Thesis Has Been Carried out Under the Microcat Funding As Part of the Scientific Programme of the French National Research Agency (ANR)

UvA-DARE (Digital Academic Repository)

Development of catalytic microreactors by plasma processes: application to wastewater treatment

Da Silva, B.T.

Publication date 2015 Document Version Final published version License Other Link to publication

Citation for published version (APA): Da Silva, B. T. (2015). Development of catalytic microreactors by plasma processes: application to wastewater treatment.

General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

Download date:28 Sep 2021

Development of catalytic

microreactors by plasma processes: application to wastewater treatment

Bradley Da Silva

Development of catalytic microreactors by plasma processes: application to wastewater treatment

The author can be reached at [email protected]

The research reported in this thesis has been carried out under the MicroCat funding as part of the scientific programme of the French National Research Agency (ANR). Development of catalytic microreactors by plasma processes: application to wastewater treatment

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Universiteit van Amsterdam en Université Pierre et Marie Curie op gezag van de Rector Magnificus prof. dr. D.C. van den Boom ten overstaan van een door het College voor Promoties ingestelde commissie, in het openbaar te verdedigen in de Agnietenkapel op woensdag 18 november 2015, te 16:00 uur

door Bradley Timothy Da Silva geboren te Panjim, Goa, India

Promotiecommissie:

Promotores: Prof. dr. D. Bonn Universiteit van Amsterdam Prof.dr. M. Tatoulian Université Pierre et Marie Curie

Overige leden: Prof.dr. P. Da Costa Université Pierre et Marie Curie Prof. dr. B. de Bruin Universiteit van Amsterdam Dr. R. Sprik Universiteit van Amsterdam Dr. N. F. Shahidzadeh Universiteit van Amsterdam Prof. Dr. T. Roques-Carmes Université de Nancy

Faculteit der Natuurwetenschappen, Wiskunde en Informatica

To my father, Froilano Da Silva

Contents Contents

ACKNOWLEDGMENTS ...... 11 ABSTRACT ...... 15 SAMENVATTING ...... 17 RÉSUMÉ ...... 19 GENERAL INTRODUCTION ...... 21 CHAPTER 1: LITERATURE REVIEW ...... 25 1. INTRODUCTION ...... 25 2. MICROFLUIDIC MATERIALS, PROPERTIES AND FABRICATION TECHNIQUES ...... 25 2.1 Glass microreactors ...... 25 2.2 Silicon microsystems ...... 26 2.3 Polymer-based microreactors ...... 27 2.4 Metal and ceramic-based microsystems ...... 30 3. ADVANCED OXIDATION PROCESSES ...... 33 4. CATALYTIC MICROREACTORS ...... 40 5. CATALYST DEPOSITION TECHNIQUES ...... 43 6. CONCLUSION ...... 46 CHAPTER 2: MATERIALS AND METHODS ...... 47 1. INTRODUCTION ...... 47 2. MICROFLUIDIC MATERIALS ...... 47 3. METAL-ORGANIC PLASMA ENHANCED CHEMICAL VAPOUR DEPOSITION PROCESS ...... 48 3.1 Description of the reactor ...... 48 3.2 General procedure for the silica-like deposition ...... 49 3.3 General procedure for the catalyst deposition ...... 50 4. CATALYTIC OZONATION PROCESS ...... 50 5. ANALYTICAL METHODS ...... 51 5.1 Surface characterizations ...... 51 5.1.1 FTIR-ATR ...... 51 5.1.2 Water contact angle measurements ...... 52 5.1.3 X-Ray diffraction...... 54 5.2 X-ray photoelectron spectroscopy ...... 55 5.2.1 Scanning Electron Microscopy ...... 57 5.2.2 Transmission Electron Microscopy ...... 58 5.2.3 Specific surface area measurements ...... 59 5.3 Analytical methods used for the liquid phases ...... 61 5.3.1 High Performance Liquid Chromatography ...... 61

7 Contents

5.3.2 Flame Atomic Absorption Spectroscopy ...... 63 5.3.3 pH-metry ...... 64 CHAPTER 3: STUDY OF THE STABILITY AND HYDROPHILICITY OF PLASMA-MODIFIED MICROFLUIDIC MATERIALS ...... 65 1. INTRODUCTION ...... 65 2. Materials and methods ...... 68 2.1 Chemicals ...... 68 2.2 Materials ...... 68 3. GENERAL PROCEDURE FOR THE SILICA-LIKE DEPOSITION USING PLASMA PROCESSES...... 69 3.1 Sputtered silica-like thin film deposition ...... 69 3.2 Deposition of SiO2-like layer by PECVD...... 70 4. RESULTS AND DISCUSSION ...... 71 4.1 Aging of silica-like coated substrates using water contact angle measurements ...... 71 4.1.1 Effect of the Ar/O2 plasma pre-treatment step ...... 71 4.1.2 Effect of the silica-like coating ...... 73 4.2 Chemical composition of the silica-like modified substrates ...... 82 4.2.1 FTIR spectroscopy...... 82 4.2.2 XPS measurements ...... 86 5. CONCLUSION ...... 93 CHAPTER 4: DEVELOPMENT OF CATALYTIC MICROREACTORS: COMPARISON OF THE PERFORMANCE OF PLASMA-DEPOSITED IRON AND COBALT OXIDES IN CATALYTIC OZONATION ...... 95 1. INTRODUCTION ...... 95 2. EXPERIMENTAL ...... 99 2.1 Elaboration of the COC microchannels ...... 99 2.2 Catalysts preparation...... 100 2.3 Thin films characterization ...... 104 2.4 Chip Assembly...... 106 2.5 Adsorption and catalytic activity ...... 106 2.6 Catalytic ozonation apparatus ...... 107 2.7 HPLC and FAAS Measurements ...... 108 3. RESULTS AND DISCUSSION ...... 109 3.1 Water stability of deposited catalysts ...... 109 3.2 Deposition of the cobalt oxide catalyst ...... 109 3.3 Deposition of the iron oxide catalyst ...... 118 4. CATALYTIC ACTIVITY MEASUREMENTS ...... 126 4.1 Results of simple ozonation of PA in batch reactor ...... 126 4.2 Results of simple ozonation of blank and silica-coated microreactor ...... 128

8 Contents

4.3 Catalytic ozonation results in iron oxide based catalytic microreactors ...... 129 4.3.1 Stability of the plasma deposited iron oxide coatings ...... 129 4.3.2 Adsorption tests results ...... 130 4.3.3 Results of catalytic ozonation tests ...... 132 4.4 Catalytic ozonation results in cobalt oxide based catalytic microreactors ...... 133 4.4.1 Stability of the plasma deposited cobalt oxide coatings ...... 133 4.4.2 Adsorption tests results ...... 135 4.4.3 Determination of the concentration of active sites for the Co3O4 catalyst ...... 136 4.4.4 Results of catalytic ozonation tests ...... 137 5. CONCLUSION ...... 142 CHAPTER 5: NUMERICAL SIMULATION OF THE PYRUVIC ACID DEGRADATION BY CATALYTIC OZONATION IN A CATALYTIC MICROREACTOR ...... 145 1. INTRODUCTION ...... 145 2. OZONE SELF-DECOMPOSITION AND SIMPLE OZONATION SIMULATION MODEL ...... 146 2.1 Description of the simple ozonation simulation model ...... 146 2.2 Simulation results of simple ozonation in batch reactor ...... 148 3. NUMERICAL SIMULATION OF THE CATALYTIC MICROSYSTEM ...... 153 3.1 Geometry and mesh used for the system ...... 153 3.2 Kinetic model used for the degradation of PA in the catalytic microreactor ...... 155 3.2.1 Steady-state model without competition of adsorbed by- products ...... 157 3.2.2 Steady-state model with competition of adsorbed by- products ...... 159 3.2.3 Simulation in the time-dependent model ...... 162 4. CONCLUSION ...... 164 GENERAL CONCLUSION AND OUTLOOKS ...... 167 LIST OF PUBLICATIONS ...... 171 BIBLIOGRAPHY ...... 173

14 Abstract Abstract

A key aspect in overcoming the energy and environmental challenges is to improve the efficiency of existing and new processes. As almost all major chemicals are nowadays produced by catalytic processes, heterogeneous catalysis plays a very active role because of environmental concerns. In order to develop more effective catalytic processes, a better understanding of the reaction pathways and kinetics is needed. In the field of wastewater treatment, catalytic ozonation is a typical example of this problem.

Catalytic microreactors can be used as innovative analytical tools for the determination of kinetics of catalytic ozonation. In this study, such catalytic microreactors were elaborated by using low pressure plasma processes for the deposition of a primer silica-like layer, then deposition and activation of the catalyst on polymer-based materials such as PDMS, NOA, THV and COC. Aging studies showed that the latter was highly stable in terms of hydrophilicity in air and water storage and displayed a rapid prototyping. The deposition and activation of iron and oxide-based catalysts demonstrated the importance of the plasma post-treatment step.

Catalytic ozonation with pyruvic acid as a refractory probe compound was performed with both catalysts. In the case of iron oxide layer, HPLC measurements demonstrated the inactivity of the latter compared to the cobalt oxide layer. With the plasma deposited cobalt oxide layer, 20 % of additional degradation was found whereas the effect was doubled (40%) when the layer was post-treated by argon plasma, indicating a synergistic effect between the surface morphology of the coating and the pyruvic acid.

A decrease in the catalytic activity for the post-treated cobalt oxide layer was found indicating a possible coverage of the active sites by intermediate products generated during catalytic ozonation. Nevertheless, the catalyst could be regenerated indicating a reversible phenomenon.

15 Abstract

A numerical study dealing with the reactions taking place on the surface of the post-treated cobalt oxide layer was carried out using the Comsol Multiphysics software. The model used to describe the degradation of pyruvic acid in a catalytic microreactor partially fitted the experimental data due to the lack of information concerning the various reactions rate constants of the intermediate species generated during the catalytic ozonation step.

Access to this information could be achieved through the use of the Coherent Anti-Stokes Raman Spectroscopy technique (CARS technique) as an online analytical tool. As perspectives, results obtained using the CARS technique would lead to an efficient tool that could predict the relevance and the direction of future improvement strategies regarding catalyzed chemical reactions.

16 Samenvatting Samenvatting

Een belangrijk aspect van het oplossen van energie- en milieuproblemen is het verbeteren van de efficiëntie van bestaande en nieuwe processen. Omdat vrijwel alle chemicaliën via katalyseprocessen geproduceerd worden, speelt heterogene katalyse een actieve rol in milieuvraagstukken. Om effectievere katalyseprocessen te ontwikkelen is een beter begrip van de reactiewegen en de kinetiek noodzakelijk. Binnen de waterzuivering is ozonisatie een typisch voorbeeld van deze problematiek

Katalytische microreactoren kunnen gebruikt worden als innovatieve analytische gereedschappen om de kinetiek van katalytische ozonisatie te bestuderen. In dit werk zijn dergelijke katalytische microreactoren uitgebreid door het gebruik van lage druk plasma processen. Deze processen maken de depositie van een primer silica-achtige laag en vervolgens de depositie en activatie van een katalysator mogelijk op polymeermaterialen zoals PDMS, NOA, THV and COC.

Verouderingsstudies lieten zien dat COC zeer stabiel waren in termen van hun hydrofiliciteit in lucht en water opslag en dat ze snelle prototypering vertoonde. De depositie en activatie van katalysators die op ijzer- of cobalt oxide zijn gebaseerd lieten duidelijk het belang zien van de plasma post-treatment stappen.

Katalytische ozonisatie met pyrodruivenzuur als een refractory probe stof is uitgevoerd met beide katalysatoren. Bij de ijzer oxide laag hebben HPLC metingen zijn passiviteit aangetoond ten opzichte van de cobalt-oxide laag. Bij de plasmadepositie van de cobalt oxide laag werd 20% extra afbraak gevonden. Dit effect werd verdubbeld (40%) bij een laag die post-treated is met argon plasma.

Dit wijst naar een synergetisch effect tussen de oppervlakte morfologie van de coating en het pyrodruivenzuur. Een afname in katalytische activiteit werd gevonden voor de post-treated cobalt oxide laag. Dit wijst naar een mogelijke

17 Samenvatting bedekking van de actieve plekken door intermediairen die ontstaan tijdens de katalytische ozonisatie.

Desalniettemin kon de katalysator herkregen worden wat wijst naar een reversibel fenomeen. Een numerieke studie van de reacties op het oppervlakte van de post- treated cobalt oxide laag is uitgevoerd met behulp van de Comsol Multipysics software. Het model dat gebruikt werd om de afbraak van pyrodruivenzuur in een katalystische microreactor was niet in staat om de experimentele data volledig te fitten. Dit kwam vanwege het gebrek aan informatie over de verschillende reactieconstanten van de intermediairen van de katalytische ozonisatie.

Deze informatie zou verkregen kunnen worden door gebruik te maken van de Coherente Anti-Stokes Raman Spectroscopie techniek (CARS) als in-situ analytisch gereedschap. In de toekomst zouden resultaten verkregen via de CARS techniek kunnen leiden tot een efficiënte methode die het mogelijk maakt de relevantie en de richting te voorspellen van toekomstige verbeteringsstrategieën van gekatalyseerde chemische reacties.

18 Résumé Résumé

Un aspect clé permettant de surmonter les défis énergétiques et environnementaux est d'améliorer l'efficacité des nouveaux procédés et ceux déjà existants. De nos jours, en raison des préoccupations environnementales, tous les produits chimiques majeurs se font par des procédés catalytiques dans laquelle la catalyse hétérogène joue un rôle très actif. Afin de développer des procédés catalytiques plus efficaces, une meilleure compréhension des cinétiques de réaction est nécessaire. Dans le domaine du traitement des eaux usées, l'ozonation catalytique est en un exemple typique.

Par conséquent, en tant qu’outils analytiques innovants, les microréacteurs catalytiques peuvent être utilisés pour la détermination de la cinétique de l'ozonation catalytique. Dans cette thèse, des microréacteurs catalytiques ont été élaborés à l'aide de procédés plasma à basse pression pour le dépôt d'une sous- couche de silice, puis pour le dépôt et l'activation du catalyseur sur des matériaux polymères tels que le PDMS, la NOA, le THV et le COC.

En plus d’afficher un prototypage rapide, les études de vieillissement ont montré la très bonne stabilité du COC en termes d’hydrophilicité à la suite d’un stockage dans l'air et de l'eau. Le dépôt et l'activation de catalyseurs à base d'oxydes de fer et de cobalt ont démontré l'importance de l'étape du post-traitement par voie plasma.

Les mesures de l’efficacité de ces catalyseurs ont été effectuées en utilisant de l'acide pyruvique, composé réfractaire à l’ozone, en tant que polluant modèle. Dans le cas de la couche d'oxyde de fer, les mesures HPLC ont montré l'inactivité de celui-ci par rapport à la couche d'oxyde de cobalt. Avec la couche d'oxyde de cobalt déposée par plasma, 20% de dégradation supplémentaire a été constaté et l'effet a été doublé (40%) lorsque la couche s’est vue post-traitée par un plasma d'argon, ce qui indique un effet synergique entre la morphologie de la surface de la couche mince et le polluant modèle.

19 Résumé

Cependant, une diminution de l'activité catalytique de la couche d'oxyde de cobalt post-traitée a pu être observée indiquant une désactivation des sites actifs par les produits intermédiaires générés pendant l'ozonation catalytique. Néanmoins, le catalyseur a pu être régénéré par l’utilisation d’eau ozonée, traduisant un phénomène réversible.

Une simulation numérique portant sur les réactions ayant lieu sur la surface de la couche d'oxyde de cobalt post-traitée a été réalisée en utilisant le logiciel Comsol Multiphysics. Le modèle utilisé pour décrire la dégradation de l'acide pyruvique dans un microréacteur catalytique s’est partiellement approchée des données expérimentales en raison du manque de données concernant les différentes constantes de réactions des espèces intermédiaires.

L'accès à ces constantes cinétiques pourrait être atteint grâce à l'utilisation de la technique de spectroscopie Raman Anti-Stokes Cohérente (technique CARS) en tant qu’outil d'analyse en temps réel. En perspectives, les résultats qui seront obtenus en utilisant la technique CARS conduiront à l’élaboration d’un outil efficace qui pourrait prédire la pertinence et les futures stratégies d'amélioration sur des réactions chimiques catalysées.

20 General Introduction General Introduction

This thesis aims at developing a catalytic microreactor as an innovative analytical tool for the determination of kinetics of catalytic ozonation in the field of wastewater treatment.

One of the objectives of this study is to develop a catalytic microfluidic device using low pressure plasma processes for the deposition and activation of the catalyst directly into a microchannel. The catalytic activity of the plasma deposited catalysts is further assessed in catalytic ozonation of a probe pollutant.

The second objective is to develop a fundamental understanding of the kinetics of a catalytic reaction by confronting the experimental results to computational fluid dynamics (CFD) simulations, in which different kinetic models describing the reaction pathway at the surface of the catalyst are tested.

In the first chapter, recent literature on microreactor technology and catalyst deposition techniques is discussed. The focus is set on materials used in the microfluidics field as well as plasma processes and heterogeneous catalysis applied to advanced oxidation processes.

The second chapter presents the different equipment and experimental protocols as well as the theoretical concepts used to achieve the objectives of the thesis. A first section describes the plasma processes techniques implemented for the deposition and activation of an iron oxide and cobalt oxide based catalysts. A second section describes the experimental set-up implemented for the measurement of the catalytic activity of such coatings. Finally, characterization techniques for the solid materials, i.e., the starting microfluidic materials and deposited catalysts as well as for the collected liquid samples are presented.

The third chapter presents the results obtained after the deposition of a silica-like layer on new classes of polymer-based microfluidic materials, achieved through

21 General Introduction the use of two different plasma processes. The aging study of the plasma coated microfluidic surfaces upon air and water storage is presented and discussed. This study allows the identification of a polymer-based microfluidic material suitable for further deposition of the catalyst.

The fourth chapter mainly focuses on the preparation and characterization of cobalt and iron oxides catalysts using a one-step plasma process. Their respective catalytic performance in an ozonation process regarding the removal rate of pyruvic acid as a probe pollutant is discussed. In addition, the role of the plasma post-treatment as well as the stability of the catalyst and its regeneration is addressed.

In the final chapter, a numerical simulation model containing the experimental data from the previous chapter is presented. A first stationary model reflecting the self-decomposition of ozone in water and further reaction with pyruvic acid is described. This model enables to assess the contribution of indirect ozonation (hydroxyl radicals generated from the self-decomposition of ozone) or direct ozonation (reaction of ozone on pyruvic acid).

Finally, a second time dependent model reflecting the ozonation of the probe pollutant in a cobalt oxide-based catalytic microreactor is elaborated in order to understand the deactivation phenomenon encountered in the previous chapter. This model is composed of a set of reactions describing the adsorption and reaction of pyruvic acid and its intermediates on the catalytic wall of the microreactor in an Eley-Rideal mechanism. Simulation results as well as the reaction kinetics are discussed.

These chapters are followed by a general discussion summarizing the results obtained and perspectives and general recommendations are given at the end of this thesis for future improvement strategies.

The research strategy is summarized in Figure 1:

22 General Introduction

Figure n°1: Research strategy.

24 Chapter 1: Literature Review Chapter 1: Literature Review

1. Introduction Microreactors with sub-millimeter flow channels have been in demand in the chemical and process industries for more than a decade now. They are used for catalyzing reactions with different types of reactants, such as homogeneous organometallic complexes [1,2], heterogeneous solid catalysts [3] or various types of biocatalysts [4-6]. They offer several advantages, such as continuous flow chemistry, isothermal reaction conditions with precise residence time, distinctive thermal and chemical kinetic behavior of reaction systems, etc. As a result various materials and fabrication techniques have been reported by many authors. 2. Microfluidic materials, properties and fabrication techniques 2.1 Glass microreactors A common fabrication method, that was devised earlier, consisted of channels etched on a silicon substrate, with a 1 μm thick silicon nitride membrane acting as the channel cap. Thin film platinum lines were used as flow and temperature sensors [7]. Later a variety of materials were used, including Foturan glass, wafer grade silicon, plastics, and metals [8]. Foturan glass is a photo structured glass manufactured from lithium aluminum silicate that is especially useful for designing microchannels and similar structures having high aspect ratios. An advantage of these reactors is that they allow very rapid mixing, so that high purity products can be obtained by suppressing unwanted side reactions [9]. They also allow efficient thermal transfer with very high heat transfer coefficients of 60,000 W/m2·K [10]. Glass/silicon microreactors, to which connections are soldered, are manufactured using two principal methods: integrated connections and modular connections [11]. In the former method, the connections are attached directly to the substrate using, for example, epoxy glues. This requires the use of pressures up to 137 bars [12] or up to 34 bars if press fittings are used with fused silica capillaries [13].

25 Chapter 1: Literature Review

The disadvantage of this method is that the fabrication process is complicated and results in a comparatively large dead volume that cannot be accessed easily. The modular connection method overcomes these drawbacks by using reusable parts which lowers the pressure requirements to as low as 6 bars and also reduces the dead volume [14,15]. However, this process requires very precise alignment between the different connecting parts and the entire casing containing the microreactor has to be provided with leak proof connections. Microreactors have also been constructed from metallic and glass solders by several groups, for example [16] and [17]. The schematic of a fabrication process using borosilicate glass wafer (Borofloat), silicon and aluminum film is shown in Figure 2 below:

Figure n°2: Manufacture of a silicon/glass substrate microreactor with capillaries soldered on the back side. Source: [11]

2.2 Silicon microsystems Microfluidic devices are also manufactured from wafer-grade silicon, one of the reasons being extensive experience developed by many manufacturers with the micromachining of integrated circuits and micro-electro-mechanical systems. The advantage of using silicon is that it is chemically inert and has a very high melting point; these properties are further improved when the surface is oxidized [18]. Single crystal silicon also has higher thermal conductivity than aluminum and can be integrated with onboard sensors that measure temperature, pressure,

26 Chapter 1: Literature Review flow velocity or other parameters [19]. This close integration allows more precise control of physical processes. The disadvantage of silicon wafers is their high manufacturing cost and also the fact that they cannot be used at very high pressures or temperature. Silicon has a melting point of 1412°C, so the devices have a safe operating envelope up to 1100°C. Although most of such microreactors are meant for use at atmospheric pressure or slightly higher, devices that can withstand pressures up to 140 bars at 80°C have also been developed [20]. Recently, a multi-channeled silicon microreactor has been fabricated using a nested potassium hydroxide etching process and this device has been found to be able to withstand highly unfavorable conditions such as exposure to fluorine vapors [21]. Etched silicon based microreactors have been used to investigate preparation mechanisms of a variety of complex organic molecules. These devices have been found to be ideal for synthesis of mono-disperse microcapsules using emulsions as templates for interfacial polymerizations [22].

2.3 Polymer-based microreactors Plastics and polymers perhaps constitute the largest base of materials from which microreactors are fabricated, due to the many advantages that these materials offer. They are often low cost materials and can be prepared easily, which has led to mass fabrication of such devices using techniques that include hot embossing [23], injection molding [24], Lithographie, Galvanoformung Abformung (LIGA) [25], deep reactive ion etching (DRIE) [26], laser ablation, photolithography, polymer-micelle incarceration [27] and microlamination. These microreactors have many applications, such as the synthesis of high purity chemical products, study of highly exothermic reactions, catalyst screening, synthesis of high- throughput materials, fuel cell fabrications, and others [28,29].

The hot embossing technique is quite simple: it is used for immobilizing palladium complexes on a cyclic olefin copolymer base, for example, polymerized N-[3-(dimethylamino) propyl] methacryl amide [30]. Ultrasonic hot

27 Chapter 1: Literature Review embossing of thermoplastic polymers has several advantages, such as: quick cycle time of a few seconds, low investment requirement, ability to change the polymer base within a few minutes and others [31]. Injection molding has been used to prepare ceramic-like microreactors from inorganic polymers such as polysilsesquioxane (POSS) and polyvinylsilazane (PVSZ). These devices exhibited high thermal stability and resistance to a variety of solvents. They were used to carry out several high temperature or pressure reactions such as the Michaelis-Arbuzov rearrangement reaction (150-170°C), the Wolff-Kishner reduction reaction (200°C), super paramagnetic Fe3O4 nanoparticle synthesis (320°C) and conversion of allyloxy-benzene to 2-allylphenol (250°C, 400 psi) [32].

Another material often used is polydimethylsiloxane (PDMS), which has low fabrication cost, low auto-fluorescence, good optical transparency, and biocompatibility with many biological and chemical reagents [33,34]. Although these microreactor chips are often used for studying polymerase chain reactions, one of the disadvantages of these devices is the formation of air bubbles during the thermocycling stage [35]. The bubbles not only induce uneven temperature profiles by acting as insulators, but they also expand at the denaturation temperature of 95°C and purge the sample out of the reactor [36-38]. Figure 3 shows a microreactor for carrying out polymerase chain reactions and how air bubble formation and expansion can push the sample out of the chamber.

28 Chapter 1: Literature Review

Figure n°3: A microreactor chip fabricated using PDMS (left). Formation of an air bubble and its subsequent expansion at elevated temperature pushes out the sample from the chamber. Source: [35]

Even though PDMS is a versatile and popular material in the fabrication of microreactors, it has a few disadvantages such as low mechanical strength, swelling in the presence of a solvent and inability to control wetting behavior. These deficiencies can be overcome by using novel materials such as Norland Optical Adhesives (NOA 81 and NOA 63) for rapid microreactor prototyping [39,40]. NOA is resistant to solvents such as toluene, biocompatible, and can be cured with UV. Low cost fabrication of multiphase flow devices using NOA have been reported, where the microfluidic channels could be made to exhibit hydrophilic or hydrophobic or hybrid properties [41]. Another such novel material is the thermoplastic copolymer Dyneon THV, which belongs to the family of fluorinated polymers. It has recently been used to manufacture microreactors using the hot embossing method and has demonstrated optical transparency, resistance to chemicals and low surface energy. The fabrication process is also low cost and requires no etching or curing [42]. A disadvantage of THV is that it also, similar to PDMS, undergoes swelling in the presence of some organic solvents.

29 Chapter 1: Literature Review

A number of authors have reported the use of cyclic olefin copolymers (COC) in the fabrication of microreactor devices. COC represents one of the new classes of polymers that are amorphous and have interesting properties such as high glass- transition temperature, good optical transparency, low shrinkage and moisture absorption, and low birefringence values [43]. Performances of microreactors fabricated from COC and a combination of borosilicate (substrate) and fused silica (capillaries) were compared by Deverall et al [44]. The chips were fabricated using UV initiated polymerization and Suzuki-Miyaura conversion percentages of iodo-benzene and 4-tolylborronic acid were compared for the devices. It was found that COC chips had lower yields compared to the silica and borosilicate chips, although both reactor types had very high yields. It was found that surface modification is required in order to attach the capillary walls to the COC surface because the surface is composed of aliphatic carbon atoms; earlier the use of 3-(trimethoxysilyl) propyl methacrylate has been proposed for this purpose [45].

Other methods of attachment have also been explored however, for example, hydrogen abstraction from the surface using benzophenone. This is the process of photografting, which occurs at ultraviolet wavelengths of 250 nm [46]. A single step method for photografting methacrylate functional groups using benzophenone has also been reported, although this process may not be suitable for all commercially available varieties of COC [47]. A method that can be used with the COC grade Topas® 6013, for example, involves a two-step photografting process where a photoinitiator is first grafted and then reacts with a monomer in order to initiate grafting from the substrate wall [48,49]. It has been proposed that this latter method can be improved by using ethanol and methanol as porogens, which decrease the amount of channel shrinkage [50].

2.4 Metal and ceramic-based microsystems Apart from polymers, metals and ceramics are another popular choice when it comes to microreactor fabrication. The techniques used for manufacturing with these materials are mostly abrasion based and include etching, machining, joining

30 Chapter 1: Literature Review and sealing, etc. An exception to such abrasive methods is selective laser melting (SLM) that makes use of computer generated 3D CAD models. Varieties of metals are used, including silver, rhodium, platinum and alloys of copper, titanium, nickel, or aluminum [51]. In case of etching, both dry and wet etching methods are low cost and can be used for fabricating sub-millimeter microchannels. This technique has been standardized in the semiconductor industries and extensively discussed in literature [52-54].The fabrication process usually consists of application of a photosensitive polymer mask material on the metal substrate and photo exposure using a primary mask. The unexposed areas are polymerized and cannot be removed by the etching solvent, forming a mask through which the metal substrate can then be etched. Variations of this process include direct writing on the mask using a laser [55]. Wet etching has the limitation that the aspect ratio cannot be more than 0.5 (depth: width), but dry or laser etching does not have this limitation. In addition, wet etching yields semicircular or elliptic channel geometries dues to the isotropic nature of the process; this is not the case for dry etching. A schematic diagram of a femtosecond pulsed laser sample for creating a photo mask on a metal microreactor is shown in Figure 4.

Figure n°4: Direct laser writing using a femtosecond pulse laser on a metal microreactor sample. The laser creates 110 femtosecond pulses with a wavelength of 800 nm and a repetition rate of 1 kHz, allowing a fabrication resolution of 2-40 μm. Source: [55]

31 Chapter 1: Literature Review

Another metal fabrication process is machining, which is especially useful for metals, such as tantalum, that can withstand corrosive structuring. Micromachining is carried out using spark erosion, either wire sparking or counter-sunk sparking, laser machining, or other precision techniques. The first two methods can be used with any material, but the stability of the metal is a factor in precision machining. Surface quality of the machined microreactor depends on the type of metal used. For example, stainless steel results in surface roughness values of 1-10 μm but brass or copper yield values as low as 30 nm [56,57]. A recent development in this area is selective laser melting, which can be used for many different types of metals and is suitable for rapid prototyping processes. The process is initiated by forming a thin base layer of metal powder and then channeling a computer controlled laser beam to melt the powder. This creates a layer of welding beads following the computerized 3D CAD model. The platform is then lowered, a new layer of metal powder is added, and the process is repeated. This results in a layer-by-layer fabrication of the desired micro structure [58-60].

Ceramic, glass and ceramic-like materials are also used in the fabrication of microreactors. The advantage of these materials is that the devices withstand very high reaction temperatures of 1000°C or higher. They do not exhibit catalytic blind activity and can often be easily integrated with catalytically active materials [61]. Glass is transparent, so it is a useful material for photochemical reactions and also in situations when fluid dynamical and other process parameters need to be inspected using optical fibers [62,63]. The disadvantage of these materials is that they are often expensive and micro fabrication of components using these materials requires specialized technology. High temperature and pressure capable microreactors have a number of applications, such as optical characterization of segmented liquid-liquid flow systems, investigation of supercritical water chemistry [64,65] and nanomaterial synthesis [66].

32 Chapter 1: Literature Review

3. Advanced Oxidation Processes One of the reaction systems that have been investigated using microreactors is a group of advanced oxidation processes (AOPs) that are useful in wastewater treatment. These AOPs are especially useful for removing industrial pollutants such as 1,4-dioxane [67]. They are also useful for transforming and destroying trace constituents, converting them completely to carbon dioxide and mineral acids, because the reactions involve oxidation through hydroxyl radical species. As a result, further processing of residual waste streams is not required [68]. The reactions can also be used to destroy trace constituents that cannot be oxidized completely by conventional oxidants, including constituents that are known to affect the endocrine system [69].

Reactions that yield hydroxyl radicals at room temperature and pressure are, in general, called oxidation processes (AOP). The advantage of AOPs is that these are able to produce high concentrations of hydroxyl radicals (HO•); these are strong oxidants that can completely oxidize most organic compounds into carbon dioxide, water and mineral acids. The HO• radical acts as a reactive electrophile, due to the presence of the unpaired electron, allowing rapid reactions with electron rich organic systems [70]. Since the reactions depend on the concentrations of the HO· species as well as the constituent that is oxidized, these are second order reactions. The second order rate constants for other oxidants are at least 3-4 times lower than those for HO• species, with the latter values for dissolved organic compounds typically reaching 108-109 L/mol·s-1[71]. Some of these rate constant values for common organic trace materials are shown in Table 1 below:

33 Chapter 1: Literature Review

Table n°1: Rate constants for hydroxyl radical in the presence of commonly occurring trace biological matter in aqueous solutions. Source: [72]

Organic Rate Constant Rate Constant Organic Compound Compound (L/mol·s-1) (L/mol·s-1)

Ammonia 9.00 x 107 Hypobromous acid 2.0 x 109

Arsenic 1.0 x 109 Hypoiodous acid 5.6 x 104 trioxide

Bromide ion 1.10 x 1010 Iodide ion 1.10 x 1010

Carbon 2.0 x 106 Iodine 1.10 x 1010 tetrachloride

Chlorate ion 1.00 x 106 Iron 3.2 x 108

Methyl tertiary butyl Chloride ion 4.30 x 109 1.6 x 109 ether (MTBE)

Chloroform 5 x 106 Nitrite ion 1.10 x 1010

N-Dimethyl CN- 7.6 x 109 4 x 108 nitrosamine

AOP reactions proceed through four principal mechanisms: addition of the hydroxyl radical, abstraction of the hydrogen atom, electron transfer and radical combination [73]. Radical addition involves an unsaturated aliphatic or aromatic organic compound and creates a radical compound that can be further oxidized [74]. Hydrogen abstraction is a slower process and creates a radical organic compound. The latter takes part in a chain reaction by reacting with oxygen, creating a peroxyl radical that can in turn react with other organic compounds [75]. The mechanism of electron transfer leads to the generation of higher valence electrons that can create an atom or free radical through oxidation of monovalent negative ions [76]. Radical combination is the process of combination of two or more radicals of a single or different species, resulting in the formation of a stable compound [77]. The mechanism of radical addition is the most common and it is

34 Chapter 1: Literature Review also known as mineralization since it results in the formation of mineral acids and salts.

Many different types of AOPs are known, including the application of ozone, UV and hydrogen peroxide (H2O2) in various combinations; Fenton’s reactions; application of ozone at elevated pH values (8-10 or above); sonolysis; supercritical water oxidation; application of UV and titanium dioxide; pulsed corona discharges; electron beam irradiation; gamma radiation; and electrohydraulic cavitation [78-81]. Not all of these are used commercially; however, the most common processes involve application of ozone, UV and H2O2 in various combinations as well as the Fenton reactions.

Each of these processes has its own application area together with its advantages and disadvantages. For example, in the case of H2O2/UV light combination, H2O2 is very stable and can be stored for long periods. On the other hand, it has poor UV absorption characteristics and requires specially designed reactors. There is also a need for treatment of residual H2O2. In case of H2O2/ozone, the advantage is that water that is opaque to UV can be treated effectively. The disadvantage is that volatile organic matter is stripped from the reactor and the process is costlier than some other common treatment processes. In case of ozone/UV combination, the residual oxidants get rapidly degraded since ozone has a short half-life of approximately 7 minutes; the disadvantage of this process is the stripping of volatile organic components and the requirement for removal of ozone in the off- gas.

