Heterogeneous (bi-)metallic catalysts for plasma catalytic conversion of carbon dioxide

(1)

(2)

(3)

(4)

(5)

I

ACKNOWLEDGEMENT

The writing of this thesis required its time and dedication from me, but it would not have been possible without the help of a collection of other people, whom I would like to thank sincerely.

Firstly, I would like to express my deepest gratitude to my counsellor, Bram Seynnaeve, for training me in conducting the experiments and guiding me through the writing. It was with him that I worked the closest during this project and it was a pleasure to do so.

Secondly, my special thanks go to my promotors, An Verberckmoes and Jeroen Lauwaert, for giving additional guidance and providing me with the feedback I required to successfully finish my thesis. I want to pay my special regards to the staff of the Industrial catalysis and adsorption technology (INCAT) department of Ghent university, in particular Nadia De Paepe. They provided me with a clean and desirable workspace, as well as anything else I required for conducting my experiments.

I am indebted to my proof-readers and friends, Liesbeth Van Damme and Martijn De Gussem, for using their precious time to help me with my English writing.

I want to sincerely thank my family for giving me the strength and love to keep going these past years. My father, for providing me with everything I would ever need, and my late mother, who used to be always there for me.

Finally, I want to thank all the professors who taught me as part of this education. The curriculum was challenging but made possible through their excellent teachings and guidance.

(6)

II

COVID-19 PREAMBLE

There are many things that can go wrong during scientific research and preventive measures are taken to eliminate most potential complications, but that a pandemic would interfere with the research conducted for this thesis was impossible to predict. The fact that it is completely beyond anyone’s abilities to do anything about it is especially frustrating. Multiple experiments were unable to be executed because of time constraints as a result of the COVID-19 pandemic. These experiments, together with the reasoning behind them, are therefore described in more detail instead, without being able to show the results, in chapter 4.

The results listed in chapter 4 were gathered during the time period of September 16, 2019, until the 12th of March 2020. On March 12, the Belgian government declared the start of the quarantine to stop the spreading of the Corona virus. As a result, most laboratories closed and all research conducted there ceased. The university laboratories were not excluded, therefore putting a stop to the research that was being conducted for this thesis.

Because of the fact that the deposition precipitation experiments only started on February 10, the results with this method of impregnation was affected the most. Only very little data was gathered, although these results were fairly promising. The impregnation of iron nitrate was executed on the day before the quarantine commenced, making the determination of the iron loading of the sample unable to be performed. Attempts to increase the iron loading were therefore not performed either. Finally, there were plans to approach dry impregnation of both iron and copper separately, either by performing the impregnation in an oven or by removing the solvent excess in the method described in chapter 3.1. Once the monometallic impregnation of iron and copper in alumina through the deposition precipitation method was optimised, attempts would have been made to perform bimetallic iron-copper impregnation of the alumina spheres. Both simultaneous and sequential impregnation, with intermediate calcination in the latter method, were still being considered. Characterisation of many samples is another important activity that was not executed because of the pandemic. Mainly the deposition precipitation samples were hardly characterised, as can be seen in chapter 4, results and discussion. This makes it impossible to confirm the supposed difference in nanoparticle size of the two impregnation methods, which was a major observation to be made in this thesis. Additionally, nitrogen sorption experiments would be performed on the alumina spheres impregnated through the deposition precipitation method to determine the mechanical integrity of the samples.

A final set of experiments that were not performed because of COVID-19 were the plasma experiments. These experiments were to be performed by a PhD student at the university of Antwerp. In the first semester, these experiments were already delayed because of difficulties in finding a PhD candidate, and by the time the student was trained enough to start experimenting, the quarantine had almost begun. If it were not for the quarantine, the different synthesized catalysts would have been tested for their viability in plasma catalysed carbon dioxide and methane conversions. Because of the lack of results from these experiments, decisions on the direction the catalysts should have progressed in were made based on the characterisation results only, which is far from optimal.

(7)

III

ABSTRACT

Supported mono- and bimetallic hematite (Fe2O3) and copper oxide (CuO) catalysts have been synthesised, using γ alumina (Al2O3) as support material. The effect of two different impregnation methods, dry impregnation and deposition precipitation, on the produced nanoparticles and homogeneity of the catalyst has been determined. Scanning electron microscopy (SEM), infrared spectroscopy (IR), X-ray diffraction (XRD), nitrogen sorption, dynamic light scattering (DLS) and atomic absorption spectroscopy (AAS) were used to characterise the prepared catalysts. Copper impregnation using the dry impregnation method with copper nitrate (Cu(NO3)2) as precursor produces a homogeneous catalyst containing CuO nanoparticles with an average size of 50 nm. Iron impregnation using the same method with iron nitrate (Fe(NO3)3) as precursor produces an inhomogeneous catalyst with a visual band some distance from the surface of the sphere with a higher concentration of hematite nanoparticles with an average size of 105 nm. Using a calcination pre-treatment on the γ-alumina support has been observed to remove the band partly, and using an acid pre-treatment of using iron citrate (Fe(III)C6H5O7) as alternative precursor removes the band completely. These observations led to higher γ-alumina surface hydroxylation, and consequent iron hydroxide precipitation, as being hypothesised as the cause of band formation during dry impregnation of iron nitrate. Sequential, bimetallic copper-iron impregnation of the alumina spheres using the dry impregnation method encounters similar challenges as the monometallic impregnations: the iron nitrate impregnation introduces inhomogeneities in the form of a dark band below the surface and a leaching out of the metal nanoparticles. Using iron citrate to perform the iron impregnation produces a visually homogeneous catalyst instead. Impregnation of copper- or iron nitrate through the deposition precipitation method, with the use of urea as precipitation agent, produces visually homogeneous results. The possible complexation of iron with urea, differences in impregnation forces during wet- and dry impregnation and mixing during impregnation in the deposition precipitation method are suspected to prevent the preliminary precipitation of iron hydroxide.

