.1 mV/s. The measured OCP in electrolyte free of ACE (red points) for Pt/C was +0.28 ± 0.03 V RHE , while all three PtRu/C were yielding Figure 9.12 Open circuit potential (OCP) measurement the same OCP in the experimental of different Pt (1-x) Ru x/C and Pt/C nanoparticles in argon purged solution of 0.2 M IPA+0.1 M HClO without error range of ca. +0.05 ± 0.01 V . 4 RHE (red points) and with 0.2 M ACE (blue points). The The latter implies that they are all standard potential calculated from thermodynamics is about equally active towards the shown as a horizontal with the dashed line. Reprinted 217 activation of IPA. According to blue with permission from . Copyright (2020) American Chemical Society. points, the addition of ACE causes an evident positive shift in the OCP value for all four catalysts, namely, the OCP for Pt/C and different ratios of PtRu/C are increased to +0.38 ± 0.01 V RHE and +0.10 ± 0.01 V RHE respectively. The latter indicates that ACE reduction participates in the value of the OCP. Again all blue points for PtRu/C are clustered together and the difference is in the experimental error range which indicates equal activity towards IPAOR. The OCP of all blue points is very close predicted by the which is shown as a horizontal dashed line. According to the half-reaction (see reaction 9.1) of conversion of IPA to ACE:

CH COCH + 2H + + 2e - CH CHOHCH (9.1) 3 3 ⇄ 3 3

The standard Gibbs free energy of the IPAOR can be calculated from of Gibbs free ○ -1 ○ -1 energy of the formation of IPA (ΔGf =-180.4 kJ mol ) and ACE (ΔGf = -155.4 kJ mol ) equal to -25.0 kJ mol -1 (see equation 9.2).

○ ○ ○ -1 ΔGCH 3COCH 3/CH 3CHOHCH 3=ΔGf, CH 3CHOHCH 3-ΔGf, CH 3COCH 3 = -25.0 kJ mol (9.2)

Consequently, the standard potential of the reaction can be calculated as the following:

○ -1 ○ ΔG -25.0 kJ mol ECH COCH /CH CHOHCH =- =- = +0.130 V vs SHE (9.3) 3 3 3 3 zF 2∙96,485 A s mol -1

91

The calculated standard potential for the ACE/IPA redox couple is equal to the equilibrium potential of +0.13 V RHE , when the activity of both IPA and ACE is unity. This means the addition of ruthenium, regardless of the ratio decreases the OCP value close to the thermodynamic value of IPA oxidation to ACE. The bias between the standard potential and measured points can be related to the deviation from the standard conditions. These measurements are another evidence of a higher activity of IPAOR on the PtRu/C catalysts compared to platinum is reflected by lower OCP (in line with the lower onset potential during sweep experiment). This difference is in good agreement with the OCV measurements during the IPA fuel cell test, namely, measured OCVs for PtRu/C or Pt/C as the anode, are 802 mV, and 670 mV respectively 202 . This supports the better performance of PtRu/C.

9.3.6 Dissolution of PtRu bimetallic alloys during sweep voltammetry 1 Aging or corrosion of the electrode is among many different mechanisms of failure of PEM-FCs caused by degradations of the membrane electrode assembly. By considering the voltammetry of the oxygen reduction reaction on Pt/C and IPA oxidation on Pt (1- x) Ru x/C as the cathodic and anodic reactions in an IPA-air fuel cell, it is expected that the potential of the anode of the IPA-PEM-FC is lower than +0.4 V RHE under the steady operation 224 . In this potential range, platinum and ruthenium are stable. However, in the time interval between shut-down and start-up, the OCV of the FC rises to the onset of the oxygen reduction reaction as soon as O 2 enters the anode compartment of the fuel cell explained by the mixed potential theory in corrosion when the anode may start to dissolve. Therefore, the metallic constitutes of Figure 9.13 Study of stability of different Pt/C Pt/C and Pt (1-x) Ru x/C were subjected to the and Pt (1-x) Ru x/C nanoparticle during cyclic dissolution test. voltammetry with a scan rate of 10 mV s -1 in Figure 9.13 displays the quantitative 0.05 M IPA + 0.1 M HClO 4 electrolyte. The dissolution of platinum and ruthenium corresponding smoothened platinum and ruthenium dissolution rate profiles are recorded recorded by real time IPC-MS in parallel to simultaneously. Reprinted with permission cyclic voltammetry. The CV was performed from 217 . Copyright (2020) American Chemical with a scan rate of 10 mV s -1 in the potential Society. range of 0 to 1.0 V RHE in the electrolyte of 0.05 M IPA + 0.1 M HClO 4. In all cases, the shape of the dissolution curves for both platinum and ruthenium (onsets and the position of the peaks) are approximately the same. Moreover, no major difference between the absolute amount of leached metal for all examined catalysts is observed except for Pt 0.33 Ru 0.67 /C which

1 The dissolution studies were conducted by Dr. Florian Speck from Electrocatalysis unit of HI-ERN. 92 shows a higher dissolution for both platinum and ruthenium. Apart from more ruthenium dissolution for Pt 0.33 Ru 0.67 /C which has the highest atomic ratio of ruthenium among all the catalysts, platinum corrodes slightly more too which may be attributed to the synergic effect of ruthenium instability. For all the examined catalysts, the dissolution rate of ruthenium is more pronounced compared to platinum. In addition, ruthenium dissolves in both anodic and cathodic and the latter is dominating. Specifically, during the positive-going scan, ruthenium starts to dissolve at +0.89 VRHE , which is in line with the onset of oxidation ruthenium at +0.74 VRHE predicted by thermodynamics 225 . Ruthenium dissolution increases in the backward scan, as the formed oxide on the surface, get reduced. Platinum begins to dissolve in the forward scan close to the vertex potential of the experiment at 0.99 V RHE which is in agreement with the onset of oxidation +0.98 VRHE from thermodynamics 225 . Anodic and cathodic peaks are not deconvoluted in this case evidenced by the single peak dissolution profile of platinum. This is because of the low UPL of the experiment which is close to the onset potential of oxidation of platinum.

9.3.7 Dissolution by changing the upper potential limit (UPL) The effect of UPL was studied by a series of sweep experiments on different

Pt/C and Pt (1-x) Ru x/C NPs in the presence of IPA. Figure 9.14 displays the total amount of dissolved platinum and ruthenium during CV in the potential range from 0 V RHE to with the different upper potential limits i.e. to 1.0, 1.2, and -1 1.4 V RHE with the scan 10 mV s in the 0.05 M IPA + 0.1 M HClO 4 electrolyte. Each data point is calculated by integrating the dissolution rate over time within a single CV. It can be seen that the dissolution of both platinum and ruthenium scale exponentially when the Figure 9.14 The integrated amount of dissolved UPL spans to higher potentials. The latter platinum and ruthenium during single cyclic is associated with the formation of more voltammetry on different Pt (1-x) Ru x/C and Pt/C oxide and subsequently reduction of the nanoparticles. The sweep experiment is from 0 VRHE to different upper potential limits (UPLs). formed oxide in the anodic and cathodic -1 Scan rate: 10 mV s . Electrolyte: 0.1 M HClO 4 + scans respectively. A more dissolution of 0.05 M IPA. Reprinted with permission from 217 . ruthenium when the atomic ratio of Copyright (2020) American Chemical Society. ruthenium in the higher in NPs is associated with the fact that the amount of the platinum as the active catalyst for IPAOR was always kept constant thus the absolute amount of ruthenium varies (e.g. at 1.4 V RHE, Pt 0.33 Ru 0.67 /C has the highest dissolution of ruthenium compared to other ratios while platinum corrosion is almost the same). The dissolution of platinum is different only at UPL=1.0 V RHE where the dissolution is negligible and the synergic effect of ruthenium is observed, namely, dissolution of platinum scales with ruthenium content. 93

9.3.8 Impact of IPA on metal dissolution In order to inspect whether the sorption of IPA or ACE has any effect on the corrosion or not, the sweep experiment was conducted in supporting electrolyte free of IPA (solid curves) and compared to the case when IPAOR takes place (dashed curves). For this experiment,

Pt 0.5 Ru 0.5 /C is chosen since it was the most promising catalyst based on the previously described sweep experiment, chronoamperometry measurement with EC-RTMS. Figure 9.15 shows the Figure 9.15 Comparison of dissolution profiles for dissolution profiles for platinum and platinum and ruthenium on Pt 0.5 Ru 0.5 /C during ruthenium during CV with a scan rate of 10 cyclic voltammetry from 0.0 to +1.0 V RHE in the mV s -1 obtained in the presence (0.05 M) presence (solid curves) and absence of IPA (dashed curves). Scan rate: 10 mV s -1. Supporting and absence of IPA in a 0.1 M HClO 4 electrolyte: 0.1 M HClO 4. Reprinted with supporting electrolyte. In the absence of permission from 217 . Copyright (2020) American IPA in supporting electrolyte, corrosion is Chemical Society. mainly attributed to the formation of metal oxide (anodic dissolution) and reduction of oxide

(cathodic dissolution). Surprisingly in the potential range of 0-1.0 V RHE , a lower dissolution rate (both anodic and cathodic) is witnessed by the addition of IPA. In another word, the addition of IPA to the electrolyte decreases substantially the dissolution by a factor of ca. 2 times and 6 times for ruthenium and platinum respectively. This inhibition can be ascribed to the suppressed formation of oxide due to the presence of the ACE ads and thereafter less reduction of oxide. Nevertheless, peak positions are maintained and no significant variation in the onset potentials is observed. These results are contrary to the dissolution of metals during the MeOH oxidation reaction, in which the presence of MeOH boosts the dissolution 226 . This might be Figure 9.16 Dissolution profile of platinum and correlated to the formation of CO ads during ruthenium of Pt 0.5 Ru 0.5 /C catalysts recorded MeOH oxidation that it favors the simultaneously during the accelerated stress test. 227 dissolution . FT-IR experiment revealed The potential program consists of 1000 cycles from that during IPAOR selective formation of 0 V RHE to different upper potential limits of 0.6 VRHE (black), 0.8 V RHE (red) and 1.0 V RHE (blue); ACE ads without accumulation of CO ads on -1 126 Scan rate of 50 mV s ; Electrolyte: 0.1 M HClO 4 + the surface is the reason for deactivation . 0.05 M IPA. Reprinted with permission from 217 . Therefore, the bulky structure of ACE and Copyright (2020) American Chemical Society. 94 the weaker interaction with the surface decreases the availability of surface sites for oxide adsorption it can explain indeed the lower dissolution of platinum and ruthenium.