Fenton’s process involves use of HO• along with ferrous ion, which is known as Fenton’s reagent. It is used for the removal of organic material such as tetrachloroethylene (PCE) and trichloroethylene (TCE). The reagent is often combined with ozone, UV or H2O2 in order to achieve greater removal efficiency. The process is particularly effective for groundwater steams that contain high iron content. Its disadvantage is that it requires maintenance of low process pH value [82]. Recently the use of ultrasound in AOPs has been shown to be technically

35 Chapter 1: Literature Review feasible, but economic viability of the process is still being studied [83]. Operational parameters for different AOP treatment methods, including concentration levels and catalysts used, are shown in Table 2. A variety of heterogeneous catalysts are used in AOPs because these help increase efficacy of treatment at negligible additional cost. These catalysts range from simple ones such as titanium dioxide and cobalt oxide to complex ones, such as tungstophosphoric acid immobilized on yttrium oxide and zirconia zeolites, copper-chitosan and aluminum oxide in Fenton processes, and perovskites of the type LaTi0.15Cu0.85O3 [84-87]. Activated C-supported Co catalysts can activate the alternate oxidant peroxymonosulfate, which yields sulfate radicals to oxidize many organic compounds [88]. The advantage of these catalysts is that they are less affected by reaction pH and can, therefore, provide a suitable alternative to the Fenton reagents commonly used. Their primary disadvantage is the high toxicity of cobalt ions, which necessitates research for a heterogeneous cobalt catalyst.

36 Chapter 1: Literature Review

Table n°2: Operational parameters, sources and catalysts used in various AOP treatment methods. Reaction Reaction Ultrasound Primary AOP Type Volume Light Source Ozone Source Catalyst Ref concentration source oxidant (mL) 12.6 μM for UV(1.66 254 nm, OSRAM, UV, mg/L), 16.7 μM for HNS 20 W/U H O 5000 NA NA 2 2 - [89] H2O2 + UV UV + H2O2 20W LP mercury 5 mM (2.194 mg/L) vapor lamp 520 kHz, Undatim H O CuO Ultrasound 600 250 μM (32.85 mg/L) NA NA 2 2 [90] ultrasonics, 6.53 mM 1 mg/ml 50W Ozonelab OL-100 model, O Ozone 150 2.5 mg/L NA NA 3 - [91] 36W at 0.75 L/min 20 mg/L flow Ozonelab 254 nm, Philips, OL-100 model, O UV + Ozone 1200 57 μM (19.95 mg/L) PL-L 18WTUV NA 3 - [92] 36W at 0.25 L/min 40 mg/L two lamps flow 254 nm, 100W MP- 20 kHz, Model Ultrasound 0.102 mM (13.40 4100 UV XL2020 NA NA - [93] + UV mg/L) mercury vapor lamp Misonix 330W 450W HANOVIA Dissolved 1% Pt on Photocatalysis 7200 15 mg/L MP-UV mercury NA NA oxygen TiO2 [94] vapor quartz lamp 9 mg/L 0.5 g/L Fenton, 35 kHz H O CuSO Ultrasound + 350 0.67 mM (63.05 mg/L) NA NA 2 2 4 [95] SODEVA 50W 60.5 mM 2.39 mM Fenton

37 Chapter 1: Literature Review

Oxide supported cobalt catalysts and heterogeneous Co3O4 systems have demonstrated high Co2+ leaching properties [96,97] while the Co/MgO system has been shown to have less Co2+ leaching propensity and greater stability [98]. The advantage of using activated carbon support for the catalyst is that carbon acts as a good adsorbent for aqueous as well as gaseous phases and the heterogeneous catalyst does not deactivate easily [85]. For these catalysts, the removal efficiency of contaminants such as phenols has been shown to be dependent on a variety of factors such as catalyst loading, oxidant concentration, and reaction temperature. Cobalt catalysts on graphite can be prepared in situ by heat decomposition of cobalt (II) nitrate and subsequent cobalt oxide crystal growth on graphite surface in the presence of a solvent such as 1-hexanol [99]. Phenol removal rates as functions of catalyst loading rate and temperature are shown in Figure 5 below:

Figure n°5: Rates of phenol removal at different catalyst loadings (left) and different temperatures (right) while using a cobalt/activated carbon catalyst. Source: [85]

Other types of support, such as porous carbon, for cobalt oxide catalysts, have also been studied. In this case, the nanoporous catalyst phase was synthesized by precarbonization of rice husks and subsequent chemical activation at 600°C. The hydroxyl radicals were generated in situ and heterogeneous Fenton oxidation process exhibited 77% COD removal capacity [100]. More complex catalytic systems, such as perovskites, have been used to compare the efficacy of AOPs such as ozone, UV, ozone/UV, H2O2/UV and others while removing pyruvic acid

38 Chapter 1: Literature Review as the contaminant. It was reported that UV combined with hydrogen peroxide led to the quickest removal of the contaminant, but highest degrees of mineralization were achieved for the combination of ozone and UV in the presence of perovskites. Hydrogen peroxide photolysis was also shown to be highly effective in removal, but the process was found to be dependent on the initial reagent concentration [101].

Another category of catalysts are heteropolyoxometalates (POM) that are often added to titanium dioxide based suspensions, colloids or matrices in order to oxidize organic matter using UV radiation [102,103]. Their disadvantage is that POMs are highly soluble in oxygenated solvents [104]. Recently, the use of POMs on zeolite support has been reported, which stabilizes charge-transfer state and transient species such as OH• and also increases the local concentration of the oxidizable substrate [105]. Electrochemical characterization of a POM encapsulated within zeolite matrix has shown that a slow redox reaction occurs in this case, which results in slow charger rates, very good stability of the heterogeneous catalyst system, and high electrolytic activity towards H2O2 reduction. Scanning electron microscopic (SEM) images of native zeolite and molybdophosphoric acid (the POM in this case) are shown in Figure 6 below:

Figure n°6: SEM images of zeolite (left) and POM encapsulated zeolite (right) catalysts. The images show good encapsulation of the POM within the zeolite substrate. Source: [105]

39 Chapter 1: Literature Review

Besides zeolites, a number of groups have reported investigations carried out using hetero polyacids and other catalysts supported on silica [106-108]. Other mesoporous supports that have been investigated include various oxides and alumina, with platinum, palladium or another noble metal acting as the main metal and copper, cobalt, indium or another metal acting as the promoter. In these cases, catalytic reduction is supposed to occur through the bimetallic catalyst combining the active sites [77]. Konova et al. showed that cobalt oxide system with excess O2 supported on alumina (CoOx/Al2O3) exhibits very high catalytic activity towards removal of volatile organic compounds in a wide temperature range of -45°C to 250°C [109]. Organic polymers such as Nafion; synthetic and natural clays such as laponite, bentonite and nontronite; acid modified clays; and porous materials such as various resins also serve as support materials [110-112]. An extensive review of catalysts and reactive species used in AOPs for removal of taste and odor forming compounds can be found in [113].

4. Catalytic Microreactors As discussed earlier, catalytic microreactors have been found to be ideal reaction vessels for carrying out many types of catalytic oxidation studies. For example, - the catalytic reduction of bromate ions (BrO3 ), a commonly found water contaminant, was studied using ruthenium catalysts on a support of carbon nanofibers (CNF) [114]. While ordinary ruthenium based catalysis has a slow kinetic rate constant [115], the use of a porous CNF bed and carrying out the reaction within the flow channels of a silicon microreactor was found to significantly increase the reaction rate, leading to more effective bromate ion removal.

Different types of microreactors have been used for studying catalytic processes, including falling film, mesh, micro-packed, membrane, and others [116-119]. For example, reaction rates during H2O2 oxidation of phenol were reported to be faster than flask while using either packed bed or catalytic wall type microreactors [120]. The packed bed microreactor was prepared by granulating a titanium-containing zeolite catalyst: titanium silicalite-1 (TS-1). The powder was

40 Chapter 1: Literature Review sieved into a size range of 100-150 μm and packed into filter capped reactor tubes having inner diameters between 1 to 2 mm. The catalytic wall microreactor was fabricated from two stainless steel plates, a catalyst plate and a plate with rectangular microchannels of 1 mm cross section. The packed bed reactor was found to accelerate the reaction rate of oxidation while the wall type reactor exhibited extended catalyst life. The regioselectivity of benzenediols was also found to be different for the two designs. Construction of the catalytic wall microreactor is shown in Figure 7 below:

Figure n°7: Schematic diagram showing construction of a catalytic wall microreactor for H2O2 led phenol oxidation. Source: [120]

Membrane microreactors are devices that combine separation through a membrane and a catalyzed reaction within one unit. Their advantage is high mass and heat transfer rates, allowing it to utilize optimal reaction conditions; at the same time lower temperature and less amount of catalyst is possible than other reactor configurations [121,122]. Membrane microreactors are used in a variety of bio- and chemical catalytic reactions by attaching catalysts to membrane pores with zeolites, carbon nanofibers, and metals as support [123-125]. The membranes are formed from nylon, PTFE or ceramic materials within one or more microchannels that can support liquid phases, with aqueous phase being separated from the organic phase [126].

41 Chapter 1: Literature Review

New hybrid membranes can also support three phase solid-liquid-gas reactions, for example, a microreactor in which CNFs were grown as catalyst support on porous stainless steel tubes. The steel-CNF hybrid membrane was used to immobilize palladium catalysts and the assembly was housed within a gas permeable polymer coating. The reactor exhibited high intrinsic nitrite reduction performance, in the absence of excess hydrogen flow, and were considered to be ideal for hydrogenation and other multiphase reactions. Membrane microreactors can be fabricated as plate- or tubular-type. In the former configuration, the microchannels are often constructed from SS-316L or similar stainless steel, or from porous silica and the catalytic membrane is formed on the walls. The fabrication process is usually based on micro-electro-mechanical system (MEMS) technology, but other processes such as LIGA and “dip pen” nanolithography (DPN) have also been developed [127].

Plate type microreactors require careful catalyst incorporation in the microchannels so that pressure drop and flow configurations are taken care of. In case of metals such as palladium, platinum or silver, the catalyst must be deposited as a thin film on a porous oxide layer that has been created using suspension coating [128-130]. The other configuration is tubular in shape, in which the flow microchannels are formed inside the tube and the separation membrane is deposited on the outside [131]. Hollow fiber membrane microreactors can be fabricated using phase inversion or sintering processes from porous aluminum oxide hollow fibers having an inner coating of palladium- aluminum oxide catalyst [132-134]. The microstructures of these devices can be altered by using different suspension compositions and spin parameters, making them suitable for different types of reactions [135,136].

Another catalytic microreactor configuration is the tube-in-tube, which can be used as a gas-liquid contactor with high throughput and good mass transfer efficiency. One such microporous device was used to study the catalytic ozonation of the azodye Acid Red 14 at different ozone flow rates. Efficiencies of decolorization as well as ozone use were found to depend on a range of operating parameters such as initial dye concentration, initial pH, gas volumetric flow rate, 42 Chapter 1: Literature Review annular channel width, micropore size, and others. For example, decolorization was observed to increase when the channel width was decreased and micropore size reduced [137]. The catalytic ozonation process was also used to compare gaseous toluene removal activities of graphene, manganese dioxide nanoparticles, and birnessite type graphene-MnO2 composites. The microreactor used Teflon and stainless steel (SS-316L) tubing and catalyst activity was found to be significantly affected by the MnO2 loading in the composite [138].

5. Catalyst deposition techniques A number of physical and chemical techniques have been developed for depositing structured catalysts on a surface [139]. The former group includes physical vapor deposition (PVD) [140], thermal oxidation [141] and anodic oxidation [142]; the latter group includes sol-gel [143], direct synthesis [144], chemical vapor deposition (CVD) [145] and plasma-enhanced chemical vapor deposition (PEVCD) [146]. Some of these methods are used to pre- or post-treat the substrates, including silicon, steel fibers, ceramics, foams, etc. [147]. Many of these methods can be used for metal-on-oxide catalyst deposition on the microreactor surface, while some are more suited to only oxide deposition and some to direct noble metal deposition on the substrate without an intervening oxide layer. An extensive list of deposition methods, supports/catalysts, substrates and operating conditions can be found in [139]. However, as shown in Figure 8, due to the increasing price of noble metals, preference is generally given to the use of metallic oxides.

43 Chapter 1: Literature Review

Figure n°8: Evolution of the stock exchange prices over the ten last years for noble metals (gold, silver, platinum) versus cobalt price. Source: Infomine.

Substrate pre-treatment is often performed to allow better immobilization of the catalytic layer and increase its operational lifetime. Anodic oxidation is used for aluminum surfaces to create a porous surface layer by applying an electric current to an electrolyte in contact with the aluminum [148]. Another such process is thermal oxidation, which is suitable for FeCrAl substrates, and is carried out at 840-900°C [149,150]. On the other hand, non-oxides such as zeolites are usually deposited through dip-coating, which results in random orientation of the zeolite crystals and is thus useful for adsorption and catalytic activities. Colloidal silica or some other binder is used in this method [151]. Another technique is direct growth of the zeolite layer (such as Sil-1, Al-ZSM5, TS-1) on a silicon microchannel, which results in complete coverage and oriented crystal growth [152]. Similarly, carbon support can be deposited on ceramics, metals and other non-porous structures using methods such as melting-carbonization, polymerization, or CVD [153,154].

44 Chapter 1: Literature Review

The suspension method is often used for depositing catalysts on ceramic monoliths, and requires the catalyst powder, a binder, acid and a solvent (often water) as the ingredients [155]. A closely related method is sol-gel deposition, in which sol formation conditions and ageing time can be controlled to obtain the required degree of branching in oligomer gel [156]. The technique can be used with a variety of precursors, such as Al[OCH(CH3)2]3, Ni(NO3)2·6H2O and o La(NO3)3·6H2O - the substrates are dipped in the sol-gel, dried at 120 C and calcined at 500oC to obtain the final structure [157].

The CVD method is perhaps the most versatile and frequently used one for catalyst deposition on a wide variety of substrates [158,159]. Its advantage is that while the chemical precursors (such as aluminum alkoxide) may be the same as those used in sol-gel methods, CVD does not require the use of a solvent. The deposition rate can be increased by using elevated temperature and low pressure, but the PECVD technique allows quick deposition even at low temperatures. This method can also be used to deposit catalysts on CNTs and powdered substrates [160]. It is a simple process in which metal-organic composite thin films can be easily obtained from organometallic precursors and the formation of agglomerates of nanostructured catalysts can be avoided [161].

A single step, dry process for obtaining CoxOy coatings using PECVD was evaluated by Guyon et al. [162]. Two variants of CVD were used: in one, plasma was created at 40 MHz and accelerated an aqueous solution of cobalt nitrate salt through a convergent nozzle; in the other metal organic PECVD (MO-PECVD) process the precursor used was cobalt carbonyl Co2(CO)8 dissolved in 1-hexene. The latter method was found to result in a highly hydrophilic layer having cobalt oxides and an organic polymer. A similar MO-PECVD process was also used by [163] in order to deposit tricobalt tetraoxide catalyst on polydimethylsiloxane microchannels. The deposited films exhibited cauliflower-like micro-cluster morphology, similar to that found by Guyon et al. [162]. This morphology and other structural characteristics, such as the ratio of Co3+ to Co2+ ions in the deposited film, were found to be increase the stability of the film as well as increase its catalytic oxidation efficiency. 45 Chapter 1: Literature Review

6. Conclusion From this review, it can be concluded that the PECVD process is ideal for the deposition of thin film catalysts such as cobalt oxide. Indeed, the low temperature obtained in low pressure plasma processes makes this technique suitable for functionalization of microfluidic material. Regarding the choice of the microfluidic materials, polymer-based microsystems such as PDMS, NOA 81, THV and COC appear to be the most suitable materials as they offer a low cost and rapid prototyping. These materials will be first investigated by depositing a silica coating using PECVD and by studying the aging of such surfaces upon air and water storage. This silica coating will be used as a support layer in order to increase the specific surface area of the catalyst and therefore enhance the catalytic activity as previously mentioned. Regarding the nature of the catalyst, the efficiency of Co3O4has been demonstrated in the literature. Thus, this low cost catalyst will be deposited and activated using a MO-PECVD process and measurements of the catalytic activity will be performed in a catalytic ozonation process.

46 Chapter 2: Materials and methods Chapter 2: Materials and methods

1. Introduction The studies contained in this thesis were carried out in 2 segments: first, an investigation on the microfluidics materials suitable for the deposition of a support layer with a low pressure plasma process was performed using a home- made device. The second concerns the deposition of a catalyst coating by plasma process and the quantification of the catalytic activity using an advanced oxidation process. Therefore, the plasma process used for the deposition of the silica-like layer and the catalyst as well as the catalytic ozonation process will be described here. The materials and analytical techniques used for the surface characterizations or for the analyses of the collected liquid samples are presented.

2. Microfluidic materials Polymer-based microfluidic materials were investigated as flat substrates in order to assess the stability of the catalytic coating in the MO-PECVD process. Flat substrates of Polydimethylsiloxane (PDMS) were prepared by pouring the monomer (Sylgard, Dow Corning) and hardener (respectively in ratio 9:1) in a 40 x 12 mm petri box. The curing was performed in an oven heated at 70°C for two hours. The PDMS samples were cut into square shapes of 2 x 2cm. Cyclic olefin copolymer (COC) films (Topas grade 6013, Tg = 130°C, 254 μm thickness) were purchased from Topas Advanced Polymers, Extrusion Lab (USA). Polymer sheets of 20 cm x 20 mm x 350 µm dimensions were obtained at 290°C from the COC pellets. The sheets were cut into 2 x 2 cm. Norland Optical Adhesive (NOA) substrates were obtained by pressing a drop of NOA 81 between a flat PDMS stamp and a microscope glass slide. The resin was then UV-cured under 7 seconds of exposure with a 365 nm UV light (Hamamatsu LC8, lamp power of 10 mW/cm²). After reticulation, the NOA covered glass slide was cut into 2 x 2 cm and cleaned with isopropanol.

47 Chapter 2: Materials and methods

THV substrates (Dyneon) grade 500 were graciously provided by the Laboratoire des Macromolécules et Microsystèmes en Biologie et Médecine. A blank bare THV wafer of 5 mm thick was cut into 10 mm x 10 mm square shapes. Polished stainless steel 316L grade plates were purchased from Goodfellow (France). These samples of 15 mm x 15 mm and 1 mm thick were used as substrates for ellipsometry experiments. All substrates were cleaned by immersion in isopropanol or/and acetone solutions followed by a high-pressure, high-purity nitrogen gas drying step.

3. Metal-Organic Plasma Enhanced Chemical Vapour Deposition Process 3.1 Description of the reactor The process used for the elaboration of the catalyst coatings is a home-made low pressure low temperature PECVD process. As shown in Figure 9, the reactor consists in a Pyrex tube of 60 cm of length and 46 mm of internal diameter. The reactor is equipped with a spray nozzle of 5 mm of diameter in order to focus the deposition on the substrate holder. Connected to a matching box, a 13.56 MHz radiofrequency generator is used to generate the plasma in the reactor via external parallel capacitive electrodes placed on the tube. Mass flow meters linked to a controller enable to vary the incoming gas flow. Vacuum is achieved by the use of a rotary vane pump and the pressure is read by means of a Pirani gauge placed at the end of the reactor. Regarding the organo-metallic precursor, it is placed in a flask above the reactor and introduced in droplets by the use of a nebulizer. The nebulizer is separated from the reactor by an electro-valve on which the opening and closing times can be adjusted. By over pressurization, the created droplets are brought in the gas phase via an introduction line which is heated above the boiling temperature of the precursor in order to avoid its condensation.

48 Chapter 2: Materials and methods

Figure n°9: Scheme of the home-made MO-PECVD process used for the deposition of thin films.

3.2 General procedure for the silica-like deposition In order to clean off the surface of the samples from eventual adsorbed atmospheric contaminants, the samples are first exposed to a 200 W argon (150 sccm) and oxygen (150 sccm) plasma treatment during 10 min with a total pressure of 1.4 mbar. After this pre-treatment step, a gas mixture composed of oxygen (300 sccm) and hexamethyldisiloxane (60 sccm) is introduced in the reactor. The organo-silicon vapor is supplied to the reactor chamber from liquid HMDSO (Merk, 98.5%) contained in a cylindrical flask, via a stainless steel line heated at 70°C. A manual valve is used for fine control of the flow. The plasma power and deposition time are respectively of 150 W and 4 mn with a total pressure of 1.7 mbar.

49 Chapter 2: Materials and methods

3.3 General procedure for the catalyst deposition After the deposition of the silica-like layer, a gas mixture composed of oxygen (300 sccm) and argon (300 sccm) is introduced in the reactor along with the organo-metallic precursor.

For the deposition of the cobalt oxide catalyst, 1g of cobalt carbonyl Co2(CO)8 (Strem Chemicals, France) dissolved in 20 mL of hexene is introduced in the flask. Dicobalt octacarbonyl is a metal organic compound solid under standard conditions which decomposes above 52 °C. The precursor reacts readily with hexene under formation of gas bubbles and a color change to deep violet occurs. For the deposition of the iron oxide catalyst, 30 mL of iron pentacarbonyl

Fe(CO)5 (Sigma-Aldrich, France) are introduced in the flask. The introduction line is heated at 70°C and the electro-valve is set at 0.9 seconds for the opening time and closes during 4 seconds. The plasma power and deposition time are respectively of 200 W and 12 mn for a total pressure of 2 mbar. Activation of the catalyst is achieved using argon plasma (300 sccm) at a power of 150 W during 30 mn.

4. Catalytic ozonation process The catalytic ozonation apparatus used in this study is a home-made system as shown in Figure 10. Ozone is delivered through the use of an ozone generator (Type COM-AD-01, Anseros) supplied by an oxygen tank. By setting the ozone generator at 100% of its full power (29 W) and by the use of a mass flow controller (MFC), ozone is sent to a bubbler containing 20 mL of ultrapure water with pyruvic acid (98%, Alfa Aesar) at concentrations ranging from 0.15 to 0.6 mmol·L-1. The ozone concentration in the liquid phase was followed online using optical fibers (Ocean Optics Maya Pro 2000) with an UV detection set at the wavelength of 259 nm (Deuterium lamp). Considering Beer-Lambert’s law (εO3 = 3300 L•mol-1•cm-1) the determined concentration of dissolved ozone was of 0.35 mmol•L-1, when the system reached thermodynamic equilibrium.

50 Chapter 2: Materials and methods

In order to have a controlled flow rate of 3.6 mL•h-1, a syringe pump (KD Scientific - KDS Legato 210) sucks the solution through the channel and 3.6 mL of the treated solution is recovered in the syringe. The ozone reaction was stopped downstream by adding 130 µL of 10 mmol•L-1 tert-Butanol (99.9%, VWR) solution to rapidly consume any unreacted O3. The magnetic stirrer ensures the homogenization between the tert-Butanol and the incoming flow arriving in the syringe. Pyruvic acid concentrations before and after reaction were performed by the use of High Performance Liquid Chromatography (HPLC).

Figure n°10: Scheme of the experimental set-up for catalytic ozonation.

5. Analytical methods 5.1 Surface characterizations 5.1.1 FTIR-ATR Infrared spectroscopy reveals the nature of the chemical bonds present on the surface of a sample. The principle of FTIR relies on the irradiation of the sample with an infrared light source. IR light is sent with a modulator that splits the infrared light into different wavelengths. At the other end of the instrument, a detector measures the amount of absorbed IR light. By data processing, a mathematical function (Fourier transform) is then applied on the collected signal and the absorbed wavelengths can be observed.

The absorbed wavelengths correspond to the vibrational frequencies of the chemical bonds between two atoms. Indeed, when exposed to infrared radiation, 51 Chapter 2: Materials and methods molecules of the sample selectively absorb radiation at given wavelengths which induce their change of dipole moment. The vibrational energy levels of these molecules transfer from the ground state to an excited state. The frequency of the absorption peak is determined by the vibrational energy gap whereas the number of absorption peaks is related to the number of vibrational freedom of the molecules present on the surface of the sample. The intensity of absorption peaks is related to the change of dipole moment and the possibility of the transition of energy levels. In conjunction with infrared spectroscopy, the use of attenuated total reflectance (ATR) enables the analysis of thin films on a substrate without further preparation. Infrared spectra were acquired using a Fourier-Transform Infrared Spectrometer (Cary 660 Spectrometer-Agilent) displayed in Figure 11, with an Attenuated Total Reflectance module (GladiATR-Pike) at wavenumbers ranging from 600 cm-1 to 4000 cm-1 with a resolution of 4 cm-1.

Figure n°11: Picture of the FTIR-ATR apparatus (Agilent).

5.1.2 Water contact angle measurements

Water contact angle (WCA) measurements are performed to assess the wetting properties of a surface. The change of the surface properties of a coated sample in terms of hydrophobicity/hydrophilicity reflects the aging of the studied surface.

52 Chapter 2: Materials and methods

The contact angle method allows access to the free energy of a surface. It allows also indicates the polar or non-polar interactions at the liquid-solid interface. The hydrophilic or hydrophobic character of a surface can be deduced from the measured angle. A low value of θ (θ < 90°) indicates a wettable surface (hydrophilic in case the liquid used is water) whereas a high value (θ > 90°) indicates a nonwettable surface (if the liquid used is water, the surface is said hydrophobic). The contact angle (θ) is measured between the baseline of the drop of a liquid and the tangent at the drop boundary on a flat solid surface as shown in Figure 12. The shape of a drop on a solid surface is conditioned by three parameters:

- γLG, the surface tension between the liquid and the gas

- γSG, the surface tension between the solid and the gas

- γSL, the surface tension between the liquid and the solid

These surface tensions are related by Young-Dupré’s equation:

후퐒퐆 = 후퐒퐋 + 후퐋퐆 ∙ 퐜퐨퐬훉

Figure n°12: Contact angle measurement principle.

In the present thesis, contact angles of MilliQ-water pendant drops on all substrates were measured using a GBX-3S Digidrop-MCAT instrument. Water droplets of 3 μL in volume were released from a syringe above the sample surface. Images of droplet formation, captured using a high-resolution camera, were analyzed using an image analysis software (Digidrop). For each treatment, contact angle measurements were performed at a minimum of 3 locations on each sample and a typical deviation of ± 5° from the mean value was observed. The

53 Chapter 2: Materials and methods reported values in this study correspond to the average of these three measurements.

5.1.3 X-Ray diffraction X-Ray powder diffraction, or XRD, is a technique used to determine the structural properties of crystalline materials. This method enables the determination of intrinsic parameters of a crystalline material such as the crystalline structure, the number of crystalline phases present, the index of the crystalline phase, the mesh refinement, the grain size and orientation, and the deformations of the crystalline network. As shown in Figure 13, XRD consists in sending a monochromatic X-ray at a given wavelength on the surface of the sample. The incident beam is reflected by the reticular planes {h,k,l} of the crystalline sample which are separated by an inter-reticular distance d. The incident angle on the surface of the sample is noted θ. Due to the induced X-ray radiation at a given frequency, an atom starts to vibrate with the same frequency by spreading the radiation in all directions. For atoms arranged in the crystal, the radiation may undergo constructive or destructive interferences according to the direction. Constructive interferences or diffraction peaks are determined by Bragg's law:

ퟐ풅 퐬퐢퐧 휽 = 풏흀

Figure n°13: Principle of the X-Ray diffraction.

54 Chapter 2: Materials and methods

Information on the structure is obtained through analysis of: - The angular position of the diffracted lines, which depend of the geometry of the crystal, the size and shape of the mesh. - The intensity of the diffracted lines, which depend of the type of atoms, their arrangement in the mesh and their crystallographic orientation. - The form of diffracted lines, which depend of the instrument, the size of the particles and their deformation.

In the present thesis, analyses were performed on a Pan Analytical-X Pert Pro (Model X’Pert Pro PW30420) XRD instrument at the wavelength of λ = 1.5418 Å

(Cu Kα1 radiation).

5.2 X-ray photoelectron spectroscopy X-ray photoelectron spectroscopy, or XPS, is a characterization technique used to determine the composition of the extreme surface of a sample (5 to 10 nm in depth). XPS is based on the irradiation of a sample with a monoenergetic X-ray beam. This beam is generated by an electron gun bombarding a metallic target (cathode). Irradiation by photons causes the ionization of the atoms located on the surface of the sample by photoelectric effect. During the relaxation of the excited atoms, electrons are emitted with an energy equal to the difference of energy between the excited state and the steady state. The energy of X-rays is of the same order than the ionization energy of the core electrons. Thus, electrons of the inner shells are mainly ejected. Measurements are performed by analyzing the kinetic energy Ek of the emitted electrons and the characteristic binding energy is calculated according to Einstein’s relation:

E binding = hν – (E kinetic + Ф) where the energy of the photons (hν) depends on the X-ray source and Φ, which is the energy required to extract the electron from the vacuum (considered as negligible), depends on the spectrometer. The energy of the photons depends of the nature of the cathode which can be aluminum (monochromatic) or magnesium

55 Chapter 2: Materials and methods

(polychromatic) with photon energies of respectively, 1486.6 eV and 1253.6 eV. The principle of X-ray photoelectron spectroscopy is presented in Figure 14. Generally, analyses of a sample are performed in two phases: first, a quick analysis called “survey” scans (3 passes) the surface using a wide range of energy (from 0 to 1400 eV). The result is a characteristic spectrum of the atoms present on the surface of the sample and relative atomic concentrations of the studied layer can be determined. Thus, pre-identification of specific peaks of desired elements can be carried out in order to obtain, in a second step, a high resolution spectrum. A high-resolution spectrum can identify and quantify the chemical bonds between atoms, by working on a low energy range (typically around 10 eV) in order to increase the resolution of the signal.

Figure n°14: Principle of X-Ray photoelectron spectroscopy.

Analyses in the present study were performed by Pascale Chevalier at the LBB laboratory (University of Laval, Quebec) with a PHI 5600 XPS spectrophotometer (Physical Electronics, Eden Prairie, MN, USA). Survey and high resolution spectra were acquired using the Kα line of a monochromatic Al (1486.6 eV) and Mg source, respectively, operated at 200 W. Analyses were

56 Chapter 2: Materials and methods performed with a 45° angle from the surface. Survey spectra were acquired from 0 to 1400 eV for 15 min. Curve fitting for the high resolution peaks was completed using XPS PEAK Software (version 4.1) by means of a least squares peak fitting procedure using a Gaussian-Lorentzian function and a Shirley baseline fitting. For each sample, three different spots were analyzed and the analyzed surface was 0.8 mm².

5.2.1 Scanning Electron Microscopy Scanning electron microscopy, or SEM, is a non-destructive analytical technique. The principle is based on the electron-matter interaction and allows obtaining images of a sample in two or three dimensions. This technique uses the transmission of secondary electrons or backscattered electrons from the material. Contrasts of the chemical composition of a material in addition to its morphology can be seen.

SEM is considered as a non-destructive method because it does not alter the surface of the sample. However, it is often necessary to deposit a thin layer of a conductive material (carbon or gold for example) on the sample in order to improve its electrical properties. An electron incident beam scans the sample surface, giving rise to a spectrum of particles or radiations: secondary electrons, backscattered electrons, Auger electrons or X-rays. More specifically the secondary electrons are created in a shock between the primary electrons contained in the beam and the atoms of the sample: a primary electron can transfer part of its energy to a weak bounded electron in the conduction band of the atom, leading to ionization due to its ejection. These electrons usually have a low energy (about 50 eV) and come from the superficial surface layers of the sample (approximately 10 first nanometers). Compared to secondary electrons, backscattered electrons have a higher energy (up to 50 keV) and come from a higher depth of the sample. These electrons are sensitive to the atomic number of the atoms in the sample: heavier atoms (those having a large number of protons) will emit more electrons than light atoms. This feature is used for the backscattered electrons analysis. Regions formed of atoms 57 Chapter 2: Materials and methods with a high atomic number will appear brighter than others: this is the phase contrast.

SEM was used in the present thesis to assess the morphology of the plasma deposited coatings. A Zeiss Ultra microscope 55 equipped with a thermally assisted field effect electron emission gun was used. This device, available in the Laboratory Interfaces and Electrochemical Systems (LISE) of the University Pierre et Marie Curie (Paris, France), provides images with a resolution down to a nanometer. The field emission gun uses a metallic cathode and applies a voltage ranging from 2 to 7 kV between the tip and the anode. This allows the production of a very high electric field (about 107 V·cm-1) at the end of the cathode.

5.2.2 Transmission Electron Microscopy Transmission Electron Microscopy (or TEM) is an analytical method used to characterize the structure and chemical composition of a solid sample. TEM consists in positioning an ultra-thin sample on a mesh grid in the middle of the trajectory of an electron beam derived from the cathode. Electrons are accelerated by the means of the electromagnetic lenses and create a monokinetic beam that interacts with the sample. The transmitted electrons constitute signals which are measured by the detector and form an image is formed. The image is magnified and focused onto a CCD camera. The input of transmission electron microscopes to optical ones comes from the wavelength of the accelerated electrons (about 1 picometer) in comparison to the wavelength of the photons of visible light (500 to 800 nm) in optical microscopes, which leads to a better observation of thin films for examples. A JEOL JEM 100CX II instrument was used in the present thesis with a Keenview CCD camera and the source was a tungsten filament with an applied tension of 100kV.

58 Chapter 2: Materials and methods

5.2.3 Specific surface area measurements The specific surface area (SSA) of a catalyst is crucial information required in the catalysis field. Specific surface area has to be differentiated from the geometric surface which doesn’t take the porosity of the catalyst into account. SSA represents the total available surface on which a molecule can adsorb and can be measured using the BET (from the name of its inventors: Brunauer, Emmet and Teller) adsorption theory.

The BET method consists in measuring the volume of a gas at the surface of a solid and determining the specific surface area of the solid in square meters per gram of catalyst. The sample is introduced into a capillary tube and undergoes a first desorption and heating step in order to remove any adsorbed atmospheric contaminants, and then the measurement step itself which is usually carried out at the temperature of liquid nitrogen in order to obtain detectable amounts of adsorption.

Based on a kinetic model that considers the surface of a solid like a network of adsorption sites or multilayer, the BET theory is an extension of Irving Langmuir’s monolayer adsorption theory that relies on the following hypotheses: - All surface sites have the same adsorption energy for the adsorbate (nitrogen or krypton gas) and each active site can be occupied only by one particle. - The adsorption of a molecule on one site occurs independently of adsorption at neighboring sites. - Adsorbates form a monolayer.

The BET theory extends Langmuir’s theory to multilayer adsorption with the following assumptions: - Gas molecules physically adsorb on a solid in infinite layers - There is no interactions between each adsorption layer - Langmuir’s theory can be applied to each layer

59 Chapter 2: Materials and methods

These assumptions lead to the following isotherm equation:

ퟏ 풄 − ퟏ 푷 ퟏ 푷풐 = ( ) + 푽 [( ) − ퟏ] 푽풎 풄 푷풐 푽풎 풄 풂 푷

Where:

 Va: volume of adsorbate at standard temperature and pressure, in mL

 P/P0: relative pressure of adsorbate  Vm: volume of adsorbate as a monolayer on the surface of the sample, in mL  C: BET constant

The equation of his isotherm requires linear plotting of 1/(푊[(푃표/푃) − 1]) against푃/푃표:

풄−ퟏ ퟏ With 풔 = and 풊 = where s represents the slope of the curve and i, the y- 푽풎 풄 푽풎 풄 intercept.

From these values, the volume of adsorbed gas and the BET constant can be deduced receptively from the slope and the y-intercept as:

풔 풄 = + ퟏ 풊 ퟏ 푽 = 풎 풔 + 풊 The total specific surface area can be calculated as following:

푽풎 × 푵 × 푺 푺 = 푨 풕풐풕풂풍 푽 푺 푺 = 풕풐풕풂풍 푩푬푻 풎 With:

23 -1  NA: Avogadro’s constant (6,022·10 mol )  S: adsorption surface of the sample in cm²

60 Chapter 2: Materials and methods

 V: molar volume of the adsorbate at standard temperature and pressure, in mL•mol  m: mass of the catalyst

Specific surface area measurements were performed by Philippe Barboux at the LCMCP (IRCP, Paris) using a Belsorp-Max apparatus (Figure 15) and by using Krypton as the adsorbate.