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

(16)

(17)

(18)

(19)

(20)

XVI

4.1. Dry impregnation ... 40 4.1.1. Monometallic copper ... 40 4.1.2. Monometallic iron ... 42 4.1.3. Bimetallic iron/copper... 52 4.2. Deposition precipitation ... 55 4.2.1. Monometallic copper ... 55 4.2.2. Monometallic iron ... 57 4.2.3. Bimetallic iron/copper... 58 4.3. Plasma experiments ... 59 5. Conclusion... 60 Reflections on sustainability ... 62 References ... 64

(22)

XVIII

LIST OF FIGURES & TABLES

1. Figures

Figure 1 Greenhouse effect; values are percentages relative to the earth-average solar constant of ~340 W/m2 _{[24] ... 1} Figure 2 Thermochemical cycle to produce syngas [45] ... 6 Figure 3 Dry reforming equilibrium [49] ... 7 Figure 4 Simplified Fischer-Tropsch reaction scheme [33] ... 9 Figure 5 Electrochemical CO2 reduction cells [59] ... 10 Figure 6 Simplified CO2 electroreduction reaction scheme on a copper catalyst [57] ... 11 Figure 7 CO2 electroreduction product distribution [62] ... 11 Figure 8 Proposed pathways of the dry reforming process; thickness of arrow line is proportional to reaction’s importance [66] ... 13 Figure 9 Thermal plasma configurations [70] ... 15 Figure 10 Glow discharge plasma [88] ... 17 Figure 11 Negative D.C. corona discharge; figure is not to scale, the plasma region takes only a small part of the interelectrode spacing [84] ... 18 Figure 12 Dielectric barrier discharge configurations [89] ... 18 Figure 13 Electron avalanche [92] ... 20 Figure 14 Heterogeneous catalytic models; with reactants A & B, substrate S, adsorbed reactants ASADS & BSADS [110] ... 24 Figure 15 Phenomena occurring during plasma catalysis [102] ... 26 Figure 16 Electron-substrate interactions during scanning electron microscopy [139] ... 33 Figure 17 Process of X-ray diffraction [141] ... 34 Figure 18 Typical BET-isotherm [108] ... 35 Figure 19 Intensity traces in a DLS measurement [143] ... 37 Figure 20 Generation of a correlation function; left: intensity trace, right: corresponding correlation function [143] ... 37 Figure 21 Experimental DBD plasma setup, drawing ... 39 Figure 22 Experimental DBD plasma setup, images ... 39 Figure 23 Dry impregnation of copper nitrate in alumina, calcined; left: 10 w% Cu, right: 20 w% Cu 40 Figure 24 XRD pattern of Copper nitrate in alumina, calcined; 20 w% Cu ... 40 Figure 25 IR spectrum of alumina (red) and copper nitrate in alumina (blue), calcined; 20 w% Cu.... 41 Figure 26 XRD pattern of CuO nanoparticles in alumina [146] ... 41 Figure 27 IR spectra of copper oxide (left) and gamma alumina (right) [148], [149] ... 42 Figure 28 DLS measurement results on copper nanoparticles of a 10 w% Cu sample ... 42 Figure 29 Dry impregnation of iron nitrate in alumina, calcined; 10 w% Fe ... 42 Figure 30 XRD pattern of iron nitrate in alumina, calcined; 10 w% Fe ... 43 Figure 31 IR spectrum of alumina (red), calcined iron nitrate in alumina (blue) and dried iron nitrate in alumina (grey); 10 w% Fe ... 43 Figure 32 XRD pattern of hematite (Fe2O3) [150] ... 43 Figure 33 IR spectra of hematite (left) and iron nitrate (right) [151], [152] ... 44 Figure 34 DLS measurement results on iron nanoparticles of a 10 w% iron sample ... 44

(23)

XIX Figure 35 SEM images of iron nitrate in alumina, calcined; 10 w% Fe ... 45 Figure 36 SEM images of the outer surface (left) and pores (right) of alumina ... 46 Figure 37 Band formation in function of time of iron nitrate in alumina, 10 w% Fe ... 46 Figure 38 BET isotherms of alumina spheres; red: shell, blue: core ... 47 Figure 39 XRD patterns of alumina spheres; red: shell, blue: core ... 48 Figure 40 XRD pattern of gamma alumina [153] ... 48 Figure 41 XRD patterns of alumina (red) and alumina after calcination pre-treatment (blue) ... 49 Figure 42 Determination of the effect of calcination pre-treatment on the acidic sites on the alumina surface ... 49 Figure 43 Dry impregnation of iron nitrate in pre-calcined alumina, calcined; 10 w% Fe ... 50 Figure 44 Dry impregnation of iron chloride (left) and iron citrate (right) in alumina, calcined; 10 w% Fe ... 50 Figure 45 XRD pattern of iron citrate in alumina, calcined; red: HCl-method, blue: NH3-method ... 50 Figure 46 Dry impregnation of iron nitrate in alumina undergone acidic pre-treatment, calcined; 10 w% Fe... 52 Figure 47 Bimetallic 10 w% Cu / 10 w% Fe catalyst, both nitrate precursors, sequential impregnation ... 52 Figure 48 DLS measurement results on iron and copper nanoparticles of a 10 w% iron, 10 w% Cu sample ... 53 Figure 49 SEM images of copper nitrate and iron nitrate in alumina, calcined; 10 w% Fe, 10 w% Cu 54 Figure 50 Bimetallic 10 w% Cu / 10 w% Fe catalyst, copper nitrate and iron citrate precursors , sequential impregnation ... 54 Figure 51 Deposition precipitation of copper nitrate in alumina, calcined; 6.8 w% Cu... 55 Figure 52 SEM image of copper nitrate in alumina, calcined; 6.8 w% Cu ... 55 Figure 53 SEM image of bimetallic Cu/Fe in alumina, calcined; 10 w% Cu, 10 w% Fe ... 56 Figure 54 Deposition precipitation of iron nitrate in alumina, calcined ... 57

2. Tables

Table 1 Summary of possible plasma reactions [6]–[8] ... 21 Table 2 Volume and internal surface area of the shell and the core of alumina spheres, calculated using the BET-method ... 47

(24)

XX

LIST OF ABBREVIATIONS

AAS atomic absorption spectroscopy A.C. alternating current

ASF Andersen, Schultz & Flory BET Brunauer, Emmett & Teller CCS carbon capture and storage CCU carbon capture and utilisation DBD dielectric barrier discharge plasma D.C. direct current

DLS dynamic light scattering

EDX energy-dispersive X-ray spectroscopy FT Fischer Tropsch process

GC-MS gas chromatography – mass spectrometry ICP inductively coupled plasma

LCA life cycle analysis

LTE local thermal equilibrium PZC point of zero charge

SEM scanning electron microscopy syngas synthesis gas

(25)

1

1. INTRODUCTION

Ever since humanity learned about global warming, many actions were taken to stop it and prevent the disastrous effects of the phenomenon from happening. In Europe for example, policies were implemented to reduce the emission of greenhouse gasses and replace non-renewable energy production with renewable alternatives. The goal is to achieve a climate neutral EU by 2050 [1], which is no easy task and the need for measures are bigger than ever.