9.3.9 Accelerated stress test (AST) To examine the durability of

Pt 0.5 Ru 0.5 /C as the best performing the AST with different conditions were carried out. This is necessary for a deeper understanding of the long term stability of the catalyst under simulated fuel cell operating conditions. Figure 9.16 displays the dissolution profile of platinum and ruthenium during speeded up degradation recorded with ICP-MS in parallel. The experimental protocol includes 1000 cycles with a scan rate of 0.5 V s -1 from 0.0 to different UPLs, namely 0.6 (black), 0.8 (red) and 1.0 V RHE (blue) . Each test was carried out on a fresh catalyst spot. Figure 9.17 Combinatorial real time monitoring Since all the experiments were of ACE and CO 2 as the reaction products of IPA performed with a very fast scan rate the time oxidation reaction via EC-RTMS and dissolution of the platinum and ruthenium during AST resolution of the ICP-MS was not meeting analyzed by time-resolved ICP-MS. The the rate of the electrochemical experiment. potential program consists of 1000 cycles from 0 -1 Therefore, the features overlap and could VRHE to +1.0 V RHE ; Scan rate of 50 mV s . not be resolved, and therefore, the observed Electrolyte for corrosion study: 0.1 M HClO 4 + 0.05 M IPA. Electrolyte for product analysis: 0.1 signal is the average of dissolution in M HClO 4 + 0.2 M IPA. Reprinted with positive- and negative-going scans. permission from 217 . Copyright (2020) At UPL=0.6 V RHE neither platinum American Chemical Society. nor ruthenium corrodes. By increasing the UPL to 0.8 V RHE , ruthenium begins to dissolve while platinum still stays stable. The total amount of dissolved ruthenium by integrating the dissolution profile over the 1000 cycles calculated was 2.3 ng. By expanding the UPL even more to 1.0 V RHE, platinum starts to dissolve slightly but the corrosion rate of ruthenium increases drastically. During 1000 cycles, The absolute amount of dissolved platinum and ruthenium are calculated by 21.0 ng and 0.2 ng respectively which shows that the ratio of the dissolved ruthenium much by a factor of 100 higher than platinum. All these results are in good agreement with the onset of the dissolution during the sweep experiment. In all cases, dissolution profiles decay with the number of cycles which is the sign of the stripping of the atoms from the surface of the NPs. The latter can be seen clearly in the case of ruthenium at

UPL=1.0 V RHE , however, the signal of ruthenium did not reach the background which indicates that it is remained either on the surface or in the bulk. Based on the calculation, 21.0 ng corresponds to 5% of the whole ruthenium content in the printed film of nanoparticles is dissolved in this experiment. The AST test in the potential range of 0.0 to +1.0 V RHE was 95

repeated with EC-RTMS and the ion currents for ACE and CO 2 were recorded. These results were combined and synchronized with dissolution profiles of platinum and ruthenium and are shown in Figure 9.17. During fast potential cycles over time, the evolution of I ACE and I CO2 occurs in short time scale is almost looking like a constant signal, since the time resolution of EC-RTMS is not sufficient to track the fast changes in the product distribution, moreover, fast switching of the potential leads to a clean surface of the catalyst which is favorable for a steady oxidation conversion of the IPA to ACE and no decay of the signal is observed. The dynamic oxidation of adsorbents residues on the surface during IPA oxidation to

ACE yields a relatively constant amount of CO 2. Here, the origin of the signal of CO 2 could not be attributed solely to oxidation of IPA but also corrosion of carbon support based on the results in section 9.3.3. Over a longer period, a slight increase in both I ACE and I CO2 is witnessed. Based on the results of the next section (9.3.10), the relatively selective dissolution of ruthenium causes a higher rate of oxidation of IPA to ACE, and consequently, more CO 2 is yielded since more oxidation may leave more adsorbates on the surface.

9.3.10 Performance of the catalyst after AST

Linear sweep voltammetry during IPAOR was carried out on the Pt 0.5 Ru 0.5 /C after AST with the same protocol i.e. scan rate of 5 mv s -1 and MSLSVs for ACE and

CO 2 were recorded (yellow curves in Figure 9.18). The acquired results were compared with Pt 0.5 Ru 0.5 /C before AST (red curves) and Pt/C (black curves).

Again a perfect match between j and IACE is witnessed for all catalysts, therefore the shape of the LSV can be attributed exclusively to the conversion of IPA to

ACE. The disappearance of the P 0.15V after AST is correlated to the relatively selective depletion of ruthenium and enrichment of platinum on the surface of

Pt 0.5 Ru 0.5 /C. However, the P 0.75V which is attributed mostly to the Pt-Pt ensembles is changed unexpectedly. By looking at yellow curves, j and I ACE both are enhanced significantly compared and the position of the peak is shifted positively to the other two catalysts, and compared to platinum, while the onset potential shifted Figure 9.18 Linear sweep voltammetry on Pt/C negatively that may be associated to (black), Pt 0.5 Ru 0.5 /C before (red) and after AST remaining ruthenium on the surface. (yellow) with the scan rate of 5 mv s -1. The voltammograms, simultaneously recorded ion Moreover, the intensity and onset of ICO2 currents for ACE and CO are shown in three are the same for catalysts in the error 2 different panels. Electrolyte: 0.1 M HClO 4 + 0.2 M 217 range. The behavior of P 0.75V can be IPA. Reprinted with permission from . Copyright explained by two hypotheses. The first is (2020) American Chemical Society. 96 that the drastic stripping of ruthenium during AST may have enhanced surface roughness and made the catalyst more porosity from the surface. However, this fails as higher exposed platinum should yield a higher amount of CO 2 which is not the case although a higher porosity can explain ACE signal enhancement. The second hypothesis is about low ruthenium content which is desirable for the second peak for the conversion of IPA to ACE. Based on this hypothesis, I CO2 can be explained, since, the ruthenium amount in the Pt 0.5 Ru 0.5 /C after AST is between Pt 0.5 Ru 0.5 /C before the AST and Pt/C, therefore the I CO2 should resemble.

9.3.12 Quantified product analysis during steady-state oxidation of IPA As it is discussed and demonstrated several times before, the analogy between mass spectrometry signal of ACE from EC-RTMS and current density of various potentiodynamic/ potentiostatic experiments indicate that the IPA oxidation has a very high degree of selectivity to ACE on all examined catalysts.

Here, this matter is investigated quantitatively by analysis of ACE and CO 2 with offline analytics after performing IPA oxidation on the best performing catalyst, namely, on

Pt 0.5 Ru 0.5 /C. Based on the mass signal of CO 2 in Figure 9.8, two exemplary oxidation potential 0.6 V RHE (close to the onset potential of CO 2 formation) and 0.9 V RHE (CO 2 forms relatively high) were chosen for this study. Moreover, according to the step experiments in

Figure 9.11 both j and I ACE decay very fast in all cases. Therefore, the classical protocol of applying a single oxidation step does not yield enough product to be detectable with offline analytics. Therefore a procedure of reactivation was applied to regenerate the poisoned catalyst after some time. A part of the employed protocol is shown in Figure 9.19 where IPA gets oxidized to ACE either 0.6 or 0.9 V RHE for 600 s and as the reaction proceeds with time the current density decays due to accumulation of the ACE ads . A short reductive potential is applied (0 V RHE for 30 s) in between to remove the ACE ads and reactivate the electrode for the next oxidative step. This protocol of oxidation/reduction was repeated 12 times to have 120 min of oxidation steps in total. For each potential (0.6 and 0.9 V RHE ) the experiment was repeated three times and the charge during oxidation was calculated by integrating the current versus time for all 10 min oxidation steps and then the overall charge considered for calculation of the faradaic efficiencies. At the end of the experiment, an aliquot of the liquid product i.e. dissolved ACE was collected from the WE compartment and analyzed with HS-GC-MS, while formed carbonate from the alkaline trap was analyzed with IC. Analysis of CO 2 as a gaseous product was not done with typical online GC, due to the transient formation of products, and dependence on the determined amount to the time of injection. The results of product analysis show that the amount of CO 2 was negligible to be detectable with IC and the amount of analyzed ACE based on faraday law is close to 100% faradaic efficiency for two examined potentials, therefore, it can be concluded that the IPA oxidation on PtRu is selective to ACE. 97

Figure 9.19 The employed potential protocol for quantification of products during steady-state oxidation of IPA and current density profiles versus time. The protocol involves the repetitive oxidative potential for 600 s (here 0.6 V RHE is used) to oxidize IPA, alternated by a 30 s reductive potential (0 V RHE ) to clean the surface from adsorbed species accumulated during the preceding oxidation step. The charge passed during the oxidative step is shown by the black integrated area only 217 during oxidation steps. Electrolyte: 0.1 M HClO 4 + 0.2 M IPA. Adapted with permission from . Copyright (2020) American Chemical Society.

9.3.11 Detection of ACE during IPA oxidation on NPs with EC-RTMS Based on the results of the previous section, ACE is the selective product of the reaction even in higher potential ranges. By considering this fact, the I ACE in the examined mass range has to have a linear relationship with the current density in electrochemistry. The latter ensures the correct comparison activity of the catalyst for IPA to ACE on the different NPs. As it is discussed in Figure 9.20 The mass spectrum of DART-TOF-MS by Chapter 4, the MCP detector linearity controlling the potential of the platinum electrode at 0.0 range is limited. Therefore, in the VRHE (no oxidation of IPA takes place) with the flow of studies where a large amount of ions electrolyte of 0.2 M IPA + 0.1 M HClO4. Instrumental from the liquid product is formed, this conditions are provided in chapter 5. matter should be checked. The detection of ACE in the m/z=59-59.1 on the Pt/C and PtRu/C 98 is not correct for three reasons. First and foremost, the amount of produced ACE during IPA oxidation is significantly higher compared to the planar platinum electrode due to the larger surface area (the current density is at least 10 times higher). Secondly, the mechanism of DART ionization in positive mode is mostly PTR. By comparing the proton affinity of ACE (812 kJ mol -1) and water (691±3 kJ mol -1), a large equilibrium constant of the protonation reaction is expected which makes the DART- TOF-MS a very sensitive detector for the determination of ACE in the protonated form. Last but not least, although in DART the extent of the fragmentation is relatively low when a large amount of an ionizable compound present in the ion source the intensity of the fragment ion can have a significant contribution in other mass to charge ratios. Here, in the case of IPA oxidation, the concentration of IPA dissolved in the electrolyte is in excess (0.2 M) and most of it does not get oxidized during electrochemistry so it enters the ion source. Based on the EI spectrum at 70 eV (hard ionization) from the NIST database 81 , IPA breaks into many fragments including m/z= 59.0-59.1 which is the mass of protonated ACE (note that the EI fragmentation pattern Figure 9.21 Representation of the overload of is valid qualitatively to some extent in DART the detector of TOF-MS for detection of ACE ion source too). during linear sweep voltammetry on Pt/C (red), Figure 9.20 shows the mass spectrum Pt 0.5 Ru 0.5 /C before (blue) and after AST (yellow). The voltammograms and during IPA oxidation on the surface of simultaneously recorded ion currents for ACE Pt 0.5 Ru 0.5 /C at 0.0 V RHE potential when no in overloaded and linear mass ranges. Scan -1 ACE is formed. Besides the main peak of the rate of 5 mv s . Electrolyte: 0.1 M HClO 4 + 0.2 IPA, at m/z=61.05 which is attributed to M IPA. protonated IPA, a significant peak at m/z=59.05 is associated fully with the fragment of IPA from the electrolyte. Therefore we can conclude that m/z=59.05 overlaps with protonated ACE in the same m/z range and it increases the background when no ACE is formed and this limits the linearity range of the selected mass range. Figure 9.21 compares I ACE during the before described AST in two mass ranges, namely m/z=60-60.1 and m/z=59-59.1 and it clarifies better the overload of MCP. The deviation from the linear range is evident in the yellow curve when the current density does not scale with the I ACE in the m/z=59-59.1. Therefore the m/z=60-60.1 was selected as the desired range for detection in this case, where the ion structure is the protonated ACE when one of the carbons is 13 C.