Figure n°15: Picture of the Belsorp-Max apparatus (Bel Japan).

5.3 Analytical methods used for the liquid phases 5.3.1 High Performance Liquid Chromatography High Performance Liquid Chromatography (HPLC) is a separation technique that enables the identification and quantification of the components in a liquid mixture. The separation relies on the relative polar affinity of the components with either the mobile phase (eluent) or the stationary phase (column).

The HPLC system is based on three main parts: - The stationary phase or the HPLC column, which is generally made of a silica gel that might be rendered non-polar by grafting hydrophobic groups. - The mobile phase or the eluent, which can be polar or non-polar depending on the type of column used. Generally, a non-polar column is used with a polar mobile phase and reversely. The mobile phase moves through the

61 Chapter 2: Materials and methods

chromatography column (the stationary phase) where the sample interacts with the stationary phase and is separated. - The detector: once the components are separated, the identification can be done by using a UV-visible detector. The analysis is done at wavelength in accordance with the desired products. Therefore, a reference sample must be injected in order to determine the corresponding wavelength and retention time.

A normal-phase chromatography was used in the present thesis (the stationary phase is polar in opposition to a reverse-phase chromatography where the stationary phase is non-polar). The column is a 4.6 x 250 mm Supelcogel H column for organic acids (Sigma-Aldrich, France) containing sulfonated polystyrene and divinylbenzene particles (9 µm of diameter). The mobile phase is MilliQ water acidified with 0.1 % of orthophosphoric acid. Injections were performed using an automatic injector (Jasco) with an injected volume of 20 µL. The parameters used for the HPLC analysis are summarized in Table 3.

Table n° 3: Parameters used for the HPLC analyzes.

Mode Isocratic

Flow rate 0.1 mL•mn-1

Pressure 30 bars

Wavelength 210 nm

Analysis time 40 mn

62 Chapter 2: Materials and methods

Figure n°16: Calibration curve for pyruvic acid (left). Picture of the HPLC apparatus (right).

5.3.2 Flame Atomic Absorption Spectroscopy Flame Atomic Absorption Spectroscopy, or FAAS, enables the qualitative and quantitative measurements of metallic species present in solution. This analytical technique uses optical methods (Beer-Lambert’s law) and is based on the absorption of photons emitted by ground state atoms. As shown in Figure 17, the sample is introduced into a graphite furnace (atomizer) and undergoes three steps: drying, decomposition and atomization. During the atomizing step (combustion), the furnace reaches high temperatures (2100 °C to 2800 °C). The light source consists in a hollow cathode made of the element being determined that emits a light beam at a specific wavelength of the desired item. As the sample passes through the flame, the light beam passes through the graphite furnace and is collected by a detector. Atoms in the ground state absorb light at the characteristic wavelengths emitted by the source. The amount of absorbed light is measured and converted into an electrical signal. This signal is then processed and the amount of the element (concentration of the element in the sample) is determined. Prior to measurements, calibration is required according to the desired metallic element. This technique uses mainly samples in liquid form and provides detection limits down to the μg·L-1 order (or ppb for some metallic elements).

63 Chapter 2: Materials and methods

In the present thesis, samples collected after catalytic ozonation in microreactors were analyzed using a Thermo Scientific unit (model 989 QZ Sollar). Analyses of cobalt and iron in solution were performed by Emmanuel Aubry at the BIOEMCO laboratory (UPMC, Paris).

Figure n°17: Scheme of the FAAS apparatus.

5.3.3 pH-metry In order to follow the variation of the concentration of protons (H+), pH of the initial and treated solutions were measured using a Mettler Toledo pH meter with an Inlab 418 electrode.

64 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials

1. Introduction Microfluidic devices have become increasingly attractive for various fields such as catalysis [163], chemical synthesis [164], biology and physics due to their reduced size and high surface area-to-volume ratio, which are characteristics that enable fast analysis, short reaction times and the potential for patterning [165]. Robust application of microfluidics requires surfaces that can be hydrophilic or hydrophobic, patterned, optically transparent and charged. As a result, surface properties play a major role in determining the microfluidic chip characteristics. Currently, the most commonly used microfluidics material is polydimethylsiloxane (PDMS): -[O-Si(CH3)2]n-. Despite exhibiting several advantageous properties, this material is extremely hydrophobic and difficult to modify. Although many studies have shown that PDMS surfaces can be modified using various techniques, the instability of the grafted or deposited coatings poses a major challenge for the fabrication of functional microchips [166]. Therefore, other polymers - especially thermoplastics displaying enhanced fabrication possibilities compared to PDMS - have been identified as alternative materials that might provide more robust coated surfaces. NOA 81 (Norland Optical Adhesive) is a thiolene-based photocurable resin that displays outstanding mechanical, chemical and optical properties for applications in the field of microfluidics [39]. NOA is a dissolvable, safe and biocompatible material with intrinsic hydrophilicity that causes microchannel filling via capillary forces [41]. Aside from the hydrophilic enhancement of the NOA 81 surface by oxygen plasma, no other report of plasma surface changes was encountered in the literature.

65 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials COC (Cyclic Olefin Copolymer) is a thermoplastic copolymer composed of norbornene (C7H10) and ethylene (H2C=CH2) groups. Other polymeric materials such as polymethyl methacrylate or polystyrene are often used as starting microfluidic materials, but the COC grade 6013S has the unique advantage of exhibiting a high glass transition temperature (Tg = 130 °C), which allows it to be shaped easily and to be coated using conventional vacuum techniques without any surface processing [167]. Moreover, COC is resistant to hydrolysis, acids, alkalis, and polar solvents. Thus, this material can withstand several processing steps including photo-patterning, wet etching and surface functionalization [168].

THV, a terpolymer composed of Tetrafluoroethylene (F2C=CF2),

Hexafluoropropylene (F2C=CF-CF3) and Vinylidene (H2C=CF2), belongs to a new class of fluorinated microfluidic materials that are well suited for droplet and organic solvent microfluidics, and that exhibits many advantages such as a low surface energy and a high resistance to chemicals [42]. The low melting temperature (165°C) of THV makes it an appropriate candidate for rapid prototyping. However, this emerging substrate lacks chemical reactivity: the THV polymer backbone presents aliphatic carbon and fluorine atoms that make the surface functionalization of this material difficult with conventional surface modification techniques. Although various techniques of polymer surface modifications by conventional or plasma processes have been reported in the literature [169-177], they lack specificity and they are not suitable for microfluidic devices because they rarely allow for selective patterning on the surface of the microchannel for purposes such as liquid phase separation [178].

Superhydrophilic wetting properties have signiﬁcant eﬀects on liquid behavior on a surface [179-182]. Therefore, control of surface energy over the microchannel walls is very desirable for enhancing the wetting properties of microchannel surfaces. Indeed, a hydrophilic layer deposition on polymers allows spatial control over water droplets generated in microsystems [183] such as water-in-oil- in-water (w/o/w) emulsions [184].

66 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials Such hydrophilic coatings can be achieved through the use of low pressure plasma deposition techniques. Plasma processes are dry, clean and environmentally-friendly and already have been used successfully for the deposition of hydrophilic coatings without exposing the materials to a high- temperature environment [185]. Among these dry processes, plasma-enhanced chemical vapor deposition (PECVD) techniques are promising for tailoring the surface properties of microfluidics materials depending to the desired application.

The effective plasma modification of polymeric material surfaces with long- enduring hydrophilic coatings invites development of microfluidics to address a range of challenges in the fields of chemistry, biology and medical technology. Moreover, depending on the nature of the chemical or gaseous precursors used for the deposition, several chemical functional groups can be grafted onto the surface depending on the desired application. For instance, PECVD based on organosilicon precursors, is one of the most appealing methods to deposit both inorganic SiO2 silica-like and SiOxCyHz polymer-like films.

Organosilicon-oxygen mixtures are widely used for thin film deposition and they are of particular interest in gas sensor applications [186] and in applications as barrier films for food [187], corrosion protection layer [188], pharmaceutical packaging for biocompatible materials [189], dielectric layers [190], microelectronics and optics [191,192]. Among the different organosilicon precursors available, tetraethoxysilane (TEOS), tetramethoxysilane (TMOS) tetramethyldisiloxane (TMDSO), hexamethyldisilazane (HMDS) and hexamethyldisiloxane (HMDSO) have been extensively studied [193-201]. Compared to silane compounds, these organosilicon precursors are non-toxic and non-explosive monomers to handle.

HMDSO is the most commonly used silane compound for silica-like deposition due to its low toxicity, low cost, and a higher room temperature vapor pressure (48 Torr) than TEOS (4 Torr). Used as a precursor in low-pressure plasmas with

67 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials oxygen, HMDSO leads to the deposition of SiO2-like layers, following this simplified chemical oxidation equation:

(CH3)3Si-O-Si (CH3)3+ 12 O2→2 SiO2+ 6 CO2+ 9 H2O [202]

Of course, this equation do not represent the plasma chemistry that occurs during the silica-like deposition process, although a recent paper proposed several dissociation processes for the fragmentation of the HMDSO monomer in a low- pressure reactor with oxygen plasma [203]. No aging studies of silica-like coated microfluidic materials have been reported using low-pressure plasma techniques. In this study, three new classes of polymers - COC, NOA 81 and THV - were investigated using two different dry processes, PECVD and sputtering, in order to produce adherent silica-like thin films. The stability of these films upon aging in air and water storage determines the most suitable material for implementation/application in a catalytic microreactor treated/fabricated by a dry plasma process.

2. Materials and methods 2.1 Chemicals Hexamethyldisiloxane (99%, HMDSO) was purchased from Merck (Germany). Acetone, isopropanol and ethanol were HPLC-grade from Sigma-Aldrich (France). All chemical reagents were used without further purification.

2.2 Materials

Cyclic olefin copolymer (COC) films (Topas® grade 6013, Tg = 130°C, 254 μm thickness) were purchased from Topas Advanced Polymers, Extrusion Lab (USA). Polymer sheets of dimensions 20 mm x 20 mm x 350 µm were obtained at 290°C from the COC pellets. The sheets were then cleaned with isopropanol and cut for further use as polymer substrates for plasma treatments.

NOA 81 substrates were obtained by pressing a drop of liquid NOA 81 monomer between a flat PDMS stamp and a glass microscope slide. The resin was then

68 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials UV-cured for 7 seconds under a 365 nm UV light (Hamamatsu LC8, lamp power of 10 mW/cm²). THV substrates (Dyneon®) grade 500 were provided by the Laboratoire des Macromolécules et Microsystèmes en Biologie et Médecine (Institut Curie, UMR168, CNRS). A 5 mm thick, blank THV wafer was cut into 10 mm x 10 mm squares for the experiment. Polished stainless steel 316L grade plates were purchased from Goodfellow (France). Samples of 15 mm x 15 mm and 1 mm thick were used as substrates for ellipsometry experiments. All substrates were cleaned by immersion in ethanol and acetone solutions followed by a high-pressure, high-purity nitrogen gas drying step.

3. General procedure for the silica-like deposition using plasma processes 3.1 Sputtered silica-like thin film deposition The deposition of the silica-like thin films was performed using a Hybrid Deposition System Hyrenique Series HSPT520 (13.56 MHz) apparatus provided by Plasmionique (Quebec, Canada). Based on prior work in our lab [168], the surfaces of the substrates were fixed on a rotating substrate holder to produce a uniform deposition. In order to remove low molecular weight fragments present on the surface of the COC, PDMS and NOA samples, the substrates were cleaned by an argon and oxygen plasma as a pre-treatment step. The chamber was first -6 pumped down to a pressure of 5.10 Torr. A mixture of Ar and O2 (2:1 mass flow ratio, Ar 20 sccm and O2 10 sccm) was then introduced into the chamber to reach the desired working pressure of 6 mTorr for 5 min at an applied power of 200 W. The pretreatment step discussed above cannot be carried out on the THV substrate due to its intrinsic chemical properties. In order to achieve subsequent functionalization, fluorinated polymers require a specific pre-treatment step consisting of nitrogen and hydrogen plasma. By selecting the appropriate nitrogen/hydrogen ratio, defluorination of the surface of fluorinated polymers can be achieved [204-206].

69 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials Therefore, based on previous work in our laboratory [207], optimal parameters for the pre-treatment step of the THV samples were determined to consist of a mixture of N2 and H2 (1/3:2/3 mass flow ratio, N2 10 sccm and H2 30 sccm) introduced into the reactor, at a working pressure of 6 mTorr for 5 min with an applied power of 75 W. After each pretreatment step, the silica-like layer was deposited by RF magnetron sputtering deposition from a 2’’ SiO2 99.995% sputtering target (Kurt J. Lesker Company, France). This target was chosen since sputtered films are known to have a composition close to this source material. The plasma was generated at an applied power of 150 W with a mixture of Ar and O2 (mass flow ratio of 6:1, Ar

30 sccm and O2 5 sccm) for a duration of 60 min.

3.2 Deposition of SiO2-like layer by PECVD Silicon oxide-like thin films were deposited using a homemade low-pressure plasma reactor. The RF plasma (13.56 MHz) was generated inside a Pyrex tube (46 mm internal diameter, 600 mm length) using capacitive coupled external electrodes [162]. PDMS, COC and NOA samples were first cleaned to remove adsorbed atmospheric contaminants using the pretreatment step described previously, via exposure to a 200 W argon (150 sccm) /oxygen (150 sccm) plasma treatment for 5 min at a total pressure of 1.4 mbar. THV substrates were pretreated by exposure to a 150 W nitrogen (75 sccm) / hydrogen (225 sccm) for 5 min at a total pressure of 1.4 mbar.

After each pretreatment step, a gas mixture composed of oxygen (300 sccm) and hexamethyldisiloxane (60 sccm) was introduced into the reactor. The monomer vapors were supplied to the reactor chamber from liquid hexamethyldisiloxane (Merk, 98.5%) contained in a cylindrical flask, via a heated stainless steel line at 70 °C. A manual valve was used for fine control of the flow. Based on previous work in our lab [208] and from the literature [209,210], the optimal parameters chosen for the silica-like deposition were 150 W and 4 min respectively for the plasma power and deposition time, at a total pressure of 1.7 mbar.

70 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials 4. Results and discussion 4.1 Aging of silica-like coated substrates using water contact angle measurements

4.1.1 Effect of the Ar/O2 plasma pre-treatment step The initial water contact angles (WCA) on the blank surfaces were equal to 60°, 90°, 100°, and 110° respectively for NOA 81, COC, THV and PDMS. As shown in Table 4, all surfaces were hydrophilic after the plasma pretreatment step. However, after 15 days of aging in air, the WCA values rapidly increased for all substrates to near the initial values of the blank substrates, whereas the substrates placed in water storage exhibited a slower recovery rate. This temporary functionalization on the uppermost surface layer was probably due to oxygen radicals generated in the plasma. Once present on these surfaces, the radicals bond with hydrogen derived from either residual air present in the reactor or from ambient air upon exposure to the atmosphere. Both argon and oxygen have been used extensively as plasmagen gases for improving the wettability of polymeric surfaces by inducing the formation of hydroxide groups on the surface, which enhances the hydrophilicity of the material [174, 177, 211]. Azioune et al. [212] noted that using argon plasma as a cleaning step leads to the removal of the contaminated layer, which favors the appearance of oxides. This contaminated layer also decomposes when it reacts with oxygen plasma, and the decomposition leads to production of carbon dioxide and water by the following reaction:

CxHyOz + (O2*, O)  CO2 + H2O

71 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials Table n°4: Evolution of water contact angles after plasma pre-treatment, in air and water storage. (PECVD: 150 W plasma discharge power, 4 min deposition time, 150 sccm oxygen flow rate and 0.3 mbar precursor pressure; sputtering: 150 W plasma discharge power, 60 min deposition time, 30 sccm Ar flow rate and

150 sccm O2 flow rate).

Water contact angle values After plasma Substrate pre-treatment Blank Aging medium After one day After 15 days step using PECVD Air storage 34° ± 3° 55° ± 3° COC 90° ± 3° <10° a) Water storage 32° ± 4° 65° ± 5° Air storage 38° ± 3° 80° ± 3° THV 100° ± 3° 35° ± 3° Water storage 30° ± 5° 75° ± 3° Air storage 42° ± 3° 55° ± 3° NOA 60° ± 3° 33° ± 3° Water storage 41° ± 3° 60° ± 4° Air storage 120° ± 10° 110° ± 6° PDMS 110° ± 3° 27° ± 3° Water storage 43° ± 3° 52° ± 3°

Water contact angle values After plasma Substrate pre-treatment Blank step in Aging medium After one day After 15 days sputtering reactor Air storage 53° ± 3° 73° ± 3° COC 90°± 3° <10° a) Water storage 32° ± 3° 80° ± 3° Air storage 46° ± 4° 88° ± 3° THV 100° ± 3° 40° ± 3° Water storage 40° ± 4° 79° ± 3° Air storage 20° ± 5° 40° ± 3° NOA 60° ± 3° <10° a) Water storage 13° ± 3° 55° ± 3° Air storage 105° ± 7° 110° ± 6° PDMS 110° ± 3° 33° ± 3° Water storage 42° ± 3° 65° ± 3° a) Superhydrophilic

The produced water molecules are then activated by the plasma, thus, favoring the formation of hydroxides.

72 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials The observed initial hydrophilicity is probably due to increase in the surface roughness following the pretreatment step due to etching at high plasma power. Roy et al. [213] evaluated nitrogen as a plasmagen gas to demonstrate that the surface of treated COC samples exhibits fewer surface roughness effects, whereas the use of argon or oxygen leads to high surface roughness and thus increased hydrophilicity relative to the blank substrate. However, in the case of PDMS, measurements clearly indicate that the surface recovers its initial hydrophobicity after both dry processes in air aging. This hydrophobic recovery has been described thoroughly in the literature [214-217] and is due to: 1°) the migration of low molecular weight fragments from the bulk to the surface, 2°) the reorientation of polar groups from the surface to the bulk phase or the reorientation of non-polar groups from the bulk to the surface, 3°) the condensation of surface hydroxyl groups. The hydrophilic character of PDMS samples stored in water seems to be retained. However, Lawton et al. [218] showed that 1°) the hydrophobic recovery of PDMS eventually occurs regardless of the storage environment and that 2°) the thickness of the layer probably plays a major role in hydrophobic recovery.

The surface of THV was rendered hydrophilic by abstraction of fluorine atoms from its surface by hydrogen and grafting of polar groups in parallel, as described by Egitto et al. [219]. In water storage, the hydrophobic recovery of the THV substrate was less extreme since more polar groups were incorporated onto the surface, as has been demonstrated by Wilson et al. [220]. Overall, though -like all substrates- the bare, pretreated surface of the THV substrate tends to recover its hydrophobicity.

Therefore, an adapted surface functionalization is crucial in order to obtain a long-lasting hydrophilicity for the surfaces of these microfluidic materials; the specific functionalization investigated here consists of a silica-like coating. 4.1.2 Effect of the silica-like coating As shown in Tables 5-7, directly following the plasma deposition, the silica-like coated COC, NOA and THV substrates became hydrophilic using both the

73 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials PECVD and sputtering techniques, with measured WCA values less than 10°. The drastic change in the surface wettability was attributed to the silica-like layer deposition. In order to assess the stability of the deposited layer, an aging study was performed by storing the silica-like coated samples in air and water for several weeks. When the PECVD technique was used to form a silica-like coating on the surface of the NOA 81, COC and THV substrates, the WCA measurements remained under 10° for up to 2 months in air storage, as shown in Figures 19-21.

The coated PDMS substrates underwent hydrophobic recovery after only a few days in air storage, as shown in Table 8, and hydrophobic recovery was slower for storage in water, with WCA values similar to those measured after the Ar/O2 pre-treatment step. In addition to the hydrophobic recovery process described previously, [221] demonstrated that an oxygen plasma treatment does not oxidize the PDMS surface uniformly. Although the coated PDMS substrate retained hydrophilic properties when stored in water, this sample was not studied further due to its limitations for use in microfluidic device fabrication; as such devices must be used in an air environment. The THV substrate coated using PECVD demonstrated a constant level of hydrophilicity in air storage, but the sputtered silica layer on THV showed a slight increase in WCA values in both air and water storage. Compared to the WCA value of 0˚ for COC and NOA directly following the sputtering deposition, the WCA value of coated THV was slightly higher (15˚).

Globally, compared to the WCA measurements performed for samples stored in air, WCA measurements of the COC, NOA and THV samples stored in water clearly increased slightly over time. This phenomenon might be explained by assuming that the adhesion of the silica-like layer is limited due to the low sticking coefficients of these polymers, especially in the case of fluorinated polymers such as THV [222]. Since the coated samples stored in air exhibit stable WCA values, the discrepancy between samples stored in the different media is

74 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials likely due to the physical effect of the medium on the coating. The polar nature of water may enable the polymeric chains beneath the coating to migrate to the surface more easily, for example, catalyzing hydrophobic recovery.

As demonstrated by Purrohit et al. [223], the formation of voids of nanodimensional sizes leads to porosity within the silicon oxide matrix. The increase in the WCA values observed for storage in water after both plasma processes might be explained by this moisture effect, which leads to water penetration [224]. In addition, Waters [225] showed that via diffusion action into these pores, water penetration leads to cracks and in some cases, to further delamination. Indeed, in the case of silica-like films deposited by PECVD, cracks were observed after storage in water, using the SEM technique as shown in Figure 18. Therefore, based on these assumptions, it can be concluded that water penetration phenomenon is amplified when the samples are stored in water and is directly related to the film thickness. In our case, ellipsometric measurements revealed a value of 30 nm of thickness for the sputtered silica-like layer whereas a value of 700 nm was found for the silica-like coating using the PECVD technique.

However, in our case, there is no delamination effect as the WCA values are stable after 60 days. Therefore, an assumption may reside in the fact that a thicker deposited layer may lead to an increase in the number of voids formed within the silica-like layer. Thus, the water penetration effect would lead an increase in the surface roughness due to the cracking of the layer. Indeed, if the penetration depth/film thickness ratio is the same, substrates coated with the PECVD silica- like layer are more likely to undergo a higher surface roughness than the ones coated using the sputtering technique, as the sputtered silica-like layer is thinner than the one deposited using the PECVD technique.

This assumption is in accordance with the measured water contact angles in air and water storage for the COC, NOA and THV substrates, as in a general trend,

75 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials their evolution curves regarding water storage exhibit a higher plateau than the ones stored in air.

(a) as deposited (b) after 15 days of water aging

Figure n°18: SEM images of PECVD silica-like coating deposited on stainless steel substrates (150 W plasma discharge power, 4 min deposition time, 150 sccm oxygen flow rate and 0.3 mbar precursor pressure).

Regarding PDMS, as shown in Figure 22, the sustained hydrophilicity in water storage is a well-known phenomenon that is frequently encountered in literature [226,227].

76 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials

Figure n°19: Evolution of water contact angles of NOA 81 with SiO2-like coatings deposited by PECVD and by sputtering in air and water storage (PECVD: 150 W plasma discharge power, 4 min deposition time, 150 sccm oxygen flow rate and 0.3 mbar precursor pressure. Sputtering: 150 W plasma discharge power, 60 min deposition time, 30 sccm Ar flow rate and 150 sccm O2 flow rate).

77 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials

Table n°5: Water contact angles measurements of NOA 81 with SiO2-like coatings deposited by PECVD and by sputtering in air and water storage. Time After Deposition Before SiO Aging x deposition Deposition 0 days 15 days 30 days 60 days medium (Blank method NOA)

PECVD 60° ± 3° <10° a) <10° a) <10° a) <10° a) Air storage Sputtering 60° ± 3° <10° a) <10° a) <10° a) <10° a)

PECVD 60° ± 3° <10° a) 28° ± 3° 44° ± 3° 45° ± 3° Water storage Sputtering 60° ± 3° <10° a) 24° ± 5° 37° ± 3° 38° ± 5° a) Superhydrophilic

Table n°6: Water contact angle measurements of COC with SiO2-like coatings deposited by PECVD and by sputtering in air and water storage. Time After Deposition Before SiO Aging x deposition Deposition 0 days 15 days 30 days 60 days medium (Blank method COC) PECVD 90° ± 3° <10° a) <10° a) <10° a) <10° a) Air storage Sputtering 90° ± 3° <10° a) 10° ± 2° 18° ± 3° 20° ± 3°

PECVD 90° ± 3° <10° a) 22° ± 5° 45° ± 4° 51° ± 3° Water storage Sputtering 90° ± 3° <10° a) 19° ± 5° 35° ± 3° 39° ± 3° a) Superhydrophilic

78 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials

Figure n°20: Evolution of water contact angles of COC with SiO2-like coatings deposited by PECVD and by sputtering in air and water storage (PECVD: 150 W plasma discharge power, 4 min deposition time, 150 sccm oxygen flow rate and 0.3 mbar precursor pressure. Sputtering: 150 W plasma discharge power, 60 min deposition time, 30 sccm Ar flow rate and 150 sccm O2 flow rate).

79 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials

Figure n°21: Evolution of water contact angles of THV with SiO2-like coatings deposited by PECVD and by sputtering in air and water storage (PECVD: 150 W plasma discharge power, 4 min deposition time, 150 sccm oxygen flow rate and 0.3 mbar precursor pressure. Sputtering: 150 W plasma discharge power, 60 min deposition time, 30 sccm Ar flow rate and 150 sccm O2 flow rate).

80 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials

Table n°7: Water contact angle measurements of THV with SiO2-like coatings deposited by PECVD and by sputtering in air and water storage. Time After Deposition Before SiO Aging x deposition Deposition 0 days 15 days 30 days 60 days medium (Blank method THV)

PECVD 100° ± 3° <10° a) <10° a) <10° a) <10° a) Air storage Sputtering 100° ± 3° 12° ± 3° 30° ± 5° 36° ± 6° 40° ± 6°

PECVD 100° ± 3° <10° a) 20° ± 4° 22° ± 3° 22° ± 4° Water storage Sputtering 100° ± 3° 15° ± 3° 20° ± 5° 28° ± 5° 32° ± 5° a) Superhydrophilic

Table n°8: Water contact angle measurements of PDMS with SiO2-like coatings deposited by PECVD and by sputtering in air and water storage. Time After Deposition Before SiO Aging x deposition Deposition 0 days 15 days 30 days 60 days medium (Blank method PDMS)

a) 128° ± 120° ± 120° ± PECVD 110° ± 3° <10° b) b) b) Air 10° 10° 10° storage Sputtering 110° ± 3° 23° ± 3° 98° ± 5° 100° ± 4° 110° ± 6°

PECVD 110° ± 3° <10° a) 20° ± 4° 44° ± 4° 64° ± 3° Water storage Sputtering 110° ± 3° 23° ± 3° 74° ± 5° 88° ± 3° 100° ± 5° a) Superhydrophilic b) Due to the superhydrophobic nature of the surface, the volume of the droplet was increased to 6 µL.

81 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials

Figure n°22: Evolution of water contact angles of PDMS with SiO2-like coatings deposited by PECVD and by sputtering in air and water storage (PECVD: 150 W plasma discharge power, 4 min deposition time, 150 sccm oxygen flow rate and 0.3 mbar precursor pressure. Sputtering: 150 W plasma discharge power, 60 min deposition time, 30 sccm Ar flow rate and 150 sccm O2 flow rate).

4.2 Chemical composition of the silica-like modified substrates 4.2.1 FTIR spectroscopy The chemical bonding structure of the deposited silica-like layer was determined using FTIR spectroscopy. The spectrum of the blank and of the silica-like coated NOA 81, COC and THV films are shown respectively in Figures 23-25.

82 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials

Figure n°23: FTIR-ATR spectra of the blank NOA 81 surface (black line) and the silica coated NOA surface either functionalized using PECVD (red line) or sputtering (blue line). (PECVD: 150 W plasma discharge power, 4 min deposition time, 300 sccm oxygen flow rate and 0.3 mbar precursor pressure. Sputtering: 150 W plasma discharge power, 60 min deposition time, 30 sccm Ar flow rate and 150 sccm O2 flow rate).

In the NOA 81 blank spectrum shown in Figure 23, the broad band between 3200 and 3500 cm-1 is due to O-H stretching of hydroxyl groups, revealing the presence of dimers or high molecular weight species. The peak around 1454 cm-1 was attributed to the wagging mode of -CH3 groups. Infrared frequencies of NOA 81 have not been previously assigned in the literature, so the absorption band located at 1733 cm-1 was assumed to be due to the carbonyl stretching of C=O, which is usually found in polymers [228]. The peak around 1681 cm-1 might correspond to an aryl derivative [229], whereas the small band in the 2800-3100 cm-1 region is probably due to C-H stretching, and the intensity of these peaks is probably related to alkyl groups [230]. In addition, the band located at 1145 cm-1 might correspond to a styrene phenyl ring [231] or to O=C-O-C antisymmetric stretching vibrations [232]. Since NOA 81 is

83 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials a mercaptoester polymer, the band located at 2571 cm-1 was attributed as the characteristic band for the S-H stretching mode of the thiol group [233].

Figure n°24: FTIR-ATR spectra of the blank COC surface (black line) and the silica coated COC surface either functionalized using PECVD (red line) or sputtering (blue line). (PECVD: 150 W plasma discharge power, 4 min deposition time, 300 sccm oxygen flow rate and 0.3 mbar precursor pressure. Sputtering: 150 W plasma discharge power, 60 min deposition time, 30 sccm Ar flow rate and 150 sccm O2 flow rate).

As shown in Figure 24, the absorption bands located at 2947 cm-1 and 2868 cm-1 on the COC blank spectrum were assigned to the carbon/hydrogen stretching vibration modes of -CH2 and -CH3 groups from the polymer backbone. The peak -1 around 1454 cm corresponds to the wagging mode of -CH3 groups.

84 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials

Figure n°25: FTIR-ATR spectra of the blank THV surface (black line) and the silica coated THV surface either functionalized using PECVD (red line) or sputtering (blue line) (PECVD: 150 W plasma discharge power, 4 min deposition time, 300 sccm oxygen flow rate and 0.3 mbar precursor pressure. Sputtering: 150 W plasma discharge power, 60 min deposition time, 30 sccm Ar flow rate and 150 sccm O2 flow rate).

The ATR-IR spectrum of the blank THV displayed in Figure 25 is dominated by -1 two broad and intense bands at 1193 and 1149 cm , which correspond to CF2 symmetric stretching vibrations [234]. For all the sputtered silica-like coated samples, the initial peaks originating from the polymer backbone were significantly lower but were still present due to the low thickness of the silica layer (30-700 nm) and the depth of measurement of the FTIR technique itself (up to 1 µm). However, the spectra of all silica-like-sputtered samples showed two bands at 1200 cm-1 and 1080 cm-1 corresponding respectively to the Si-O-Si asymmetrical and symmetrical stretching modes, and a characteristic vibrational band at 809 cm-1 corresponding to the Si-O-Si bending mode [235].

85 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials On the infrared spectra of the substrates coated using PECVD, the similar characteristic band corresponding to the Si-O-Si bending mode was attributed to the band located at 808 cm-1. Compared to sputtered substrates, the position of the Si-O-Si stretching bands were shifted to lower wavenumbers set at 1080 cm-1 and 1160 cm-1, respectively, for the asymmetrical and symmetrical stretching bands [236]. This shift observed in the Si-O-Si absorption band is due to the differing nature of the two plasmas. In sputtering, the plasma generated is a low-pressure discharge with energetic Ar ion bombardment that occurs during the deposition of the layer. In PECVD, plasma with low ion energies is produced. Therefore, the shift of the Si-O-Si absorption band may be due to decreased ion bombardment during PECVD, leading to a thick layer [237]. This hypothesis is reinforced by ellipsometric measurements, which show that PECVD produces a thicker coating than sputtering.

For all IR spectra, the broad absorption region around 3400 cm-1 was assigned to the hydroxyl group stretching mode. This band is related to the presence of Si-OH groups, as the interaction between the SiO2 layer and the moisture in ambient air lead to the formation of -OH groups at the surface of the coating [238]. Furthermore, the presence of Si-OH groups is revealed by the vibrational band located at 963 cm-1. The FTIR results demonstrate the presence of a silica-like coating for the COC, NOA and THV substrates, for both types plasma deposition processes investigated here. Further, all the substrates, coated using other method, displayed the same spectrum after 60 days of storage in water (results not shown here).

4.2.2 XPS measurements

The surface chemical compositions of the SiO2-like films deposited by sputtering and PECVD were analyzed by XPS analysis.

86 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials

Blank COC Blank NOA 81

Figure n°26: XPS C1s spectra of the surface of blank COC and NOA 81.

Figure 26 shows the curve fitting for C1s high resolution spectra of blank COC and NOA 81. The C1s spectra of the blank films were satisfactorily fitted by a combination of three distinct peaks assigned to C-C/C-H moieties (BE = 285.0 eV), C-O (BE = 286.5 eV for COC and 286.3 eV for NOA), and -COOH/COOR (BE = 289.0 eV). All components were assigned with the same full width at half- maximum and their respective positions were not fixed. These values and the corresponding assignments are in good agreement with data previously published in the literature [236].

The C-C/C-H peak corresponds to carbon bonded to carbon or to hydrogen in the polymer backbone. However, the C-O and COOH/COOR peaks of the blank substrates, attributed to the carbon-oxygen bond, reveal that these polymer surfaces are not inert to oxidation when in contact with air atmosphere, as observed in [239]. The atomic compositions extracted from the survey spectra of the coated and blank COC are presented in Table 9.

87 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials Table n°9: Elemental atomic concentrations extracted from XPS on COC substrates before and after deposition of the SiO2-like layer (PECVD: 150 W plasma discharge power, 4 min deposition time, 150 sccm oxygen flow rate and 0.3 mbar precursor pressure. Sputtering: 150 W plasma discharge power, 60 min deposition time, 30 sccm Ar flow rate and 150 sccm O2 flow rate). Layer C O Si O/C Si/C COC Samples composition (%) (%) (%) Ratio Ratio related to Si

Blank surface 86.7 13.3 - 0.15 - -

After Ar/O2 plasma pre-treatment step 73.5 25.3 1.2 0.34 0.016 - using PECVD

After SiO PECVD x 5.2 67.9 26.9 13.05 5.17 Si O C deposition 2.52 0.19

After Ar/O2 plasma pre-treatment step 73.2 25.4 1.4 0.34 0.019 - using sputtering

After sputtered SiOx 11.0 62.7 26.3 5.7 2.39 Si O2.38 C0.41

C1s high resolution spectrum Si2p high resolution spectrum

Figure n°27:XPS C1s and Si2p high resolution spectra of the surface of COC coated with silica-like layer by PECVD (PECVD: 150 W plasma discharge power, 4 min deposition time, 150 sccm oxygen flow rate and 0.3 mbar precursor pressure).

88 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials

Figure 27 shows the curve fitting with C1s and Si2p high-resolution spectra of the silica-like coated COC using PECVD. It has to be noted that the coated NOA 81 and THV samples display the same spectrum as the COC one when using

PECVD or sputtering. The C1s high resolution peak seen in Figure 27 was fitted with a single peak at 285 eV, reflecting possible carbon contamination of the upper surface of the coating.

Moreover, the Si2p high-resolution spectrum was satisfactorily fitted with a single peak located at 103.9 eV and corresponding to the Si-O bonds [168]. This single peak is in good agreement with previously published data and confirms the deposition of silica-like film in the same composition range for all the substrates. Thus, these results confirm the presence of silica-like film on the coated COC, THV and NOA 81 substrates as indicated in Tables 9-11. For all the sputtered silica-like coated substrates, though, the carbon content extracted from the survey spectrum was also observed in the ATR-FTIR spectrum (presence of carbonyl groups between 1600 cm-1 and 1750 cm-1). This carbon content might result from the incorporation of contaminants during the sputtering deposition process, or they might originate from the ablation/ re-deposition of the carbon atoms of the blank substrate itself.