A large contributor to the problem is the excessive use of fossil fuels as an energy source [2]. If untouched, fossil fuels act as a form of carbon sequestration, where carbon is trapped in the earth and cannot contribute to global warming. When used for energy production however, a large part of the carbon is eventually released into the atmosphere as carbon dioxide (CO2), which is one of the most problematic greenhouse gasses [3].

Greenhouse gasses contribute to global warming through a phenomenon called the greenhouse effect. When solar radiation penetrates the atmosphere to reach the earth, part of this radiation is absorbed by the earth’s surface, another part is reflected. The absorbed radiation is mainly reemitted as infrared radiation, which is further absorbed or reflected by the greenhouse gasses present in the atmosphere, trapping its energy around the earth [4]. A diagram of the earth’s radiation balance is shown in Figure 1.

Figure 1 Greenhouse effect; values are percentages relative to the earth-average solar constant of ~340 W/m2 _[5]

The greenhouse effect is a naturally occurring process that is required for life to exist on the planet earth, since it would be too cold without it. When highly increased however, the temperatures will rise, which can have problematic effects e.g. the rise of the sea level and extreme weather. The amount of greenhouse gasses in the atmosphere have hugely increased in the last decades [6], to the point where high increases of natural disasters are observed worldwide [7]. Lowering the amount of greenhouse gases in the atmosphere is therefore necessary to prevent the worsening of these evident effects.

To come back to the topic of fossil fuels, it is clear that these sequestered carbon sources are best left alone to prevent the increase of carbon dioxide in the atmosphere. In fact, one of the frequently used methods to lower carbon dioxide emissions is called carbon sequestration. Another is referred to as carbon utilisation.

(26)

2 The term carbon capture and sequestration (CCS) is used when referring to the process of capturing CO2 and storing it in a suitable location for extended periods of time [8], while carbon capture and utilisation (CCU) is used for the process of capturing CO2 and converting it into useful chemicals and fuels [9]. The former is usually done in underground cave systems with porous rock formations and provides for an effective way to prevent carbon dioxide from entering the atmosphere. However, CCS has no economic benefits and is therefore not a sustainable way of addressing the problem [9]. In contrast, CCU does have an economically viable side, as fuels and chemicals can be produced while still effectively reducing carbon dioxide emissions. This makes the method much more sustainable and thus favourable over CCS. Additionally, the fuels produced through CCU can act as an alternative to the fossil fuels previously addressed, thereby preventing this naturally sequestered carbon from entering the atmosphere. However, to ensure the economic viability of CCU, the carbon dioxide conversion processes must be sufficiently optimised, for converting what is essentially invaluable waste into usable chemicals of even fuels is not so easily done [9].

Producing fuels from carbon dioxide is not an easy task, because CO2 conversions into bigger molecules do not occur spontaneously. The fact that CO2 is a by-product of many exothermic reactions, like the oxidation of organic molecules (e.g. the burning of carbohydrates), suggests this. The exothermicity of these reactions indicates that energy will be needed to revert these reactions, where CO2 is used to produce larger organic molecules. This is demonstrated by photosynthesis of plants, where sunlight provides the energy to convert CO2 and water (H2O) in carbohydrates. In a more practical setup, different sources can be used to provide this energy. High temperatures (thermal), electricity (electrochemical) or plasma technology can be used to increase the reactivity of CO2 and enable its conversions. In this thesis, plasma is used for this purpose. It is a relatively new method that is less researched, but it shows high potential.

Plasma is a state of matter first described by Crookes in 1879 and named after corpuscle-free blood by Langmuir in 1928, as the former reminded him of the latter, being a fluid in which a variety of different particles move around. The plasma state is defined as an ionised gas containing positive and negative ions, as well as electrons, but a variety of other species are usually present, e.g. excited atoms, radicals, and short-wave radiation, resulting in a mixture with high reactivity. This reactive mixture can be generated through different ionisation methods to dissociate and ionise the molecules in a non-ionised gas. Two groups of ionisation methods can be differentiated: thermal techniques, using an arc, spark or flame, or non-thermal methods, where a strong electric field is used [10]. Once this reactive mixture is formed, the unstable species will recombine in countless different ways to produce stable molecules. Without the presence of a catalyst, the end mixture consists for the most part out of small organic molecules. For example, when a CO2 – methane (CH4) mixture is sent through an uncatalyzed plasma, the end mixture mainly consists of carbon monoxide (CO), hydrogen gas (H2) and unconverted CO2 and CH4 [11]–[13]. CO and H2 can be used in various applications, a mixture of the two gasses, for example, is generally referred to as syngas and can be used to produce a variety of chemicals, e.g. methanol and carbohydrates, in additional processes.

However, with the use of catalysts, the selectivity in the productions of other, larger chemicals can be increased in the plasma reactions. This allows for the productions of these chemicals from CO2 and

(27)

3 CH4 in a single step, instead of the need for an additional process to convert syngas into those chemicals. The fact that plasma conversions enable the reduction of a two-step process to a single step shows the high potential, both in energy efficiency and resource requirements, of the plasma technology. The selectivity in which chemicals are produced depends on what catalyst is used during the conversion process. There are many possible catalysts, e.g. zeolites, metals, metal oxides, starch, etc., that can be considered and they all have a different effect on the plasma reactions, although some elements lend themselves better for this purpose then others, as research can uncover.