99

9.4 Conclusions The elucidation of some aspects of the mechanism of electrooxidation of IPA on the surface of platinum and PtRu as for an IPA fuel cell in acidic conditions was demonstrated. The main conclusions are: (i) The combination of EC-RTMS and EC-IRRAS on extended planar platinum

electrodes revealed that ACE is the major product, and CO 2 is formed only in traces. (ii) EC-IRRAS confirms the presence of ACE not only in the bulk solution but also on the

electrode surface as well as it shows that CO 2 is formed only at higher potential ranges. (iii) The suppression of the rate of ACE formation during step experiments was associated

with the poisoning of the catalyst by the gradual accumulation of ACE ads on the surface rather than CO ads . (iv) Deactivation of the catalyst was studied further during a chronoamperometry measurement with EC-RTMS. The results indicate that there is a strong relationship between the rate of formation of ACE and the surface coverage of adsorbates e.g.

ACE ads on platinum. (v) The ACE ads can be removed from the surface either by exposure to high potentials in the oxide region or to potentials at the Hupd region. (vi) The activity, selectivity, and stability of the IPA oxidation reaction were investigated

on commercially available Pt/C and Pt (1-x) Ru x/C catalysts which were characterized by time-resolved product and dissolution analysis. PA is oxidized selectively to ACE on

all NPs regardless of the composition of the catalyst and only traces of CO 2 were detected in higher potential regions relevant to the IPA-PEM-FC operation evidenced by EC-RTMS. (vii) The appearance of a lower oxidation peak on PtRu was decreasing the overpotential of the IPA oxidation in the relevant operation potential range of the IPA-PEM-FC which would increase the power output of the fuel cell, however, in a narrow potential range and the activity drops may due to formation of ruthenium oxide and a deeper understanding in situ surface characterization techniques is required.

(viii) The origin of formed CO 2 was attributed partially to corrosion of the carbon support and mainly to the oxidation of strongly adsorbed species during IPA oxidation. Quantification of ACE during steady-state oxidation of IPA confirmed that the amount

of CO 2 was negligible, i.e. below the detection limit of the employed analytical technique. (ix) Ruthenium is an inactive catalyst for IPAOR. (x) The reaction is accompanied by the accumulation of ACE which decreases the availability of active sites and thus suppresses the IPA oxidation for all NPs. (xi) Platinum is more prone to ACE poisoning than the PtRu catalysts. (xii) OCP measurements on all NPs demonstrated that the addition of ruthenium decreases the overpotential of IPAOR to the potential close to predicted thermodynamic value. (xiii) Even though PtRu catalysts outperform platinum from activity measurements, the corrosion study showed that inferior stability of ruthenium compared to platinum especially during the long-term run of the fuel cell. (xiv) The presence of IPA in electrolyte decreases the dissolution rate. 100

(xv) The combination of product analysis and dissolution studies during and after the

accelerated stress test demonstrates that activity and stability of Pt 0.5 Ru 0.5 /C do not come along and the depletion of ruthenium from the surface does seem to alter the

onset potential as well as a substantial change in peak intensity of both j and I ACE .

101

10 Summary and outlook

In the present work, electrochemical real-time mass spectrometry (EC-RTMS) was demonstrated for the continuous determination of liquid and gaseous products during electrochemical reactions shortly after formation. The offered temporal resolution (in the order of seconds) and the excellent sensitivity (in the ppb-range) enable the observation of transients in product formation under potentiodynamic/potentiostatic conditions. EC-RTMS pushes the boundaries of the analysis for the complex mixture of products in electrosynthesis, and it offers a tool to study the mechanism of the reactions. Moreover, it provides a broad variety of possibilities e.g. for the high throughput screening of activity/selectivity which can accelerate the development of robust, efficient, and selective materials in future studies under operando conditions. The methodology can cover the detection of a vast range of products independent of their vapor pressure or the presence of salts, which are not accessible with any of the existing methodologies. The capabilities of the presented method were demonstrated successfully for various electrochemical reactions, ranging from the CO 2 reduction reaction (CO 2RR) to the oxidation of alcohols. The presented methodology is a solid basis and framework for further investigations and optimizations. There is a large room for improvements such as the time response of the technique and the decrease of the broadening effect of signals in both mass spectrometers by modification of the inlet system. EC-RTMS does not describe necessarily the combination of two as-presented mass spectrometry techniques; according to the conditions, the mass spectrometers, especially ion sources can be different. Instead, EC-RTMS describes the concept of using mass spectrometry techniques for the characterization of liquid and gaseous products simultaneously. This piece of work provides the platform to exploit not only aqueous electrochemistry but also non-aqueous electrochemistry, potentially including battery research. The application can be extended to electrosynthesis with organic reactions where the ionization mechanism of liquid products can be different and hence it needs further thorough study. The proof of concept for semi-quantitative comparison of intensities of the same compound for different surfaces is showed successfully. Full quantification of products is not trivial and it should be the subject of further studies. Coupling other electrochemical cells under well-defined mass transport conditions (like a rotating disk electrode) to the EC-RTMS can create further opportunities for performing product analysis under well-controlled conditions and with well-defined surfaces e.g. single-crystal electrodes. The scope of applicability is not limited to classical electrochemical cell configurations, but also can be extended to the analysis of the products in reactors, e.g. fuel cells. SFC-EC-RTMS can be part of a holistic approach to exploit the capability of a quick assessment of the potential-dependent product distribution on different electrocatalytic materials. Once the best performing material is found, study the reaction can go further with long term electrolysis complemented with quantitative product analysis to study the kinetic parameters of the reaction under steady-state conditions. The optimization of mass transport, cost and economical assessment for designing a reactor is the last stage that needs information from two previous steps. 102

Last but not least, electrochemical engineering is not well implemented in the chemical industry for the production of sustainable, renewable chemicals. The main reason is the financial barrier to outcompete the efficient and well-established catalytic processes, and, also it may be associated with the level of maturity of catalytic synthesis compared to electrosynthesis. It is hoped that this field of research becomes more attractive especially by financial support from sustainable energy policy. Apart from the development in the direction of electrochemistry at ambient conditions, high pressure and temperature electrochemistry can open new directions in heterogeneous catalysis by adding the electrochemical potential as another control parameter, additional to temperature and pressure.

103

Bibliography

1. Van Geem, K. M.; Galvita, V. V.; Marin, G. B., Making chemicals with electricity. Science 2019, 364 (6442), 734-735.

2. Whitesides, G. M.; Crabtree, G. W., Don't forget long-term fundamental research in energy. Science 2007, 315 (5813), 796-8.

3. The Energy of the Future, Fourth “Energy Transition” Monitoring Report – Summary ; The Federal Ministry for Economic Affairs and Energy (BMWi): 2015.

4. Xu, Y.; Ramanathan, V.; Victor, D. G., Global warming will happen faster than we think. Nature 2018, 564 (7734), 30-32.

5. Lewis, N. S.; Nocera, D. G., Powering the planet: Chemical challenges in solar energy utilization. Proc. Natl. Acad. Sci. 2006, 103 (43), 15729-15735.

6. Bardi, U., Peak oil: The four stages of a new idea. Energy 2009, 34 (3), 323-326.

7. Schüth, F., Chemical compounds for energy storage. Chem. Ing. Tech. 2011, 83 (11), 1984-1993.

8. Yan, R.; Saha, T. K.; Bai, F.; Gu, H., The of the 2016 South Australia blackout: a catastrophic event in a high renewable network. IEEE Trans. Power Syst. 2018, 33 (5), 5374-5388.

9. Top 10 Emerging Technologies 2019 ; The World Economic Forum: 2019.

10. Agora-energiewende, Power Generation and Consumption Germany 2018. www.agora-energiewende.de 2019 .

11. De Luna, P.; Quintero-Bermudez, R.; Dinh, C.-T.; Ross, M. B.; Bushuyev, O. S.; Todorović, P.; Regier, T.; Kelley, S. O.; Yang, P.; Sargent, E. H., Catalyst electro- redeposition controls morphology and for selective carbon dioxide reduction. Nat. Catal. 2018, 1 (2), 103.

12. Evans, A.; Strezov, V.; Evans, T. J., Assessment of utility energy storage options for increased renewable energy penetration. Renew. Sust. Energ. Rev. 2012, 16 (6), 4141-4147. 104

13. Guittet, M.; Capezzali, M.; Gaudard, L.; Romerio, F.; Vuille, F.; Avellan, F., Study of the drivers and asset management of pumped-storage power plants historical and geographical perspective. Energy 2016, 111 , 560-579.

14. Schlögl, E. b. R., Chemical Energy Storage . Walter de Gruyter GmbH: 2013.

15. Eberle, U.; Felderhoff, M.; Schueth, F., Chemical and physical solutions for hydrogen storage. Angew. Chem. Int. Ed. 2009, 48 (36), 6608-6630.

16. Michaelis, J.; Plötz, P.; Gnann, T.; Wietschel, M., 3. Vergleich alternativer Antriebstechnologien Batterie-, Plug-in Hybrid-und Brennstoffzellenfahrzeug. ALTERNATIVE 2014 , 51.

17. Markiewicz, M.; Zhang, Y.; Bösmann, A.; Brückner, N.; Thöming, J.; Wasserscheid, P.; Stolte, S., Environmental and health impact assessment of Liquid Organic Hydrogen Carrier (LOHC) systems–challenges and preliminary results. Energy Environ. Sci. 2015, 8 (3), 1035-1045.

18. Benasla, M.; Hess, D.; Allaoui, T.; Brahami, M.; Denaï, M., The transition towards a sustainable energy system in Europe: What role can North Africa's solar resources play? Energy Strategy Rev. 2019, 24 , 1-13.

19. Knies, G., Clean Power from Deserts: The DESERTEC Concept for Energy, Water and Climate Security . Protext-Verlag: 2008.

20. Benasla, M.; Allaoui, T.; Brahami, M.; Denai, M.; Sood, V. K., HVDC links between North Africa and Europe: Impacts and benefits on the dynamic performance of the European system. Renew. Sust. Energ. Rev. 2018, 82 , 3981-3991.

21. Harris, N., . Edtech Press: 2019.

22. Urmee, T.; Harries, D.; Schlapfer, A., Issues related to rural electrification using renewable energy in developing countries of Asia and Pacific. Renew. Energy 2009, 34 (2), 354-357.

23. Möhle, S.; Zirbes, M.; Rodrigo, E.; Gieshoff, T.; Wiebe, A.; Waldvogel, S. R., Modern Electrochemical Aspects for the Synthesis of Value Added Organic Products. Angew. - Chem. Int. Ed. 2018, 57 (21), 6018-6041. 105

24. Wiebe, A.; Gieshoff, T.; Möhle, S.; Rodrigo, E.; Zirbes, M.; Waldvogel, S. R., Electrifying organic synthesis. Angew. Chem. Int. Ed. 2018, 57 (20), 5594-5619.