89 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials Table n°10: Elemental atomic concentrations extracted from XPS on NOA 81

substrates before and after deposition of SiO2 layer (PECVD: 150 W plasma discharge power, 4 min deposition time, 150 sccm oxygen flow rate and 0.3 mbar precursor pressure. Sputtering: 150 W plasma discharge power, 60 min deposition

time, 30 sccm Ar flow rate and 150 sccm O2 flow rate).

Layer C O N S Si O/C O/Si NOA 81 Samples composition (%) (%) (%) (%) (%) Ratio Ratio related to Si

Blank surface 64.7 24.5 4.3 3.7 2.8 0.37 8.75 -

After Ar/O2 plasma pre-treatment step 43.6 37.7 9.2 6.1 3.4 0.86 11.08 - using PECVD

After SiO PECVD x 8.4 65.6 - - 26.0 7.81 2.52 Si O C deposition 2.52 0.32

After Ar/O2 plasma pre-treatment step 49 36.4 8.0 4.8 1.8 0.74 20.22 - using sputtering

After sputtered 8.8 64.5 0.7 - 26.0 7.33 2.48 Si O2.48 C0.34N0.03 SiOx

After N2/H2 plasma pre-treatment Blank THV

Figure n°28: XPS C1s high-resolution spectra of the surface of THV before and

after the N2/H2 plasma pre-treatment step by PECVD (150 W plasma discharge power, 5 min treatment time, nitrogen (75 sccm) / hydrogen (225 sccm), total pressure of 1.4 mbar).

90 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials As shown in Figure 28, the typical XPS pattern for the THV substrate reveals the presence of -CF3, -CF2, -CF, and -C-CF peaks, which are characteristic of fluorinated polymers. These peaks confirm the presence of fluorine and carbon atoms in the polymer backbone. For the blank THV (Figure 28), the C1s region was calibrated to the CF2 peak at 292 eV and to the C-C/C-H peak at 285 eV according to [240]. The C1s spectrum was fitted with five spectral components assigned to C-H and C-C (BE = 285 eV), C-CF and C-OH (BE = 287.27 eV), CF

(BE = 289.67 eV), CF2 (BE = 292 eV) and CF3 (BE = 294.07 eV). In conjunction with FTIR spectra (Figure 25), these data indicate the presence of CF2 bonds in the polymer backbone. Moreover, high intensity of C-H and C-C at 285.0 eV in the C1s peak was observed. This high intensity could be the result of a contamination in the fabrication process of THV sheets, as reported in literature [241].

91 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials

Table n°11: Elemental atomic concentrations extracted from XPS on THV

substrates before and after deposition of SiO2 layer (PECVD: 150 W plasma discharge power, 4 min deposition time, 300 sccm oxygen flow rate and 0.3 mbar precursor pressure. Sputtering: 150 W plasma discharge power, 60 min deposition

time, 30 sccm Ar flow rate and 150 sccm O2 flow rate).

Layer C O Si N F O/C C/F THV Samples composition (%) (%) (%) (%) (%) Ratio Ratio related to Si

Blank surface 45.0 4.2 - - 50.8 0.09 0.88 -

After N2/H2 plasma pre-treatment step 58.5 16.0 3.2 2.2 20.1 0.27 2.91 - using PECVD

After N2/H2 plasma and SiOx coating Si O C 6.7 65.7 25.5 - 2.1 9.80 3.19 2.57 0.26 deposited by F0.08 PECVD

After N2/H2 plasma pre-treatment step 57.8 6.5 1.5 - 34.2 0.11 1.69 - using sputtering

After N2/H2 plasma pre-treatment and Si O C 10.5 60.4 24.4 - 4.7 5.75 2.23 2.47 0.43 sputtered SiOx F0.19 deposition

According to the surface compositions shown in Table 11, it appears that sputtering increases incorporation of fluorine atoms into the deposited layer, as compared to PECVD. The sputtering technique accomplishes these results by inducing a re-deposition of the fluorine atoms present at the topmost surface layer

during the N2/H2 plasma pre-treatment step. A similar phenomenon occurs for the silica-like film deposited by PECVD, but the fluorine incorporation is less pronounced due to the difference of parameters (pressure, flow rates, gas phase

92 Chapter 3: Study of the stability and hydrophilicity of plasma-modified microfluidic materials composition, etc…) in both processes. Nevertheless, these results are consistent with the WCA measurements. 5. Conclusion Under optimal conditions, it has been demonstrated that the deposition of a silica- like film using two different dry processes accomplished incorporation of superhydrophilic layer onto three different classes of microfluidic material. COC, NOA and THV were successfully rendered hydrophilic (and remained thus for several weeks) by means of a silica layer deposition, confirmed by FTIR and XPS. The aging of the silica-like deposited layer was assessed by WCA measurements and the stability of the coating in air and water storage was demonstrated. However, air aging revealed the instability of such coatings on PDMS surfaces due to the well-documented hydrophobic recovery of this material. Between these new classes of materials, the silica-like coated COC appears to be the most promising for microfluidic applications because of its high stability in terms of hydrophilicity in air and water storage and rapid prototyping compared to THV and NOA. Therefore, COC will be used as the starting material for the elaboration of the catalytic microreactor while the deposited silica-like thin film in the PECVD process will be used as a support layer for the adhesion between the catalyst and the COC substrate. However, an optimization could be performed on the sputtering technique by studying the different mechanisms involved in the plasma chemistry. Here, neither the plasma chemistry nor the organosilicon precursor fragmentation was studied here. Nevertheless, this can be achieved through the use of in-situ characterization techniques such as Optical Emission Spectroscopy (OES) or Mass Spectroscopy.

94 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

1. Introduction A key aspect in overcoming the energy and environmental challenges is to improve the efficiency of existing and new processes. As almost all major chemicals are nowadays produced by catalytic processes, an improved efficiency usually entails the optimization of the catalyst and/or reactor. Among such processes, heterogeneous catalysis plays a very active role because of environmental concerns. In order to develop more effective catalytic processes, a better understanding of the reaction pathways and kinetics is needed. However, detailed measurement of the reaction kinetics of fast chemical reactions are often proved difficult, either because the reaction is simply too fast for conventional processes and exceeds typical reactant mixing times, or because the fast reaction kinetics are obscured by a rate-limiting step (the ability of the reactants/reaction products to diffuse to the catalyst surface) that determines the overall reaction kinetics. For this reason, accurate reaction pathways and their associated kinetics still remain largely unknown in the field of catalysis.

Catalytic ozonation is a typical example of this problem. In the field of water and wastewater treatment, chemical oxidation of organic pollutants is one of the major steps of drinking water treatment as it enables the removal of residual organic pollutants that were not eliminated in previous steps. As a trend towards cleaner processes, advanced oxidation processes (AOPs) are becoming more and more attractive. Among the different existing AOPs, the use of ozone is one of the most promising technologies. Indeed, compared to chlorine, the use of ozone

95 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation as the oxidizing agent exhibits many advantages, since it demonstrates additional antibacterial and antiviral actions [242]. The effective usage of ozone in the above mentioned fields is vital and therefore, extensive research aiming at the utilization of catalytic ozonation has been undertaken [243-245]. Several research groups have reported the high efficiency of catalytic ozonation in the removal of organic contaminants in water [246-252]. However, despite a high number of papers published in this field, there is a lack of understanding of the mechanisms governing catalytic ozonation processes and also deactivation of catalysts.

As a result, catalytic ozonation still remains in the sphere of laboratory experiments with only a few attempts at technological applications. The major problem lies in contradictory mechanisms governing catalytic process proposed by different research groups [253]. Some authors suggest radical pathways involving ozone decomposition and formation of hydroxyl radicals whereas some others indicate that catalytic oxidation proceeds via different pathways, which do not involve hydroxyl radicals. Another important issue is adsorption of ozone and organics as it is still not clear how, or if, ozone adsorbs on the surface of the catalyst and whether this process leads to ozone decomposition followed by the formation of surface-bound or free radicals. In addition, it is also not known whether adsorption of organics on the surface of catalyst plays a crucial role in the present catalytic process. Several authors reported high adsorption of organic molecules on catalysts whereas others indicated that this process is not required for efficient catalytic oxidation. Therefore, microfluidic devices could be a very useful tool to investigate the mechanism of catalytic ozonation.

The past decade has seen significant advances in the design and use of microfluidic devices to perform chemical analysis and reactions in micron-sized channels and reactors for a broad range of applications in various fields such as physics, chemistry and biotechnology [254-257]. Due to their reduced size and high surface-area-to-volume ratio, such devices offer fast analysis, a high

96 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation potential for dedicated patterning [258] and short reaction times. The study of reaction mechanisms in microchannels is a promising avenue since the time evolution of products and intermediates reactions can be separated spatially due to the flow in the microchannel [259].

Regarding the type and nature of catalysts that has to be used, nanoclusters of noble metals providing high values of specific surface areas seem to be ideal candidates [260,261]. The most promising method would be to incorporate such catalytic nanoparticles directly onto the wall of a microchannel. Although this has already been attempted, a lowered reactivity due to a small active surface, as well as the leaching of the metals during the chemical reactions were both reported. Therefore, applications of such methods are severely limited [262,263].

To overcome these issues, one possible solution is to decompose ozone in order to produce highly reactive HO• radicals. This can be achieved by the use of catalysts such as ionic metals (e.g. Mn2+, Fe2+, Fe3+, Cr2+, Cr3+…) in homogeneous catalysis [264] or by the use of noble metals in heterogeneous catalysis [265], the latter of course being preferred from a process point of view. Thus, most heterogeneous catalytic ozonation studies refer to metals supported on powders [266]. Unfortunately, the use of powders requires an extra step of separation in order to retrieve and regenerate the catalyst. Therefore, an immobilized catalyst with a desired lifetime adapted to the process, i.e. that can perform a high number of turnovers, is generally preferred.

In the field of microfluidics, a number of catalytic microreactors [267-269] have recently been developed to immobilize the catalyst by different methods such as “packed-bed” microreactors [270], the insertion of palladium membranes inside microreactors [271] or even the use of polymer brushes to form a polymeric nanostructure enabling the covalently bond of an organic catalyst on the walls of the microreactor [272]. However, given the high price of noble metals, the use of metal oxides is encouraged. In addition, achievement of such catalytic microreactors raises the issue of the nature of the material used for their

97 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation elaboration. Indeed, for purposes of cost reduction, polymeric-based materials are commonly used in the field of microfluidics.

Nevertheless, these catalysts can’t be coated on such materials using standard methods in light of the fact that these polymers would melt under the high annealing temperature required for the activation of the catalyst. Therefore, Plasma Enhanced Chemical Vapor Deposition (PECVD) processes are one of the techniques that can be used to achieve the deposition of catalytic thin films on such microfluidic materials. Despite the fact that PECVD processes operate at low temperatures, these techniques display many other advantages such as no requirement for solvents and surface activation of catalyst at low temperatures. Indeed, from previous works [273-276], it was demonstrated that plasma processes could lead to the activation of a catalytic coating at low temperatures.

Compared to conventional methods such as impregnation and calcination, the main advantage is that this technique allows structuring the coatings at the micrometric or even the nanometric level what could enhance greatly the catalytic efficiency compared to conventional catalysts [277-280]. In fact, the active phase on the catalyst surface must be highly dispersed over a large specific surface area and the specific activity has to be maximized to ensure an efficient reaction in heterogeneous catalysis. The other advantages are that the chemical composition and the oxidation degree of the metallic elements can be easily tuned by modifying the process parameters such as the gas composition, the discharge power and the treatment time.

The aim of this study is to deposit a thin catalytic layer into a polymer-based microreactor using a Metal-Organic Plasma Enhanced Chemical Vapor deposition process (MO-PECVD). As demonstrated in the previous chapter, Cyclic Olefin Copolymer (COC) appears to be one of the most suitable microfluidic materials which not only can enable rapid prototyping but also withstand plasma processes. It is therefore presently used as the starting material. Regarding the deposition and activation of the catalytic coating, both steps are

98 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation performed in a single process, i.e., without removing the substrate from the deposition reactor.

Cobalt and iron oxide-based catalysts were chosen for their low cost and ability to catalyze oxidation reactions [281]. Once, the catalytic microreactors obtained, their efficiency are assessed in a catalytic ozonation process. The specific main challenge residing here is the comprehension of the mechanism that governs catalytic ozonation. Studies on the mechanism have mostly been made on relatively simple reactors until now. Unfortunately, there is still a lack of generic model for predicting the efficiency of the catalyst in microreactors.

The present chapter is mainly focused on the preparation and characterization of

Co3O4 and Fe2O3 catalysts using a one-step plasma process as well as their performance in an ozonation process regarding the removal rate of pyruvic acid, catalyst stability and regeneration.

2. Experimental 2.1 Elaboration of the COC microchannels The first step of the fabrication process of the Cyclic Olefin Copolymer (COC) microchannels is given in Figure 29. COC pellets and thin films (Topas® 6013S- 04, Tg = 130 °C, 254 µm thick) were purchased from Topas Advanced Polymers. A few grams of the pre-polymerized pellets are placed into an aluminum mold containing the channel design.

Figure n° 29: Scheme of the hot embossing step for the elaboration of the blank microreactor.

99 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation The mold was fabricated by micromilling (Minitech Machinery, USA) and the feature sizes of the straight microchannel were 15 mm in length, 500 µm in width and 100 µm of depth. The mold used in this study also contained microchannels of 50 mm in length for future studies.

The hot embossing step consists in inserting the mold in a heating hydraulic press (Specac) in which the pellets are consequently melted at 170°C for 10 min. Then a typical pressure of 2 MPa is applied for 10 min in order to obtain chips without bubbles trapped inside the polymer. Finally, after being cooled down to room temperature, the COC plate containing the microchannels is released from the aluminum mold as shown in Figure 30.

Figure n° 30: Pictures of the hot embossing step for the elaboration of the blank microreactor. From left to right: aluminum mold, COC pellets into the mold, final COC plate obtained.

2.2 Catalysts preparation Prior to the deposition of both catalysts using the MO-PECVD technique, connection holes were drilled with a 4mm bit and tapped to receive the future plastic connectors. The COC samples were cleaned with ethanol and dried under an argon flow at room temperature. Finally, a shadow-mask was applied on the edges of the channel for future bonding as shown in Figure 31.

100 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

Figure n° 31: Scheme of the overview of the elaboration of the catalytic microreactor.

In order to clean off the surface of the samples from adsorbed atmospheric contaminants, the samples were first exposed to a 200 W Argon (150 sccm) / Oxygen (150 sccm) plasma treatment during 10 min for a total pressure of 1.4 mbar. After this pre-treatment step, a silica layer was deposited in order to increase the adhesion between the catalyst and the surface of the channel but also in order to obtain a homogeneous oxide passivation layer. The silicon oxide thin film was deposited using a homemade low-pressure plasma reactor. The radiofrequency plasma (13.56 MHz) was generated inside a Pyrex tube (46 mm in internal diameter, 600 mm in length) by the use of capacitive coupled external electrodes (Figure 32).

A gas mixture composed of oxygen (300 sccm) and hexamethyldisiloxane (60 sccm) was introduced into the reactor. This system can be used for both liquid and vapor precursors which are introduced into the reactor. The monomer vapors were supplied to the reactor chamber from liquid hexamethyldisiloxane (Merk,

101 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation 98.5%) contained in a cylindrical flask, via a heated stainless steel line at 70 °C. A manual valve was used for fine control of the flow. From previous works in our laboratory [162-163], the plasma power and deposition time were respectively of 150 W and 4 min for a total pressure of 1.7 mbar.

Cobalt oxide and iron oxide thin ﬁlms were deposited using the same reactor with a total pressure of 110 Pa and a glow discharge power of 200 W by the use of a carrier gas (Ar: 300 sccm and O2: 300 sccm) as shown in Figure 32. The chosen precursor in this process is Octacarbonyl Dicobalt (Strem Chemicals, France) also called cobalt carbonyl Co2(CO)8 which is solid under standard conditions. In order to improve the precursor injection into the system, 1 g of the precursor was dissolved in 25 mL of hexene in order to produce fine droplets by the use of an ultrasonic nebulizer. These drops are then evaporated through a heated electro- valve and introduced into the reactor which leads to a pulsed introduction of the precursor in the plasma reactor (t on = 0.9 s, t off = 4 s) for a total treatment time of 720 s. The electro-valve was heated to 70 °C in order to improve the evaporation of the precursor through the introduction line. No heating was applied to the substrate.

The ﬁlm was deposited on a 2 cm x 2 cm COC film (thickness: 254 µm) previously polished and treated with Ar/O2 plasma (1:1 ratio) with a total ﬂow rate of 300 sccm and a discharge power of 200 W. The preliminary pre-treatment is carried out to remove eventual adsorbed atmospheric contaminants.

102 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

Figure n° 32: Scheme of the experimental set-up of the MO-PECVD for deposition of Co3O4 and Fe2O3 catalysts.

The Fe2O3 thin film was deposited using the same device as described previously. The main changes here reside in the use of a bubbler instead of a nebulizer and in the application of cycles to avoid the lack of adhesion of the layer onto the sample and therefore, achieve the oxidation step. The overall catalyst deposition process consists in 6 cycles where a cycle corresponds to a deposition step of 2 mn with Ar and O2 alternated with an oxidation step of 2 mn. The cycle parameters are summarized in Table 12.

103 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

Table n°12: Experimental parameters used for the deposition of iron oxide catalysts. Pre- Deposition Post- Parameters Oxidation treatment step treatment Time 5 mn 2 mn 2 mn 30 mn Ar flow rate 150 sccm 200 sccm - 300 sccm

O2 flow rate 150 sccm 100 sccm 100 sccm - Discharge power 150 W 200 W 200 W 150 W

After deposition of the metallic oxides, both samples were post-treated using Ar plasma (300 sccm) at a power of 150 W during 30 mn and compared to stainless steel substrates annealed at 500 °C and 900 °C, respectively for the plasma deposited cobalt and iron oxides. These parameters were chosen in accordance with previous works from our laboratory which demonstrated that the use of an argon plasma lead to more active species and thus, a higher catalytic activity [282].

2.3 Thin films characterization In order to assess the stability and hydrophilicity of the coated surfaces, water contact angles measurements were performed on the flat coated substrates. Contact angles of MilliQ-water on all substrates were measured using a GBX-3S Digidrop-MCAT instrument. Water droplets of 3 μL in volume were released from a syringe above the sample surface. Images of droplet formation, captured using a high-resolution camera, were analyzed using an image analysis software (Digidrop). For each treatment, contact angle measurements were performed at a minimum of 3 locations on each sample and a typical deviation of ±5° from the mean value was observed. The reported values correspond to the average of these three measurements.

Infrared spectra were carried using a Fourier-Transform Infrared Spectrometer (Cary 660 Spectrometer-Agilent) with an Attenuated Total Reflectance module

104 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation (GladiATR-Pike). For each spectrum, 44 scans were accumulated with a spectral resolution of 4 cm-1 between 600 and 4000 cm-1. The surface morphology and thickness of the films were characterized by using a field emission scanning electron microscopy (SEM) (Leica S440) with a 20- 300,000x magniﬁcation at a resolution of 4.5 nm. X-ray diffraction (XRD) was performed using a Pan Analytical-X Pert Pro apparatus with a Cu Kα1 radiation source (8027.8 eV) in order to determine whether the plasma deposited films are amorphous or crystalline and in the latter case, measure the crystallites size.

Samples for the transmission electron microscopy investigations were prepared by pouring the dispersed Co3O4 and Fe2O3 particles onto a 300-mesh copper grid using a JEOL JEM 2100F apparatus (JEOL, Japan) operating at an accelerating voltage of 200 kV. Analyses of the crystals were performed on a diffractometer D5000 X-ray using a Cu Kα1 radiation source of 35 kV and the scan was ranged from 10° to 80° with 0.02 steps. The crystallographic nature of the individual particles was indicated by the microdiffraction patterns measured by TEM.

XPS spectra were recorded using a PHI 5600 XPS spectrometer (Physical Electronics, Eden Prairie, USA). Survey and high resolution spectra were acquired using the Kα line of a monochromatic Al (1486.6 eV) and Mg source, respectively, operated at 200 W. Analyses were performed with a 45° angle from the surface. Survey spectra were acquired from 0 to 1400 eV for 15 min. Curve fitting for the high resolution Co2p and Fe2p peaks was completed using XPS PEAK Software (version 4.1) by means of a least squares peak fitting procedure using a Gaussian-Lorentzian function and a Shirley baseline fitting. For each sample, three different spots were analyzed and the analyzed surface was evaluated at 0.005 cm2.

105 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation 2.4 Chip Assembly In order to enhance the polymer chain mobility, a mixture of cyclohexane and hexadecane (3:1 volume ratio) was spread on the edges of the COC microchannel to be bonded [283]. The top cover and channel were aligned under an optical microscope and bonded together inside a heated hydraulic press (Specac) at 110 °C for 3 min with a total applied pressure of 0.6 MPa as shown in Figure 33.

Figure n° 33: Scheme of the catalytic microreactor assembly (left). Picture of the sealed catalytic microreactor containing the plasma post-treated cobalt oxide layer (right).

By connecting Teflon tubing (Fisher Scientific) to plastic connectors (F-125H, Fisher Scientific) aligned with the drilled holes of the channel, the solution was then introduced into the catalytic microreactor.

2.5 Adsorption and catalytic activity The efficiency of the catalytic microreactor was determined by comparing the degradation of a probe pollutant between a blank microreactor (i.e. without catalyst) and the catalytic microreactor; both microsystems having the same dimensions. Catalytic ozonation was studied here by selecting pyruvic acid as the refractory pollutant. Pyruvic acid (PA) was chosen as a probe pollutant since it is known to have low reaction rates with molecular ozone (kO3/AP = 0.13 and 0.98 L•mol-1•s-1 at a pH = 1 and 7 respectively [284]) whereas with hydroxyl radicals, PA has a high reaction rate (kHO•/PA = 1.2•108 L•mol-1•s-1 [285] and hence to pose problems in water purification [286]. In addition, the use of acidic pH prevents the decomposition of ozone [287]. Adsorptions tests were carried by

106 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation flushing the aqueous solution of PA through the catalytic microreactor without ozone.

2.6 Catalytic ozonation apparatus

Figure n° 34: Scheme of the experimental set-up used for catalytic ozonation.

The catalytic ozonation apparatus showed in Figure 34 consists of an ozone generator (Type COM-AD-01, Anseros) supplied by an oxygen tank. By setting the ozone generator at its full power (29 W) and by the use of a mass flow controller (MFC), ozone is sent to a bubbler containing 20 mL of ultrapure water with pyruvic acid (98%, Alfa Aesar) at a concentration of 0.3•10-3 mol•L-1. The ozone concentration in the liquid phase was followed online using optical fibers (Ocean Optics Maya Pro 2000) with an UV detection set at the wavelength of 259 -1 - nm (Deuterium lamp). Considering Beer-Lambert’s law (εO3 = 3300 L•mol •cm 1) the determined concentration of dissolved ozone was of 0.35•10-3 mol•L-1, when the system reached thermodynamic equilibrium. In order to have a controlled flow rate of 3.6 mL•h-1, a syringe pump (KD Scientific - KDS Legato 210) sucks the solution through the channel and 3.6 mL of the treated solution is recovered in the syringe. The ozone reaction was stopped downstream by adding 130 µL of 10•10-3 mol·L-1 tert-Butanol (99.9%, VWR) solution to rapidly consume any unreacted O3 [288]. The magnetic stirrer ensures the homogenization between the tert-Butanol and the incoming flow arriving in the syringe.

107 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation The percentage of degradation of the catalytic ozonation with both oxides was obtained by measuring the pyruvic acid concentrations before and after reaction using High Performance Liquid Chromatography (HPLC).

2.7 HPLC and FAAS Measurements Using a Star Varian chromatograph, HPLC analyses were performed in an isocratic mode with 20 µL portion of each treated sample. The initial PA and treated samples were analyzed on a 250 X 4.6 mm Supelcogel H column (Sigma- Aldrich) for organic acids, with acidified Milli-Q water (0.1% Orthophosphoric acid, Sigma-Aldrich) as the eluent, at a flow rate of 0.1 mL·mn-1. The UV detection was set at 210 nm at room temperature in order to analyze organic acids.

In order to assess the stability of such coatings upon catalytic ozonation, analyses of metallic cobalt and iron in solution were performed by Flame Atomic Absorption Spectrometry (FAAS) (Thermo Scientiﬁc, Solaar M) with a detection limit of 1 µg·L-1 for both metallic elements.

108 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation 3. Results and discussion 3.1 Water stability of deposited catalysts The measured water contact angle on the COC blank surface was equal to 90° whereas the cobalt oxide and iron oxide coated COC films were hydrophilic with a measured water contact angle (WCA) value less than 10°. An aging study was performed by storing the coated samples in air and water for several weeks. The WCA values remained under a value of 10° for up to 2 months, thereby meaning a high wettability for cobalt and iron oxides coatings. The aging study regarding the silica-like coated COC substrates can be found in the previous chapter.

3.2 Deposition of the cobalt oxide catalyst The chemical bonding structure of the deposited thin film was determined using FTIR-ATR spectroscopy and compared to a stainless substrate annealed at 500°C.

109 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation Figure n°35: FTIR-ATR spectra of cobalt oxide deposited by MO-PECVD on COC as deposited (green line), post-treated by argon plasma (red line), on the Stainless Steel substrate annealed at 500 °C (blue line).

As shown in Figure 35, the FTIR spectra of all deposited samples reveal the presence of 4 main absorption bands. The bands near 1321 cm-1 and 1423 cm-1 were assigned to C-O stretching and C-O-H bending due to the presence of carboxyl groups. This is confirmed by the presence of the C=O stretching band of the carboxyl groups located near 1700 cm-1. The band located at 657 cm-1 is characteristic of the Co-O bond stretching vibration [289]. The large absorption band located between 3000 and 3500 cm-1 is assigned to -OH stretching and corresponds to water adsorption on the surface of the sample. It has to be noted that absorption bands corresponding to the stretching of C-C and C-H are hidden by this large band. Thus, it can be concluded that cobalt oxide and an organic matrix with carboxyl groups are present on this surface. This organic matrix probably comes from the carbonyl groups contained in the precursor or either from the solvent used during the deposition, as hexene is known to polymerize in plasma processes [290].

Compared to the initial deposited cobalt oxide layer, the plasma post-treated and annealed samples display a more intense Co-O bond stretching band whereas the bands corresponding to carboxyl groups diminish. In addition, the band corresponding to water adsorption tends to narrow when the sample is plasma post-treated or annealed indicating a phase transformation [291]. This phenomenon probably indicates that the polymer matrix formed during the plasma deposition is etched during the argon plasma post-treatment or calcination, leading to a surface exposure of the cobalt oxide, as confirmed by SEM micrographs shown in Figure 36.

From the SEM images, the surfaces of the samples show nearly the same surface morphology with increase in grain size and reveal the presence of nanoclusters having a typical cauliflower-like shape coating which is characteristic of the 110 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation cobalt oxide layer deposited in similar conditions. The plasma post-treated and annealed samples shown in Figures 36 (b) and (c), exhibit finer nanoparticles on these surfaces than the initial deposited layer (Figure 36 (a)). Moreover, it can be seen that when the sample is post-treated with argon plasma, sharper nanoclusters are formed whereas when the sample undergoes an annealing step, nearly all nanoclusters change to a nano-catkin structure as shown on Figure 36 (f) and in previous works [163]. The cause of such particle refinement is probably due to the ion-bombardment generated in the argon plasma.

(a) (d)

(b) (e)

111 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

Figure n° 36: SEM images of Co3O4: (a), (b) and (c) are respectively the as- deposited, plasma post-treated and annealed samples. (d), (e) and (f) are the respective magnification of the surface layer.

Ellipsometry measurements were attempted to measure the thickness of the deposited layer. However, the black color of the cobalt oxide layer prevented accurate thickness determination by this technique. Therefore, the thickness of the layer was evaluated by using SEM images taken directly in the coated microchannel as shown in Figure 37.

(a) (b)

Figure n° 37: SEM images of plasma post-treated Co3O4 in the COC microchannel: (a) Side view of the layer in the channel and (b) side view of the layer on the inner surface.

112 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

From the previous SEM images, it can be seen that the thickness of the plasma post-treated cobalt oxide layer ranges between 3-6 µm in the microchannel while ranging between 700 and 800 nm on the inner surface. In addition, a typical columnar growth due to the crystallization can be observed and was previously proved by Klepper et al [292].

The crystalline structure of the deposited cobalt oxide layer was then investigated using XRD in order to determine the crystallinity of such thin films.

Figure n° 38: X-Ray Diffractograms of cobalt oxide samples on stainless steel substrates. Patterns are shifted vertically for better visualization.

As observed in Figure 38, the crystalline structure was determined by XRD measurements with the catalyst deposited on stainless steel substrates in order to remove the broad features of the amorphous COC patterns (results not shown here). As cobalt oxide can be encountered under three stable phases, XRD patterns for cubic CoO (JCPDS 43-1004), hexagonal Co2O3 (JCPDS 2-770) and spinel Co3O4 (JCPDS 42-1467) were reported here. It can be noted that for all samples, no signals of the silica-like layer were found, which reflects the

113 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation amorphous nature of the silica-like underlayer coated on the surface of the samples prior to the catalyst deposition.

The XRD spectrum of the untreated cobalt oxide thin layer reveals the presence of 4 peaks with three of them located at 2θ = 43.2°, 50.7° and 74.6° corresponding to the stainless steel substrate structures [293].A small peak located at 2θ =36.8° can be assigned to the [311] lattice plane of spinel Co3O4 [294] or to the [111] lattice plane of cubic CoO. In the case of the plasma post-treated sample, this peak increases whereas another intense peak located at 2θ =31.2° appears. The latter can be either assigned to the

[220] lattice plane of Co2O3 [295] or to the [220] lattice plane of Co3O4.

Besides the peak located at 2θ =36.8°, no other peaks related to the CoO phase was observed, which probably indicates that the cobalt oxide layer undergoes a phase transformation from CoO to the Co2O3 or the Co3O4 phase. However, the intensity of this peak and the presence of peaks located at 2θ = 19°, 44.8°, 59.4° and 65.2° respectively corresponding to the [110], [440], [422], [511] and [440] lattice planes of Co3O4 obviously shows the prevalence of this phase.

Furthermore, it is still unclear whether the Co2O3 compound exists in a form of a stable crystal. As for the plasma post-treated sample, the XRD spectrum of the annealed sample display sharper and more intense reflections of the Co3O4 phase. The mean particle size was calculated by applying the following Debye-Scherrer equation [296,297]: 푲 × 흀 푫 = 휷 × 풄풐풔 휽

 K is the Scherrer constant, a value of 0.9 is generally taken  λ is the X-ray wavelength  β is the full width at half maximum of the peaks  θ is the measured angle

114 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation Values of 15 and 10 nm were found for the mean particle size respectively for the as-deposited and plasma post-treated cobalt oxide samples.

However, these XRD analyses do not lead to a definitive assignment as they don’t fully allow discriminating the accurate chemical composition of the annealed or the plasma post-treated samples. Nevertheless, these results demonstrate that cobalt oxide with a preferred crystalline growth orientation is obtained in the layer, and thus, in accordance with the previous SEM observations (Figures 36- 37).

Figure n° 39: XPS High Resolution spectra of Co2p for as deposited cobalt oxide layer and plasma post-treated on COC substrates.

As shown in Figure 39, the Co2p high resolution spectra was fitted by two components at 780 eV and 796 eV, respectively for the Co2p3/2 and Co2p1/2 for the as deposited and plasma post-treated cobalt oxide layer samples.

115 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

As reported by Voß et al. [298], the Co2p3/2 and Co2p1/2 photoelectron lines of Co(II) compounds exhibit a pronounced shoulder on their high energy side (ranging from 784 to 792 eV) which can be assigned to a shake-up process [299].

The satellite peaks in the Co2p spectra are an important signal to discriminate the bonding valence of the cobalt oxide compounds. The lower intensity of the shake- up satellites at 9 eV from the main spin–orbit components of original sample showed that the plasma post-treated cobalt oxide layer was Co3O4 and not CoO since its characteristic shake-up peaks are not observed [300-302].The visible chemical shift occurring for the Co2p peaks indicate that the ionic balance state of

Co2p is different for each crystallized Co3O4 ﬁlm [303].

Table n°13: Atomic composition of cobalt oxide samples extracted from XPS.

Sample C (%) O (%) Co (%) O/Co C/Co

As deposited 25.3 50.7 24 2.11 1.05

Post-treated by argon plasma 20.6 51 28.4 1.79 0.72

As indicated in Table 13, the atomic content of the samples obtained for the cobalt oxide thin films present traces of neither silicon nor metallic compounds from the silica-like coating or the underlying substrate. The ratio of C/Co decreases when the sample is plasma post-treated, thus indicating that the polymer matrix is being etched by the argon bombardment generated during plasma. In the same way, the O/Co ratio decreases indicating a phase transformation. However, this ratio is higher than the one expected for pure stoichiometric Co3O4 (1.33) justified by the presence of carbonyl species. Therefore, it can be concluded that the effect of the argon plasma post-treatment leads to a crystalline phase transformation which is further confirmed by the BET measurements.

From the previous SEM images and possibly assuming that the Co3O4 nanoparticles and the pores are in a spherical shape, the mean particle diameter can be estimated according to the following equation:

116 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation ퟔ × ퟏퟎퟑ 풅푩푬푻 = 흆 × 푺푩푬푻

 dBET is the average diameter of a spherical particle (in nm) 3  ρ is the theoretical density of Co3O4 (6.11 g/cm ) 2  SBET is the specific surface area of the catalyst layer (in m /g)

The average pore diameter can be calculated from the Barret-Joyner-Halenda (BJH) adsorption method [304], given by the following equation: ퟑ ퟒ × 푽풑풐풓풆 × ퟏퟎ 풅풑풐풓풆 = 푺푩푬푻

 dpore is the mean diameter of a spherical pore (in nm) 3  Vpore is the pore volume (in cm /g) 2  SBET is the specific surface area of the catalyst layer (in m /g)

As shown in Table 14, the BET surface increases from 85 m²/g for the as deposited cobalt oxide layer to 104 m²/g for the argon plasma post-treated sample, thus confirming the expected crystallization effect of the argon plasma post-treatment. In addition, the estimated average pore size decreases with the plasma post-treatment at a lower value than the mean diameter size of the particle.

The mean particle sizes calculated from the surface area data are 11 nm and 9 nm respectively for the untreated and plasma post-treated samples; values which are in good agreement with the XRD results. However, regarding the deposited silica layer, calculations from the BET measurement and estimated from the BJH methods indicate an average pore diameter of 5.4 µm with a mean particle size of 7.1 µm. These values are most probably inaccurate as the previous BJH equation is applicable to microporous and mesoporous materials, and thus not applicable to the deposited silica-like layer due its low specific surface area value (0.4 m2/g).