The elements that have the most potential to be used as catalysts are the transition metals, because of the unique electron exchanging properties they possess [14]. It is therefore not surprising that transition metals are used in many catalytic chemical conversions today. This is no different for plasma reactions, where the use of different transition metals as catalyst result in the production of different molecules during CO2 and CH4 conversion. There are two ways in which these metal catalysts give rise to unique end mixtures: firstly, they lower the activation energy of only specific reactions, increasing their occurrence kinetically. Secondly, they alter the physical properties of the plasma, changing the electric and magnetic fields wherein the reactions take place.

To effectively use a metal as a heterogenous catalyst, it is typically immobilized on a stable and porous support material, e.g. silica or alumina, to maximise the catalysts’ surface area and thus optimise its catalytic activity. For the use in plasma configurations specifically, a spherical configuration of such a support material is used because it allows for the catalyst to be held in place in the chamber where the plasma is maintained. Multiple methods exist to cover the support material with the desired transition metal(s), e.g. dry impregnation and deposition precipitation, which results in a different morphology of the prepared catalyst. The morphology of a catalytic material is an important factor that needs to be considered during heterogeneous catalysis, as specific morphological structures can act as active sites [15], and must therefore be taken into account. Nanoparticles are highly effective, because of the high surface area that they provide [16].

Apart from the used impregnation method, there are many additional parameters that influence the catalyst properties. Differences in metal concentration, nature of the support, impregnation time, drying program, the order in which multimetallic catalysts are synthesized, etc. all have an effect. The challenge is to find the optimal adjustments to every parameter that give the most uniform and reproducible catalyst with the best performance in the studied plasma reaction. The same can be said about the parameters of the plasma reactions themselves, where even the stacking of the catalyst in the plasma has a big influence on which products are obtained. This makes the design of a setup to achieve the highest possible conversion, and later recreating that setup, very challenging.

The project covered in this thesis encompasses specific methods to achieve the previously described goals. The dry reforming process, in which a mixture of CO2 and CH4 is used to produce syngas, is carried out in a dielectric barrier discharge (DBD) plasma, where monometallic and bimetallic catalysts based on copper (Cu), iron (Fe) and nickel (Ni) are used to alter the selectivity towards the formation of multi-carbon species instead of CO and H2. γ-alumina spheres with a diameter of 1.8 mm are used as a carrier to enable the placement of catalyst nanoparticles into the plasma reactor. Formation of copper and/or iron nanoparticles on the surface of the support material is carried out using the incipient wetness impregnation and deposition precipitation methods.

(28)

4

2. LITERATURE REVIEW

2.1. Carbon dioxide conversion

During this project, an alternative way will be studied to the presently used techniques with which larger molecules can be synthesized using CO2 as the carbon source. The goal is to find more efficient pathways using relatively new concepts, namely plasma reactions. It is therefore useful to have an idea of the other techniques used in current industrial processes, so that a comparison can be made. An overview will be presented in this chapter. Firstly, however, the thermodynamics of the observed reactions are explained in order to understand the requirement of specialized techniques to achieve the set goal.

The fact that CO2 is produced in many exothermic reactions suggest that energy is required to revert these reactions and produce (oxygenated) hydrocarbons from CO2. The reality is a little more complicated. When looking at the thermodynamic data of these reactions, it becomes clear that they are thermodynamically spontaneous at any temperature in the direction where CO2 is produced. To illustrate this, the reaction enthalpy (∆𝑟𝐻, equation (2)) and -entropy (∆𝑟𝑆, equation (3)) of the oxidation of ethanol, as shown in equation (1), is calculated using Hess’ law.

2 𝐶2𝐻5𝑂𝐻 (𝑙) + 7 𝑂2 (𝑔) → 4 𝐶𝑂2 (𝑔) + 6 𝐻2𝑂 (𝑔) (1) ∆𝑟𝐻 = 4 ∆𝑓𝐻(𝐶𝑂2) + 6 ∆𝑓𝐻(𝐻2𝑂) − 7 ∆𝑓𝐻(𝑂2) − 2 ∆𝑓𝐻(𝐶2𝐻5𝑂𝐻) (2) = [4 ∙ (−393,51) + 6 ∙ (−241,82) − 7 ∙ 0 − 2 ∙ (−277,69)] 𝑘𝐽 𝑚𝑜𝑙 = −2469,58 𝑘𝐽 𝑚𝑜𝑙 ∆𝑟𝑆 = 4 𝑆𝑚(𝐶𝑂2) + 6 𝑆𝑚(𝐻2𝑂) − 7 𝑆𝑚(𝑂2) − 2 𝑆𝑚(𝐶2𝐻5𝑂𝐻) (3) = [4 ∙ 213,74 + 6 ∙ 188,83 − 7 ∙ 205,138 − 2 ∙ 160,7] 𝐽 𝑚𝑜𝑙 𝐾 = 230,574 _{𝑚𝑜𝑙 𝐾}𝐽

The direction in which a reaction will propagate is determined by the difference in Gibbs free energy of the reagents and the products. It can be calculated with the differences in enthalpy and entropy using equation (4).

∆𝐺 = ∆𝐻 – 𝑇∆𝑆 (4)

A negative value for the Gibbs free energy indicates a thermodynamically spontaneous reaction. The calculated enthalpy (equation (2)) of the reaction is negative, indicating the exothermicity of the reaction. For the Gibbs free energy to be positive at higher temperatures, the entropy of the reaction would have to be negative. Since the entropy (equation (4)) is positive however, the Gibbs free energy will be negative at any temperature.

This indicates that these reactions cannot be reverted by simply adding thermal energy. Different methods are required to work around this problem such as thermal techniques, electrochemical processes, or the use of plasmas.

(29)

5

2.1.1. Syngas production through splitting

In most techniques to convert CO2 using high temperatures, carbon monoxide (CO) is first produced as an intermediate. In contrast with further steps, where larger molecules are produced, this first step can be made thermodynamically spontaneous by simply adding thermal energy. From there, CO can be split further to produce elementary carbon through the Boudouard reaction (equation (6)), or it can be used to produce hydrocarbons and oxygenates with hydrogen gas (H2) [17]. A mixture of CO and H2 is usually referred to as synthesis gas (syngas). The relatively high reactivity of this mixture compared to CO2 can be used to produce bigger molecules in pathways that can be made thermodynamically spontaneous in ways that will be explained later, in chapter 2.1.3.