25. Foit, S. R.; Vinke, I. C.; de Haart, L. G.; Eichel, R. A., Power to Syngas: An - - Enabling Technology for the Transition of the Energy System? Angew. Chem. Int. Ed. 2017, 56 (20), 5402-5411.

26. Kwon, Y.; Schouten, K. J. P.; van der Waal, J. C.; de Jong, E.; Koper, M. T., Electrocatalytic conversion of furanic compounds. ACS Catal. 2016, 6 (10), 6704-6717.

27. Sherbo, R. S.; Delima, R. S.; Chiykowski, V. A.; MacLeod, B. P.; Berlinguette, C. P., Complete electron economy by pairing electrolysis with hydrogenation. Nat. Catal. 2018, 1 (7), 501.

28. Dinh, C.-T.; Burdyny, T.; Kibria, M. G.; Seifitokaldani, A.; Gabardo, C. M.; de Arquer, F. P. G.; Kiani, A.; Edwards, J. P.; De Luna, P.; Bushuyev, O. S., CO2 electroreduction to ethylene via hydroxide-mediated copper catalysis at an abrupt interface. Science 2018, 360 (6390), 783-787.

29. Simões, M.; Baranton, S.; Coutanceau, C., Electrochemical valorisation of glycerol. ChemSusChem 2012, 5 (11), 2106-2124.

30. Nitopi, S.; Bertheussen, E.; Scott, S. B.; Liu, X.; Engstfeld, A. K.; Horch, S.; Seger, B.; Stephens, I. E.; Chan, K.; Hahn, C., Progress and Perspectives of Electrochemical CO2 Reduction on Copper in Aqueous Electrolyte. Chem. Rev. 2019, 119 (12), 7610-7672.

31. Li, X.; Anderson, P.; Jhong, H.-R. M.; Paster, M.; Stubbins, J. F.; Kenis, P. J., Greenhouse gas emissions, energy efficiency, and cost of synthetic fuel production using electrochemical CO2 conversion and the Fischer–Tropsch process. Energy Fuels 2016, 30 (7), 5980-5989.

32. Nilges, P.; Schröder, U., Electrochemistry for biofuel generation: production of furans by electrocatalytic hydrogenation of furfurals. Energy Environ. Sci. 2013, 6 (10), 2925-2931.

33. Ma, X.; Jiang, C.; Xu, H.; Richardson, S. In-cylinder optical study on combustion of DMF and DMF fuel blends ; 0148-7191; SAE Technical Paper: 2012.

34. Gray, E. M.; Webb, C.; Andrews, J.; Shabani, B.; Tsai, P.; Chan, S., Hydrogen storage for off-grid power supply. Int. J. Hydrog. Energy 2011, 36 (1), 654-663. 106

35. Horn, E. J.; Rosen, B. R.; Baran, P. S., Synthetic organic electrochemistry: an enabling and innately sustainable method. ACS Cent. Sci. 2016, 2 (5), 302-308.

36. Faraday, M., VI. Experimental researches in electricity.-Seventh Series. Philos. Trans. R. Soc. Lond. 1834, (124), 77-122.

37. Kolbe, H., Zersetzung der Valeriansäure durch den elektrischen Strom. Annal. Chem. Pharm. 1848, 64 , 339-341.

38. Cardoso, D. S.; Šljukić, B.; Santos, D. M.; Sequeira, C. s. A., Organic electrosynthesis: from laboratorial practice to industrial applications. Org. Process Res. Dev. 2017, 21 (9), 1213-1226.

39. Winiwarter, A.; Silvioli, L.; Scott, S. B.; Enemark-Rasmussen, K.; Sariç, M.; Trimarco, D. B.; Vesborg, P. C.; Moses, P. G.; Stephens, I. E.; Seger, B., Towards an atomistic understanding of electrocatalytic partial hydrocarbon oxidation: propene on palladium. Energy Environ. Sci. 2019, 12 (3), 1055-1067.

40. Siahrostami, S.; Verdaguer-Casadevall, A.; Karamad, M.; Deiana, D.; Malacrida, P.; Wickman, B.; Escudero-Escribano, M.; Paoli, E. A.; Frydendal, R.; Hansen, T. W.,

Enabling direct H 2O2 production through rational electrocatalyst design. Nat. Mater. 2013, 12 (12), 1137-1143.

41. Lan, R.; Irvine, J. T.; Tao, S., Synthesis of ammonia directly from air and water at ambient temperature and pressure. Sci. Rep. 2013, 3, 1145.

42. Yang, S.; Verdaguer-Casadevall, A.; Arnarson, L.; Silvioli, L.; Čolić, V.; Frydendal, R.; Rossmeisl, J.; Chorkendorff, I.; Stephens, I. E., Toward the decentralized electrochemical production of H 2O2: a focus on the catalysis. ACS Catal. 2018, 8 (5), 4064- 4081.

43. Weinacht, T.; Pearson, B. J., Time-Resolved Spectroscopy: An Experimental Perspective . CRC Press: 2018.

44. Sasaki, K.; Marinkovic, N.; Isaacs, H. S.; Adzic, R. R., -based in situ characterization of carbon-supported platinum and platinum monolayer electrocatalysts. ACS Catal. 2016, 6 (1), 69-76. 107

45. Deng, Y.; Handoko, A. D.; Du, Y.; Xi, S.; Yeo, B. S., In situ of copper and copper oxide surfaces during electrochemical oxygen evolution reaction: identification of CuIII oxides as catalytically active species. ACS Catal. 2016, 6 (4), 2473- 2481.

46. Liang, Y.; McLaughlin, D.; Csoklich, C.; Schneider, O.; Bandarenka, A. S., The nature of active centers catalyzing oxygen electro-reduction at platinum surfaces in alkaline media. Energy Environ. Sci. 2019, 12 (1), 351-357.

47. Hodnik, N.; Dehm, G.; Mayrhofer, K. J., Importance and challenges of electrochemical in situ liquid cell electron microscopy for energy conversion research. Acc. Chem. Res. 2016, 49 (9), 2015-2022.

48. Topalov, A. A.; Katsounaros, I.; Auinger, M.; Cherevko, S.; Meier, J. C.; Klemm, S. O.; Mayrhofer, K. J., Dissolution of platinum: limits for the deployment of electrochemical energy conversion? Angew. Chem. Int. Ed. 2012, 51 (50), 12613-12615.

49. Kortlever, R.; Shen, J.; Schouten, K. J. P.; Calle-Vallejo, F.; Koper, M. T., Catalysts and reaction pathways for the electrochemical reduction of carbon dioxide. J. Phys. Chem. Lett. 2015, 6 (20), 4073-4082.

50. Raicheva, S.; Sokolova, E.; Christov, M., Electrochemical behaviour of oxygen- containing aliphatic compounds adsorbed on platinum in acid medium: Part I. Methanol. J. Electroanal. Chem. Interf. Electrochem. 1984, 175 (1-2), 167-181.

51. Kuhl, K. P.; Cave, E. R.; Abram, D. N.; Jaramillo, T. F., New insights into the electrochemical reduction of carbon dioxide on metallic copper surfaces. Energy Environ. Sci. 2012, 5 (5), 7050-7059.

52. Li, C. W.; Ciston, J.; Kanan, M. W., Electroreduction of carbon monoxide to liquid fuel on oxide-derived nanocrystalline copper. Nature 2014, 508 , 504-507.

53. Katsounaros, I.; Ipsakis, D.; Polatides, C.; Kyriacou, G., Efficient electrochemical reduction of nitrate to nitrogen on tin at very high cathodic potentials. Electrochim. Acta 2006, 52 , 1329-1338.

54. Barwe, S.; Weidner, J.; Cychy, S.; Morales, D. M.; Dieckhöfer, S.; Hiltrop, D.; Masa, J.; Muhler, M.; Schuhmann, W., Electrocatalytic oxidation of 5- 108

(hydroxymethyl)furfural using high-surface-area nickel boride. Angew. Chem. Int. Ed. 57 (35), 11460-11464.

55. Bertheussen, E.; Verdaguer Casadevall, A.; Ravasio, D.; Montoya, J. H.; Trimarco, - D. B.; Roy, C.; Meier, S.; Wendland, J.; Nørskov, J. K.; Stephens, I. E., Acetaldehyde as an intermediate in the electroreduction of carbon monoxide to ethanol on oxide derived copper. - Angew. Chem. Int. Ed. 2016, 55 (4), 1450-1454.

56. Verdaguer-Casadevall, A.; Li, C. W.; Johansson, T. P.; Scott, S. B.; McKeown, J. T.; Kumar, M.; Stephens, I. E.; Kanan, M. W.; Chorkendorff, I., Probing the active surface sites for CO reduction on oxide-derived copper electrocatalysts. J. Am. Chem. Soc. 2015, 137 (31), 9808-9811.

57. Kwon, Y.; Koper, M. T., Combining voltammetry with HPLC: application to electro- oxidation of glycerol. Anal. Chem. 2010, 82 (13), 5420-5424.

58. Santasalo-Aarnio, A.; Kwon, Y.; Ahlberg, E.; Kontturi, K.; Kallio, T.; Koper, M. T., Comparison of methanol, ethanol and iso-propanol oxidation on Pt and Pd electrodes in alkaline media studied by HPLC. Electrochem. Commun. 2011, 13 (5), 466-469.

59. Tong, Y. J., In situ electrochemical nuclear magnetic resonance spectroscopy for electrocatalysis: Challenges and prospects. Curr. Opin. Electrochem. 2017, 4 (1), 60-68.

60. Wang, H.; Löffler, T.; Baltruschat, H., Formation of intermediates during methanol oxidation: A quantitative DEMS study. J. Appl. Electrochem. 2001, 31 (7), 759-765.

61. Löffler, M.; Khanipour, P.; Kulyk, N.; Mayrhofer, K. J. J.; Katsounaros, I., Insights into Liquid Product Formation during Carbon Dioxide Reduction on Copper and Oxide- Derived Copper from Quantitative Real-Time Measurements. ACS Catal. 2020, 10 (12), 6735-6740.

62. Kasian, O.; Grote, J. P.; Geiger, S.; Cherevko, S.; Mayrhofer, K. J., The common intermediates of oxygen evolution and dissolution reactions during water electrolysis on iridium. Angew. Chem. Int. Ed. 2018, 57 (9), 2488-2491.

63. Baltruschat, H., Differential electrochemical mass spectrometry. J. Am. Soc. Mass Spectrom. 2004, 15 , 1693-1706. 109

64. Skoog, D. A.; Holler, F. J.; Crouch, S. R., Principles of instrumental analysis . Cengage learning: 2017.

65. Lee, J. K.; Kim, S.; Nam, H. G.; Zare, R. N., Microdroplet fusion mass spectrometry for fast reaction kinetics. Proc. Natl. Acad. Sci. 2015, 112 (13), 3898-3903.

66. Brown, T. A.; Chen, H.; Zare, R. N., Detection of the Short Lived Cation - Intermediate in the Electrooxidation of N, N Dimethylaniline by Mass Spectrometry. Angew. - Chem. Int. Ed. 2015, 54 (38), 11183-11185.