117 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation Table n°14: BET measurements of cobalt oxide deposited by MO-PECVD on COC substrates with a silica-like underlayer. Total Average Developed Pore S pore Mean particle Sample BET Surface Area / volume (m2/g) diameter size (nm) b) Geometric (cm3/g) (nm) a) surface Ratio Deposited 0,4 700 5400 0,52 7100 SiO2 As deposited 85 3300 11 0,23 11 Co3O4 Deposited Co O post- 3 4 104 3800 8 0,21 9 treated by Ar plasma

a) Estimated by the BJH method b) Calculated from the BET measurements

3.3 Deposition of the iron oxide catalyst On the FTIR spectrum of the iron oxide coated sample, as shown in Figure 40, a characteristic band of the Fe-O bond stretching can be found at 670 cm-1 [305,306]. The bands near 1100 cm-1 and 1500 cm-1 were respectively assigned to C-O stretching and C=O stretching due to the presence of carboxyl groups. The absorption bands located at 2945 cm-1 and 2867 cm-1 are assigned to the

carbon/hydrogen stretching vibration modes of -CH2 and -CH3 groups from the polymer backbone. For all the substrates, the band located between 2340 cm-1 and -1 2360 cm is assigned to the CO2 elongation. The large absorption band located between 3000 and 3500 cm-1 is assigned to -OH stretching, due to water molecules being adsorbed just like for the cobalt catalyst spectrum. Here again, we can conclude that iron oxide and an organic matrix with alcohol and carboxyl groups are present on the surface.

118 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

Figure n° 40: FTIR spectra of iron oxide as deposited (red line), post-treated by argon plasma (green line), plasma deposited then annealed (red line).

119 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

a) d)

b) e)

c) f) Figure n°41: SEM images of plasma deposited iron oxide. Plane views of a) as deposited b) post-treated by argon plasma, c) annealed at 900 °C; d), e) and f) are the corresponding side views.

SEM micrographs displayed in Figure 41 reflects the effect of the plasma post- treatment on the coated surface. The surface of the thin films presents the same

120 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation characteristic columnar growth with a cauliflower-like shape morphology. As for the cobalt oxide layer, due to the pronounced red color of the deposits, the thickness of the layer was evaluated from the SEM micrographs and ranges between 5 and 6 µm.

Figure n°42: XRD patterns of the deposited iron oxide coatings. Patterns are shifted vertically for better visualization.

From the XRD patterns of the iron oxide coatings shown in Figure 42, the plasma deposited iron oxide layer appear. XRD patterns of the polymorphs of Fe2O3. It has to be noted that the magnetite Fe3O4 XRD pattern is very similar to the maghemite (γ-Fe2O3) one. The as-deposited iron oxide and plasma post-treated sample present similar peaks (2θ= 30°, 32°, 35°, 44°) than the hematite (α-Fe2O3) phase. However in the case of the post-treated sample, small peaks appearing at 2θ= 33° and 44° indicates the presence of the maghemite phase (γ-Fe2O3). We can possible assume that a mixture of these two phases occur during the plasma post- treatment whereas for the annealed sample, it can clearly be seen that it is the hematite phase which occurs on the surface. 121 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

By applying the Debye-Scherrer equation, values of 50 and 40 nm were found for the mean particle size respectively for the as-deposited and plasma post-treated iron oxide samples. Moreover, due to this high size values compared to the cobalt oxide samples, we can assume that the iron oxide layer didn’t undergo a high crystallization rate.

(a) (b)

Indeed, from the TEM micrographs shown in Figure n°43(a) and (b), two phases seem to appear on the surface of the iron oxide surface treated by argon plasma. Crystallized nanoparticles of iron oxide having a hexagonal form, which are represented by dark dots in the bright-field micrograph (Figure 43(a)) and by white dots on the dark-field micrograph (Figure 43(c)), appear along nanoparticles having an intermediate shape between an amorphous structure and a tetragonal crystalline one.

122 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation Moreover, it can be clearly seen that these nanoparticles aggregate and stack on each other. The hexagonal crystalline structure clearly indicates the presence of

α-Fe2O3 whereas γ-Fe2O3 has a tetragonal crystalline structure [307]. The average size of the nanoparticles was estimated to 10 nm. By using the diffraction mode, the values of the diffraction angles were determined to be 30°, 35° and 44° (2θ) which clearly indicate the presence of γ-

Fe2O3 [308,309].

Figure n°44: XPS High Resolution spectra of the iron oxide layer as deposited by MO-PECVD, argon plasma post-treated and annealed samples.

As shown in Figure 44, the iron oxide samples were fitted with two main peaks:

Fe2p3/2 (711 eV) and Fe2p1/2 (725 eV) [310]. The XPS spectrum of the annealed sample was reported here in order to understand the effect of the argon plasma post-treatment, as to the best of our knowledge, no comparison between both activation techniques were reported in the literature. Regarding the annealed sample, the clear presence of a satellite peak (719 eV) confirms the iron oxide phase to be α-Fe2O3 as it corresponds to the characteristic shake-up peak observed 3+ for Fe in α-Fe2O3 [311].

123 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation These results are in accordance with the fact that the phase transformation of the

γ-Fe2O3 phase to the α-Fe2O3 one occurs at high temperatures (above 400°C) [312]. Therefore, as for the cobalt oxide layer, the argon plasma post-treatment enables here the substitution of the calcination step by inducing a phase transformation and a surface refinement of the plasma deposited layers.

Table n°15: Atomic contents of plasma deposited iron oxides on COC substrates extracted from XPS for the as-deposited and argon plasma post-treated sample, on stainless steel substrate for the annealed sample. Sample C (%) O (%) Fe (%) Mn (%) Cr (%) O/Fe C/Fe As 22.9 55.3 21.8 - - 2.54 1.05 deposited Post- treated by 17.5 58.1 24.4 - - 2.38 0.71 argon plasma Annealed 11.9 60.2 7.6 4.7 15.6 7.92 1.56 at 900°C

As indicated in Table 15, when the sample undergoes argon plasma post- treatment, the atomic percentage of iron increases from 21.8 % to 24.4% whereas the carbon content decreases. The O/Fe ratio passes from 2.54 to 2.38 leading towards a ratio of an ideal Fe2O3 layer (1.5). However, for the annealed sample, atomic contents of manganese and chromium were identified. The latter appear on the surface of the sample due to their migration from the stainless steel substrate. Indeed, the calcination temperature is close enough to the processing temperature of stainless steel (1000°C).

Therefore, the effect of the plasma post-treatment cannot be compared to the effect of calcination. Nevertheless, from the previous characterization results, it can be assumed that the argon plasma post-treatment leads to a crystallization of the initial deposited layer in the same way as for the cobalt oxide layer. BET measurements for the as deposited and plasma post-treated iron oxide coatings were reported in Table 16.

124 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

Table n°16: BET Measurements for iron oxide deposited by plasma (as deposited and plasma post-treated). Total Developed Average Pore Mean S Surface Area / pore Sample BET volume particle (m2/g) Geometric diameter (cm3/g) size (nm) b) surface Ratio (nm) a) Deposited 0,4 700 5400 0,52 7100 SiO2 As deposited 26 3600 64 0,41 44 Fe2O3 Deposited Fe O post- 2 3 35 4900 41 0,36 33 treated by Ar plasma

a) Estimated by the BJH method b) Calculated from the BET measurements

Compared to the Co3O4 thin film samples, the measured BET of the iron oxides samples before and after the plasma post-treatment present a value of approximately 4 times. In the same way, the average pore diameter and mean particle size are at least 3 times greater. This is probably due to the non-fully crystallized particles present on the surface of the layer; assumption which is in accordance with the previous observations from the XRD measurements and TEM images. From the previous characterizations, we can conclude that in the case of the iron oxide catalyst, the argon plasma post-treatment does not lead to a full crystallization and phase transformation of the deposited iron oxide sample. Thus, unlike the cobalt oxide layer, the argon plasma post-treatment step cannot be considered as a substitution of the calcination step for the deposited iron oxide layer within the current experimental parameters. A possible assumption may reside in the difference of bond energies between both catalysts. Indeed, a value of 368 kJ/mol for the enthalpy of formation of the Co-O bond was reported versus a value of 409 kJ/mol for the Fe-O bond [313]. The temperature generated at the surface of the sample by ion bombardment during the plasma post-treatment step

125 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation is not sufficient enough to induce a full phase transformation as the energy required for the formation of a Fe-O bond is higher than the Co-O one.

However, an effect of the plasma post-treatment can be assumed as the pore volume seem has the same order of magnitude than the cobalt oxide catalyst. Therefore, catalytic activity measurements for the deposited iron oxide layer are expected to be low.

4. Catalytic activity measurements 4.1 Results of simple ozonation of PA in batch reactor Prior to the catalytic activity measurements of both metallic oxides, the kinetics of the decomposition of ozone in the pyruvic acid solution was assessed in the batch reactor. Samples of initial and final solutions of pyruvic acid after reaction with ozone spaced by 5 minutes were taken during one hour. Degradation of pyruvic acid was calculated using the HPLC technique.

The percentage of degradation was calculated from the HPLC chromatograms with the following equation: [퐏퐀] − [퐏퐀] % 퐝퐞퐠퐫퐚퐝퐚퐭퐢퐨퐧 = ퟎ × ퟏퟎퟎ [퐏퐀]ퟎ

 [PA]0 is the initial concentration of pyruvic acid  [PA] is the concentration of pyruvic acid after treatment

In relationship with the HPLC calibration curve previously established (see Chapter 2), both concentrations were obtained by reporting the value of the calculated area of the pyruvic acid signal in the equation. The reported pyruvic acid concentrations obtained after each treatment correspond to a mean average value for three distinct experiments. The evolution of the concentration of PA obtained experimentally in the batch reactor is reported in Figure 45.

126 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

Figure n°45: Evolution of the concentration of PA in the batch reactor as a -3 -1 -3 -1 function of time (pH= 3, [PA]0 = 0.3•10 mol•L , [O3] = 0.3•10 mol•L ) obtained experimentally.

From the previous figure, it can be seen that a final value of 30% ± 4% of degradation was found experimentally after one hour of ozonation with an initial PA concentration of 0.3•10-3 mol•L-1 and at pH = 3. The production of ozone being constant, a first-order reaction rate can be considered with the following equation:

ퟎ −풌풂풑풑×풕 푪푷푨(풕) = 푪푷푨풆

-1 With: 풌풂풑풑 = 풌푶ퟑ, 푷푨 × [푶ퟑ] in s

By using reaction rate values from the literature [284]in the same range of -1 -1 -1 - concentration for pyruvic acid, i.e., kO3,AP = 1.3•10 L•mol •s and 9.8•10 1L•mol-1•s-1 respectively at pH = 1 and 7, the corresponding decay values are 13 % and 65 % after one hour of ozonation.

127 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation Therefore, the experimental value of 30 % obtained for the PA decay is in the same order of magnitude and possibly follows a first order reaction.

4.2 Results of simple ozonation of blank and silica-coated microreactor Simple ozonation on pyruvic acid was carried out in a blank COC microreactor and in a plasma deposited silica-coated microreactor in order to assess the effect of the COC microchannel and silica layer.

Figure n°46: Simple ozonation performance in blank and silica coated microreactor. Error bars indicate standard deviations for 3 replicate measurements.

As shown in Figure 46, it can be seen that an average value of 30 ± 5% of degradation for simple ozonation of pyruvic acid is reached after one hour in the blank microreactor. Compared to the previous value of 30 % ± 4 % of degradation in the batch reactor, it can be concluded that the effect of the Teflon tube connecting the batch to the inlet of the microreactor is negligible. It has to be noted that the change in the initial concentration of pyruvic acid in the stock solution strongly contributes to the reported standard deviation as a value of 0.1 % was found for the standard deviation of HPLC.

128 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation Regarding the results of simple ozonation carried out in the silica-like coated microreactor, it appears that the latter do not exhibit a catalytic activity as the value of 32.9 % ± 2% of degradation is enclosed in the standard deviation of the blank microreactor. In light of these results, it can be concluded that the blank microreactor and silica coated microreactor do not exhibit a catalytic activity.

4.3 Catalytic ozonation results in iron oxide based catalytic microreactors 4.3.1 Stability of the plasma deposited iron oxide coatings FAAS measurements were performed after each adsorption and catalytic ozonation tests at the same flow rate, i.e., 3.6 mL•h-1. Adsorption tests consisted in passing a solution of pyruvic acid (0.3•10-3 mol•L-1) in the catalytic microreactor during one hour. It has to be noted that washing steps consisting in passing Milli-Q water at a flow rate of 3.6 mL•h-1during one hour were performed between each measurement. Results of the FAAS measurements are reported in Figures 47-48.

Results indicate that iron oxide coatings are stable upon adsorption, catalytic ozonation and washing steps as the only detectable amounts were found in the initial washing steps. Values of 0.11 and 0.19 ppm of metallic iron were found respectively for the as deposited and plasma post-treated catalyst after adsorption tests. However, the detected amount of iron in the initial wash solution may find its origin in the leaching of iron oxide nanoparticles of the Fe2O3 superficial layer, inherent to the MO-PECVD process.

In addition, knowing that a value of 60 ppm for 3.6 mL of solution can be calculated if the complete leaching of the deposited Fe2O3 layer is considered, the detected amount of iron in the initial washing step represents less than 0.2 and 0.3 % of the maximum iron amount that can be leached respectively for the as deposited and plasma post-treated iron oxide layer. Therefore, it can be concluded

129 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation that during the experiments, no leaching of iron from the deposited Fe2O3 layer occurred.

Figure n°47: Leaching results for the iron oxide layer deposited by plasma (as deposited and plasma post-treated) after adsorption tests and washing steps.

Figure n°48: Leaching results for the iron oxide layer deposited by plasma (as deposited and plasma post-treated) after catalytic ozonation tests and washing steps. 4.3.2 Adsorption tests results

Adsorption values for the Fe2O3 catalyst were obtained by passing pyruvic acid in the microchannel at a flow rate of 3.6 mL•h-1 during one hour and by determining

130 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation the final pyruvic acid concentration by the use of HPLC. Between each adsorption test, a washing step was performed as described in section 4.3.1.

Figure n°49: Adsorption tests for the iron oxide thin films deposited by plasma (as deposited and plasma post-treated). Error bars indicate standard deviation for 3 replicate catalytic activity measurements for each run.

As shown in Figure 49, the results indicate that the blank microreactor presents here a low adsorption value (4.5 % ± 1%) which probably comes from the increase in the initial surface roughness of the COC top cover during the hot embossing step. Indeed, according to Leech [314], a value of 10 nm was found for the average surface roughness for the COC 6013 grade embossed at 170 °C, i.e., in similar conditions as for our microchannel.

Therefore, the iron oxide based catalytic microreactors do not exhibit a higher pyruvic acid adsorption compared to the blank microreactor. Moreover, a decrease in the adsorbed pyruvic acid concentration depending on the number of runs can be even observed. At first, this decrease is not significant as the obtained values are enclosed in the standard deviation values of the blank microreactor and could be considered as a reproducibility issue.

131 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation Nevertheless, these results could also be possibly explained by a decrease in the number of adsorption sites due to the sustained adsorption of pyruvic acid on the surface of the catalyst. The washing step would be therefore not sufficient enough to remove all the pyruvic acid molecules adsorbed on the catalyst.

4.3.3 Results of catalytic ozonation tests

Figure n°50: Catalytic ozonation performance for the iron oxide layer deposited by plasma (as deposited and plasma post-treated). Error bars indicate standard deviation for 3 replicate catalytic activity measurements for each run.

From the values reported in Figure 50, it can be seen that the Fe2O3 catalyst prepared by MO-PECVD do not present any catalytic activity as, globally, the values are close to the blank (30 ± 5%) and silica-like coated microreactors (33 ± 2%). Although these values are enclosed in the standard deviation of the blank microreactor measurements, they seem to diminish in the same way than the adsorption tests results. Therefore, it can be concluded that the deposited Fe2O3 layer do not present a catalytic activity and that the obtained values are due to poor reproducibility.

132 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation The absence of catalytic activity could reside in the fact that the plasma post- treatment step does not manage to achieve an adequate crystallization in comparison to the deposited cobalt oxide layer, leading to a nonexistent catalytic activity as previously expected from the characterization results of the iron oxide layer.

Indeed, Rubashov et al. [315] studied Fe2O3 as a potential catalyst for the decomposition of ozone. The catalytic activity of Fe2O3 was reported to be nonexistent when the catalyst was in the form of aggregated particles whereas when it was in the form of dispersed particles, the catalyst was shown to be efficient for the decomposition of ozone. Moreover, the stability of Fe2O3 was relatively low and the formation of oxygen directly associated with the surface led to the poisoning of the catalyst in the manner of a Mars-Van Krevelen mechanism due to lattice oxygen [316]. Given the previous characterization results, we can conclude that the plasma post-treatment conditions used here lead to a non-efficient catalyst in ozonation of pyruvic acid.

4.4 Catalytic ozonation results in cobalt oxide based catalytic microreactors 4.4.1 Stability of the plasma deposited cobalt oxide coatings

In the same way than the iron oxide catalyst, leaching of the deposited Co3O4 layer was assessed after performing adsorption and ozonation of PA using FAAS measurements.

133 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

Figure n°51: Leaching results for the cobalt oxide layer deposited by plasma (as deposited and plasma post-treated) after adsorption tests and washing steps.

Figure n°52: Leaching results for the cobalt oxide layer deposited by plasma (as deposited and plasma post-treated) after catalytic ozonation tests and washing steps.

As shown in Figures 51-52, both prepared cobalt oxides present a strong stability towards adsorption and catalytic ozonation tests as well as towards washing steps, as only the initial washing step present a sufficient amount of cobalt that can be detected by FAAS measurements just like in the case of the iron oxide catalyst.

134 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation In a similar way, by considering a full leaching of the cobalt oxide layer, a value of 97 ppm can be calculated in 3.6 mL of solution. Thus, the leaching measured after the initial washing step only represents 0.59 % and 0.3 % of the total catalyst layer, respectively for the as-deposited cobalt oxide layer and plasma post-treated sample. Therefore, it can be concluded that the plasma deposited cobalt oxides are also stable upon adsorption and catalytic ozonation tests.

4.4.2 Adsorption tests results

Figure n°53: Adsorption tests for cobalt oxide thin films deposited by plasma (as deposited and plasma post-treated). Error bars indicate standard deviations for 3 replicate catalytic activity measurements for each case.

As shown in Figure 53, the adsorption values of PA on the surface of both cobalt oxides is close to the blank microreactor one as previously found for the iron oxide layer. The works of Gumuchian [317] showed an adsorption value of 33 % using 3 g of a commercial Co3O4 powder (measured specific surface area of 11.8 m²/g) with a

135 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation starting PA concentration of 5 mmol•L-1for a S/Q ratio of 200 000 m²/mol; S/Q given by the following equation:

푺 풎풄풂풕풂풍풚풔풕 × 푺푩푬푻 = 푸 [푷푨] × 푽풔풐풍풖풕풊풐풏

In the present study, the S/Q ratio was determined to be equal to 12 000 and 14 000 m²/mol respectively for the as-deposited cobalt oxide and plasma post- treated cobalt oxide layer. Based on the Langmuir isotherm established by Gumuchian for cobalt oxide, it can be concluded that the reported values of 4.6 % and 3.7% are relevant as the S/Q here is 10 orders of magnitude lower. Once again, a decrease in the adsorbed pyruvic acid concentration depending on the number of runs can be even observed and could be considered as a reproducibility issue. Nevertheless, these results could also be possibly explained by a decrease in the number of adsorption sites due to the sustained adsorption of pyruvic acid on the surface of the catalyst. The washing step would be therefore not sufficient enough to remove all the pyruvic acid molecules adsorbed on the catalyst.

4.4.3 Determination of the concentration of active sites for the Co3O4 catalyst Based on the previous BET measurements of the plasma post-treated cobalt oxide layer, the concentration of active sites of the catalyst can be determined by a simple ratio between the amount of adsorbed pyruvic acid and the determined real surface area of the catalyst:

풏풂풅풔풐풓풃풆풅 풑풚풓풖풗풊풄 풂풄풊풅 푪 = 풔 푺

Indeed, with the simple adsorption tests which consisted in passing pyruvic acid in the catalytic microreactor, a value of 3.7% ± 0.3% of adsorbed pyruvic acid was determined by HPLC measurements.

136 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation Consequently, with an initial concentration of pyruvic acid of 0.3 mol•L-1 and a flow of 3.6 mL•h-1, the total amount of adsorbed pyruvic acid after one hour corresponds to a value of 5•10-8 mol.

From the BET measurements of the plasma post-treated layer, a value of 3•10-2 m² was found for the total developed surface area. Therefore, with the previous equation, a value of 1.8•10-6 mol/m² was found for the concentration of actives sites. Moreover, in her thesis, Gumuchian found a similar value (1.9•10-6 mol/m²) using commercial Co3O4 in the form of nanoparticles (500 nm of diameter) which could lead to the assumption that the adsorption of pyruvic acid do not depend of the morphological structure. However, this assumption will not be further developed here.

In addition, a simple calculation can be done in order to evaluate the steric hindrance induced by pyruvic acid on the surface of the catalyst by using density functional theory (DFT) model evaluation and more specifically the total polar surface area (TPSA). Indeed, TPSA is a computer based simulation that takes into account the polar surface of a molecule. In our case, the pyruvic acid molecule is assumed to be adsorbed on the surface of the catalyst via the polar functions. A value of 3•10-6 mol/m² was found considering a polar surface of 54•10-19 m² with the TPSA measurements. These results indicate that at saturation, the quantity of pyruvic acid adsorbed on the catalyst surface is the maximum possible due to steric hindrance.

Regarding the equilibrium constant KA, between the adsorption/desorption rate of PA on the surface of the catalyst, the latter was determined by the reporting the initial concentration of PA (0.3 mol/m3) on the Langmuir isotherm obtained by

Gumuchian and a value of KA = 2 mg/g was found.

4.4.4 Results of catalytic ozonation tests

137 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation

Figure n°54: Catalytic ozonation performance for cobalt oxide thin film deposited by plasma (as deposited and plasma post-treated). Error bars indicate standard deviations for 3 replicate catalytic activity measurements for each case.

As shown in Figure 54, it can be seen that the cobalt based catalysts prepared by MO-PECVD present a higher catalytic activity than the iron oxide based one. Moreover, the effect of the plasma post-treatment can be clearly observed as the microreactor with the plasma post-treated catalytic layer exhibits a higher activity (71 ± 1%) compared to the initial cobalt oxide layer (47.6 ± 1%). Actually, the decay of the pyruvic acid concentration due to catalysis adds an additional 40 % ± 6% of degradation as measurements performed in the initial blank microreactor showed a value of 30.7 ± 5%.

Unfortunately, for both Co3O4 microreactors, a decrease in the catalytic activity can be noticed despite the intermediate washing steps. However, a final washing step consisting in sending a solution of Milli-Q water with dissolved ozone (0.3 mM) during one hour at a flow rate of 3.6 mL•h-1 in the catalytic microreactor lead to a regeneration of the catalyst. In other words, a re-use of the catalytic

138 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation microreactor is possible as the decrease of catalytic activity is not an irreversible phenomenon.

In order to understand the decrease in the catalytic activity of the cobalt oxides, a first assumption regarding an eventual poisoning of the catalyst can be made, as no leaching of the deposited layer was found by FAAS measurements.

This assumption in regards to the deactivation of the cobalt oxide catalyst may reside in the decrease of number of actives sites due to adsorption of oxidation products produced during ozonation of pyruvic acid. From the literature [318], by-products of the ozonation of pyruvic acid may be identified as acetic acid and oxalic acid. Thus, due to blocking of the active sites, these adsorbed products would lead to a poisoning effect preventing new flowing pyruvic molecules to access the catalyst actives sites and further be decomposed upon reaction with ozone. A possible explanation of the regeneration step can be explained by the fact that the dissolved ozone molecules will react with the adsorbed species. Thus, reaction of the latter with ozone would lead to the cleaning of the active sites of the Co3O4 catalyst. The latter assumption implies that ozone reacts preferentially with pyruvic acid to the detriment of the adsorbed oxidation products. The comparison of the reactivity of pyruvic acid and its oxidation products with ozone supports this assumption. Indeed, the reaction rate constants of direct ozonation are 3•10-5 and 4•10-2 L•mol-1•s-1 respectively for acetic acid and oxalic acid [319] whereas a reaction rate constant of 3.3•10-1 L•mol-1•s-1 was experimentally determined for pyruvic acid.

Another assumption regarding the deactivation of the catalyst may be due to the change in the oxidation state of the active sites during the different runs. This assumption may be addressed as a Mars-van Krevelen mechanism type where the intermediate species originating from the ozone self-decomposition poison the catalyst.

139 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation Assuming that the degradation of pyruvic acid is essentially due to heterogeneous catalysis, the catalytic activity will be related to the exposed facets of the cobalt oxide nanoparticles. The cobalt oxide obtained here is of the Co3O4 spinel-type consisting in octahedrally coordinated Co3+ and tetrahedrally coordinated Co2+ [320].

The [110] plane of Co3O4 has been reported to be the most catalytically active one as it favors Co3+ sites [321,322]. Indeed, this assumption was validated recently by calculating the surfaces energies of an ideal cobalt oxide crystal based on the density function theory (DFT) [323]. Su et al. demonstrated that the [111] 3+ plane provides more active Co sites. Nevertheless, the [110] plane of the Co3O4 spinel-type nanoparticle plane has the lowest surface energy, thus suggesting the most stable facet. Based on the previous XRD measurements and in accordance with the results from the literature, we can hypothetically presume that the [110] plane is responsible for the catalytic activity.

Therefore, the deactivation of the catalyst can possibly be explained by the considering the intermediates species generated during the self-decomposition of 2- •- ozone. As an example, different species such as O2 and O2 [324] with respective standard reduction potentials of -2.8V and -0.66V [325] may play the role of reducing agents. •- Considering the O2 specie, reduction of the cobalt oxide can be written as: •− ퟑ+ ퟐ+ 푶ퟐ + 푪풐 → 푶ퟐ + 푪풐

Derived from the Nernst equation, the equilibrium constant can be calculated as follows: (푬푶푪풐ퟑ+, 푪풐ퟐ+) − (푬푶푶 , 푶•−) 퐥퐨퐠 푲 = ퟐ ퟐ ퟎ. ퟎퟓퟗ

A value of 7.3•1041 was found indicating that this reaction is strongly promoted. Assuming this specie comes from the following self-decomposition of ozone [326], generation of the oxygen superoxide specie can be written as follows:

140 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation − •− • 푶ퟑ + 푶푯 → 푶ퟐ + 푯푶ퟐ

According to the works of Rao et al. [327] in an aqueous solutions, reduction of •- the O2 specie was found to be not reversible in equilibrium conditions.

The regeneration can further be explained by considering the standard reduction 3+/ 2+ potentials of E° (Co Co ) =1.81 V and E° (O3/O2) = 2.07 V with the following equation: ퟐ+ ퟑ+ ퟐ푶ퟑ + 푪풐 → ퟑ푶ퟐ + 푪풐

Like previously, the equilibrium constant can be calculated as follows: ퟐ × ((푬푶푶 , 푶 ) − (푬푶푪풐ퟑ+, 푪풐ퟐ+)) 퐥퐨퐠 푲 = ퟑ ퟐ ퟎ. ퟎퟓퟗ

A value of 6.5•108 was found for the previous equilibrium constant indicating that this reaction is less promoted than the reaction between oxygen superoxide and the catalyst in pure ozonation.

In addition, assuming that ozone reacts in an equimolar ratio of 1 with PA, this regeneration step requires 2 times more ozone than in the catalytic ozonation step, indicating a possible decrease in the availability of ozone molecules to effectively regenerate the surface during the catalytic ozonation step.

However, the reaction between ozone and the oxygen superoxide must also be considered:

− •− • 푶ퟑ + 푶푯 → 푶ퟐ + 푯푶ퟐ With k = 70 L•mol-1•s-1 from [326]

•− 푶ퟑ + 푶ퟐ → 푶ퟑ + 푶ퟐ

With k = 9•107 L•mol-1•s-1 from [328]

141 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation Therefore, from these rate constant values, it can be seen that ozone can oxidize the Co2+ present on the surface to Co3+ as the reaction between ozone and the superoxide specie is more rapid than the generation of this latter, leading to a regeneration of the catalytic layer.

In the light of these results and assuming that the reaction between the superoxide specie and PA is favored, it can be concluded that the previous formulated assumption (deactivation of the layer is assumed to be due to the reaction of the oxygen superoxide with the cobalt oxide layer) during the catalytic ozonation tests is probably wrong.

Unfortunately, to the best of our knowledge no values were found in the literature for the reaction rate constant for the latter reaction. Therefore, the most probable hypothesis for the deactivation of the layer resides in the fact that acetic acid and oxalic acid are adsorbed as by-products resulting from the catalytic ozonation of pyruvic acid on the surface of the cobalt oxide catalyst.

5. Conclusion A MO-PECVD plasma process was used to elaborate catalytic microreactors by depositing and activating iron and oxide-based catalysts at low temperatures. Characterization results showed that the plasma post-treatment step successfully induced a phase transition change and a structural refinement of the surface of the initial deposited layer. The efficiency of both catalysts was assessed in a catalytic ozonation process with pyruvic acid as a refractory probe compound. In the case of iron oxide layer, HPLC measurements demonstrated the inactivity of such catalyst even for the plasma post-treated sample. One assumption resides in the conditions used for the plasma post-treatment that are not fully optimized for the iron oxide layer in order to successfully achieve an adequate crystallization. In comparison to the deposited cobalt oxide layer, the decomposition rate of an organic pollutant with ozone was increased in the presence of a cobalt oxide layer.

142 Chapter 4: Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation In presence of the plasma deposited cobalt oxide layer, 20 % of additional degradation was found whereas when the layer was post-treated by argon plasma, the effect was doubled (40%), indicating a synergistic effect between the surface morphology of the coating and PA. These results are very promising considering the low residence time of only 0.75 second in the microreactor. Moreover, the use of plasma processes for the deposition catalyst offers interesting perspectives in the depollution field as it minimizes the potential nanotoxicity of the surface and suggests a long-term viability of the coating as demonstrated with FAAS measurements.

However, a decrease of the catalytic activity for the post-treated cobalt oxide layer was found indicating a possible coverage of the active sites by intermediate products generated during catalytic ozonation. Nevertheless, the catalyst was regenerated by passing ozone dissolved in Milli-Q water indicating that this phenomenon is reversible. Therefore, deactivation could possibly be avoided by increasing the O3/PA ratio.

Further investigations regarding the kinetics of catalytic heterogeneous ozonation will be studied in the next chapter by performing computer simulations using the Comsol Multiphysics software.

143

144 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor

1. Introduction Ozone is often used in industry for the disinfection of a great number of micropollutants, as well as in improving the taste, smell and destruction of colors and pathogenic germs [329-330]. The mechanisms of its reaction with different pollutants have been extensively studied over the last 25 years [319,324,331]. Once dissolved in water, ozone decomposes and generally leads to the formation of unstable elements with a lifetime that may extend from a few milliseconds to several days. However, its decomposition mostly leads to the formation of hydroxyl radicals with a high oxidation potential [332]. The self-decomposition of ozone in water depends on the temperature, its concentration in water, on the alkalinity of the solution to be treated and pH.

Indeed, it has been demonstrated that an increase in the pH accelerates the self- decomposition and that ozone behaves as a precursor agent of hydroxyl radicals [333].This indirect reaction and non-selective attack may lead to a full mineralization. In contrast, at a lower pH, ozone will behave as a disinfecting agent, by reacting directly on the pollutants in solution. In this case, the reaction is selective and may lead to an accumulation of by-products in the solution to be treated. In order to increase the decomposition of ozone into hydroxyl radicals, ozone has often been coupled with UV light and/or H2O2 [334,335].

These types of AOPs are efficient for the elimination of a large number of pollutants which are difficult to be degraded by simple ozonation such as carboxylic acids. However, they require a heavy maintenance and a good

145 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor optimization in order to avoid the formation of scavengers for hydroxyl radicals [336-338]. Therefore, a possible solution may reside in the use of catalytic ozonation where a supported catalyst such as activated carbon or metallic oxides is introduced in the reaction system. In spite of an impressive number of papers devoted to catalytic ozonation [339- 341], no clear consensus on the mechanisms of degradation of organic pollutants was reached. The reason can be that the mechanisms differ according to (i) the nature of the pollutant, (ii) the nature of the catalyst and (iii) the solution properties such as pH and conductivity.

The objective of this study is to propose a Computational Fluid Dynamics (CFD) model in order to find suitable mechanisms describing the degradation of an ozone refractory compound, pyruvic acid (PA), in the presence of ozone in a catalytic microreactor containing a plasma-deposited cobalt oxide coating. The model will be elaborated on the basis of the previous experimental dataset (Chapter 4) and by taking into account deactivation of the catalyst.

2. Ozone self-decomposition and simple ozonation simulation model 2.1 Description of the simple ozonation simulation model As shown in the previous chapter, the mixed solution of dissolved ozone and pyruvic acid passes through the catalytic microreactor, via the use of a syringe pump, by pulling backward the mixed solution from the batch reactor into a syringe. However, before performing a numerical simulation of the catalytic microreactor, kinetics in the bulk have to be determined. For this purpose, the numerical modeling of the self-decomposition of ozone and further reaction with pyruvic acid was assessed by elaborating a 0D model. This model takes into account direct (ozone as the reactant) and indirect (hydroxyl radicals as reactive species) ozonation of pyruvic acid. In addition, assuming that oxalic acid and acetic acid

146 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor are possibly by-products of the oxidation of pyruvic acid, indirect and direct ozonation of the latter were also taken into account.

The model used for ozone self-decomposition derives from [342] whereas kinetics for direct and indirect ozonation of pyruvic acid derives from [87]. The model used in this study is presented in Table 17.

Table n°17: Simplified kinetic model describing the self-decomposition of ozone and reaction with pyruvic acid. N° Reaction Reaction rate constant Reference O  HO O  HO -1 -1 1 3 2 2 k1 = 70 L•mol •s [326] HO H  O 5 -1 2 2 2 k2 = 3.2•10 s [71]   3 H  O   k = 2•1010 L•mol-1•s-1 [71] 2 HO2 3 O  O O  O 7 -1 -1 4 3 2 3 2 k4 = 9•10 L•mol •s [328] O  H O HO OH  O 7 -1 -1 5 3 2 2 k5 = 5.2•10 L•mol •s [319]  6 PA OH  AC 7 -1 -1  k6 = 3.1•10 L•mol •s [71] 7 PA OH  AOX  6 -1 -1 8 AC OH CO2  H 2O k7 = 1.4•10 L•mol •s [285]  7 -1 -1 9 AOX  OH CO2  H 2O k8 = 1.6•10 L•mol •s [285] 10 PA  O  AC 3 -1 -1 -1 k9 = 1.3•10 L•mol •s [284] PA  O  AOX 11 3 AC  O CO  H O -5 -1 -1 12 3 2 2 k10 = 3•10 L•mol •s [284] AOX  O CO  H O -1 -1 13 3 2 2 k11 = 0.4 L•mol •s [284]

In this model, ozone self-decomposition leads to the formation of two types of •- • radicals: the oxygen super-oxide (O2 ) and the hydroxyl radical OH. To the best of our knowledge, the first one is not selective and its reaction mechanism was poorly studied, whereas the hydroxyl radical action mechanism was thoroughly studied and the literature is rich in matters of kinetic constant values [71]. Therefore, the oxygen super-oxide will not be taken as an oxidizing agent in our simulations.

147 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor

The reaction of ozone on pyruvic acid (PA) was taken here as direct (reactions 10-11) and indirect (reactions 6-7) and by assuming that the degradation of pyruvic acid leads to two kind of intermediate species: oxalic acid (AOX) and acetic acid (AC). In addition, direct or indirect ozonation of the intermediate pollutants is further assumed to lead to full mineralization. 3 As initial data for the simulation, initial concentrations of PA and O3 (0.3 mol/m ) as well as the pH value (pH = 3) are known.