The first step in both processes is the same: the production of CO from CO2. The most straightforward way to do this is by splitting CO2 in CO and oxygen gas (O2) at high temperatures, of which the reaction equation is shown in equation (1)(5).

2 𝐶𝑂2 (𝑔) ⇋ 2 𝐶𝑂 (𝑔) + 𝑂2 (𝑔) (5)

For this reaction to occur towards the formation of CO, temperatures above 2000 K are required. A higher conversion of 60 % can be achieved by using even higher temperatures of 3000 – 3500 K, or by using a semipermeable membrane to selectively remove O2 from the reaction mixture once it is formed. Early experiments show the use of zirconia-based membranes in this process, as it has already been used to produce H2 from water in a similar process. Using this membrane, a conversion of 22 % can be achieved at a temperature below 2000 K [18]. More recent research reports the use of ceramic membranes coated with lead (Pb) and nickel (Ni) catalysts to increase the conversion to 52 % at temperatures below 1000 °C [19].

CO can then be used to convert carbon in two ways: either directly through further splitting, or with H2 as syngas. The latter will be discussed further down in this chapter, after single-step processes for syngas production are explained. The splitting of CO is executed using the Boudouard reaction, shown in equation (6).

2 𝐶𝑂 (𝑔) ⇋ 𝐶 (𝑠) + 𝐶𝑂2 (𝑔) (6)

This is an extensively researched reaction, as it is a side reaction occurring during the production of other products, e.g. in furnaces where ceramics are fired or metallurgic processes are conducted, with unwanted carbon deposition as a result. A study reported the reaction from occurring in a temperature range of 400 – 700 °C in the presence of iron oxides, with a maximum rate around 550 °C [20]. In the context of carbon conversion, the reaction can be used for carbon sequestration/utilisation pathways both as the production of CO from CO2 and the production of solid carbon from CO. Even though thermodynamically this reaction would spontaneously form carbon, its high activation energy prevents it from occurring without a catalyst. Ferromagnetic metals like cobalt (Co) and nickel (Ni) are used for this purpose. It should be noted that solid carbon is, though a good way for sequestration, not a valuable material to produce. It’s often more desirable to convert carbon into fuels and chemicals, since these can be used in further processes [21].

(30)

6 Before going over to other processes where syngas can be produced in a single step, the process of thermochemical H2O splitting to produce H2 should be addressed. Similar to CO2 splitting (equation (5)), H2O is split in H2 and O2 at high temperatures. The reaction equation is shown in equation (7).

2 𝐻2𝑂 (𝑔) ⇋ 2 𝐻2 (𝑔) + 𝑂2 (𝑔) (7)

Temperatures above 2500 K are required for this reaction to occur towards the formation of H2 [22]. As with CO2 splitting, higher conversions can be achieved by removing the produced O2 from the reactive mixture with a semipermeable membrane. This is even more important in H2O splitting since O2 and H2 can form an explosive mixture.

Starting from the 1960s, research has been conducted to find more energy efficient ways to execute this reaction in the form of thermochemical cycles [23]. These are alternative ways to perform a chemical reaction, but in multiple steps. This makes it possible to bypass any difficulty in the initial reaction procedure, for example when the products are very reactive and need to be separated or when extreme temperatures are required. There are countless possible cycles to consider, but only a select few are successful in being more efficient then the initial chemical reaction [22], [24]–[27]. Some of these promising thermochemical cycles are based on redox pair systems, where a metal transitions between its oxidized and reduced state during the multiple steps of these cycles. The metal undergoes thermal reduction on one end and is oxidized by bonding oxygen on the other end. H2O can be split using such a thermochemical cycle together with CO2 (equation (7) and (5) respectively) to produce syngas in a coupled redox pair system [28]. The cycle is shown in Figure 2.

Figure 2 Thermochemical cycle to produce syngas [29]

The first part of the cycle is the production of syngas from steam and CO2, where the metal is oxidised to bond the excess oxygen atoms from the CO2 and H2O molecules and producing CO and H2 respectively. The oxidised metal can then be reduced to release O2 by heating in the second part of the cycle. The reduced metal can be recycled to repeat the cycle. The fact that lower thermal energies are required for the splitting of CO2 and H2O this way makes the redox pair system more favourable then the single step process. Additionally, the cycle prevents an explosive mixture of H2 and O2 from being formed, since the oxidised metal can be reduced to release O2 when separated from the syngas mixture [28], [29].

(31)

7 The metal redox pairs used during this process can both be based on metal oxides / metal systems, or metal oxide / metal oxide systems. The extent of the reduction plays a crucial part in the efficiency of the pair. Reduction to a more reduced oxide is less efficient than all the way to the elementary metal, where the oxidation state is zero. A higher possible dissociation means more oxygen can be released during one cycle, resulting in higher syngas yields per mass of redox material. Many redox pairs have been researched in the past, each with its advantages and disadvantages [29]. Examples are volatile pairs, e.g. ZnO/Zn and SnO2/SnO, that enter gaseous states during thermal reduction, and non-volatile pairs, e.g. ferrites and hercynites, that remain solid during the complete cycle. Reduction temperatures depend on the type of redox pair and range from 1400 K to 2000 K [28].

2.1.2. Syngas production through dry reforming

An alternative way to directly produce syngas from CO2 in a single step is through dry reforming, where a mixture of methane (CH4) and CO2 is used to produce syngas. The reaction equation is shown in equation (8).

𝐶𝐻4 (𝑔) + 𝐶𝑂2 (𝑔) ⇋ 2 𝐶𝑂 (𝑔) + 2 𝐻2 (𝑔) (8) This method for producing syngas has been researched since 1928 [30], but it recently gained more attention because of the environmental challenges humanity is faced with. CH4 is emitted during multiple industrial processes in the petrochemical and refinery sectors as a waste gas [31]. Both CH4 and CO2 contribute heavily to global warming when present in the atmosphere (EPA, 2017). A pathway to use both gases to produce syngas is therefore very desirable [32].