67. Dempster, A., A new method of positive ray analysis. Phys. Rev. 1918, 11 (4), 316.

68. Mayhew, A. M. E. C. A., Proton Transfer Reaction Mass Spectrometry, Principles and Applications . Wiley: 2014.

69. Field, F. H., Chemical ionization mass spectrometry. Acc. Chem. Res. 1968, 1 (2), 42- 49.

70. Edtbauer, A.; Hartungen, E.; Jordan, A.; Hanel, G.; Herbig, J.; Jürschik, S.; Lanza, M.; Breiev, K.; Märk, L.; Sulzer, P., Theory and practical examples of the quantification of

CH 4, CO, O 2, and CO 2 with an advanced proton-transfer-reaction/selective-reagent-ionization instrument (PTR/SRI-MS). Int. J. Mass Spectrom. 2014, 365 , 10-14.

71. Venter, A.; Nefliu, M.; Cooks, R. G., Ambient desorption ionization mass spectrometry. Trends Anal. Chem. 2008, 27 (4), 284-290.

72. Cody, R. B.; Laramée, J. A.; Durst, H. D., Versatile new ion source for the analysis of materials in open air under ambient conditions. Anal. Chem. 2005, 77 (8), 2297-2302.

73. Domin, M.; Cody, R., Ambient ionization mass spectrometry . Royal Society of Chemistry: 2014.

74. Harris, D. C.; Lucy, C. A., Quantitative Chemical Analysis . 9th ed.; W. H. Freeman and Company: 2016.

75. Hajslova, J.; Cajka, T.; Vaclavik, L., Challenging applications offered by direct analysis in real time (DART) in food-quality and safety analysis. Trends Anal. Chem. 2011, 30 (2), 204-218. 110

76. Cody, R. B.; Laramée, J. A.; Nilles, J. M.; Durst, H. D., Direct analysis in real time (DART) mass spectrometry. JEOL news 2005, 40 (1), 8-12.

77. Dane, A. J.; Cody, R. B., Selective ionization of in powdered milk by using argon direct analysis in real time (DART) mass spectrometry. Analyst 2010, 135 (4), 696-699.

78. Baldwin, K., Metastable helium: Atom optics with nano-grenades. Contemp. Phys. 2005, 46 (2), 105-120.

79. NIST Chemistry Web Book, http://webbook.nist.gov/chemistry/ . http://webbook.nist.gov/chemistry/ .

80. Dong, Y., Direct Analysis in Real Time Mass Spectrometry: Principles and Practices of DART-MS . John Wiley & Sons: 2017.

81. Cody, R. B., Observation of molecular ions and analysis of nonpolar compounds with the direct analysis in real time ion source. Anal. Chem. 2008, 81 (3), 1101-1107.

82. Jordan, A.; Haidacher, S.; Hanel, G.; Hartungen, E.; Herbig, J.; Märk, L.; Schottkowsky, R.; Seehauser, H.; Sulzer, P.; Märk, T., An online ultra-high sensitivity Proton-transfer-reaction mass-spectrometer combined with switchable reagent ion capability (PTR+ SRI− MS). Int. J. Mass Spectrom. 2009, 286 (1), 32-38.

83. Herbert, C. G.; Johnstone, R. A., Mass spectrometry basics . CRC press: 2002.

84. Perez, S.; Eichhorn, P.; Barcelo, D., Applications of time-of-flight and mass spectrometry in environmental, food, doping, and forensic analysis . Elsevier: 2016; Vol. 71.

85. Kotiaho, T.; Lauritsen, F. R.; Choudhury, T. K.; Cooks, R. G.; Tsao, G. T., Membrane introduction mass spectrometry. Anal. Chem. 1991, 63 (18), 875A-883A.

86. Krogh, E. T.; Gill, C. G., Membrane introduction mass spectrometry (MIMS): a versatile tool for direct, real time chemical measurements. J. Mass Spectrom. 2014, 49 (12), - 1205-1213.

87. Lauritsen, F. R.; Kotiaho, T., Advances in membrane inlet mass spectrometry (MIMS). Crit. Rev. Anal. Chem. 1996, 15 (4), 237-264. 111

88. Wolter, O.; Heitbaum, J., Differential electrochemical mass spectroscopy (DEMS) — a new method for the study of electrode processes. Ber. Bunsenges. Physik. Chem. 1984, 88 (1), 2-6.

89. Bruckenstein, S.; Gadde, R. R., Use of a porous electrode for in situ mass spectrometric determination of volatile electrode reaction products. J. Am. Chem. Soc. 1971, 93 (3), 793-794.

90. Housmans, T. H.; Wonders, A. H.; Koper, M. T., Structure sensitivity of methanol electrooxidation pathways on platinum: an on-line electrochemical mass spectrometry study. J. Phys. Chem. B 2006, 110 (20), 10021-31.

91. Wonders, A. H.; Housmans, T. H.; Rosca, V.; Koper, M. T., On-line mass spectrometry system for measurements at single-crystal electrodes in hanging meniscus configuration. J. Appl. Electrochem. 2006, 36 (11), 1215-1221.

92. Wang, H.; Jusys, Z.; Behm, R., Ethanol electrooxidation on a carbon-supported Pt catalyst: reaction kinetics and product yields. J. Phys. Chem. B 2004, 108 (50), 19413-19424.

93. Schouten, K. J. P.; Kwon, Y.; van der Ham, C. J. M.; Qin, Z.; Koper, M. T. M., A new mechanism for the selectivity to C-1 and C-2 species in the electrochemical reduction of carbon dioxide on copper electrodes. Chem. Sci. 2011, 2 (10), 1902-1909.

94. Clark, E. L.; Bell, A. T., Direct Observation of the Local Reaction Environment during the Electrochemical Reduction of CO2. J. Am. Chem. Soc. 2018, 140 (22), 7012-7020.

95. A technical comparison of SIFT-MS and GC-MS. Syft Technologies Ltd: 2012.

96. Lobaccaro, P.; Mandal, L.; Motapothula, M. R.; Sherburne, M.; Martin, J.; Venkatesan, T.; Ager, J. W., Initial Application of Selected Ion Flow Tube Mass - - Spectrometry to Real Time Product Detection in Electrochemical CO2 Reduction. Energy - Technol. 2018, 6 (1), 110-121.

97. Mandal, L.; Yang, K. R.; Motapothula, M. R.; Ren, D.; Lobaccaro, P.; Patra, A.; Sherburne, M.; Batista, V. S.; Yeo, B. S.; Ager, J. W., Investigating the Role of Copper Oxide in Electrochemical CO2 Reduction in Real Time. ACS Appl. Mater. Interfaces 2018, 10 (10), 8574-8584. 112

98. Milligan, D. B.; Francis, G. J.; Prince, B. J.; McEwan, M. J., Demonstration of selected ion flow tube MS detection in the parts per trillion range. Anal. Chem. 2007, 79 (6), 2537-2540.

99. Blake, R. S.; Monks, P. S.; Ellis, A. M., Proton-transfer reaction mass spectrometry. Chem. Rev. 2009, 109 (3), 861-896.

100. Hambitzer, G.; Heitbaum, J., Electrochemical thermospray mass spectrometry. Anal. Chem. 1986, 58 (6), 1067-1070.

101. Volk, K. J.; Yost, R. A.; Brajter-Toth, A., On-line electrochemistry/thermospray/ as a new approach to the study of redox reactions: the oxidation of uric acid. Anal. Chem. 1989, 61 (15), 1709-1717.

102. Lindon, J. C.; Tranter, G. E.; Koppenaal, D., Encyclopedia of spectroscopy and spectrometry . Academic Press: 2016.

103. Bond, A. M.; Colton, R.; D'Agostino, A.; Downard, A. J.; Traeger, J. C., A role for electrospray mass spectrometry in electrochemical studies. Anal. Chem. 1995, 67 (10), 1691- 1695.

104. Van Berkel, G. J.; Zhou, F., Characterization of an electrospray ion source as a controlled-current . Anal. Chem. 1995, 67 (17), 2916-2923.

105. Pramanik, B. N.; Ganguly, A. K.; Gross, M. L., Applied electrospray mass spectrometry: practical spectroscopy series . CRC Press: 2002; Vol. 32.

106. Zhao, W.; Jusys, Z.; Behm, R. J. r., Complete quantitative online analysis of methanol electrooxidation products via electron impact and electrospray ionization mass spectrometry. Anal. Chem. 2012, 84 (13), 5479-5483.

107. Kertesz, V.; Van Berkel, G. J., Minimizing analyte electrolysis in an electrospray emitter. J. Mass Spectrom. 2001, 36 (2), 204-210.

108. Oberacher, H.; Pitterl, F.; Erb, R.; Plattner, S., Mass spectrometric methods for monitoring redox processes in electrochemical cells. Mass Spectrom. Rev. 2015, 34 (1), 64- 92. 113

109. Zettersten, C. On-line Electrochemistry Electrospray Ionisation Mass Spectrometry: Method Development and Applications. Acta Universitatis Upsaliensis, 2009.

110. Takats, Z.; Wiseman, J. M.; Gologan, B.; Cooks, R. G., Mass spectrometry sampling under ambient conditions with desorption electrospray ionization. Science 2004, 306 (5695), 471-473.

111. Miao, Z.; Chen, H., Direct analysis of liquid samples by desorption electrospray ionization-mass spectrometry (DESI-MS). J. Am. Soc. Mass Spectrom. 2009, 20 (1), 10-19.

112. Benassi, M.; Wu, C.; Nefliu, M.; Ifa, D. R.; Volný, M.; Cooks, R. G., Redox transformations in desorption electrospray ionization. Int. J. Mass Spectrom. 2009, 280 (1-3), 235-240.

113. Cindric, M.; Matysik, F.-M., Coupling electrochemistry to capillary electrophoresis- mass spectrometry. Trends Anal. Chem. 2015, 70 , 122-127.

114. Scholz, R.; Palatzky, P.; Matysik, F.-M., Simulation of oxidative stress of guanosine and 8-oxo-7, 8-dihydroguanosine by electrochemically assisted injection–capillary electrophoresis–mass spectrometry. Anal. Bioanal. Chem. 2014, 406 (3), 687-694.

115. Beißmann, S.; Buchberger, W.; Hertsens, R.; Klampfl, C. W., High-performance liquid chromatography coupled to direct analysis in real time mass spectrometry: Investigations on gradient elution and influence of complex matrices on signal intensities. J. Chromatogr. A 2011, 1218 (31), 5180-5186.

116. Chang, C.; Xu, G.; Bai, Y.; Zhang, C.; Li, X.; Li, M.; Liu, Y.; Liu, H., Online coupling of capillary electrophoresis with direct analysis in real time mass spectrometry. Anal. Chem. 2012, 85 (1), 170-176.

117. Eberherr, W.; Buchberger, W.; Hertsens, R.; Klampfl, C. W., Investigations on the coupling of high-performance liquid chromatography to direct analysis in real time mass spectrometry. Anal. Chem. 2010, 82 (13), 5792-5796.