2.2 Simulation results of simple ozonation in batch reactor A numerical simulation with the previous established model (Table 17) was carried out.

Figure n°55: Simulations results of the evolution of the concentrations of PA,

AOX and AC in the batch reactor as a function of time (pH= 3, [PA]0= 0.3 3 3 mol/m , [O3] = 0.3 mol/m ).

As shown in Figure 55, after one hour of simple ozonation, the concentration of pyruvic acid reaches a value of 0.195 mol/m3 corresponding to 35% of

148 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor degradation of PA. This value has to be compared with the experimental one obtained in similar conditions.

In the previous chapter, it was shown that a final value of 30% of degradation was found experimentally after one hour of ozonation with an initial PA concentration of 0.3 mol/m3 and at pH = 3. By comparing the overall shapes of the evolution of PA concentration between the simulation results and the one obtained by the previous HPLC measurements, it can be concluded that the model previously established fits the experimental results.

The model was then used to determine the respective roles of direct and indirect ozonation according to the value of pH. Indirect ozonation is normally due to the reaction of hydroxyl radicals on PA (reaction n°6 in Table 17). However, the production of hydroxyl radicals is severely limited by the first step (reaction n°1 in Table 17) as the reaction rate constant (70 L•mol-1•s-1) is the lowest in ozone self-decomposition.

Figure 56 was obtained by reporting the following reaction rates for indirect and direct ozonation performed after one hour:

− 풗풊풏풅풊풓풆풄풕 = 풌ퟏ[푶푯 ][푶ퟑ]

풗풅풊풓풆풄풕 = 풌ퟏퟎ[푶ퟑ][푷푨]

149 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor

Figure n°56: Contribution of the reaction rate between indirect and direct ozonation as a function of pH (domain located between 3 and 9).

The simulation results shown in Figure 56 indicate that the contribution of ozone (direct ozonation) is more important than the contribution of hydroxyl radicals (indirect ozonation) for an acidic pH ranging between 3 and 7 for an initial concentration of PA of 0.3 mol/m3. At a pH = 3, it can be seen that the ratio of the two reaction rates is very low (1.8•10-5) indicating a poor contribution of hydroxyl radicals (indirect ozonation). The latter is increased when pH increases from 7 to 9, the ratio becoming greater than 1 at a pH = 8 until a value of 18 is reached at a pH = 9.

Therefore, the reaction rates ratio increasing along with pH indicates that ozone decomposes more rapidly in a basic solution than in an acidic one. Moreover, by - increasing the pH, the probability of encounter between O3 and OH in the reaction system increases, leading to a greater production of hydroxyl radicals. Thus, the latter will not contribute in degrading PA and its intermediates at a low pH.

150 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor These results are in agreement with previous studies indicating that ozonation self-decomposition is slower in acidic solutions [287] and confirm that indirect ozonation can be neglected in our simulations.

Based on the assumption that hydroxyl radicals poorly contribute in the ozonation of pyruvic acid as stated in the literature [286], the following reaction rate can be considered:

풌푶ퟑ,푷푨 푷푨 + 푶ퟑ → 푷풓풐풅풖풄풕풔

With the reaction rate: 풓 = 풌푶ퟑ, 푷푨 × [푷푨] × [푶ퟑ]

Figure n°57: First-order kinetic of the simple ozonation of pyruvic acid in batch 3 3 reactor (pH= 3, [PA]0= 0.3 mol/m , [O3] = 0.3 mol/m ).

As shown in Figure 57, the pyruvic acid decay was fitted with a linear regression curve. Indeed, ozone being produced in a continuous mode, an analysis of the decay can be assimilated to a first-order kinetic assuming the following reaction:

풓풂풑풑 = 풌풂풑풑 × [푷푨]

151 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor -1 With: 풌풂풑풑 = 풌푶ퟑ, 푷푨 × [푶ퟑ] in s

The differential equation can be written as:

풅[푷푨] 풓 = − = 풌 × [푷푨] 풂풑풑 풅풕 풂풑풑

And by rearranging the previous equation:

풅[푷푨] − = 풌 × 풕 [푷푨] 풂풑풑

Finally, integration of the latter leads to the following equation:

[푷푨] 퐥퐧 = −풌풂풑풑 × 풕 [푷푨]ퟎ

Therefore, the variation of the PA concentration as a function of time inside the batch reactor can be written as follows:

ퟎ −풌풂풑풑×풕 푪푷푨(풕) = 푪푷푨풆

−ퟒ −ퟏ With: 풌풂풑풑 = ퟏ • ퟏퟎ 퐬 .

A value of 1•10-4 s-1 at a pH 3 with a good correlation (R² = 0.98) was found for the apparent first-order rate constant. The global reaction rate corresponding to a second-order reaction can be therefore calculated and a value of 3.3•10-1 L•mol- 1•s-1 is obtained. This value is in good agreement (same order of magnitude) with the ones reported in the literature in the same range of concentration for pyruvic -1 -1 -1 -1 -1 -1 acid and pH (kO3,AP = 1.3•10 L•mol •s and 9.8•10 L•mol •s respectively at pH = 1 and 7 [284] ).

In addition, from Figure 55, it can be seen that the kinetic model chosen here leads to an accumulation of acetic acid in the batch reactor. Here, at a pH = 3 after one hour, the average production of AC reaches a value of 5.3•10-2 mol/m3 for an initial concentration of 0.30 mol/m3 of pyruvic acid. Regarding the concentration of oxalic acid, the model gives a low and almost constant concentration of the oxalic acid in solution as shown in Figure 55. The latter reached a value of 9.75•10-5 mol/m3 for an initial concentration of 0.3

152 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor mol/m3 of pyruvic acid. In fact, as shown in Figure 58 (magnification of the concentration of oxalic acid shown in Figure 55), oxalic acid is produced rapidly and then decreases in time.

Figure n°58: Evolution of the concentration of AOX in the batch reactor as a 3 3 function of time (pH= 3, [PA]0= 0.3 mol/m , [O3]0 = 0.3 mol/m ).

Therefore, the role of indirect ozonation being negligible, the previous set of equations can be reduced to the direct ozonation of PA but also accelerates the convergence of our model in Comsol. 3. Numerical simulation of the catalytic microsystem 3.1 Geometry and mesh used for the system The numerical simulations were carried out using a 2D geometry where a 100 µm X 1.5 cm rectangle which represents the side view of the microreactor as shown in Figure 59.

153 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor

Figure n°59: Geometrical implementation of the catalytic microreactor in Comsol Multiphysics.

Top Catalytic cover wall

Figure n°60: Magnification of the mesh used for the simulations in Comsol Multiphysics.

As shown in Figure 60, the mesh used for the computer simulation was refined near the surface for precision of calculation purposes. The model used for the simulations was a simple convection-diffusion that can be found in the Chemical Engineering module of Comsol Multiphysics. The kinetic models used for the computer simulations are described hereafter.

154 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor

3.2 Kinetic model used for the degradation of PA in the catalytic microreactor As demonstrated by Alvarez et al. [341], the contribution of indirect ozonation can be neglected in acidic conditions in the same range of concentrations used in the present study and with the same nature of catalyst, i.e. Co3O4. Alvarez showed that the conversion rate of pyruvic acid increased along with pH giving rise to the indirect way of ozonation through hydroxyl radicals.

The previous kinetic model was simplified by considering only direct ozonation of adsorbed PA and its adsorbed intermediates formed during the reaction. A scenario following an Eley-Rideal mechanism has been chosen to describe the

O3/CO3O4/PA interaction.

In this scenario, the various pollutants (noted hereafter Xi) are adsorbed on the catalyst and react with the dissolved ozone. This can be described by the following equations:

155 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor Table n°18: Kinetic model of catalytic ozonation chosen as an Eley-Rideal scenario.

Reaction Adjustable parameters N°

푘푎푑푠푃퐴 → ∗ 푘푎푑푠푃퐴 (푃퐴) + (∗) (푃퐴) 퐾퐴 = 1 ← 푘푑푒푠푃퐴 푘푑푒푠푃퐴

k ∗ q1 ∗ k 2 (푃퐴) + 푂3 → (퐴푂푋) q1

k ∗ raox ∗ k 3 (퐴푂푋) + 푂3 → (푃푟표푑푢푐푡) raox

푘푑푒푠퐴푂푋 → (퐴푂푋)∗ (퐴푂푋) + (∗) 푘 퐴푂푋 and 푘 퐴푂푋 4 ← 푎푑푠 푑푒푠 푘푎푑푠퐴푂푋

푘푞1 ∗ ∗ k 5 (푃퐴) + 푂3 → (퐴퐶) q1

푘푑푒푠퐴퐶 → (퐴퐶)∗ (퐴퐶) + (∗) 푘 퐴퐶 and 푘 퐴퐶 6 ← 푎푑푠 푑푒푠 푘푎푑푠퐴퐶

In the present model, the adsorption and desorption rate constants of oxalic acid are adjustable parameters as well as the reaction rate constant between the adsorbed pyruvic acid (kq1), adsorbed oxalic acid (kraox) and ozone. Acetic acid was supposed to be a stable specie which is not further degraded by ozone as shown in Gumuchian’s thesis [317].

In addition, Gumuchian also demonstrated the low adsorption of acetic acid (less than 1 %) on the surface of cobalt oxide powders (500 nm of diameter). Therefore, acetic acid will be assumed to be completely desorbed from the surface of the catalyst by fixing the adsorption rate constant of pyruvic acid, -2 -1 kadsAC, at 1 mol•m •s and consequently, the desorption rate constant, kdesAC at 10 mol•m-2•s-1.

156 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor As demonstrated in the previous chapter, the adsorption/desorption equilibrium constant of PA (Ka) and the concentration of active sites (Cs) are known and are respectively equal to 2 mg/g and 2•10-6 mol•m-2.However, simulations in Comsol being performed according to a flat geometry, the value for the concentration of active sites was fixed at 2•10-3 mol•m-2 as the geometric surface/developed surface ratio is decreased by a 103 factor as stated by the previous BET measurements (Chapter 4).

As shown in Table n°18, a number of adjustable parameters are left to be determined or varied in the present numerical model in an attempt to follow the deactivation mechanism of the post-treated cobalt oxide layer obtained experimentally. The adjustable parameters are summarized in Table 19.

Table n°19: Set of adjustable parameters used for the simulations.

Adjustable Parameter Description Reaction rate constant of ozone on adsorbed kq1 pyruvic acid Reaction rate constant of ozone on adsorbed kraox oxalic acid Adsorption rate constant of oxalic acid on the kads AOX surface of the catalyst Desorption rate constant of oxalic acid on the kdes AOX surface of the catalyst

3.2.1 Steady-state model without competition of adsorbed by-products A first simulation using the stationary mode was carried out by considering the direct ozonation on pyruvic acid with equations 1-2 of Table 18 and by applying the following reaction rate at the surface of the catalyst:

퐤퐪ퟏ퐊퐀퐂퐎 퐂퐬 퐂퐏퐀 퐯 = ퟑ ퟏ + 퐊퐀퐂퐏퐀

The present model relies on the assumption that ozone reacts very rapidly with the molecules of PA which are adsorbed on the surface of the catalyst and that

157 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor there is no competition between adsorbed intermediate species and PA; adsorption of the latter PA being rapid. Based on the previous assumptions, this first model allows to determine the value of the reaction rate constant kq1, which is the reaction rate constant of the direct ozonation on adsorbed pyruvic acid in a stationary mode.

Figure n°61: Degradation of pyruvic acid versus reaction rate constant of direct ozonation on adsorbed pyruvic acid.

As shown in Figure 61, the degradation of pyruvic acid reaches a plateau at a value of 53%; plateau that reflects the diffusion regime. However, as mentioned in the previous chapter, it was found that the decomposition of pyruvic acid in the presence of the Co3O4 catalyst reached a value of 71% ± 3% after one hour of reaction, whereas for the blank microreactor (e.g. without catalyst), a degradation of 31% ± 5 % was found. The degradation due to catalysis reaches a value of 40% ± 8 % and it can be concluded that the catalytic ozonation reaction studied in the microreactor is not limited by diffusion with the previous assumptions, (i.e. no competition between adsorbed intermediate species and PA during the first run). -1 -1 Therefore, a value of 0.3 mol •L•s was found for kq1 corresponding to a degradation of 40 % of the initial pyruvic acid concentration.

158 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor 3.2.2 Steady-state model with competition of adsorbed by-products Once the reaction rate constant of ozone on adsorbed pyruvic acid determined, a second model was elaborated by including adsorbed by-products resulting from the oxidation of PA.

Knowing the values of KA, the concentration of active sites and Kq1, the reaction rate constant between adsorbed pyruvic acid and ozone, the other adjustable parameters were varied in order to retrieve the deactivation kinetic of the cobalt oxide catalyst found experimentally. As a first assumption, the value for the equilibrium constant of AOX was fixed as equal to the PA one, i.e. KA = 2, leaving kraox as the only adjustable parameter as indicated in Table 19.

Table n°19: Parameters used for the catalytic ozonation simulation. kads PA 1 kads AOX 1 kdes PA kads PA/Ka kdes AOX kads AOX/Ka Ka 2 kads AC 0.1 kq1 0,3 kdes AC 10 Adjustable kraox Cs 0.002 parameter

Figure n°62: Evolution of the degradation of pyruvic acid as a function of the reaction rate constant between ozone and adsorbed oxalic acid.

159 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor

As shown in Figure 62, initially, when there is no reaction between ozone and adsorbed oxalic acid, a value of 31.5 % of degradation is reached. The steady- state model reflecting long reaction times, the latter value indicates the non- deactivation of the catalyst. However, a value of 14 % of degradation was found experimentally after three hours of catalytic ozonation with the plasma post- treated cobalt oxide catalyst. Here, it can be seen that an increase of the reaction rate constant do not lead to a significant change in the degradation of pyruvic acid (4.4 % of variation). Therefore, the equilibrium constant chosen previously for AOX is not accurate, meaning that this chemical specie does not adsorb with the same rate as AP on the surface of the catalyst, thus, in opposition with the initial assumption. Consequently, a simulation with the assumption that AOX is irreversibly adsorbed on the catalyst surface (kdes AOX = 0) was performed. The values of the different kinetic rate constants taken in this new simulation are summarized in Table 20 hereafter:

Table n°20: Parameters used for the numerical simulation of the degradation of PA in the catalytic microreactor.

kads PA 1 kads AOX 1

kdes PA 0.5 kdes AOX 0

Ka 2 kads AC 0.1

kq1 0,3 kdes AC 10 Adjustable kraox Cs 0.002 parameter

160 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor

Figure n°63: Evolution of the degradation of pyruvic acid as a function of the reaction rate constant between ozone and adsorbed oxalic acid.

Figure n°64: Magnification of the evolution of the degradation of pyruvic acid as a function of the reaction rate constant between ozone and adsorbed oxalic acid.

According to Figure 63, when the rate constant of direct ozonation between pyruvic acid and ozone is rapid, a maximum steady sate value of 26 % is reached. The aim of this simulation being the determination of the minimum value of kraox, it can be seen in Figure 64, that for a value of kraoxequal to 1•10-2 mol- 1•L•s-1, the steady-state degradation reaches a value of 15 %. Therefore, the value

161 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor of the reaction rate constant between oxalic acid and ozone was fixed equal to 1•10-2 mol-1•L•s-1, the final percentage of degradation obtained experimentally after 3 runs being 14 %.

3.2.3 Simulation in the time-dependent model After determining the previous adjustable parameters, a numerical simulation was performed in a time-dependent model and compared to the experimental results. The parameters chosen for the different constants are summarized in Table 21.

Table n°21: Parameters used for the numerical simulation of the degradation of PA in the catalytic microreactor.

kads PA 1 kads AOX 1 kdes PA 0.5 kdes AOX 0 Ka 2 kads AC 0.1 kq1 0,3 kdes AC 10 kraox 0.01 Cs 0.002

Figure n°65: Evolution of the average degradation of pyruvic acid as a function of time at the outlet of the catalytic microreactor.

162 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor The simulation result of the PA decay displayed in Figure 65 show that a degradation of 25 % is reached after a few seconds whereas the deactivation of the catalyst occurs rapidly reaching a plateau at a value of 11.9 % of degradation after 800 seconds. In the first steady-state model (shown in section 3.2.1), a value of 0.3 mol-1•L•s-1 was fixed for kq1 in order to obtain a degradation value of 40 % of the initial PA concentration. The latter value should be reached as no competition of the adsorbed by-products occurs at the beginning of the catalytic ozonation. However, only a value of 25 % of degradation is reached in the time-dependent model.

The values of the degradation of pyruvic acid should be 40, 30 and 14 % respectively after 1, 2 and 3 hours of reaction corresponding to the runs performed in the microreactor. The simulation results show that to the expected steady-state value of 15 % is reached after only 400 seconds of reaction. Normally, the PA degradation should tend to a value of 15 % after 3 hours of reaction. Furthermore, results from the numerical simulation indicate that a stationary state is reached rapidly which is not the case experimentally.

These differences may probably be explained by convergence issues between the stationary models and the time-dependent one. Another explanation may reside in the fact that the deactivation occurs so rapidly that the value of 40 % of degradation of the initial PA expected in the absence of deactivation of the catalyst will probably never be observed in the time-dependent model. Nevertheless, the desorption rate constant for oxalic acid onto the surface of the catalyst was varied in a new simulation in order to assess the initial assumption based on the irreversible bonding of oxalic acid on the surface of the catalyst.

163 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor

Figure n°66: Evolution of the average degradation of pyruvic acid at the outlet of the catalytic microreactor as a function of time and desorption rate constant of oxalic acid on the catalyst.

From Figure 66, it can be seen that when the desorption rate constant of oxalic acid onto the surface of the catalyst increases with a value ranging from 0 to 1•10- 1 m-2•s-1, the steady-state value of PA degradation percentage increases from 11.9 to 18.2% and this steady-state value is reached more rapidly. In other words, this indicates that when the desorption rate constant increases, oxalic acid is less retained on the surface of the catalyst; thus leading to more available active sites. Therefore, the variation of the desorption rate constant do not enable to get closer to the experimental results.

4. Conclusion A numerical simulation of the catalytic ozonation of pyruvic acid was performed in a microreactor using the Comsol Multiphysics software. A first stationary model was used to evaluate the role of direct or indirect ozonation of pyruvic acid in the batch reactor. Results showed that direct ozonation mainly contributes to

164 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor the degradation of pyruvic acid. The set of equations was reduced to direct ozonation in second stationary model assuming an Eley-Rideal scenario.

With this second model, the reaction rate of the adsorbed pyruvic acid and ozone was determined and implemented in a time dependent model. The time-dependent model was developed to retrieve the experimental results showing the deactivation of the plasma post-treated cobalt oxide layer in catalytic ozonation of pyruvic acid. Results showed that the model partially fitted the results obtained in the catalytic microreactor as the value of 15 % of degradation was found after 800 seconds instead of the 3 hours expected from the experimental results. A first explanation would consist in reviewing the initial assumption that catalytic ozonation undergoes an Eley-Rideal mechanism. Recently, De Souza et al [343] proposed an interesting numerical model in order to estimate the key parameters of chemical reactions independently of the catalytic mechanism used. By lack of time, de Souza’s numerical model was not used in our system but seems to be quite promising to investigate for future works. Moreover, the difference between the simulation results and the experimental ones could be due to the lack of information concerning the different adsorption and desorption reactions rate constants. In order to perform a deeper study, additional information on these reaction rate constants should be obtained, for example, by the use of the temperature jump method [344].

Furthermore, the approach proposed in this study demonstrates that kinetics of catalyzed reactions can be determined in microreactors instead of using conventional batch reactors. For instance, microreactors allow to control of the flow of the reactants which is not the case in conventional batch reactors. Although the use of microreactors strongly limits the diffusion phenomenon as shown in the present chapter, diffusion can still occur for a catalyzed reaction in a given microreactor geometry. This can be overcome by achieving a downscaling of the microreactor. Nevertheless, the downscaling of a microreactor is limited due to the

165 Chapter 5: Numerical simulation of the pyruvic acid degradation by catalytic ozonation in a catalytic microreactor technological implementation of such devices. Actually, as an example, a minimal value of 50 µm for the height of a Cyclic Olefin Copolymer based microreactor can be reached.

Finally, it would be interesting to confront the numerical method proposed in this study with an online analytical technique that could allow the monitoring of the chemical reactions occurring in the microreactor. Coherent Anti-Stokes Raman Scattering (CARS) spectroscopy appears to be one of the most promising techniques that could describe the different reaction pathways occurring at the surface of the catalyst, by direct observation of the time dependent evolution of the spatial concentration profiles of both reactants and products in a microreactor. Relying on their inherent vibrational signature using Raman-based microscopy, former studies have demonstrated that this technique could lead to label-free, chemically specific, quantitative determination of local concentrations of reactants and products, with 3D submicron spatial resolution (400 nm in x and y directions, 1.5 µm in z direction) [259,345-347]. Therefore, this quantitative in- situ method opens new avenues for monitoring reactions in microfluidic environments in order to develop a numerical predictive tool. This predictive tool would allow determining the synergistic effect between a given catalyst and the pollutant in catalytic ozonation and more generally, the reactants in catalyzed reactions.

166 General conclusion and outlooks General conclusion and outlooks

The objective of the work presented in the present thesis was to elaborate a catalytic microreactor by plasma processes in order to develop an innovative analytical tool providing unique information on reaction kinetics of catalysed reactions. In the present thesis, synergetic effects between ozone and different catalysts for the oxidation of organic pollutants in water were studied.

In the first Chapter, information on the commonly used microfluidic materials as well as catalysts used in heterogeneous catalytic ozonation and ozone refractory pollutants were retrieved from the literature in order to implement a catalytic microreactor using low pressure plasma processes. Polymer-based materials such as PDMS, NOA, THV and COC were identified as suitable microfluidic materials that could possibly withstand low pressure plasma processes. Regarding the nature of the catalysts, iron oxide and cobalt oxide were chosen to be deposited in a low pressure plasma process. Pyruvic acid was chosen as a refractory probe pollutant in order to assess the efficiency of the different catalysts in catalytic ozonation.

Chapter 2 covered the preparation of the thin films. The low pressure plasma processes techniques used for the deposition of the support layer as well as for the catalyst and its activation were described. After a brief theoretical introduction, analytical techniques for the surface characterizations or for the analyses of the collected liquid samples employed in the present work were described.

In Chapter 3, a screening among the polymer-based microfluidic materials identified in the previous chapters was performed by depositing a primer silica- like thin film, acting as a support layer between the catalyst and the polymer substrate, using two different plasma processes, i.e., PECVD and sputtering. Assessed by WCA measurements in air and water storage, aging results showed that the silica-like layer successfully rendered COC, NOA, PDMS and THV hydrophilic for several weeks. However, air aging revealed the instability of the

167 General conclusion and outlooks silica layer on PDMS surfaces due to the well-known hydrophobic recovery of this material. In opposition, the silica-like coated COC appeared to be the most promising for microfluidic applications because of its high stability in terms of hydrophilicity in air and water storage and rapid prototyping compared to THV and NOA. As a consequence, COC was chosen as the starting material for the elaboration of the catalytic microreactor.

Although the aging study show good results in terms of long-lasting hydrophilicity for the different polymer-based substrates, the different mechanisms involved in the plasma chemistry remains to be studied through the use of in-situ characterization techniques such as Optical Emission Spectroscopy (OES) or Mass Spectroscopy.

As shown in Chapter 4, a MO-PECVD plasma process was used to deposit and activate iron and oxide-based catalysts at low temperatures. Characterization results demonstrated the importance of the plasma post-treatment step, as the latter successfully induced a phase transition change and a structural refinement of the surface of the initial deposited layer for both catalysts. Efficiency of both catalysts was assessed performing catalytic ozonation with pyruvic acid as a refractory probe compound.

In the case of iron oxide layer, HPLC measurements demonstrated the inactivity of the latter, even for the plasma post-treated sample, indicating that the conditions used for the plasma post-treatment were not fully optimized for the iron oxide layer to successfully achieve an adequate crystallization.

In comparison, the decomposition rate of an organic pollutant with ozone was increased in the presence of a cobalt oxide layer. With the plasma deposited cobalt oxide layer, 20 % of additional degradation was found whereas the effect

168 General conclusion and outlooks was doubled (40%) when the layer was post-treated by argon plasma, indicating a synergistic effect between the surface morphology of the coating and PA.

Moreover, the use of plasma processes for the deposition catalyst offers interesting perspectives in the depollution field as it minimizes the potential nanotoxicity of the surface and suggests a long-term viability of the coating as demonstrated with FAAS measurements. A decrease of the catalytic activity for the post-treated cobalt oxide layer was found indicating a possible coverage of the active sites by intermediate products generated during catalytic ozonation. Nevertheless, it was shown that the catalyst could be regenerated by flushing ozone dissolved in Milli-Q water, indicating that this phenomenon is reversible. Therefore, experiments with a higher ozone concentration are suggested for a better understanding of the catalytic effect on the product selectivity.

In Chapter 5, a numerical study dealing with the reactions taking place on the surface of the post-treated cobalt oxide layer during catalytic ozonation was carried out using the Comsol Multiphysics software. It was shown that the model used to describe the degradation of pyruvic acid in a catalytic microreactor only partially fitted the experimental data.

This could be due to the lack of information concerning the different adsorption and desorption reactions rate constants of the intermediate species generated during the catalytic ozonation step. Indeed, catalytic reactions involve numerous steps which are adsorption, reaction and desorption. Therefore, it is suggested to refine the CFD model developed during this work in order to establish the details of the reaction network between the adsorbed species and gas phase reactions.

At the end of this work, it should be recognized that the complexity of the reaction mechanisms in catalytic microreactors does not allow a simplified analysis. Fundamental advances on numerical models will be required before complete models are successful in modelling the chemical phenomena related to

169 General conclusion and outlooks interfaces and for instance, providing access to the information on activation energies for various catalyst/reagent couple.

However, this can be achieved through the use of the Coherent Anti-Stokes Raman Spectroscopy technique (CARS technique) as an online analytical tool. Results obtained by the CARS technique would lead to a model that could be used as a tool to predict the relevance and the direction of future improvement strategies regarding catalyzed chemical reactions.

170 List of publications List of publications

 Da Silva, B., Schelcher, G., Winter, L., Guyon, C., Tabeling, P., Bonn, D. and Tatoulian, M. Study of the stability and hydrophilicity of plasma- modified microfluidic materials, in preparation.  Da Silva, B., Schelcher, G., Guyon, C., Ognier, S., Da Costa, P., Bonn, D. and M. Tatoulian, Development of catalytic microreactors: comparison of the performance of plasma-deposited iron and cobalt oxides in catalytic ozonation, in preparation.  Da Silva, B., Habibi, M., Ognier, S., Mostafavi-Amjad, J., Khalesifard, H. R. M., Tatoulian, M. and Bonn, D. Silver nanocluster-based catalytic microreactor for water purification, in preparation.  Ladner, Y., D'Orlyé, F., Perréard, C., Da Silva, B., Guyon, C., Tatoulian, M. et al. (2014). Surface Functionalization by Plasma Treatment and Click Chemistry of a New Family of Fluorinated Polymeric Materials for Microfluidic Chips. Plasma Processes and Polymers, 11(6),  Morscheidt, W., Cavadias, S., Rousseau, F. and Da Silva, B. (2013). Pollution of the Rhine River: An introduction to numerical modelling. Education For Chemical Engineers, 8(4), e119-e123.  Ladner, Y., d'Orlyé, F., Perréard, C., Da Silva, B., Guyon, C. and Tatoulian, M. et al. (2013). Surface Functionalization of COC Microfluidic Materials by Plasma and Click Chemistry Processes. Plasma Processes and Polymers, 10(11), 959-969.  Chen, G., Guyon, C., Zhang, Z., Da Silva, B., Da Costa, P. and Ognier, S. et al. (2013). Catkin liked nano-Co3O4 catalyst built-in organic microreactor by PEMOCVD method for trace CO oxidation at room temperature. Microfluidics and Nanofluidics, 16(1-2), 141-148.

171

172 Bibliography Bibliography

1 Behr, A. (2005). Organometallic Compounds and Homogeneous Catalysis. Ullmann’s Encyclopedia of Industrial Chemistry. Weinheim: Wiley-VCH, 1-43.

2 Cornils, B., Herrmann, W.A., Horvath, I.T., Leitner, W., Mecking, S., Olivier-Bourbigou, H., Vogt, D.(2005). Multiphase Homogeneous Catalysis.Weinheim: Wiley-VCH.

3 Knozinger, H., Kochoefl, K., 2005. Heterogeneous Catalysis and Solid Catalysts. Ullmann’s Encyclopedia of Industrial Chemistry. Weinheim: Wiley-VCH, 1-117.

4 Aehle, W. (2004). Enzymes in industry. Weinheim: Wiley-VCH, 1-274.

5 Liese, A., Lütz, S. (2004). Non-immobilized Biocatalysts in Industrial Fine Chemical Synthesis. Ullmann’s Encyclopedia of Industrial Chemistry. Weinheim: Wiley-VCH.

6 Prenosil, J.E., Kut, O.M., Dunn, I.J., Heinzle, E. (2007). Immobilized Biocatalysts. Ullmann’s Biotechnology and Biochemical Engineering. Weinheim: Wiley-VCH, 683- 734.

7 Srinivasan, R., Hsing, I., Berger, P., Jensen, K., Firebaugh, S. and Schmidt, M. et al. (1997). Micromachined reactors for catalytic partial oxidation reactions. Aiche Journal, 43(11), 3059-3069.

8 Mills, P. L. Quiram, D. J., Ryley, J. F. (2007). Microreactor technology and process miniaturization for catalytic reactions - A perspective on recent developments and emerging technologies. Chemical Engineering Science 62, 6992-7010.

9 Watts, P., Pascali, G. and Salvadori, P. (2012).Positron Emission Tomography Radiosynthesis in Microreactors. Journal Of Flow Chemistry, 2(2), 37-42.

10 Morini, G. L. (2004). Single-phase convective heat transfer in microchannels: a review of experimental results. International Journal of Thermal Sciences, 43 (7), 631-651.

11 Trachsel, F., Hutter, C., Rohr, P. R. (2008) Transparent silicon/glass microreactor for high-pressure and high-temperature reactions. Chemical Engineering Journal 135S, S309- S316.

12 A.V. Pattekar, M.V. Kothare.(2003).Novel microfluidic interconnectors for high temperature and pressure applications, J. Micromech. Microeng. 13, 337-345.

13 Gray, B., Jaeggi, D., Mourlas, N., van Drieënhuizen, B., Williams, K., Maluf, N. and Kovacs, G. (1999).Novel interconnection technologies for integrated microfluidic systems. Sensors and Actuators A: Physical, 77(1), 57-65.

14 Nittis, V., Fortt, R., Legge, C. and de Mello, A. (2001). A high-pressure interconnect for chemical microsystem applications. Lab On A Chip, 1(2), 148.

15 Tze-Jung Yao, Sangwook Lee, Weileun Fang and Yu-Chong Tai,.(2000). Micromachined rubber O-ring micro-fluidic couplers. Proceedings IEEE Thirteenth Annual International Conference On Micro Electro Mechanical Systems.

173 Bibliography

16 Ratner, D., Murphy, E., Jhunjhunwala, M., Snyder, D., Jensen, K. and Seeberger, P. (2005).Microreactor-based reaction optimization in organic chemistry glycosylation as a challenge. Chem. Commun., 578-580.

17 Yen, B., Günther, A., Schmidt, M., Jensen, K. and Bawendi, M. (2005).A microfabricated gas-liquid segmented flow reactor for high-temperature synthesis: the case of CdSe quantum dots. Angew.Chem. Int. 44, 5447-5451.

18 Henriksen, T., Olsen, J., Vesborg, P., Chorkendorff, I. and Hansen, O. (2009).Highly sensitive silicon microreactor for catalyst testing. Rev. Sci. Instrum., 80(12), 124101.

19 Petralia, S., Verardo, R., Klaric, E., Cavallaro, S., Alessi, E. and Schneider, C. (2013). In-Check system: A highly integrated silicon Lab-on-Chip for sample preparation, PCR amplification and microarray detection of nucleic acids directly from biological samples. Sensors and Actuators B: Chemical, 187, 99-105.

20 Hartman, R. and Jensen, K. (2009).Microchemical systems for continuous-flow synthesis. Lab On A Chip, 9(17), 2495.

21 de Mas, N., Schmidt, M. and Jensen, K. (2014). Nested potassium hydroxide etching and protective coatings for silicon-based microreactors. J. Micromech. Microeng., 24(3), 035011.

22 Longstreet, A. and McQuade, D. (2013). Organic Reaction Systems: Using Microcapsules and Microreactors to Perform Chemical Synthesis. Accounts Of Chemical Research, 46(2), 327-338.

23 Sun, Y., Luo, Y. and Wang, X. (2012).Micro energy director array in ultrasonic precise bonding for thermoplastic micro assembly. Journal Of Materials Processing Technology, 212(6), 1331-1337.

24 Watts, P. and Haswell, S. (2005).The Application of Microreactors for Small Scale Organic Synthesis. Chemical Engineering and Technology, 28(3), 290-301.

25 Ukita, Y., Asano, T., Fujiwara, K., Matsui, K., Takeo, M. and Negoro, S. et al. (2008).Application of vertical microreactor stack with polystylene microbeads to immunoassay. Sensors and Actuators A: Physical, 145-146, 449-455.

26 Liu, H., Gong, H., Ramalingam, N., Jiang, Y., Dai, C. and Hui, K. (2007).Micro air bubble formation and its control during polymerase chain reaction (PCR) in polydimethylsiloxane (PDMS) microreactors. J. Micromech. Microeng., 17(10), 2055- 2064.

27 Ng, J., Nie, Y., Chuah, G. and Jaenicke, S. (2010). A wall-coated catalytic capillary microreactor for the direct formation of hydrogen peroxide. Journal Of Catalysis, 269(2), 302-308.

28 Jin, J., Dong, L., Zhang, K. and Liu, J. (2012).Recent Advances in Microfluidic Synthesis. Chin. J. Org. Chem., 32(1), 201.

174 Bibliography

29 Sotowa, K., Irie, K., Fukumori, T., Kusakabe, K. and Sugiyama, S. (2007). Droplet Formation by the Collision of Two Aqueous Solutions in a Microchannel and Application to Particle Synthesis. Chemical Engineering and Technology, 30(3), 383-388.

30 Li, H., Gao, X., Ding, H., Yang, M. and Pu, Q. (2012). Highly efficient Suzuki cross- coupling reaction within an open channel plastic microreactor immobilized with palladium complexes. Microfluidics and Nanofluidics, 12(6), 981-989.

31 Sackmann, J., Burlage, K., Gerhardy, C., Memering, B., Liao, S. and Schomburg, W. (2015). Review on ultrasonic fabrication of polymer micro devices. Ultrasonics, 56, 189- 200.

32 Ren, W., Perumal, J., Wang, J., Wang, H., Sharma, S. and Kim, D. (2014). Whole ceramic-like microreactors from inorganic polymers for high temperature or/and high pressure chemical syntheses. Lab Chip, 14(4), 779-786.

33 Ranjit Prakash, A., Adamia, S., Sieben, V., Pilarski, P., Pilarski, L. and Backhouse, C. (2006).Small volume PCR in PDMS biochips with integrated fluid control and vapour barrier. Sensors and Actuators B: Chemical, 113(1), 398-409.