Without a catalyst, a minimum temperature of 643 °C is required for dry reforming to occur. Many catalysts can be used to lower the required temperature, but in an industrial setting, supported nickel (Ni) catalysts provide the most economical alternative at the present [32]. They are, however, more prone to deactivation compared to noble metals. Deactivation occurs by carbon deposition through methane decomposition and the Boudouard reaction, shown in equation (6). The thermodynamic stability of the different molecules in dry reforming in function of temperature is shown in Figure 3 [33].

(32)

8 Another side reaction, apart from the Boudouard reaction and methane decomposition, is the formation of water from CO2 and H2. The highest amount of carbon formation is reported to occur at temperatures ranging from 560 – 700 °C, since CH4 only decomposes at temperatures above 560 °C and the Boudouard reaction does not occur above 700 °C (Berry, Ames & Snow, 1956). Formation of water provides no additional complications other than decreasing the H2 yield. Figure 3 shows an increasing yield of syngas at higher temperatures, particularly above 700 °C. When using an easily deactivated catalyst like nickel, it is recommended to work in a temperature range of 700 °C and above. With more noble catalysts that deactivate less easily, lower temperatures can be used [32]. An alternative way to lower the required temperature for dry reforming, is to combine it with other syngas producing processes like steam reforming and the partial oxidative reforming of CH4. In these processes, called tri-reforming, H2O and O2 are used with CH4 to produce syngas respectively [34]. The tri-reforming reaction equation is shown in equation (9).

20 𝐶𝐻4 (𝑔) + 𝐻2𝑂 (𝑔) + 9 𝑂2 (𝑔) + 𝐶𝑂2 (𝑔) ⇋ 21 𝐶𝑂 (𝑔) + 41 𝐻2 (𝑔) (9) This reaction has a much lower reaction enthalpy as compared to the dry reforming process, indicating that lower temperatures are required. Additionally, catalyst deactivation through carbon formation via the Boudouard reaction is reduced, increasing catalyst life and process efficiency. Much lower quantities of CO2 are consumed in this reaction relative to CH4 however, making the combined reaction less ideal for the conversion of CO2, and thus less relevant to this project. The reaction can still provide a viable technique to convert CO2 when large amounts of CH4 are available. Furthermore, the produced ratio of H2:CO in the syngas of 2:1 is ideal for further application in the Fischer-Tropsch process to produce hydrocarbons [32].

2.1.3. Production of fuels from syngas

Now that the pathways to produce syngas from CO2 are discussed, it is useful to demonstrate how fuels can be produced from syngas. Diesel and other mixtures of hydrocarbons can be produced through the Fischer–Tropsch (FT) process. Oxygenated hydrocarbons can be produced using different catalytic processes and through syngas fermentation [17], [35].

The FT-process is a widely used process to produce short and long hydrocarbons from syngas. The process is very complicated and the final steps are not yet completely understood. Nevertheless, a simplification of the understood pathways is described. The first part of the process consists of multiple steps, of which adsorption and dissociation of syngas on the catalytic surface and the formation of single carbon hydrogenates on the catalytic surface are key. During the final part of the process, longer chains are formed from these single carbon species on the catalytic surface [17]. A simplified reaction scheme is shown in Figure 4.

A key step in the FT-process is the dissociation of CO. From the adsorbed carbon, CHX species are produced that undergo further coupling to grow chains. Experiments show that the dissociation of CO is, depending on the nature of the catalyst, influenced by the presence of kink and step sites, as well as surface vacancy sites on the catalytic surface. Kink and step sites are structural sites formed during catalyst synthesis that are suspected to act as the active sites in some catalytic systems [36].

(33)

9

Figure 4 Simplified Fischer-Tropsch reaction scheme [17]

The C-C coupling after the dissociation of CO that leads to chain growth is the most complicated step of the process. How exactly the coupling manifests itself is still unknown. The length of the produced hydrocarbon cannot be controlled and follows the Anderson - Shultz - Flory (ASF) statistical distribution, shown in equation (10) [17].

𝑀𝑛 = (1 − 𝛼)𝛼𝑛−1 (10)

Where 𝑛 is the number of carbon atoms in the chain and 𝛼 the chain growth probability. ASF distribution is typically adopted by step-growth polymerisation processes (e.g. polycondensation and polyaddition). Since FT follows a related mechanism as step-growth polymerisation, it has a similar distribution. ASF shows that smaller chains with less carbon atoms are more likely to be formed when 𝛼 is small, and the other way around. 𝛼 is determined by many factors like nature, nanoparticle size and crystallography of the catalyst. Choice of the right catalyst is therefore crucial to increase the selectivity of the desired chain lengths. Usually, the aim is to produce as high a chain length as possible, since long carbon chains can be cracked into shorter chains with a better selectivity [17].

From the statistical mixture of hydrocarbons as result of the FT-process, high quality fuels can be produced by further purification. Typical processes used for this are hydrocracking and oligomerization [37]. Hydrocracking is the process of selectively breaking long chain alkanes into shorter ones in the presence of H2 and a catalyst. Products like diesel and gasoline are often produced using this method [38]. Oligomerization is the process of making short polymers, also known as oligomers, from short hydrocarbon chains, producing longer alkanes. The process requires multiple reaction steps to finally produce alkanes with the desired length. During each step, a very limited amount of polymerisation is achieved, inhibited by added suppressors to control the reaction [39]. Oxygenates can also be produced from syngas in the FT- process with the use of specific catalysts that enable carbon-oxygen bonding. More commonly however, other catalytic processes are used to produce methanol, from which a multitude of other oxygenates can be synthesized. Present day catalyst consisting of copper (Cu), zinc (Zn) oxides and/or alumina (Al2O3) have a 99% selectivity of producing methanol from syngas [40]. Alternatively, syngas can be fermented by acetogenic bacteria to produce oxygenates. These bacteria follow the reductive acetyl-CoA pathway to synthesize acetate from syngas, from which ethanol, butanol and butanediol can be formed in further pathways. The end product depends on the bacteria used during fermentation [35].