118. Saang'onyo, D.; Selby, G.; Smith, D. L., Validation of a direct analysis in real time mass spectrometry (DART-MS) method for the quantitation of six carbon sugars in a saccharification matrix. Anal. Methods 2012, 4 (10), 3460-3465. 114

119. Yu, S.; Crawford, E.; Tice, J.; Musselman, B.; Wu, J.-T., without sample cleanup or chromatography: the evaluation and initial implementation of direct analysis in real time ionization mass spectrometry for the quantification of drugs in biological matrixes. Anal. Chem. 2008, 81 (1), 193-202.

120. Ren, L.; Han, Y.; Zhang, Y.; Zhang, Y.; Meng, X.; Shi, Q., Spray injection direct analysis in real time (DART) ionization for petroleum analysis. Energy Fuels 2016, 30 (6), 4486-4493.

121. Khanipour, P.; Löffler, M.; Reichert, A. M.; Haase, F. T.; Mayrhofer, K. J. J.; Katsounaros, I., Electrochemical Real-Time Mass Spectrometry (EC-RTMS): Monitoring Electrochemical Reaction Products in Real Time. Angew. Chem. Int. Ed. 2019, 58 (22), 7273- 7277.

122. Haschke, S.; Wu, Y.; Bashouti, M.; Christiansen, S.; Bachmann, J., Engineering nanoporous (III) oxide into an effective water oxidation electrode. ChemCatChem 2015, 7 (16), 2455-2459.

123. Haschke, S.; Pankin, D.; Petrov, Y.; Bochmann, S.; Manshina, A.; Bachmann, J., Design Rules for Oxygen Evolution Catalysis at Porous Iron Oxide Electrodes: A 1000 Fold - Current Density Increase. ChemSusChem 2017, 10 (18), 3644-3651.

124. Khanipour, P.; Haschke, S.; Bachmann, J.; Mayrhofer, K. J. J.; Katsounaros, I.,

Electrooxidation of saturated C 1-C3 primary alcohols on platinum: Potential-resolved product analysis with electrochemical real-time mass spectrometry (EC-RTMS). Electrochim. Acta 2019, 315 , 67-74.

125. Faisal, F.; Bertram, M.; Stumm, C.; Waidhas, F.; Brummel, O.; Libuda, J., Preparation of Complex Model Electrocatalysts in Ultra-High Vacuum and Transfer into the Electrolyte for Electrochemical IR Spectroscopy and other Techniques. Rev. Sci. Instrum. 2018, 89 (11), 114101.

126. Waidhas, F.; Haschke, S.; Khanipour, P.; Fromm, L.; Görling, A.; Bachmann, J.; Katsounaros, I.; Mayrhofer, K. J. J.; Brummel, O.; Libuda, J., Secondary Alcohols as Rechargeable Electrofuels: Electrooxidation of at Pt Electrodes. ACS Catal. 2020, 10 (12), 6831-6842. 115

127. Cherevko, S.; Topalov, A. A.; Katsounaros, I.; Mayrhofer, K. J., Electrochemical dissolution of gold in acidic medium. Electrochem. Commun. 2013, 28 , 44-46.

128. Grote, J.-P.; Zeradjanin, A. R.; Cherevko, S.; Savan, A.; Breitbach, B.; Ludwig, A.; Mayrhofer, K. J., Screening of material libraries for electrochemical CO2 reduction catalysts– Improving selectivity of Cu by mixing with Co. J. Catal. 2016, 343 , 248-256.

129. Resnick, P. R.; Buck, W. H., Teflon AF Amorphous Fluoropolymers. In Modern Fluoropolymers , Wiley: 1997; pp 397-419.

130. Karlsson, J. M.; Haraldsson, T.; Laakso, S.; Virtanen, A.; Mäki, M.; Ronan, G.; Van Der Wijngaart, W. In PCR on a PDMS-based microchip with integrated bubble removal , 16th International Solid-State Sensors, Actuators and Microsystems Conference, IEEE: 2011; pp 2215-2218.

131. Baker, R. W., Membrane technology and applications . John Wiley & Sons: 2012.

132. Chang, C.; Zhou, Z.; Yang, Y.; Han, Y.; Bai, Y.; Zhao, M.; Liu, H., Normal phase LC coupled with direct analysis in real time MS for the chiral analysis of 4 - (methylnitrosamino) 1 (3 pyridyl) 1 butanol and jasmonic acid. Electrophoresis 2012, 33 - - - - - (22), 3387-3393.

133. Zhang, Y.; Li, X.; Nie, H.; Yang, L.; Li, Z.; Bai, Y.; Niu, L.; Song, D.; Liu, H., Interface for online coupling of surface plasmon resonance to direct analysis in real time mass spectrometry. Anal. Chem. 2015, 87 (13), 6505-6509.

134. Todolí, J.-L.; Mermet, J.-M., Evaluation of a direct injection high-efficiency nebulizer (DIHEN) by comparison with a high-efficiency nebulizer (HEN) coupled to a cyclonic spray chamber as a liquid sample introduction system for ICP-AES. J. Anal. At. Spectrom. 2001, 16 (5), 514-520.

135. Todoli, J.-L.; Mermet, J.-M., Liquid sample introduction in ICP spectrometry: A practical guide . Elsevier: 2011.

136. Gross, J. H., Direct analysis in real time—a critical review on DART-MS. Anal. Bioanal. Chem. 2014, 406 (1), 63-80.

137. Hill, S. J., Inductively Coupled Plasma Spectrometry and its Applications . John Wiley & Sons: 2008; Vol. 8. 116

138. Le Pogam, P.; Legouin, B.; Le Lamer, A. C.; Boustie, J.; Rondeau, D., Analysis of the cyanolichen Lichina pygmaea metabolites using in situ DART MS: from detection to - of mycosporine serinol. J. Mass Spectrom. 2015, 50 (3), 454-462.

139. Newsome, G. A.; Ackerman, L. K.; Johnson, K. J., Humidity affects relative ion abundance in direct analysis in real time mass spectrometry of hexamethylene triperoxide diamine. Anal. Chem. 2014, 86 (24), 11977-11980.

140. Song, L.; Gibson, S. C.; Bhandari, D.; Cook, K. D.; Bartmess, J. E., Ionization mechanism of positive-ion direct analysis in real time: a transient microenvironment concept. Anal. Chem. 2009, 81 (24), 10080-10088.

141. Goebbert, D. J.; Wenthold, P. G., Water dimer proton affinity from the kinetic method: dissociation energy of the water dimer. Eur. J. Mass. Spectrom. 2004, 10 (6), 837-845.

142. Wróblewski, T.; Ziemczonek, L.; Alhasan, A.; Karwasz, G., Ab initio and density functional theory calculations of proton affinities for volatile organic compounds. Eur. Phys. J. Spec. Top. 2007, 144 (1), 191-195.

143. Stönner, C.; Derstroff, B.; Klüpfel, T.; Crowley, J. N.; Williams, J., Glyoxal measurement with a proton transfer reaction time of flight mass spectrometer (PTR TOF - - MS): characterization and calibration. J. Mass Spectrom. 2017, 52 (1), 30-35.

144. Vlasenko, A.; Macdonald, A.; Sjostedt, S.; Abbatt, J., Formaldehyde measurements by Proton transfer reaction–Mass Spectrometry (PTR-MS): correction for humidity effects. Atmos. Meas. Tech. 2010, 3 (4), 1055-1062.

145. Padilla, M.; Baturina, O.; Gordon, J. P.; Artyushkova, K.; Atanassov, P.; Serov, A.,

Selective CO 2 electroreduction to C 2H4 on porous Cu films synthesized by sacrificial support method. J. CO2 Util. 2017, 19 , 137-145.

146. Coskun, H.; Aljabour, A.; De Luna, P.; Farka, D.; Greunz, T.; Stifter, D.; Kus, M.; Zheng, X.; Liu, M.; Hassel, A. W., Biofunctionalized conductive polymers enable efficient

CO 2 electroreduction. Sci. Adv. 2017, 3 (8), e1700686.

147. Hori, Y., Electrochemical CO 2 reduction on metal electrodes. In Modern aspects of electrochemistry no. 42 , Springer: 2008; pp 89-189. 117

148. Li, C. W.; Kanan, M. W., CO 2 reduction at low overpotential on Cu electrodes resulting from the reduction of thick Cu 2O films. J. Am. Chem. Soc. 2012, 134 (17), 7231- 7234.

149. Cave, E. R.; Shi, C.; Kuhl, K. P.; Hatsukade, T.; Abram, D. N.; Hahn, C.; Chan,

K.; Jaramillo, T. F., Trends in the catalytic activity of hydrogen evolution during CO 2 electroreduction on transition metals. ACS Catal. 2018, 8 (4), 3035-3040.

150. Ledezma-Yanez, I.; Wallace, W. D. Z.; Sebastián-Pascual, P.; Climent, V.; Feliu, J. M.; Koper, M. T., Interfacial water reorganization as a pH-dependent descriptor of the hydrogen evolution rate on platinum electrodes. Nat. Energy 2017, 2 (4), 17031.

151. Wang, Y.; Wang, Z.; Dinh, C.-T.; Li, J.; Ozden, A.; Kibria, M. G.; Seifitokaldani,

A.; Tan, C.-S.; Gabardo, C. M.; Luo, M., Catalyst synthesis under CO 2 electroreduction favours faceting and promotes renewable fuels electrosynthesis. Nat. Catal. 2019 , 1-9.

152. Lee, S. Y.; Jung, H.; Kim, N.-K.; Oh, H.-S.; Min, B. K.; Hwang, Y. J., Mixed copper states in anodized Cu electrocatalyst for stable and selective ethylene production from

CO 2 reduction. J. Am. Chem. Soc. 2018, 140 (28), 8681-8689.

153. Schouten, K. J. P.; Kwon, Y.; van der Ham, C. J. M.; Qin, Z.; Koper, M. T. M., A new mechanism for the selectivity to C1 and C2 species in the electrochemical reduction of carbon dioxide on copper electrodes. Chem. Sci. 2011, 2, 1902-1909.

154. Hori, Y.; Murata, A.; Takahashi, R., Formation of hydrocarbons in the electrochemical reduction of carbon dioxide at a copper electrode in aqueous solution. J. Chem. Soc., Faraday Trans. 1989, 85 , 2309-2326.

155. Seidel, Y.; Schneider, A.; Jusys, Z.; Wickman, B.; Kasemo, B.; Behm, R., Mesoscopic mass transport effects in electrocatalytic processes. Faraday Discuss. 2009, 140 , 167-184.

156. Ma, M.; Djanashvili, K.; Smith, W. A., Controllable hydrocarbon formation from the electrochemical reduction of CO 2 over Cu nanowire arrays. Angew. Chem. Int. Ed. 2016, 55 (23), 6680-6684. 118

157. Schouten, K. J. P.; Gallent, E. P.; Koper, M. T., The influence of pH on the reduction of CO and CO 2 to hydrocarbons on copper electrodes. J. Electroanal. Chem. 2014, 716 , 53- 57.

158. Velasco-Vélez, J.-J. s.; Jones, T.; Gao, D.; Carbonio, E.; Arrigo, R.; Hsu, C.-J.; Huang, Y.-C.; Dong, C.-L.; Chen, J.-M.; Lee, J.-F., The role of the copper oxidation state in the electrocatalytic reduction of CO 2 into valuable hydrocarbons. ACS Sustain. Chem. Eng. 2018, 7 (1), 1485-1492.