34 Niu, Z., Chen, W., Shao, S., Jia, X. and Zhang, W. (2006).DNA amplification on a PDMS-glass hybrid microchip. J. Micromech. Microeng., 16(2), 425-433.

35 Liu, H., Gong, H., Ramalingam, N., Jiang, Y., Dai, C., & Hui, K. (2007). Micro air bubble formation and its control during polymerase chain reaction (PCR) in polydimethylsiloxane (PDMS) microreactors. J. Micromech. Microeng., 17(10), 2055- 2064.

36 Toriello, N., Liu, C. and Mathies, R. (2006).Multichannel Reverse Transcription- Polymerase Chain Reaction Microdevice for Rapid Gene Expression and Biomarker Analysis. Analytical Chemistry, 78(23), 7997-8003.

37 Cady, N., Stelick, S., Kunnavakkam, M. and Batt, C. (2005). Real-time PCR detection of Listeria monocytogenes using an integrated microfluidics platform. Sensors and Actuators B: Chemical, 107(1), 332-341.

38 Nakayama, T., Kurosawa, Y., Furui, S., Kerman, K., Kobayashi, M. and Rao, S. et al. (2006). Circumventing air bubbles in microfluidic systems and quantitative continuous- flow PCR applications. Anal Bioanal Chem, 386(5), 1327-1333.

39 Bartolo, D., Degré, G., Nghe, P. and Studer, V. (2008).Microfluidic stickers. Lab Chip, 8(2), 274-279.

40 Kim, S., Yang, Y., Kim, M., Nam, S., Lee, K. and Lee, N. et al. (2007). Simple Route to Hydrophilic Microfluidic Chip Fabrication Using an Ultraviolet (UV)-Cured Polymer. Adv. Funct. Mater., 17(17), 3493-3498.

41 Wägli, P., Homsy, A. and de Rooij, N. (2011).Norland optical adhesive (NOA81) microchannels with adjustable wetting behavior and high chemical resistance against a range of mid-infrared-transparent organic solvents. Sensors and Actuators B: Chemical, 156(2), 994-1001.

175 Bibliography

42 Begolo, S., Colas, G., Viovy, J. and Malaquin, L. (2011). New family of fluorinated polymer chips for droplet and organic solvent microfluidics. Lab Chip, 11(3), 508-512.

43 Shin, J., Park, J., Liu, C., He, J. and Kim, S. (2005). Chemical structure and physical properties of cyclic olefin copolymers (IUPAC Technical Report).Pure and Applied Chemistry, 77(5).

44 Deverell, J., Rodemann, T., Smith, J., Canty, A. and Guijt, R. (2011). UV initiated formation of polymer monoliths in glass and polymer microreactors. Sensors and Actuators B: Chemical, 155(1), 388-396.

45 Abele, S., Nie, F., Foret, F., Paull, B. and Macka, M. (2008).UV-LED photopolymerised monoliths. The Analyst, 133(7), 864.

46 Rånby, B., Yang, W. and Tretinnikov, O. (1999).Surface photografting of polymer fibers, films and sheets. Nuclear Instruments and Methods In Physics Research Section B: Beam Interactions With Materials and Atoms, 151(1-4), 301-305.

47 Stachowiak, T., Rohr, T., Hilder, E., Peterson, D., Yi, M., Svec, F. and Fréchet, J. (2003). Fabrication of porous polymer monoliths covalently attached to the walls of channels in plastic microdevices. Electrophoresis, 24(21), 3689-3693.

48 Mair, D., Geiger, E., Pisano, A., Fréchet, J. and Svec, F. (2006). Injection molded microfluidic chips featuring integrated interconnects. Lab On A Chip, 6(10), 1346.

49 Stachowiak, T., Mair, D., Holden, T., Lee, L., Svec, F. and Fréchet, J. (2007). Hydrophilic surface modification of cyclic olefin copolymer microfluidic chips using sequential photografting. J. Sep. Sci., 30(7), 1088-1093.

50 Yu, C., Xu, M., Svec, F. and Fréchet, J. (2002). Preparation of monolithic polymers with controlled porous properties for microfluidic chip applications using photoinitiated free- radical polymerization. J. Polym. Sci.A Polym. Chem., 40(6), 755-769.

51 Wirth, T. (2013).Microreactors in Organic Chemistry and Catalysis. Weinheim: Wiley.

52 Brandner, J. (2012). Microfabrication in metals, ceramics and polymers. Russ J Gen Chem, 82(12), 2025-2033.

53 J. Yoshida (2003), Microreactors:Epoch-making Technology for Synthesis. CMC Publishing Co., Ltd., Tokyo, Japan.

54 Hoera, C., Ohla, S., Shu, Z., Beckert, E., Nagl, S. and Belder, D. (2014). An integrated microfluidic chip enabling control and spatially resolved monitoring of temperature in micro flow reactors. Anal Bioanal Chem, 407(2), 387-396.

55 Day, D. and Gu, M. (2009). A microfluidic microreactor for the synthesis of gold nanorods. Nanotechnology, 20(16), 169801-169801.

56 Walker, J. (2004). Machining fundamentals. Tinley Park, Ill.: Goodheart-Willcox Co.

176 Bibliography

57 Vittoriosi, A., Hecht, K. and Brandner, J.J. (2012) Surface characterization for microstructured systems: methods and examples. Jahrbuch Oberflachentechnik68, 199.

58 Fischer, P., Romano, V., Blatter, A. and Weber, H. (2005). Highly precise pulsed selective laser sintering of metallic powders. Laser Phys. Lett., 2(1), 48-55.

59 Mitsos, A. and Barton, P. (2009). Microfabricated power generation devices. Weinheim: Wiley-VCH.

60 Bártolo, P. (2008).Virtual and rapid manufacturing. London: Taylor and Francis.

61 Marre, S., Adamo, A., Basak, S., Aymonier, C. and Jensen, K. (2010).Design and Packaging of Microreactors for High Pressure and High Temperature Applications. Industrial and Engineering Chemistry Research, 49(22), 11310-11320.

62 Iliescu, C., Chen, B. and Miao, J. (2008).On the wet etching of Pyrex glass. Sensors and Actuators A: Physical, 143(1), 154-161.

63 Mazurczyk, R., El Khoury, G., Dugas, V., Hannes, B., Laurenceau, E. and Cabrera, M. et al. (2008). Low-cost, fast prototyping method of fabrication of the microreactor devices in soda-lime glass. Sensors and Actuators B: Chemical, 128(2), 552-559.

64 Loppinet-Serani, A., Aymonier, C. and Cansell, F. (2008).Current and Foreseeable Applications of Supercritical Water for Energy and the Environment. Chemsuschem, 1(6), 486-503.

65 Brunner, G. (2009). Near and supercritical water. Part II: Oxidative processes. The Journal Of Supercritical Fluids, 47(3), 382-390.

66 Marre, S., Park, J., Rempel, J., Guan, J., Bawendi, M. and Jensen, K. (2008).Supercritical Continuous-Microflow Synthesis of Narrow Size Distribution Quantum Dots. Adv. Mater., 20(24), 4830-4834.

67 Zenker, M., Borden, R. and Barlaz, M. (2004). Biodegradation of 1,4-Dioxane Using Trickling Filter. J. Environ. Eng., 130(9), 926-931.

68 Asano, T. (2007).Water reuse. New York: McGraw-Hill.

69 Rosenfeldt, E. and Linden, K. (2004).Degradation of Endocrine Disrupting Chemicals Bisphenol A, Ethinyl Estradiol and Estradiol during UV Photolysis and Advanced Oxidation Processes. Environmental Science and Technology, 38(20), 5476-5483.

70 Eckenfelder, W., Bowers, A. and Roth, J. (1994).Chemical oxidation--technologies for the nineties. Lancaster: Technomic Pub. Co.

71 Buxton, G., Greenstock, C., Helman, W., Ross, A. and Tsang, W. (1988). Critical Review of Rate Constants for Reactions of Hydrated Electrons, Hydrogen Atoms and Hydroxyl Radicals in Aqueous Solution, Journal Of Physical and Chemical Reference Data, 17(2), 513.

72 Crittenden, J. (2005). Water treatment principles and design. Hoboken, N.J.: J. Wiley.

177 Bibliography

73 Oppenländer, T. (2003).Photochemical purification of water and air. Weinheim: Wiley- VCH.

74 Belgiorno, V., Naddeo, V. and Rizzo, L. (2011).Water, wastewater and soil treatment by advanced oxidation processes. lulu.com.

75 Kavanaugh, M. (2003).Removal of MTBE with advanced oxidation processes. Denver, CO: Awwa Research Foundation.

76 Rao, D. (2012). Wastewater treatment. Boca Raton, FL: CRC Press.

77 Litter, M., Candal, R. and Meichtry, J. (2014). Advanced Oxidation Technologies. Hoboken: Taylor and Francis.

78 Adewuyi, Y. (2005). Sonochemistry in environmental remediation I: Combinative and Hybrid Sonophotochemical Oxidation Processes for the Treatment of Pollutants in Water. Environmental Science and Technology, 39(10), 3409-3420.

79 Jain, R., Camarillo, M. K., Stringfellow, W. T. (2014). Drinking Water Security for Engineers, Planners and Managers. Oxford: Elsevier, Inc.

80 Cheremisinoff, N. P., Davletshin, A. (2015). Hydraulic Fracturing Operations: Handbook of Environmental Management Practices. New York: Wiley.

81 Mahamuni, N. and Adewuyi, Y. (2010). Advanced oxidation processes (AOPs) involving ultrasound for waste water treatment: A review with emphasis on cost estimation. Ultrasonics Sonochemistry, 17(6), 990-1003.

82 Namkung, K., Burgess, A., Bremner, D. and Staines, H. (2008).Advanced Fenton processing of aqueous phenol solutions: A continuous system study including sonication effects. Ultrasonics Sonochemistry, 15(3), 171-176.

83 Adewuyi, Y. and Oyenekan, B. (2007).Optimization of a Sonochemical Process Using a Novel Reactor and Taguchi Statistical Experimental Design Methodology. Industrial and Engineering Chemistry Research, 46(2), 411-420.

84 Gaya, U. and Abdullah, A. (2008). Heterogeneous photocatalytic degradation of organic contaminants over titanium dioxide: A review of fundamentals, progress and problems. Journal Of Photochemistry and Photobiology C: Photochemistry Reviews, 9(1), 1-12.

85 Shukla, P., Wang, S., Sun, H., Ang, H. and Tadé, M. (2010). Activated carbon supported cobalt catalysts for advanced oxidation of organic contaminants in aqueous solution. Applied Catalysis B: Environmental, 100(3-4), 529-534.

86 Sawant, D., Vinu, A., Mirajkar, S., Lefebvre, F., Ariga, K. and Anandan, S. et al. (2007). Silicotungstic acid/zirconia immobilized on SBA-15 for esterifications. Journal Of Molecular Catalysis A: Chemical, 271(1-2), 46-56.

178 Bibliography

87 Rivas, F., Carbajo, M., Beltrán, F., Acedo, B. and Gimeno, O. (2006). Perovskite catalytic ozonation of pyruvic acid in water: Operating conditions influence and kinetics. Applied Catalysis B: Environmental, 62(1-2), 93-103.

88 Ling, S., Wang, S. and Peng, Y. (2010).Oxidative degradation of dyes in water using 2+ 2+/ Co /H2O2 and Co peroxymonosulfate. Journal Of Hazardous Materials, 178(1-3), 385- 389.

89 Hirvonen, A., Tuhkanen, T. and Kalliokoski, P. (1996). Formation of chlorinated acetic acids during UV/H2O2-oxidation of ground water contaminated with chlorinated ethylenes. Chemosphere, 32(6), 1091-1102.

90 Destaillats, H., Alderson, T. and Hoffmann, M. (2001). Applications of Ultrasound in NAPL Remediation: Sonochemical Degradation of TCE in Aqueous Surfactant Solutions. Environmental Science and Technology, 35(14), 3019-3024.

91 Nakano, Y., Okawa, K., Nishijima, W. and Okada, M. (2003).Ozone decomposition of hazardous chemical substance in organic solvents. Water Research, 37(11), 2595-2598.

92 Tezcanli-Geuyer and G., Ince, N. H. (2004), Individual and combined effects of ultrasound, ozone and UV irradiation: a case study with textile dyes. Ultrasonics, 42, 603- 609.

93 Sato, C. and Yao, J. (2006).Simultaneous and Sequential Photosonolysis of TCE and PCE. J. Environ. Eng., 132(1), 32-41.

94 Crittenden, J., Liu, J., Hand, D. and Perram, D. (1997).Photocatalytic oxidation of chlorinated hydrocarbons in water. Water Research, 31(3), 429-438.

95 Entezari, M., Pétrier, C. and Devidal, P. (2003). Sonochemical degradation of phenol in water: a comparison of classical equipment with a new cylindrical reactor. Ultrasonics Sonochemistry, 10(2), 103-108.

96 Chen, X., Chen, J., Qiao, X., Wang, D. and Cai, X. (2008). Performance of nano- Co3O4/peroxymonosulfate system: Kinetics and mechanism study using Acid Orange 7 as a model compound. Applied Catalysis B: Environmental, 80(1-2), 116-121.

97 Yang, Q., Choi, H., Chen, Y. and Dionysiou, D. (2008). Heterogeneous activation of peroxymonosulfate by supported cobalt catalysts for the degradation of 2,4-dichlorophenol in water: The effect of support, cobalt precursor and UV radiation. Applied Catalysis B: Environmental, 77(3-4), 300-307.

98 Zhang, W., Tay, H., Lim, S., Wang, Y., Zhong, Z. and Xu, R. (2010).Supported cobalt oxide on MgO: Highly efficient catalysts for degradation of organic dyes in dilute solutions.Applied Catalysis B: Environmental, 95(1-2), 93-99.

99 Shi, P., Zhu, S., Zheng, H., Li, D. and Xu, S. (2013). Supported Co3O4 on expanded graphite as a catalyst for the degradation of Orange II in water using sulfate radicals. Desalination and Water Treatment, 52(16-18), 3384-3391.

179 Bibliography

100 Karthikeyan, S., Boopathy, R. and Sekaran, G. (2015). In situ generation of hydroxyl radical by cobalt oxide supported porous carbon enhance removal of refractory organics in tannery dyeing wastewater. Journal Of Colloid and Interface Science, 448, 163-174.

101 Rivas, F., Carbajo, M., Beltrán, F., Gimeno, O. and Frades, J. (2008). Comparison of different advanced oxidation processes (AOPs) in the presence of perovskites. Journal Of Hazardous Materials, 155(3), 407-414.

102 Gao, J., Cao, S., Tay, Q., Liu, Y., Yu, L. and Ye, K. et al. (2013). Molecule-Based Water-Oxidation Catalysts (WOCs): Cluster-Size-Dependent Dye-Sensitized Polyoxometalates for Visible-Light-Driven O2 Evolution. Sci. Rep., 3.

103 Balaska, A., Samar, M. and Grid, A. (2014). Phenol photodegradation process assisted with Wells-Dawson heteropolyacids. Desalination and Water Treatment, 54(2), 382-392.

104 Okuhara, T., Mizuno, N., Misono, M. (1996). Catalytic chemistry of heteropoly compounds. Adv. Catal., 41, 113-252.

105 Pourbeyram, S., Moosavifar, M. and Hasanzadeh, V. (2014).Electrochemical characterization of the encapsulated polyoxometalates (POMs) into the zeolite. Journal Of Electroanalytical Chemistry, 714-715, 19-24.

106 Vázquez, P., Pizzio, L., Cáceres, C., Blanco, M., Thomas, H. and Alesso, E. et al. (2000).Silica-supported heteropolyacids as catalysts in alcohol dehydration reactions. Journal Of Molecular Catalysis A: Chemical, 161(1-2), 223-232.

107 van Grieken, R., Aguado, J., López-Muñoz, M. and Marugán, J. (2002). Synthesis of size-controlled silica-supported TiO2 photocatalysts. Journal Of Photochemistry and Photobiology A: Chemistry, 148(1-3), 315-322.

108 Wei, W., Hao, S., Zhou, J., Huang, M. and Jiang, Y. (2004).Catalytic behavior of silica- supported chitosan-platinum-iron complex for asymmetric hydrogenation of ketones. Polym. Adv. Technol., 15(5), 287-290.

109 Konova, P., Stoyanova, M., Naydenov, A., Christokova, S. and Mehandjiev, D. (2006). Catalytic oxidation of VOCs and CO by ozone over alumina supported cobalt oxide. Applied Catalysis A: General, 298, 109-114.

110 Cheng, M., Song, W., Ma, W., Chen, C., Zhao, J., Lin, J. and Zhu, H. (2008). Catalytic activity of iron species in layered clays for photodegradation of organic dyes under visible irradiation. Applied Catalysis B: Environmental, 77(3-4), 355-363.

111 Iurascu, B., Siminiceanu, I., Vione, D., Vicente, M. and Gil, A. (2009).Phenol degradation in water through a heterogeneous photo-Fenton process catalyzed by Fe- treated laponite. Water Research, 43(5), 1313-1322.

112 Zhang, S., Liang, S., Wang, X., Long, J., Li, Z. and Wu, L. (2011). Trinuclear iron cluster intercalated montmorillonite catalyst: Microstructure and photo-Fenton performance. Catalysis Today, 175(1), 362-369.

180 Bibliography

113 Antonopoulou, M., Evgenidou, E., Lambropoulou, D. and Konstantinou, I. (2014). A review on advanced oxidation processes for the removal of taste and odor compounds from aqueous media. Water Research, 53, 215-234.

114 Thakur, D., Tiggelaar, R., Weber, Y., Gardeniers, J., Lefferts, L. and Seshan, K. (2011). Ruthenium catalyst on carbon nanofiber support layers for use in silicon-based structured microreactors. Part II: Catalytic reduction of bromate contaminants in aqueous phase. Applied Catalysis B: Environmental, 102(1-2), 243-250.

115 Mills, A. and Meadows, G. (1995). Heterogeneous redox catalysis: A novel route for removing bromate ions from water. Water Research, 29(9), 2181-2185.

116 Yeong, K., Gavriilidis, A., Zapf, R. and Hessel, V. (2003).Catalyst preparation and deactivation issues for nitrobenzene hydrogenation in a microstructured falling film reactor. Catalysis Today, 81(4), 641-651.

117 Illner, S., Hofmann, C., Löb, P. and Kragl, U. (2014).A Falling-Film Microreactor for Enzymatic Oxidation of Glucose. Chemcatchem, 6(6), 1748-1754.

118 Wang, K., Johnston, P. and Brown, R. (2014).Comparison of in-situ and ex-situ catalytic pyrolysis in a micro-reactor system. Bioresource Technology, 173, 124-131.

119 Tan, X. and Li, K. (2013).Membrane microreactors for catalytic reactions. J. Chem. Technol. Biotechnol., 88(10), 1771-1779.

120 Yube, K., Furuta, M. and Mae, K. (2007).Selective oxidation of phenol with hydrogen peroxide using two types of catalytic microreactor. Catalysis Today, 125(1-2), 56-63.

121 Aran, H., Chinthaginjala, J., Groote, R., Roelofs, T., Lefferts, L., Wessling, M. and Lammertink, R. (2011). Porous ceramic mesoreactors: A new approach for gas-liquid contacting in multiphase microreaction technology. Chemical Engineering Journal, 169(1- 3), 239-246.

122 Garcıa-Garcıa F. R., Kingsbury, B. F. K., Rahman, M. A., Li. K. (2012). Asymmetric ceramic hollow fibres applied in heterogeneous catalytic gas phase reactions. Catalysis Today, 193, 20-30. 123Pina, M., Mallada, R., Arruebo, M., Urbiztondo, M., Navascués, N., de la Iglesia, O. and Santamaria, J. (2011).Zeolite films and membranes.Emerging applications. Microporous and Mesoporous Materials, 144(1-3), 19-27.

124 Aran, H., Pacheco Benito, S., Luiten-Olieman, M., Er, S., Wessling, M. and Lefferts, L. et al. (2011).Carbon nanofibers in catalytic membrane microreactors. Journal Of Membrane Science, 381(1-2), 244-250.

125 Ye, S., Hamakawa, S., Tanaka, S., Sato, K., Esashi, M. and Mizukami, F. (2009). A one-step conversion of benzene to phenol using MEMS-based Pd membrane microreactors. Chemical Engineering Journal, 155(3), 829-837.

126 Asanomi, Y., Yamaguchi, H., Miyazaki, M. and Maeda, H. (2011).Enzyme- Immobilized Microfluidic Process Reactors. Molecules, 16(12), 6041-6059.

181 Bibliography

127 Wan, Y., Chau, J., Gavriilidis, A. and Yeung, K. (2001).Design and fabrication of zeolite-based microreactors and membrane microseparators. Microporous and Mesoporous Materials, 42(2-3), 157-175.

128 Conant, T., Karim, A. and Datye, A. (2007).Coating of steam reforming catalysts in non-porous multi-channeled microreactors. Catalysis Today, 125(1-2), 11-15.

129 Conant, T., Karim, A., Rogers, S., Samms, S., Randolph, G. and Datye, A. (2006). Wall coating behavior of catalyst slurries in non-porous ceramic microstructures. Chemical Engineering Science, 61(17), 5678-5685.

130 Stefanescu, A., van Veen, A., Mirodatos, C., Beziat, J. and Duval-Brunel, E. (2007).Wall coating optimization for microchannel reactors. Catalysis Today, 125(1-2), 16-23.

131 Yamamoto, S., Hanaoka, T., Hamakawa, S., Sato, K. and Mizukami, F. (2006).Application of a microchannel to catalytic dehydrogenation of cyclohexane on Pd membrane. Catalysis Today, 118(1-2), 2-6.

132 Tan, X., Liu, S. and Li, K. (2001). Preparation and characterization of inorganic hollow fiber membranes.Journal Of Membrane Science, 188(1), 87-95.

133 Drioli, E. and Giorno, L. (2010). Comprehensive membrane science and engineering. Amsterdam: Elsevier Science., 253-273.

134 Aran, H., Klooster, H., Jani, J., Wessling, M., Lefferts, L. and Lammertink, R. (2012). Influence of geometrical and operational parameters on the performance of porous catalytic membrane reactors. Chemical Engineering Journal, 207-208, 814-821.

135 Tan, X., Liu, N., Meng, B. and Liu, S. (2011). Morphology control of the perovskite hollow fibre membranes for oxygen separation using different bore fluids. Journal Of Membrane Science, 378(1-2), 308-318.

136 Kingsbury, B. and Li, K. (2009).A morphological study of ceramic hollow fibre membranes. Journal Of Membrane Science, 328(1-2), 134-140.

137 Gao, M., Zeng, Z., Sun, B., Zou, H., Chen, J. and Shao, L. (2012). Ozonation of azo dye Acid Red 14 in a microporous tube-in-tube microchannel reactor: Decolorization and mechanism. Chemosphere, 89(2), 190-197.

138 Hu, M., Hui, K. and Hui, K. (2014).Role of graphene in MnO2/graphene composite for catalytic ozonation of gaseous toluene. Chemical Engineering Journal, 254, 237-244.

139 Meille, V. (2006). Review on methods to deposit catalysts on structured surfaces. Applied Catalysis A: General, 315, 1-17.

140 Ahn, S., Choi, I., Kwon, O. and Kim, J. (2014). Hydrogen production through the fuel processing of liquefied natural gas with silicon-based micro-reactors. Chemical Engineering Journal, 247, 9-15.

182 Bibliography

141 Yu, X., Tu, S., Wang, Z. and Qi, Y. (2006).Development of a microchannel reactor concerning steam reforming of methanol. Chemical Engineering Journal, 116(2), 123-132.

142 Ni, Z., Seebauer, E. and Masel, R. (2005). Effects of Microreactor Geometry on Performance: Differences between Posted Reactors and Channel Reactors. Industrial and Engineering Chemistry Research, 44(12), 4267-4271.

143 Danion, A., Disdier, J., Guillard, C., Païssé, O. and Jaffrezic-Renault, N. (2006). Photocatalytic degradation of imidazolinone fungicide in TiO2-coated optical fiber reactor. Applied Catalysis B: Environmental, 62(3-4), 274-281.

144 Włoch, E., Łukaszczyk, A., Żurek, Z. and Sulikowski, B. (2006).Synthesis of ferrierite coatings on the FeCrAl substrate. Catalysis Today, 114(2-3), 231-236.

145 Chen, H., Chen, L., Lu, Y., Hong, Q., Chua, H., Tang, S. and Lin, J. (2004). Synthesis, characterization and application of nano-structured Mo2C thin films. Catalysis Today, 96(3), 161-164.

146 Berenguer-Murcia, A., Rebrov, E., Cabaj, M., Wheatley, A., Johnson, B., Robertson, J. and Schouten, J. (2009).Confined palladium colloids in mesoporous frameworks for carbon nanotube growth.J Mater Sci, 44(24), 6563-6570.

147 Meille, V., Pallier, S., Santacruzbustamante, G., Roumanie, M. and Reymond, J. (2005).Deposition of Al2O3 layers on structured supports for the design of new catalytic reactors. Applied Catalysis A: General, 286(2), 232-238.

148 Gao, Y., Chen, M., Zhang, T. and Zheng, X. (2011). A novel method for the growth of ZSM-5 zeolite membrane on the surface of stainless steel. Materials Letters, 65(17-18), 2789-2792.

149 Seo, Y., Yu, S., Cho, S. and Song, K. (2003). The catalytic heat exchanger using catalytic fin tubes.Chemical Engineering Science, 58(1), 43-53.

150 Giani, L., Cristiani, C., Groppi, G. and Tronconi, E. (2006). Washcoating method for Pd/γ-Al2O3 deposition on metallic foams. Applied Catalysis B: Environmental, 62(1-2), 121-131.

151 Avila, P., Montes, M. and Miró, E. (2005).Monolithic reactors for environmental applications. Chemical Engineering Journal, 109(1-3), 11-36.

152 Mies, M., Rebrov, E., Jansen, J., Decroon, M. and Sschouten, J. (2007). Method for the in situ preparation of a single layer of zeolite Beta crystals on a molybdenum substrate for microreactor applications. Journal Of Catalysis, 247(2), 328-338.

153 García-Bordejé, E., Lázaro, M., Moliner, R., Álvarez, P., Gómez-Serrano, V. and Fierro, J. (2006). Vanadium supported on carbon coated honeycomb monoliths for the selective catalytic reduction of NO at low temperatures: Influence of the oxidation pre- treatment. Carbon, 44(3), 407-417.

183 Bibliography

154 Jarrah, N., van Ommen, J. G. and Lefferts, L. (2006). Mechanistic aspects of the formation of carbon-nanofibers on the surface of Ni foam: A new microstructured catalyst support. Journal Of Catalysis, 239(2), 460-469.

155 Eskandarloo, H., Badiei, A., Behnajady, M. and Ziarani, G. (2015). UV-LEDs assisted preparation of silver deposited TiO2 catalyst bed inside microchannels as a high efficiency microphotoreactor for cleaning polluted water. Chemical Engineering Journal, 270, 158- 167.

156 Rahman, M., García-García, F., Hatim, M., Kingsbury, B. and Li, K. (2011).Development of a catalytic hollow fibre membrane micro-reactor for high purity H2 production. Journal Of Membrane Science, 368(1-2), 116-123.

157 Liguras, D., Goundani, K. and Verykios, X. (2004). Production of hydrogen for fuel cells by catalytic partial oxidation of ethanol over structured Ni catalysts. Journal Of Power Sources, 130(1-2), 30-37.

158 Picher, M., Anglaret, E., Arenal, R. and Jourdain, V. (2011).Processes Controlling the Diameter Distribution of Single-Walled Carbon Nanotubes during Catalytic Chemical Vapor Deposition. ACS Nano, 5(3), 2118-2125.

159 Thakur, D., Tiggelaar, R., Gardeniers, J., Lefferts, L. and Seshan, K. (2013). Silicon based microreactors for catalytic reduction in aqueous phase: Use of carbon nanofiber supported palladium catalyst. Chemical Engineering Journal, 227, 128-136.

160 Schimpf, S., Lucas, M., Mohr, C., Rodemerck, U., Brückner, A. and Radnik, J. et al. (2002).Supported gold nanoparticles: in-depth catalyst characterization and application in hydrogenation and oxidation reactions. Catalysis Today, 72(1-2), 63-78.

161 Tyczkowski, J., Kapica, R. and Łojewska, J. (2007). Thin cobalt oxide films for catalysis deposited by plasma-enhanced metal-organic chemical vapor deposition. Thin Solid Films, 515(16), 6590-6595.

162 Guyon, C., Barkallah, A., Rousseau, F., Giffard, K., Morvan, D. and Tatoulian, M. (2011).Deposition of cobalt oxide thin films by plasma-enhanced chemical vapour deposition (PECVD) for catalytic applications. Surface and Coatings Technology, 206(7), 1673-1679.

163 Chen, G., Guyon, C., Zhang, Z., Da Silva, B., Da Costa, P. and Ognier, S. et al. (2013). Catkin liked nano-Co3O4 catalyst built-in organic microreactor by PEMOCVD method for trace CO oxidation at room temperature. Microfluidics and Nanofluidics, 16(1-2), 141- 148.

164 Watts, P. and Haswell, S. (2009).Chemical Synthesis in Microreactors. Methods In Molecular Biology, 583, 109-120.

165 Sollier, E., Murray, C., Maoddi, P. and Di Carlo, D. (2011).Rapid prototyping polymers for microfluidic devices and high pressure injections. Lab On A Chip, 11(22), 3752.

166 Zhou, J., Khodakov, D., Ellis, A. and Voelcker, N. (2011).Surface modification for PDMS-based microfluidic devices. Electrophoresis, 33(1), 89-104.

184 Bibliography

167 Jena, R., Yue, C. and Lam, Y. (2011). Micro fabrication of cyclic olefin copolymer (COC) based microfluidic devices. Microsystem Technologies, 18(2), 159-166.

168 Schelcher, G., Guyon, C., Ognier, S., Cavadias, S., Martinez, E. and Taniga, V. et al. (2014). Cyclic olefin copolymer plasma millireactors. Lab On A Chip, 14(16), 3037.

169 Liu, J. and Lee, M. (2006).Permanent surface modification of polymeric capillary electrophoresis microchips for protein and peptide analysis. Electrophoresis, 27(18), 3533- 3546.

170 Friedrich, J. (2011). Mechanisms of Plasma Polymerization - Reviewed from a Chemical Point of View. Plasma Processes and Polymers, 8(9), 783-802.

171 Friedrich, J., Mix, R., Schulze, R., Meyer‐Plath, A., Joshi, R. and Wettmarshausen, S. (2008). New Plasma Techniques for Polymer Surface Modification with Monotype Functional Groups. Plasma Processes and Polymers, 5(5), 407-423.

172 Wettmarshausen, S., Kühn, G., Hidde, G., Mittmann, H. and Friedrich, J. (2007). Plasmabromination - the Selective Way to Monotype Functionalized Polymer Surfaces. Plasma Processes and Polymers, 4(9), 832-839.

173 Laib, S. and MacCraith, B. (2007).Immobilization of Biomolecules on Cycloolefin Polymer Supports.Analytical Chemistry, 79(16), 6264-6270.

174 Hwang, S., Tseng, M., Shu, J. and Her Yu, H. (2008).Surface modification of cyclic olefin copolymer substrate by oxygen plasma treatment. Surface and Coatings Technology, 202(15), 3669-3674.

175 Roy, S., Yue, C. and Lam, Y. (2011). Influence of plasma surface treatment on thermal bonding and flow behavior in Cyclic Olefin Copolymer (COC) based microfluidic devices. Vacuum, 85(12), 1102-1104.

176 Xu, B., Yan, X., Xu, J. and Chen, H. (2012). One step high quality poly(dimethylsiloxane)-hydrocarbon plastics bonding. Biomicrofluidics, 6(1), 016507.

177 Nikolova, D., Dayss, E., Leps, G. and Wutzler, A. (2004).Surface modification of cycloolefinic copolymers for optimization of the adhesion to metals. Surface and Interface Analysis, 36(8), 689-693.

178 Schäfer, A. (2006). Nanofiltration-principles and applications. Oxford: Elsevier.

179 Bulmuş, V., Ayhan, H. and Pişkin, E. (1997).Modified PMMA monosize microbeads for glucose oxidase immobilization. The Chemical Engineering Journal and The Biochemical Engineering Journal, 65(1), 71-76.

180 Walker, G. and Beebe, D. (2002).A passive pumping method for microfluidic devices. Lab On A Chip, 2(3), 131.

181 Zimmermann, M., Schmid, H., Hunziker, P. and Delamarche, E. (2007). Capillary pumps for autonomous capillary systems. Lab Chip, 7(1), 119-125.

185 Bibliography

182 Lynn, N. and Dandy, D. (2009). Passive microfluidic pumping using coupled capillary/evaporation effects. Lab On A Chip, 9(23), 3422.

183 Mishchenko, L., Khan, M., Aizenberg, J. and Hatton, B. (2013).Spatial Control of Condensation and Freezing on Superhydrophobic Surfaces with Hydrophilic Patches. Adv. Funct. Mater., 23(36), 4577-4584.

184 Yan, J., Bauer, W., Fischlechner, M., Hollfelder, F., Kaminski, C. and Huck, W. (2013). Monodisperse Water-in-Oil-in-Water (W/O/W) Double Emulsion Droplets as Uniform Compartments for High-Throughput Analysis via Flow Cytometry. Micromachines, 4(4), 402-413.

185 Wrobel, M. A. and Wertheimer, M. R. (1990). Plasma-polymerized organosilicons and organometallics, R. d’Agostino (ed.), Academic Press, New York, 163-268.

186 Arapan, L., Alexieva, G., Avramov, I., Radeva, E., Strashilov, V., Katardjiev, I. and Yantchev, V. (2011). Highly Mass-Sensitive Thin Film Plate Acoustic Resonators (FPAR).Sensors, 11(12), 6942-6953.

187 Kim, S., Song, E., Jo, K., Yun, T., Moon, M. and Lee, K. (2013).Composite oxygen- barrier coating on a polypropylene food container.Thin Solid Films, 540, 112-117.

188 Angelini, E., d'Agostino, R., Fracassi, F., Grassini, S. and Rosalbino, F. (2002). Surface analysis of PECVD organosilicon films for corrosion protection of steel substrates. Surface and Interface Analysis, 34(1), 155-159.

189 Pramatarova, L., Radeva, E. Pecheva, E., Hikov, T., Krasteva, N., Dimitrova, R., Mitev, D., Montgomery, P., Sammons R. and Altankov, G. (2011). Biomimetics, Pramatarova, L. (ed.) InTech, 297-320.

190 Brusa, R. (2004). Porosity in low dielectric constant SiOCH films depth profiled by positron annihilation spectroscopy. J. Appl. Phys., 95(5), 2348.

191 Tsai, C., Li, Y., Cheng, I. and Chen, J. (2013).O2/HMDSO-Plasma-Deposited Organic- Inorganic Hybrid Film for Gate Dielectric of MgZnO Thin-Film Transistor. Plasma Processes and Polymers, 11(1), 89-95.

192 Mota, R., Galvão, D., Durrant, S., De Moraes, M., de Oliveira Dantas, S. and Cantão, M. (1995). HMDSO plasma polymerization and thin film optical properties. Thin Solid Films, 270(1-2), 109-113.

193 Magni, D., Deschenaux, C., Hollenstein, C., Creatore, A. and Fayet, P. (2000). Oxygen diluted hexamethyldisiloxane plasmas investigated by means of in situ infrared absorption spectroscopy and mass spectrometry. Journal Of Physics D: Applied Physics, 34(1), 87-94.