(34)

10

2.1.4. Electrochemical carbon conversion

Apart from the thermal techniques discussed in this chapter so far, electrochemical methods are practiced for converting CO2. During these methods, electricity is used as an energy source for the reduction of CO2 in a process called electrolysis to produce valuable molecules like formic acid and ethanol [41]. The process can both be performed in dissolved state, where CO2 is dissolved in a basic solution as carbonate (CO32-), or in a gas diffusion electrode. The former has the disadvantage of the low solubility of CO2 in water. To increase the CO2 activity along the catalyst, high temperatures or pressures are used. The latter does not have this disadvantage. CO2, in a gaseous state, is sent along the cathode, which is deposited on a gas diffusion layer in contact with a basic solution [42]. Figure 5 shows three different, commonly-used configurations.

Figure 5 Electrochemical CO2 reduction cells [43]

In Figure 5a, the H-cell configuration is shown, where CO2 is in a dissolved state. Figure 5b depicts a gas diffusion electrode where the catholyte is a mobile phase. Figure 5c shows a gas diffusion electrode where the catholyte is fixed on a support. Even though CO2 is fed to the electrochemical cell in a gaseous phase in the last two configurations, the reaction mechanism is the same as in the H-cell since CO2 dissolves in the catholyte before reaction along the cathode can occur. Apart from the fact that the gaseous configurations are not limited by the low solubility of CO2 in water, since reaction occurs immediately after dissolving, they have an additional advantage concerning the CO2 diffusion rate towards the cathode. This happens much faster in a gas then in a liquid, with roughly a three magnitude difference between the two [43].

In all three configurations, the reduction of CO2 occurs at the cathode while water is oxidised at the anode. Simultaneous reduction of water at the cathode, where H2 is produced, cannot be controlled. The cathode is made from a material with a catalytic effect for the formation of the desired fuels. Gold (Au) and silver (Ag) can be used to produce CO with good efficiencies. Formic acid can be produced with tin (Sn) and indium (In) catalysts. Copper (Cu) is the only catalyst that has been reported to have this effect on the formation of C1 – C2 products [44]. Differences in morphology of the copper catalyst can increase the selectivity for the formation of different molecules [45]. The use of copper catalysts to produce products like ethene and ethanol will now be elaborated upon. A simplified reaction scheme is shown in Figure 6.

(35)

11

Figure 6 Simplified CO2 electroreduction reaction scheme on a copper catalyst [41]

The reaction scheme shows that methanol cannot be produced during the electrochemical reduction of CO2. This is peculiar since copper catalysts provide a high selectivity towards the formation of methanol in non-aqueous CO conversions, as described previously in chapter 2.1.3. The first reduction step, even before CO is formed, consists of the formation of formic acid. A second reduction step produces CO, which is only weakly adsorbed on the catalyst. Partial desorption of CO cannot be prevented. From there, the pathway splits to produce different intermediates that lead to the formation of methane, ethene and ethanol [41].

Selectivity towards the different products can further be altered by changing the applied potential in the electrochemical cell. For the production of CH4, a theoretical potential of +0,17 V vs. RHE is required when the thermodynamic data is consulted. In reality however, much higher (negative) potentials of -0,8 V are needed. The reason for this considerable overpotential is unknown, but it makes the process very inefficient. Research towards the cause of this phenomenon can reveal possibilities for optimisation to greatly increase the efficiency of the process. Figure 7 shows a typical example of the distribution of some of the products at different potentials. Faradaic yield is a relative measure of quantity of formation, where the number of electrons needed to form the chemical is taken into account [46].

(36)

12 The formation of formic acid and CO has a peak around -0,85 V. Since these two products are more efficiently produced with other catalysts, the potential is further increased to form CH4 and ethene at -1,0 V or higher. The decrease in H2 production at higher potentials is caused by the increasing coverage of the catalyst’s active sites by CO [46].

As previously mentioned, CO can selectively be produced by electrochemical CO2 reduction on gold (Au) and silver (Ag) catalysts. Since water is also reduced on the cathode during this process, with H2 production as a result, it is possible to make syngas in an alternative way. Since gold and silver are expensive materials to use as an electrode in these processes, other catalysts can be used to reduce CO2 to CO fairly selectively, like nickel (Ni) and zirconium (Zr) catalysts [47]. The catalysts have a lower yield, but they still provide the most economical option. The electrochemical cells used in the electrochemical syngas production strongly resemble the cells shown in Figure 5 [48].

2.1.5. Plasma catalytic carbon conversion

The conversion of carbon dioxide with the use of plasma catalysis is a fairly new technology compared to the previously described methods. Plasma conversions can be used as an alternative to multiple of the different, previously described conversion processes: it can be used to split CO2, as in the process of thermal CO2-splitting (2.1.1); to produce syngas from carbon dioxide and methane or from the simultaneous splitting of H2O and CO2, as an alternative to the dry reforming process (2.1.2) and thermochemical cycles respectively; or it can be used to convert the reactants into other chemicals in a single step, as is the case with electrochemical conversion (2.1.4).

The main reason for why plasma catalytic conversion processes are not favoured over the other methods at the present, is the fact that the energy efficiency of the plasma configurations is not as high. The electrochemical techniques and thermochemical cycles (2.1.1) provide the best energy efficiencies: water splitting can be performed with an energy efficiency of up to 75 % in electrochemical cells and carbon dioxide to syngas conversions, which can be performed in thermochemical cycles with a solar-to-fuel efficiency of up to 20 %, would need a plasma energy efficiency of at least 60 % in order to be competitive. Plasma conversions, on the other hand, only achieve energy efficiencies of up to 20 % at the moment [49].

Optimization is therefore required for plasma technology to be viable, although the lack of the disadvantages encountered in the electrochemical and thermochemical methods, e.g. the low solubility of CO2 in water and the inability to produce anything but syngas, respectively, shows that plasma technology still has high potential. To illustrate the ability to produce more than just syngas in carbon conversion processes with plasma technology, a proposed reaction scheme of the dry reforming process, when performed in a plasma, is shown in Figure 8.