159. Yan, X.; Sun, J.; Meng, Y., Experimental insight into the chemical corrosion mechanism of copper with an oil-in-water emulsion solution. RSC Adv. 2018, 8 (18), 9833- 9840.

160. Lamy, C.; Lima, A.; LeRhun, V.; Delime, F.; Coutanceau, C.; Léger, J.-M., Recent advances in the development of direct alcohol fuel cells (DAFC). J. Power Sources 2002, 105 (2), 283-296.

161. Rinaldi, R.; Schüth, F., Design of solid catalysts for the conversion of biomass. Energy Environ. Sci. 2009, 2 (6), 610-626.

162. Gomes, J.; Profeti, D.; Deiner, L. J., Influence of the Particle Size Distribution on the Activity and Selectivity of Carbon Supported Platinum Nanoparticle Catalysts for Ethanol - Electrooxidation. ChemElectroChem 2014, 1 (3), 655-662.

163. Lai, S. C.; Koper, M. T., Electro-oxidation of ethanol and acetaldehyde on platinum single-crystal electrodes. Faraday Discuss. 2009, 140 , 399-416.

164. Lee, C.-G.; Ojima, H.; Umeda, M., Electrooxidation of C 1 to C 3 alcohols with Pt and Pt–Ru sputter deposited interdigitated array electrodes. Electrochim. Acta 2008, 53 (7), 3029- 3035.

165. Şen, F.; Gökağaç, G., Pt nanoparticles synthesized with new surfactants: improvement in C 1–C 3 alcohol oxidation catalytic activity. J. Appl. Electrochem. 2014, 44 (1), 199-207.

166. Wang, J.; Wasmus, S.; Savinell, R., Evaluation of Ethanol, 1 Propanol, and 2 - - Propanol in a Direct Oxidation Polymer Electrolyte Fuel Cell A Real Time Mass - - Spectrometry Study. J. Electrochem. Soc. 1995, 142 (12), 4218-4224. 119

167. Santasalo, A.; Vidal-Iglesias, F.; Solla-Gullón, J.; Berná, A.; Kallio, T.; Feliu, J., Electrooxidation of methanol and 2-propanol mixtures at platinum single crystal electrodes. Electrochim. Acta 2009, 54 (26), 6576-6583.

168. Bittins Cattaneo, B.; Wilhelm, S.; Cattaneo, E.; Buschmann, H.; Vielstich, W., - Intermediates and products of ethanol oxidation on platinum in acid solution. Ber. Bunsenges. Physik. Chem. 1988, 92 (11), 1210-1218.

169. Sun, S.-G.; Yang, D.-F.; Tian, Z.-W., In situ FTIR studies on the adsorption and oxidation of n-propanol and isopropanol at a platinum electrode in sulphuric acid solutions. J. Electroanal. Chem. Interf. Electrochem. 1990, 289 (1-2), 177-187.

170. Cantane, D. A.; Ambrosio, W. F.; Chatenet, M.; Lima, F. H. B. d., Electro-oxidation of ethanol on Pt/C, Rh/C, and Pt/Rh/C-based electrocatalysts investigated by on-line DEMS. J. Electroanal. Chem. 2012, 681 , 56-65.

171. Batista, E.; Malpass, G. R. P.; Motheo, A. d. J.; Iwasita, T., New insight into the pathways of methanol oxidation. Electrochem. Commun. 2003, 5 (10), 843-846.

172. Wang, H.; Wingender, C.; Baltruschat, H.; Lopez, M.; Reetz, M., Methanol oxidation on Pt, PtRu, and colloidal Pt electrocatalysts: a DEMS study of product formation. J. Electroanal. Chem. 2001, 509 (2), 163-169.

173. Czakó, G.; Nagy, B.; Tasi, G.; Somogyi, Á.; Šimunek, J.; Noga, J.; Braams, B. J.; Bowman, J. M.; Császár; , A. G., Proton affinity and enthalpy of formation of formaldehyde. Int. J. Quantum Chem. 2009, 109 (11), 2393-2409.

174. Hunter, E. P. L.; Lias, S. G., Evaluated gas phase basicities and proton affinities of molecules: An update. J. Phys. Chem. Ref. Data 1998, 27 (3), 413-656.

175. Housmans, T. H. M.; Wonders, A. H.; Koper, M. T. M., Structure sensitivity of methanol electrooxidation pathways on platinum: An on-line electrochemical mass spectrometry study. J. Phys. Chem. B 2006, 110 (20), 10021-10031.

176. Jusys, Z.; Kaiser, J.; Behm, R. J., Composition and activity of high surface area PtRu catalysts towards adsorbed CO and methanol electrooxidation: A DEMS study. Electrochim. Acta 2002, 47 (22), 3693-3706. 120

177. Zhao, W.; Jusys, Z.; Behm, R. J., Complete quantitative online analysis of methanol electrooxidation products via electron impact and electrospray ionization mass spectrometry. Anal. Chem. 2012, 84 (13), 5479-5483.

178. Malpass, G.; Motheo, A., Galvanostatic oxidation of formaldehyde-methanol solutions on Ti/Ru 0.3 Ti 0.7 O2 electrodes using a filter-press cell. J. Appl. Electrochem. 2001, 31 (12), 1351-1357.

179. Iwasita, T.; Pastor, E., A dems and FTir spectroscopic investigation of adsorbed ethanol on polycrystalline platinum. Electrochim. Acta 1994, 39 (4), 531-537.

180. Iwasita, T.; Rasch, B.; Cattaneo, E.; Vielstich, W., A sniftirs study of ethanol oxidation on platinum. Electrochim. Acta 1989, 34 (8), 1073-1079.

181. Wang, H.; Jusys, Z.; Behm, R. J., Ethanol electrooxidation on a carbon-supported Pt catalyst: reaction kinetics and product yields. J. Phys. Chem. B 2004, 108 (50), 19413-19424.

182. Sun, S.; Halseid, M. C.; Heinen, M.; Jusys, Z.; Behm, R., Ethanol electrooxidation on a carbon-supported Pt catalyst at elevated temperature and pressure: A high- temperature/high-pressure DEMS study. J. Power Sources 2009, 190 (1), 2-13.

183. Wang, Q.; Sun, G.; Jiang, L.; Xin, Q.; Sun, S.; Jiang, Y.; Chen, S.; Jusys, Z.;

Behm, R., Adsorption and oxidation of ethanol on colloid-based Pt/C, PtRu/C and Pt 3Sn/C catalysts: In situ FTIR spectroscopy and on-line DEMS studies. Phys. Chem. Chem. Phys. 2007, 9 (21), 2686-2696.

184. Perez, J. M.; Beden, B.; Hahn, F.; Aldaz, A.; Lamy, C., “In situ” infrared reflectance spectroscopic study of the early stages of ethanol adsorption at a platinum electrode in acid medium. J. Electroanal. Chem. Interf. Electrochem. 1989, 262 (1), 251-261.

185. Willsau, J.; Heitbaum, J., Analysis of adsorbed intermediates and determination of surface potential shifts by dems. Electrochim. Acta 1986, 31 (8), 943-948.

186. Schmiemann, U.; Müller, U.; Baltruschat, H., The influence of the surface structure on the adsorption of ethene, ethanol and cyclohexene as studied by DEMS. Electrochim. Acta 1995, 40 (1), 99-107. 121

187. Wang, H.; Jusys, Z.; Behm, R. J., Ethanol electro-oxidation on carbon-supported Pt,

PtRu and Pt 3Sn catalysts: A quantitative DEMS study. J. Power Sources 2006, 154 (2), 351- 359.

188. Pastor, E.; Wasmus, S.; Iwasita, T.; Arévalo, M. C.; González, S.; Arvia, A. J.,

Spectroscopic investigations of C 3 primary alcohols on platinum electrodes in acid solutions: Part I. n-propanol. J. Electroanal. Chem. 1993, 350 (1), 97-116.

189. Schnaidt, J.; Heinen, M.; Jusys, Z.; Behm, R. J., Oxidation of 1-propanol on a Pt film electrode studied by combined electrochemical, in situ IR spectroscopy and online mass spectrometry measurements. Electrochim. Acta 2013, 104 , 505-517.

190. Wang, J.; Wasmus, S.; Savinell, R. F., Evaluation of Ethanol, 1-Propanol, and 2- Propanol in a Direct Oxidation Polymer-Electrolyte Fuel Cell. J. Electrochem. Soc. 1995, 142 , 4218-4224.

191. Gootzen, J. F. E.; Wonders, A. H.; Visscher, W.; van Veen, J. A. R., Adsorption of

C3 alcohols, 1-butanol, and ethene on platinized platinum as studied with FTIRS and DEMS. Langmuir 1997, 13 (6), 1659-1667.

192. Bockris, J. M., A hydrogen economy. Science 1972, 176 (4041), 1323-1323.

193. Yang, J.; Sudik, A.; Wolverton, C.; Siegel, D. J., High capacity hydrogen storage materials: attributes for automotive applications and techniques for materials discovery. Chem. Soc. Rev. 2010, 39 (2), 656-675.

194. Von Wild, J.; Friedrich, T.; Cooper, A.; Toseland, B.; Muraro, G.; TeGrotenhuis, W.; Wang, Y.; Humble, P.; Karim, A., Liquid Organic Hydrogen Carriers (LOHC): An auspicious alternative to conventional hydrogen storage technologies. Proceedings WHEC 2010 , 189-197.

195. Zhang, Y.; Evans, J. R.; Yang, S., Corrected values for boiling points and enthalpies of vaporization of elements in handbooks. J. Chem. Eng. Data 2011, 56 (2), 328-337.

196. Hua, T.; Ahluwalia, R.; Peng, J.-K.; Kromer, M.; Lasher, S.; McKenney, K.; Law, K.; Sinha, J., Technical assessment of compressed hydrogen storage tank systems for automotive applications. Int. J. Hydrog. Energy 2011, 36 (4), 3037-3049. 122

197. Dillon, A. C.; Jones, K.; Bekkedahl, T.; Kiang, C.; Bethune, D.; Heben, M., Storage of hydrogen in single-walled carbon nanotubes. Nature 1997, 386 (6623), 377.

198. Rosi, N. L.; Eckert, J.; Eddaoudi, M.; Vodak, D. T.; Kim, J.; O'Keeffe, M.; Yaghi, O. M., Hydrogen storage in microporous metal-organic frameworks. Science 2003, 300 (5622), 1127-1129.

199. Klerke, A.; Christensen, C. H.; Nørskov, J. K.; Vegge, T., Ammonia for hydrogen storage: challenges and opportunities. J. Mater. Chem. 2008, 18 (20), 2304-2310.

200. Teichmann, D.; Arlt, W.; Wasserscheid, P.; Freymann, R., A future energy supply based on Liquid Organic Hydrogen Carriers (LOHC). Energy Environ. Sci. 2011, 4 (8), 2767- 2773.

201. Preuster, P.; Papp, C.; Wasserscheid, P., Liquid organic hydrogen carriers (LOHCs): toward a hydrogen-free hydrogen economy. Acc. Chem. Res. 2016, 50 (1), 74-85.