194 Ricci, M., Dorier, J., Hollenstein, C. and Fayet, P. (2010).Influence of Argon and Nitrogen Admixture in HMDSO/O2 Plasmas onto Powder Formation. Plasma Processes and Polymers, (8), 108-117.

186 Bibliography

195 Leu, G., Brockhaus, A. and Engemann, J. (2003). Diagnostics of a hexamethyldisiloxane/oxygen deposition plasma. Surface and Coatings Technology, 174- 175, 928-932.

196 Goullet, A., Vallée, C., Granier, A. and Turban, G. (2000). Optical spectroscopic analyses of OH incorporation into SiO2 films deposited from O2/tetraethoxysilane plasmas. Journal Of Vacuum Science and Technology A: Vacuum, Surfaces and Films, 18(5), 2452.

197Hegemann, D., Brunner, H. and Oehr, C. (2003).Plasma treatment of polymers for surface and adhesion improvement. Nuclear Instruments and Methods In Physics Research Section B: Beam Interactions With Materials and Atoms, 208, 281-286.

198 Huang, C., Liu, C., Su, C., Hsu, W. and Wu, S. (2009). Investigation of atmospheric- pressure plasma deposited SiOx films on polymeric substrates. Thin Solid Films, 517(17), 5141-5145.

199 Croci, S., Pêcheur, A., Autran, J., Vedda, A., Caccavale, F., Martini, M. and Spinolo, G. (2001). SiO2 films deposited on silicon at low temperature by plasma-enhanced decomposition of hexamethyldisilazane: Defect characterization. Journal Of Vacuum Science and Technology A: Vacuum, Surfaces and Films, 19(5), 2670.

200 Korzec, D., Theirich, D., Werner, F., Traub, K. and Engemann, J. (1995). Remote and direct microwave plasma deposition of HMDSO films: comparative study. Surface and Coatings Technology, 74-75, 67-74.

201 Chung, T., Kang, M., Chung, C. and Kim, Y. (2009). Effects of process parameters on the properties of silicon oxide films using plasma enhanced chemical vapor deposition with tetramethoxysilane. Current Applied Physics, 9(3), 598-604.

202 Farag, Z., Krüger, S., Hidde, G., Schimanski, A., Jäger, C. and Friedrich, J. (2013). Deposition of thick polymer or inorganic layers with flame-retardant properties by combination of plasma and spray processes. Surface and Coatings Technology, 228, 266- 274.

203 Blanchard, N., Hanselmann, B., Drosten, J., Heuberger, M. and Hegemann, D. (2014).Densification and Hydration of HMDSO Plasma Polymers. Plasma Processes and Polymers, 12(1), 32-41.

204 Mavadat, M., Ghasemzadeh-Barvarz, M., Turgeon, S., Duchesne, C. and Laroche, G. (2013).Correlation between the Plasma Characteristics and the Surface Chemistry of Plasma-Treated Polymers through Partial Least-Squares Analysis. Langmuir, 29(51), 15859-15867.

205 Pringle, S., Joss, V. and Jones, C. (1996). Ammonia Plasma Treatment of PTFE Under Known Plasma Conditions. Surface and Interface Analysis, 24(12), 821-829.

206 Sarra-Bournet, C., Ayotte, G., Turgeon, S., Massines, F. and Laroche, G. (2009).Effects of Chemical Composition and the Addition of H2 in a N2 Atmospheric Pressure Dielectric Barrier Discharge on Polymer Surface Functionalization.Langmuir, 25(16), 9432-9440.

187 Bibliography

207 Lewis, F. (2010). Étude sur les propriétés d'adhérence des revêtements fluorocarbonés déposés par plasma sur l'acier inoxydable 316L pour les stents coronariens. Québec: Université Laval.

208 Petit-Etienne, C. (2007). Dépôt d'oxyde de silicium par procédé plasma hors équilibre à basse pression et à pression atmosphérique sur de l'acier. Tel-00367151. HAL-CNRS.

209 Kleiman, J. and Iskanderova, Z. (2003).Protection of materials and structures from space environment. Dordrecht: Kluwer Academic Publishers, pp 443-450.

210 G. Ram Prasad,T, Stephen Daniels, D.C. Cameron, B.P. McNamara, Elizabeth Tully, R. O’Kennedy, Surf. Coat. Technol 200 (2005) 1031-1035

211 Bhattacharya, S., Datta, A., Berg, J. and Gangopadhyay, S. (2005). Studies on surface wettability of poly(dimethyl) siloxane (PDMS) and glass under oxygen-plasma treatment and correlation with bond strength. Journal Of Microelectromechanical Systems, 14(3), 590-597.

212 Azioune, A., Marcozzi, M., Revello, V. and Pireaux, J. (2007). Deposition of polysiloxane-like nanofilms onto an aluminium alloy by plasma polymerized hexamethyldisiloxane: characterization by XPS and contact angle measurements. Surface and Interface Analysis, 39(7), 615-623.

213 Roy, S. and Yue, C. (2011). Surface Modification of COC Microfluidic Devices: A Comparative Study of Nitrogen Plasma Treatment and its Advantages Over Argon and Oxygen Plasma Treatments. Plasma Processes and Polymers, 8(5), 432-443.

214 Bodas, D. and Khan-Malek, C. (2007). Hydrophilization and hydrophobic recovery of PDMS by oxygen plasma and chemical treatment-An SEM investigation. Sensors and Actuators B: Chemical, 123(1), 368-373.

215 McDonald, J. and Whitesides, G. (2002). Poly(dimethylsiloxane) as a Material for Fabricating Microfluidic Devices. Accounts Of Chemical Research, 35(7), 491-499.

216 Barbier, V., Tatoulian, M., Li, H., Arefi-Khonsari, F., Ajdari, A. and Tabeling, P. (2006). Stable Modification of PDMS Surface Properties by Plasma Polymerization: Application to the Formation of Double Emulsions in Microfluidic Systems. Langmuir, 22(12), 5230-5232.

217 Kim, J., Chaudhury, M. and Owen, M. (2000). Hydrophobic Recovery of Polydimethylsiloxane Elastomer Exposed to Partial Electrical Discharge. Journal Of Colloid and Interface Science, 226(2), 231-236.

218 Lawton, R., Price, C., Runge, A., Doherty, W. and Saavedra, S. (2005). Air plasma treatment of submicron thick PDMS polymer films: effect of oxidation time and storage conditions. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 253(1-3), 213-215.

219 Egitto, F. and Matienzo, L. (1990). Modification of polytetrafluoroethylene and polyethylene surfaces downstream from helium microwave plasmas. Polymer Degradation and Stability, 30(3), 293-308.

188 Bibliography

220 Wilson, D., Williams, R. and Pond, R. (2001).Plasma modification of PTFE surfaces. Part II: Plasma-treated surfaces following storage in air or PBS. Surface and Interface Analysis, 31(5), 397-408.

221 Fritz, J. and Owen, M. (1995).Hydrophobic Recovery of Plasma-Treated Polydimethylsiloxane. The Journal Of Adhesion, 54(1-4), 33-45.

222 Feldsien, J., Kim, D. and Economou, D. (2000). SiO2 etching in inductively coupled C2F6 plasmas: surface chemistry and two-dimensional simulations. Thin Solid Films, 374(2), 311-325.

223 Purohit, V., Mielczarski, E., Mielczarski, J. and Akesso, L. (2013).Evidence of coexistence of micro and nanoporosity of organo-silica polymeric films deposited on silicon by plasma deposition. Materials Chemistry and Physics, 141(2-3), 602-612.

224 Tsuchiya, T., Inoue, A. and Sakata, J. (2000).Tensile testing of insulating thin films; humidity effect on tensile strength of SiO2 films. Sensors and Actuators A: Physical, 82(1- 3), 286-290.

225 Waters, P. (2005). The effects of moisture on thin film delamination and adhesion. Tampa, Fla.: University of South Florida.

226 Tan, S., Nguyen, N., Chua, Y. and Kang, T. (2010).Oxygen plasma treatment for reducing hydrophobicity of a sealed polydimethylsiloxane microchannel. Biomicrofluidics, 4(3), 032204.

227 Zhao, L., Lee, J. and Sen, P. (2012). Long-term retention of hydrophilic behavior of plasma treated polydimethylsiloxane (PDMS) surfaces stored under water and Luria- Bertani broth. Sensors and Actuators A: Physical, 181, 33-42.

228 Shriner, R. (2004). The systematic identification of organic compounds. Hoboken, NJ: Wiley.

229 Bongiovanni, R., Sangermano, M., Malucelli, G. and Priola, A. (2005). UV curing of photoinitiator-free systems containing bismaleimides and diacrylate resins: bulk and surface properties. Progress In Organic Coatings, 53(1), 46-49.

230 Nørbygaard, T. and Berg, R. (2004). Analysis of Phthalate Ester Content in Poly(vinyl chloride) Plastics by Means of Fourier Transform Raman Spectroscopy. Applied Spectroscopy, 58(4), 410-413.

231 Miliani, C., Ombelli, M., Morresi, A. and Romani, A. (2002). Spectroscopic study of acrylic resins in solid matrices. Surface and Coatings Technology, 151-152, 276-280.

232 Bauer, F., Decker, U., Naumov, S. and Riedel, C. (2014). Photoinitiator-free UV curing and matting of acrylate-based nanocomposite coatings: Part 3. Progress In Organic Coatings, 77(6), 1085-1094.

189 Bibliography

233 Yang, D. and Wakamatsu, T. (1996).Polarized Grazing Angle FTIR Study of Molecular Orientation and Bonding of Thioglycolates at the Metal Interface. Surface and Interface Analysis, 24(12), 803-810.

234 Mihaly, J., Sterkel, S., Ortner, H. M.,Kocsis, L.,Hajba, L., Furdyga, E.,Minka,J. (2006). Croatica Chemica Acta, (79), 497-501.

235 Ma, K., Reza, F., Saaem, I. and Tian, J. (2009).Versatile surface functionalization of cyclic olefin copolymer (COC) with sputtered SiO2 thin film for potential BioMEMS applications. Journal Of Materials Chemistry, 19(42), 7914.

236 Walkiewicz-Pietrzykowska, A., Cotrino, J. and González-Elipe, A. (2005).Deposition of Thin Films of SiOxCyH in a Surfatron Microwave Plasma Reactor with Hexamethyldisiloxane as Precursor. Chemical Vapor Deposition, 11(6-7), 317-323.

237 Casserly, T. and Gleason, K. (2005).Chemical Vapor Deposition of Organosilicon Thin Films from Methylmethoxysilanes. Plasma Processes and Polymers, 2(9), 679-687.

238 Perry, C. and Li, X. (1991).Structural studies of gel phases. Part 2: Infrared spectroscopic study of silica monoliths; the interaction of water with surface species. Faraday Trans., 87(24), 3857.

239 Roy, S., Yue, C., Lam, Y., Wang, Z. and Hu, H. (2010). Surface analysis, hydrophilic enhancement, ageing behavior and flow in plasma modified cyclic olefin copolymer (COC)-based microfluidic devices. Sensors and Actuators B: Chemical, 150(2), 537-549.

240 Khairallah, Y., Arefi, F. and Amouroux, J. (1994). Surface fluorination of polyethylene films by different CF4 glow discharges: effects of frequency and electrode configuration. Thin Solid Films, 241(1-2), 295-300.

241 Ladner, Y., D'Orlyé, F., Perréard, C., Da Silva, B., Guyon, C. and Tatoulian, M. et al. (2014).Surface Functionalization by Plasma Treatment and Click Chemistry of a New Family of Fluorinated Polymeric Materials for Microfluidic Chips. Plasma Processes and Polymers, 11(6), 518-523.

242 von Gunten, U. (2003). Ozonation of drinking water: Part I. Oxidation kinetics and product formation. Water Research, 37(7), 1443-1467.

243 Gracia, R., Aragües, J. and Ovelleiro, J. (1996).Study of the Catalytic Ozonation of Humic Substances in Water and Their Ozonation Byproducts. Ozone: Science and Engineering, 18(3), 195-208.

244 Allemane, H., Delouane, B., Paillard, H. and Legube, B. (1993). Comparative Efficiency Of Three Systems (O3, O3/H2O2 and O3/TiO2) For The Oxidation Of Natural Organic Matter In Water. Ozone: Science and Engineering, 15(5), 419-432.

245 Cooper, C. and Burch, R. (1999).An investigation of catalytic ozonation for the oxidation of halocarbons in drinking water preparation. Water Research, 33(18), 3695- 3700.

190 Bibliography

246 Andreozzi, R., Caprio, V., Insola, A., Marotta, R. and Tufano, V. (1998). The ozonation of pyruvic acid in aqueous solutions catalyzed by suspended and dissolved manganese. Water Research, 32(5), 1492-1496.

247 Volk, C., Roche, P., Joret, J. and Paillard, H. (1997). Comparison of the effect of ozone, ozone-hydrogen peroxide system and catalytic ozone on the biodegradable organic matter of a fulvic acid solution. Water Research, 31(3), 650-656.

248 Andreozzi, R., Insola, A., Caprio, V., Marotta, R. and Tufano, V. (1996). The use of manganese dioxide as a heterogeneous catalyst for oxalic acid ozonation in aqueous solution. Applied Catalysis A: General, 138(1), 75-81.

249 Ma, J. and Graham, N. (1997). Preliminary Investigation of Manganese-Catalyzed Ozonation for the Destruction of Atrazine. Ozone: Science and Engineering, 19(3), 227- 240.

250 Legube, B. and Karpel Vel Leitner, N.(1999). Catalytic ozonation: a promising advanced oxidation technology for water treatment. Catalysis Today, 53(1), 61-72.

251 Vel Leitener, N., Gombert, B., Legube, B. and Luck, F. (1998). Impact of catalytic ozonation on the removal of a chelating agent and surfactants in aqueous solution. Water Science and Technology, 38(4-5), 203-209.

252 Al‐Hayek, N., Legube, B. and Doré, M. (1989). Ozonation catalytique (Fe III/A12O3) du phénol et de ses produits d'ozonation. Environmental Technology Letters, 10(4), 415- 426.

253 Nawrocki, J. and Kasprzyk-Hordern, B. (2010). The efficiency and mechanisms of catalytic ozonation. Applied Catalysis B: Environmental, 99(1-2), 27-42.

254 Jensen, K., Reizman, B. and Newman, S. (2014). Tools for chemical synthesis in microsystems. Lab On A Chip, 14(17), 3206.

255 Pennemann, H., Hessel, V. and Löwe, H. (2004). Chemical microprocess technology- from laboratory-scale to production. Chemical Engineering Science, 59(22-23), 4789- 4794.

256 deMello, A. (2006). Control and detection of chemical reactions in microfluidic systems. Nature, 442(7101), 394-402.

257 Yetisen, A., Akram, M. and Lowe, C. (2013). Paper-based microfluidic point-of-care diagnostic devices. Lab On A Chip, 13(12), 2210.

258 Ghosh, A., Ganguly, R., Schutzius, T. and Megaridis, C. (2014). Wettability patterning for high-rate, pumpless fluid transport on open, non-planar microfluidic platforms. Lab Chip, 14(9), 1538-1550.

259 Schafer, D., Squier, J., Maarseveen, J., Bonn, D., Bonn, M. and Müller, M. (2008). In Situ Quantitative Measurement of Concentration Profiles in a Microreactor with Submicron Resolution Using Multiplex CARS Microscopy. J. Am. Chem. Soc., 130(35), 11592-11593.

191 Bibliography

260 Greenway, G., Haswell, S., Morgan, D., Skelton, V. and Styring, P. (2000). The use of a novel microreactor for high throughput continuous flow organic synthesis. Sensors and Actuators B: Chemical, 63(3), 153-158.

261 Ouyang, X. and Besser, R. (2003). Development of a microreactor-based parallel catalyst analysis system for synthesis gas conversion. Catalysis Today, 84(1-2), 33-41.

262 Ueno, M., Suzuki, T., Naito, T., Oyamada, H. and Kobayashi, S. (2008). Development of microchannel reactors using polysilane-supported palladium catalytic systems in capillaries. Chemical Communications, (14), 1647.

263 Costantini, F., Bula, W., Salvio, R., Huskens, J., Gardeniers, H., Reinhoudt, D. and Verboom, W. (2009). Nanostructure Based on Polymer Brushes for Efficient Heterogeneous Catalysis in Microreactors. J. Am. Chem. Soc., 131(5), 1650-1651.

264 Pines, D. and Reckhow, D. (2002). Effect of Dissolved Cobalt(II) on the Ozonation of Oxalic Acid. Environmental Science and Technology, 36(19), 4046-4051.

265 Lin, J., Kawai, A. and Nakajima, T. (2002). Effective catalysts for decomposition of aqueous ozone. Applied Catalysis B: Environmental, 39(2), 157-165.

266 Kasprzyk-Hordern, B. (2003). Catalytic ozonation and methods of enhancing molecular ozone reactions in water treatment. Applied Catalysis B: Environmental, 46(4), 639-669.

267 Baxendale, I., Griffiths-Jones, C., Ley, S. and Tranmer, G. (2006). Microwave-Assisted Suzuki Coupling Reactions with an Encapsulated Palladium Catalyst for Batch and Continuous-Flow Transformations. Chemistry - A European Journal, 12(16), 4407-4416.

268 Uozumi, Y., Yamada, Y., Beppu, T., Fukuyama, N., Ueno, M. and Kitamori, T. (2006). Instantaneous Carbon-Carbon Bond Formation Using a Microchannel Reactor with a Catalytic Membrane. J. Am. Chem. Soc., 128(50), 15994-15995.

269 Kobayashi, J. (2004). A Microfluidic Device for Conducting Gas-Liquid-Solid Hydrogenation Reactions. Science, 304(5675), 1305-1308.

270 Lee, K., Min, S. and Jang, J. (2010). Preparation of a Catalytic Membrane Reactor with Palladium Nanoparticles Supported by a Packed-Bed Silica Nanosupporter for Gas-Phase Methanol Oxidation. Small, 6(21), 2378-2382.

271 Yamada, Y., Watanabe, T., Beppu, T., Fukuyama, N., Torii, K. and Uozumi, Y. (2010). Palladium Membrane-Installed Microchannel Devices for Instantaneous Suzuki-Miyaura Cross-Coupling. Chemistry - A European Journal, 16(37), 11311-11319.

272 Munirathinam, R., Ricciardi, R., Egberink, R., Huskens, J., Holtkamp, M. and Wormeester, H. et al. (2013). Gallium-containing polymer brush film as efficient supported Lewis acid catalyst in a glass microreactor. Beilstein Journal Of Organic Chemistry, 9, 1698-1704.

192 Bibliography

273 Foix, M., Guyon, C., Tatoulian, M. and Costa, P. (2013). Microwave Plasma Treatment for Catalyst Preparation: Application to Alumina Supported Silver Catalysts for SCR NOx by Ethanol. Modern Research In Catalysis, 02(03), 68-82.

274 Bartolomeu, R., Foix, M., Fernandes, A., Tatoulian, M., Ribeiro, M., Henriques, C. and da Costa, P. (2011). Fluidized bed plasma for pre-treatment of Co-ferrierite catalysts: An approach to NOx abatement. Catalysis Today, 176(1), 234-238.

275 Mu, X., Bartmann, U., Guraya, M., Busser, G., Weckenmann, U., Fischer, R. and Muhler, M. (2003). The preparation of Pd/SiO2 catalysts by chemical vapor deposition in a fluidized-bed reactor. Applied Catalysis A: General, 248(1-2), 85-95.

276 Suvanto, s. and Pakkanen, T. (1997). A fluidized bed reactor for low temperature deposition. Journal Of Molecular Catalysis A: Chemical, 125(2-3), 91-96.

277 Jeong, Y., Lee, J., Ha, S. and Kim, S. (2009). Fabrication of cobalt-organic composite thin film via plasma-enhanced chemical vapor deposition for antibacterial applications. Thin Solid Films, 517(9), 2855-2858.

278 Dhar, R., Pedrow, P., Liddell, K., Quentin Ming, Moeller, T. and Osman, M. (2005). Plasma-enhanced metal-organic chemical vapor deposition (PEMOCVD) of catalytic coatings for fuel cell reformers. IEEE Trans. Plasma Sci., 33(1), 138-146.

279 Foix, M., Guyon, C., Tatoulian, M. and Da Costa, P. (2010). Study of the use of fluidized bed plasma reactors for the treatment of alumina supported palladium catalyst: Application for SCR NOx by CH4 in stationary sources. Catalysis Communications, 12(1), 20-24.

280 Foix, M., Guyon, C., Tatoulian, M. and Da Costa, P. (2010). Fluidized Bed Plasmas Reactor for Catalyst Synthesis and Pretreatment. Application for Pollution Abatement in Stationary and Mobile Sources. AMR, 89-91, 118-123.

281 S Pasko, S., Hubert-Pfalzgraf, L., Abrutis, A. and Vaissermann, J. (2004). Synthesis and molecular structures of cobalt(II) β-diketonate complexes as new MOCVD precursors for cobalt oxide films. Polyhedron, 23(5), 735-741.

282 Foix Le Beuvant, M. (2011). Développement de nouveaux procédés de synthèses de catalyseurs sur poudres en plasma lits fluidisés. HAL-CNRS.

283 Miserere, S., Mottet, G., Taniga, V., Descroix, S., Viovy, J. and Malaquin, L. (2012). Fabrication of thermoplastics chips through lamination based techniques. Lab On A Chip, 12(10), 1849.

284 Schöne, L. and Herrmann, H. (2013). Kinetic measurements on the reactivity of hydrogen peroxide and ozone towards small atmospherically relevant aldehydes, ketones and organic acids in aqueous solution. Atmospheric Chemistry and Physics Discussions, 13(10), 25537-25566.

285 Ervens, B., Gligorovski, S. and Herrmann, H. (2003). Temperature-dependent rate constants for hydroxyl radical reactions with organic compounds in aqueous solutions. Phys. Chem. Chem. Phys., 5(9), 1811-1824.

193 Bibliography

286 Li, W., Qiang, Z., Zhang, T. and Cao, F. (2012). Kinetics and mechanism of pyruvic acid degradation by ozone in the presence of PdO/CeO2. Applied Catalysis B: Environmental, 113-114, 290-295.

287 Buffle, M., Schumacher, J., Meylan, S., Jekel, M. and von Gunten, U. (2006). Ozonation and Advanced Oxidation of Wastewater: Effect of O3 Dose, pH, DOM and HO•-Scavengers on Ozone Decomposition and HO• Generation. Ozone: Science and Engineering, 28(4), 247-259.

288 Tizaoui, C., Grima, N. and Derdar, M. (2009). Effect of the radical scavenger t-butanol on gas-liquid mass transfer. Chemical Engineering Science, 64(21), 4375-4382.

289 Kandalkar, S., Dhawale, D., Kim, C. and Lokhande, C. (2010). Chemical synthesis of cobalt oxide thin film electrode for supercapacitor application. Synthetic Metals, 160(11- 12), 1299-1302.

290 Morales, P., Gómez, L., Olayo, M., Cruz, G., Palacios, C., Morales, J. and Ordoñez, E. (2009). Polyethylene Obtained by Plasma Polymerization of Hexene. Macromolecular Symposia, 283-284(1), 13-17.

291 Petitto, S., Marsh, E., Carson, G. and Langell, M. (2008). Cobalt oxide surface chemistry: The interaction of CoO(100), Co3O4(110) and Co3O4(111) with oxygen and water. Journal Of Molecular Catalysis A: Chemical, 281(1-2), 49-58.

292 Klepper, K., Nilsen, O. and Fjellvåg, H. (2007). Growth of thin films of Co3O4 by atomic layer deposition. Thin Solid Films, 515(20-21), 7772-7781.

293 Szymańska, A., Oleszak, D., Grabias, A., Rosiński, M., Sikorski, K., Kazior, J. and Kurzydłowski, K. J. (2004). Phase transformations in ball milled AISI 316L stainless steel powder and the microstructure of the steel obtained by its sintering. Rev. Adv. Mater. Sci, 8, 143-146.

294 Will, G., Masciocchi, N., Parrish, W. and Hart, M. (1987). Refinement of simple crystal structures from synchrotron radiation powder diffraction data. J Appl Cryst, 20(5), 394- 401.

295 Athawale, A., Majumdar, M., Singh, H. and Navinkiran, K. (2010). Synthesis of Cobalt Oxide Nanoparticles/Fibres in Alcoholic Medium using y-ray Technique. Defence Science Journal, 60(5), 507-513.

296 Burton, A., Ong, K., Rea, T. and Chan, I. (2009). On the estimation of average crystallite size of zeolites from the Scherrer equation: A critical evaluation of its application to zeolites with one-dimensional pore systems. Microporous and Mesoporous Materials, 117(1-2), 75-90.

297 Scherrer, P., Gottingen, N.G.W. (1918). Nachr. Ges. Wiss. Goeltingen.Math.-Phys. Kl (2), 96.

298 Voß, M., Borgmann, D. and Wedler, G. (2002). Characterization of Alumina, Silica and Titania Supported Cobalt Catalysts. Journal Of Catalysis, 212(1), 10-21.

194 Bibliography

299 Dittmar, A., Kosslick, H., Müller, J. and Pohl, M. (2004). Characterization of cobalt oxide supported on titania prepared by microwave plasma enhanced chemical vapor deposition. Surface and Coatings Technology, 182(1), 35-42.

300 Jiménez, V., Espinós, J. and González-Elipe, A. (1998). Control of the stoichiometry in the deposition of cobalt oxides on SiO2. Surface and Interface Analysis, 26(1), 62-71.

301 Haber, J. and Ungier, L. (1977). On chemical shifts of ESCA and Auger lines in cobalt oxides. Journal Of Electron Spectroscopy and Related Phenomena, 12(3), 305-312.

302 Barreca, D., Massignan, C., Daolio, S., Fabrizio, M., Piccirillo, C., Armelao, L. and Tondello, E. (2001). Composition and Microstructure of Cobalt Oxide Thin Films Obtained from a Novel Cobalt(II) Precursor by Chemical Vapor Deposition. Chemistry Of Materials, 13(2), 588-593.

303 Vaz, C.A.F., Henrich, V.E., Ahn, C.H., Altman, E.I. (2009). Growth and characterization of thin epitaxial films, Journal of Crystal Growth, 311(9), 2648-265.

304 Su, D., Xie, X., Munroe, P., Dou, S. and Wang, G. (2014). Mesoporous hexagonal Co3O4 for high performance lithium ion batteries. Sci. Rep., 4, 6519.

305 Wu, G., Tan, X., Li, G. and Hu, C. (2010). Effect of preparation method on the physical and catalytic property of nanocrystalline Fe2O3. Journal Of Alloys and Compounds, 504(2), 371-376.

306 Wu, Y. and Wang, X. (2011). Preparation and characterization of single-phase α-Fe2O3 nano-powders by Pechini sol-gel method. Materials Letters, 65(13), 2062-2065.

307 Demianets, L., Pouchko, S. and Gaynutdinov, R. (2003). Fe2O3 single crystals: hydrothermal growth, crystal chemistry and growth morphology. Journal Of Crystal Growth, 259(1-2), 165-178.

308 Sahoo, S., Agarwal, K., Singh, A., Polke, B. and Raha, K. (2011). Characterization of γ- and α-Fe2O3 nano powders synthesized by emulsion precipitation-calcination route and rheological behaviour of α-Fe2O3. International Journal of Engineering, Science and Technology, 2(8), 118-126.

309 Lv, A., Hu, C., Nie, Y. and Qu, J. (2010). Catalytic ozonation of toxic pollutants over magnetic cobalt and manganese co-doped γ-Fe2O3. Applied Catalysis B: Environmental, 100(1-2), 62-67.

310 Han, Q., Liu, Z., Xu, Y., Chen, Z., Wang, T. and Zhang, H. (2007). Growth and Properties of Single-Crystalline γ-Fe2O3 Nanowires. J. Phys. Chem. C, 111(13), 5034- 5038.

311 Xin, J., Jia-jia, C., Jian-hui, X., Yi-ning, S., You-zuo, F., Min-sen, Z. and Quan-feng, D. (2012). Fe2O3 xerogel used as the anode material for lithium ion batteries with excellent electrochemical performance. Chemical Communications, 48(59), 7410.

195 Bibliography

312 Orolínová, Z., Mockovčiaková, A., Zeleňák, V. and Myndyk, M. (2012). Influence of heat treatment on phase transformation of clay–iron oxide composite. Journal Of Alloys and Compounds, 511(1), 63-69.

313 Benson, S. (1965). III - Bond energies. J. Chem. Educ., 42(9), 502.

314 Leech, P., Zhang, X. and Zhu, Y. (2010). Hot Embossing of Microfluidic Channel Structures in Cyclic Olefin Copolymers. MRS Proc., 1272.

315 Rubashov, A.M., Pogorelov, V.V., Strahov, B.V. (1972) Catalytic activity of Fe2O3 in decomposition of ozone. Russ. J. Phys. Chem., 46(9), 2283-2291.

316 Rubashov, A.M, Strahov, B.V. (1973). Formation of oxygen species on Fe2O3 surface during ozonation. Russ. J. Phys. Chem., 47(8), 2115-2122.

317 Gumuchian, D., (2014). Développement et étude d'un procédé d'oxydation avancée de traitement de l'eau pour composés organiques résistants par couplage plasma/catalyse.Tel-01139999. HAL-CNRS.

318 Xie, Y. (2004). Disinfection byproducts in drinking water. Boca Raton: Lewis Publishers, 20-24.

319 Hoigné, J. and Bader, H. (1983). Rate constants of reactions of ozone with organic and inorganic compounds in water-II. Water Research, 17(2), 185-194.

320 Jansson, J., Palmqvist, A. E. C., Fridell, E., Skoglundh, M., Osterlund, L, Thormahlen, P, Langer V., J. (2002). On the Catalytic Activity of Co3O4 in Low-Temperature CO Oxidation. Journal Of Catalysis, 211(2), 387-397.

321 Xie, X., Li, Y., Liu, Z., Haruta, M. and Shen, W. (2009). Low-temperature oxidation of CO catalysed by Co3O4 nanorods. Nature, 458(7239), 746-749.

322 Thormählen, P., Skoglundh, M., Fridell, E. and Andersson, B. (1999). Low- Temperature CO Oxidation over Platinum and Cobalt Oxide Catalysts. Journal Of Catalysis, 188(2), 300-310.

323 Su, D., Dou, S. and Wang, G. (2014). Single Crystalline Co3O4 Nanocrystals Exposed with Different Crystal Planes for Li-O2 Batteries. Sci. Rep., 4, 5767.

324 Buehler, R., Staehelin, J. and Hoigne, J. (1984). Ozone decomposition in water studied - - by pulse radiolysis. 1. Perhydroxyl (HO2)/hyperoxide (O2 ) and HO3/O3 as intermediates. The Journal Of Physical Chemistry, 88(12), 2560-2564.

325 Sawyer, D. (1991). Oxygen chemistry. New York: Oxford University Press, 73-74.

326 Staehelin, J. and Hoigne, J. (1982). Decomposition of ozone in water: rate of initiation by hydroxide ions and hydrogen peroxide. Environmental Science and Technology, 16(10), 676-681.

327 Rao, P. and Hayon, E. (1975). Redox potentials of free radicals. IV. Superoxide and - hydroperoxy radicals. O2 and HO2. The Journal Of Physical Chemistry, 79(4), 397-402.

196 Bibliography

328 Taube, H. and Bray, W. (1941). Additions and Corrections - Chain Reactions in Aqueous Solutions Containing Ozone, Hydrogen Peroxide and Acid. J. Am. Chem. Soc., 63(12), 3551-3551.

329 Gottschalk, C., Libra, J. and Saupe, A. (2010). Ozonation of water and waste water. Weinheim: Wiley-VCH.

330 Rice, R. (1999). Ozone in the United States of America-State-Of-The-Art. Ozone: Science and Engineering, 21(2), 99-118.

331 Sehested, K., Corfitzen, H., Holcman, J. and Hart, E. (1998). On the Mechanism of the Decomposition of Acidic O3 Solutions, Thermally or H2O2-Initiated. J. Phys. Chem. A, 102(16), 2667-2672.

332 Hoigne J., Chemistry of aqueous ozone and transformation of pollutants by ozonation and advanced oxidation processes, in: Hrubec, J., Baldauf, G. and Brauch, H. (1995). The handbook of environmental chemistry. Berlin, Germany: Springer-Verlag.

333 Oguz, E., Tortum, A. and Keskinler, B. (2008). Determination of the apparent rate constants of the degradation of humic substances by ozonation and modeling of the removal of humic substances from the aqueous solutions with neural network. Journal Of Hazardous Materials, 157(2-3), 455-463.

334 Acero, J. L., von Gunten, U. (2001). Characterization of oxidation processes: ozonation and the AOP O3/H2O2. Journal of American Water Works Association, 93, 90-100.

335 Vogelpohl, A. (2007). Applications of AOPs in wastewater treatment. Water Science and Technology, 55(12), 207.

336 Bruchet, A., Costentin E. and von Guten, U. (1996). Bromate formation in advanced oxidation processes. Journal American Water Works Association, 88(6), 53-65.

337 Tabrizi, G. and Mehrvar, M. (2004). Integration of Advanced Oxidation Technologies and Biological Processes: Recent Developments, Trends and Advances. Journal Of Environmental Science and Health, Part A- Toxic/Hazardous Substances and Environmental Engineering, 39(11-12), 3029-3081.

338 Andreozzi, R., Caprio, V., Marotta, R. and Tufano, V. (2001). Kinetic modeling of pyruvic acid ozonation in aqueous solutions catalyzed by Mn(II) and Mn(IV) ions. Water Research, 35(1), 109-120.

339 Valdés, H. and Zaror, C. (2006). Heterogeneous and homogeneous catalytic ozonation of benzothiazole promoted by activated carbon: Kinetic approach. Chemosphere, 65(7), 1131-1136.

340 Beltrán, F., Rivas, F. and Monterodeespinosa, R. (2004). A TiO2/Al2O3 catalyst to improve the ozonation of oxalic acid in water. Applied Catalysis B: Environmental, 47(2), 101-109.

197 Bibliography

341 Álvarez, P., Beltrán, F., Pocostales, J. and Masa, F. (2007). Preparation and structural characterization of Co/Al2O3 catalysts for the ozonation of pyruvic acid. Applied Catalysis B: Environmental, 72(3-4), 322-330.

342 Mizuno, T., Tsuno, H. and Yamada, H. (2007). Development of Ozone Self- Decomposition Model for Engineering Design. Ozone: Science & Engineering, 29(1), 55- 63.

343 de Souza, L., Haida, H., Thévenin, D., Seidel-Morgenstern, A. and Janiga, G. (2013). Model selection and parameter estimation for chemical reactions using global model structure. Computers and Chemical Engineering, 58, 269-277.

344 Atkins, P. and De Paula, J. (2006). Atkins' Physical chemistry. Oxford: Oxford University Press, 805-807.

345 Vartiainen, E., Rinia, H., Müller, M., & Bonn, M. (2006). Direct extraction of Raman line-shapes from congested CARS spectra. Opt. Express, 14(8), 3622.

346 Wurpel, G. and Müller, M. (2006). Water confined by lipid bilayers: A multiplex CARS study. Chemical Physics Letters, 425(4-6), 336-341.

347 Bonn, M., Bakker, H., Rago, G., Pouzy, F., Siekierzycka, J., Brouwer, A. and Bonn, D. (2009). Suppression of Proton Mobility by Hydrophobic Hydration. J. Am. Chem. Soc., 131(47), 17070-17071.

198