202. Sievi, G.; Geburtig, D.; Skeledzic, T.; Bosmann, A.; Preuster, P.; Brummel, O.; Waidhas, F.; Montero, M. A.; Khanipour, P.; Katsounaros, I.; Libuda, J.; Mayrhofer, K. J. J.; Wasserscheid, P., Towards an efficient liquid organic hydrogen carrier fuel cell concept. Energy Environ. Sci. 2019, 12 (7), 2305-2314.

203. Markiewicz, M. E.; Hebert, D. M.; Bergens, S. H., Electro-oxidation of 2-propanol on platinum in alkaline electrolytes. J. Power Sources 2006, 161 (2), 761-767.

204. Liu, Y.; Zeng, Y.; Liu, R.; Wu, H.; Wang, G.; Cao, D., Poisoning of acetone to Pt and Au electrodes for electrooxidation of 2-propanol in alkaline medium. Electrochim. Acta 2012, 76 , 174-178.

205. Okanishi, T.; Katayama, Y.; Ito, R.; Muroyama, H.; Matsui, T.; Eguchi, K., Electrochemical oxidation of 2-propanol over platinum and palladium electrodes in alkaline media studied by in situ attenuated total reflection infrared spectroscopy. Phys. Chem. Chem. Phys. 2016, 18 (15), 10109-10115.

206. Cheng, F.; Wang, H.; Sun, Z.; Ning, M.; Cai, Z.; Zhang, M., Electrodeposited fabrication of highly ordered Pd nanowire arrays for alcohol electrooxidation. Electrochem. Commun. 2008, 10 (5), 798-801. 123

207. Zalineeva, A.; Serov, A.; Padilla, M.; Martinez, U.; Artyushkova, K.; Baranton, S.; Coutanceau, C.; Atanassov, P., Nano-structured Pd-Sn catalysts for alcohol electro-oxidation in alkaline medium. Electrochem. Commun. 2015, 57 , 48-51.

208. Cheng, Y.; Liu, Y.; Cao, D.; Wang, G.; Gao, Y., Effects of acetone on electrooxidation of 2-propanol in alkaline medium on the Pd/Ni-foam electrode. J. Power Sources 2011, 196 (6), 3124-3128.

209. Serov, A.; Martinez, U.; Falase, A.; Atanassov, P., Highly active PdCu catalysts for electrooxidation of 2-propanol. Electrochem. Commun. 2012, 22 , 193-196.

210. Choi, Y.; Sinev, I.; Mistry, H.; Zegkinoglou, I.; Roldan Cuenya, B., Probing the dynamic structure and of Au nanocatalysts during the electrochemical oxidation of 2-propanol. ACS Catal. 2016, 6 (5), 3396-3403.

211. Xing, S.; Liu, Z.; Xue, Q.; Yin, S.; Li, F.; Cai, W.; Li, S.; Chen, P.; Jin, P.; Yao, H., Rh nanoroses for isopropanol oxidation reaction. Appl. Catal. B 2019, 259 , 118082.

212. Pastor, E.; González, S.; Arvia, A. J., Electroreactivity of isopropanol on platinum in acids studied by DEMS and FTIRS. J. Electroanal. Chem. 1995, 395 (1), 233-242.

213. Sun, S.-G.; Lin, Y., Kinetics of isopropanol oxidation on Pt (111), Pt (110), Pt (100), Pt (610) and Pt (211) single crystal electrodes-: Studies of in situ time-resolved FTIR spectroscopy. Electrochim. Acta 1998, 44 (6-7), 1153-1162.

214. Leung, L.-W. H.; Chang, S.-C.; Weaver, M. J., Real-time FTIR spectroscopy as an electrochemical mechanistic probe: electrooxidation of ethanol and related species on well- defined Pt (111) surfaces. J. Electroanal. Chem. Interf. Electrochem. 1989, 266 (2), 317-336.

215. Sokolova, E.; Christov, M., Electrochemical behaviour of oxygen-containing aliphatic compounds adsorbed on platinum in acid medium: Part III. Secondary alcohols. J. Electroanal. Chem. Interf. Electrochem. 1984, 175 (1-2), 195-205.

216. Christov, M.; Sokolova, E., Electrochemical behaviour of oxygen-containing aliphatic compounds adsorbed on platinum in acid medium: Part II. Primary alcohols. J. Electroanal. Chem. Interf. Electrochem. 1984, 175 (1-2), 183-193.

217. Khanipour, P.; Speck, F. D.; Mangoufis-Giasin, I.; Mayrhofer, K. J. J.; Cherevko, S.; Katsounaros, I., Electrochemical oxidation of isopropanol on platinum-ruthenium 124 nanoparticles studied with real-time product and dissolution analytics. ACS Appl. Mater. Interfaces 2020, 12 (30), 33670–33678.

218. Gasteiger, H. A.; Markovic, N.; Ross Jr, P. N.; Cairns, E. J., Methanol electrooxidation on well-characterized platinum-ruthenium bulk alloys. J. Phys. Chem. 1993, 97 (46), 12020-12029.

219. Rodrigues, I. d. A.; De Souza, J. P. I.; Pastor, E.; Nart, F. C., Cleavage of the C− C Bond during the Electrooxidation of 1-Propanol and 2-Propanol: Effect of the Pt Morphology and of Codeposited Ru. Langmuir 1997, 13 (25), 6829-6835.

220. Qi, Z.; Kaufman, A., Liquid-feed direct oxidation fuel cells using neat 2-propanol as fuel. J. Power Sources 2003, 118 (1-2), 54-60.

221. Gasteiger, H. A.; Markovic, N.; Ross, P. N.; Cairns, E. J., Carbon monoxide electrooxidation on well-characterized platinum-ruthenium alloys. J. Phys. Chem. 1994, 98 (2), 617-625.

222. McBreen, J.; Mukerjee, S., In Situ X Ray Absorption Studies of a Pt Ru - - Electrocatalyst. J. Electrochem. Soc. 1995, 142 (10), 3399-3404.

223. Sun, S.-G.; Lin, Y., Kinetic aspects of oxidation of isopropanol on Pt electrodes investigated by in situ time-resolved FTIR spectroscopy. J. Electroanal. Chem. 1994, 375 (1- 2), 401-404.

224. Cherevko, S.; Zeradjanin, A. R.; Topalov, A. A.; Kulyk, N.; Katsounaros, I.; Mayrhofer, K. J. J., Dissolution of Noble Metals during Oxygen Evolution in Acidic Media. ChemCatChem 2014, 6, 2219-2223.

225. Pourbaix, M., Atlas of Electrochemical Equilibria in Aqueous Solutions . 1st ed.; Pergamon Press: Oxford, 1966.

226. Jovanovic, P.; Selih, V. S.; Sala, M.; Hocevar, S.; Ruiz-Zepeda, F.; Hodnik, N.; Bele, M.; Gaberscek, M., Potentiodynamic dissolution study of PtRu/C electrocatalyst in the presence of methanol. Electrochim. Acta 2016, 211 , 851-859.

227. Topalov, A. A.; Zeradjanin, A. R.; Cherevko, S.; Mayrhofer, K. J. J., The impact of dissolved reactive gases on platinum dissolution in acidic media. Electrochem. Commun. 2014, 40 , 49-53. 125

126

Appendix

Publication list: (A) Published manuscripts

1. Peyman Khanipour , Florian D. Speck, Iosif Mangoufis-Giasin, Karl J.J. Mayrhofer, Serhiy Cherevko, and Ioannis Katsounaros, Electrochemical Oxidation of Isopropanol on Platinum-Ruthenium Nanoparticles Studied With Real-Time Product and Dissolution Analytics, ACS Appl. Mater. Interfaces 12 (2020) 33670−33678.

2. Mario Löffler, Peyman Khanipour , Nadiia Kulyk, Karl J.J. Mayrhofer, and Ioannis Katsounaros, Insights Into Liquid Product Formation During Carbon Dioxide Reduction on Copper and Oxide-Derived Copper From Quantitative Real-Time Measurements, ACS Catal. 10 (2020) 6735−6740.

3. Fabian Waidhas, Sandra Haschke, Peyman Khanipour , Lukas Fromm, Andreas Görling, Julien Bachmann, Ioannis Katsounaros, Karl J. J. Mayrhofer, Olaf Brummel, and Jörg Libuda, Secondary Alcohols as Rechargeable Electrofuels: Electrooxidation of Isopropyl Alcohol at Pt Electrodes, ACS Catal. 10 (2020) 6831−6842.

4. Peyman Khanipour , Sandra Haschke, Julien Bachmann, Karl J.J.Mayrhofer, and Ioannis Katsounaros, Electrooxidation of Saturated C1-C3 Primary Alcohols on Platinum: Potential-Resolved Product Analysis with Electrochemical Real-Time Mass Spectrometry (EC-RTMS), Electrochim. Acta 315 (2019) 67-74.

5. Gabriel Sievi, Denise Geburtig, Tanja Skeledzic, Andreas Bösmann, Patrick Preuster, Olaf Brummel, Fabian Waidhas, María A. Montero, Peyman Khanipour , Ioannis Katsounaros, Jörg Libuda, Karl J. J. Mayrhofer, and Peter Wasserscheid, Towards an Efficient Liquid Organic Hydrogen Carrier Fuel Cell Concept, Energy Environ. Sci. 12 (2019) 2305-2314.

6. Peyman Khanipour , Mario Löffler, Andreas M. Reichert, Felix T. Haase, Karl J. J. Mayrhofer, and Ioannis Katsounaros, Electrochemical Real-Time Mass Spectrometry (EC-RTMS): Monitoring Electrochemical Reaction Products in Real Time, Angew. Chem. Int. Ed. 58 (2019) 7273–7277. Very important paper (VIP) Featured in the Front Cover: Angew. Chem. Int. Ed. 58 (2019) 7145–7496. Highlighted with: Marçal Capdevila-Cortada, Products in Real Time , Nat. Catal., 2 (2019) 280 .

(B) Manuscripts in preparation (tentative titles)

1. Peyman Khanipour , Andreas M. Reichert, Pascal Hauenstein, Felix Derleth, Simon Thiele, Karl J.J. Mayrhofer, and Ioannis Katsounaros, A Holistic Approach for the Selective and Sustainable Electrosynthesis of Acrolein.

127

2. Ricarda Kloth, Peyman Khanipour , Karl J. J. Mayrhofer, and Ioannis Katsounaros Implementation of a Closed Ionization Interface for the Analysis of Liquid Sample Streams with Direct Analysis in Real Time Mass Spectrometry.

3. Iosif Mangoufis-Giasin, Peyman Khanipour , Federico Calle-Vallejo, Olaf Brummel, Karl Mayrhofer, and Ioannis Katsounaros, Electrochemical Oxidation of 2-Propanol Over Noble Metals in Alkaline Media Studied With Real Time-Resolved Product Analytics.

4. Andreas M. Reichert, Walter A. P. Villaroel, Peyman Khanipour , David McLaughlin, Sebastian Bochmann, Federico Calle-Vallejo, and Ioannis Katsounaros, Elucidating the Role of Glyoxal in the CO 2 Reduction Reaction on Copper and Oxide- Derived Copper.