The reactants of the dry reforming process, being CO2 and CH4, are shown to the left of the figure. The arrows show possible reaction pathways, where the thickness of the arrow is proportional to the ‘importance’ of the reaction, although they are no indication of the yield of the products. In the introduction, it was already mentioned that CO and H2, or syngas, is produced the most in plasma conversions without the use of a catalyst. Other potential products that can be recognised are methanol (CH3OH), formaldehyde (CH2O), ethane (C2H6), ethylene (C2H4), acetylene (C2H2), propane

(37)

13 (C3H8) and propylene (C3H6). The unboxed molecules are intermediates that are not stable and thus cannot exist as products [50].

Figure 8 Proposed pathways of the dry reforming process; thickness of arrow line is proportional to reaction’s importance [50]

The CH4 pathway shows its dissociation as a result of electron impact, where unstable CH3 and CH2 radicals are produced. The former recombines to form higher carbon species, e.g. ethane, propane and methanol, while the latter reacts with CO2 to form formaldehyde. CO2 is dissociated, also through electron impact, with the formation of CO and oxygen as a result. CO does not react further, except for its dissociation to produce more oxygen in small amounts, resulting in the high CO yields observed during plasma conversions. Oxygen initiates the formation of higher oxygenates, e.g. acetaldehyde (CH3CHO) and ketene (CH2CO) in small amounts, as indicated by the thin arrows [50].

Multiple dehydrogenation reactions on the non-oxygenated higher carbon species result in the production of unsaturated hydrocarbons and hydrogen gas, which, together with the dissociation of CH4, explains the high H2 yields observed during plasma conversion experiments. The very thick arrow leading from ethylene to a C2H5 radical indicates the high rate of this reaction, mainly resulting from the relatively high density of C2H5. This radical then reacts further to form ethane or propane, reportedly only acting as the intermediate of the backwards reaction of ethylene formation [50]. Since this thesis is centred around plasma technology, it will be more thoroughly covered than the carbon conversion methods described in previous chapters. Chapter 2.2 summarises the different plasma configurations employed today, as well as the chemistry of plasma reactions. Chapter 2.3 deals with the effect of a catalyst on the plasma reactions and the plasma itself, as well as the effect of the plasma on heterogenous catalysts.

(38)

14

2.2. Plasma technology

Plasma is in essence an ionized gas, but it differs significantly enough from an unionized gas for it to be described as a different state of matter. This ionized gas contains multiple highly reactive species e.g. free electrons, positively and negatively charged ions, excited atoms and radicals [10]. A plasma can be used to provide the reactivity required to make otherwise thermodynamically unfavourable reactions occur. The mechanisms behind this enhanced reactivity are very complex, comprising of many different phenomena such as electromagnetic fields and plasma waves [51]. Simply put however, the molecules are dissociated in the aforementioned reactive species and recombined to produce new compounds [52]. In order to have a basic understanding of the mechanisms, the different techniques to generate a plasma are described, after which some of the more important mechanisms, and the ways to control them, are explained.

There are multiple ways to generate a plasma, but the general principle of ionization is the same in every configuration: electrons are accelerated in different ways in order to ionise the molecules in the gas by colliding with them [10]. The different techniques used to accelerate the electrons can be split into two major groups: thermal and non-thermal [53]. Whether a plasma can be called thermal or not is closely related to the degree of ionisation achieved in the plasma. The higher the temperature, the faster the species in the gas will move and the more collisions will occur, resulting in more ionisation. Alternatively, a higher degree of ionisation will result in a rise in temperature because more species are accelerated in the applied electric field.

An important distinction to make when looking at plasma is the difference in speed at which the electrons and the heavier particles (e.g. ions, molecules & radicals) move. Because electrons are so light, they are much more easily accelerated in, say, a magnetic field than the heavier species. The observable temperature of the plasma, however, is only determined by the speed at which the heavier species move. This means that in a non-thermal plasma, the electrons move much faster than the heavier particles, while all species move at about the same speed in a thermal plasma. This phenomenon is usually referred to as the thermal equilibrium of the plasma and determines some of its properties.

2.2.1. Thermal techniques

Thermally generated plasmas are generally produced by sending a gas, called the plasma gas, at a high velocity through a high temperature discharge or an inductive coil. This high temperature discharge can be seen as a constant flow of electrons from one electrode to another. When the plasma gas comes in contact with this electron flow, the neutral molecules will collide with these electrons and ionise. An inductive coil on the other hand transfers the electric energy of a high voltage current running through the coil to the plasma gas in a process called inductive coupling. Magnetic fields generated by the coil accelerate the free electrons in the plasma gas, which collide with neutral species to initiate ionisation. The high degree of ionisation achieved in these configuration results in a very rapid temperature rise. Temperatures can range from 8 000 – 30 000 K. Nitrogen gas (N2), oxygen gas (O2), argon (Ar), helium (He) and hydrogen gas (H2), or mixtures of these, are often used as plasma gas [54], [55]. The different configurations are shown in Figure 9.

(39)

15

Figure 9 Thermal plasma configurations [54]

The first two configurations, shown Figure 9a and b, use a direct current (D.C.) electric arc to ionise the plasma gas. High voltages over two electrodes are required to maintain the arc, which is first generated by a high voltage discharge between the two electrodes. This discharge creates a conductive path through which the electrons ‘jump’ from the cathode to the anode, resulting in a visual arc that generates temperatures around 12 000 – 30 000 K. The difference between the two arc configurations is the placement of the anode (+). In Figure 9a, the anode is placed close to the cathode at the entrance of the plasma chamber. As a result, the arc is relatively small and positioned at the nozzle of the cathode, commonly referred to as a non-transferred arc. In Figure 9b, the anode is positioned at the other side of the plasma chamber, resulting in a distance between the anode and the cathode. Hence, the arc encompasses the whole chamber, which is called a transferred arc. While a non-transferred arc is only used to generate the plasma, a transferred arc will also come in direct contact with the reaction mixture and thus contribute in other ways to the observed reactions [52], [54]–[57]. Transferred D.C. arc plasmas can be used in cutting, welding and waste removal applications, where ultra-high temperatures are required. Non-transferred D.C. plasmas are mainly used for spraying and synthesis of powders. Plasma spraying is a commonly used technique to create uniform coatings that are heat and corrosion resistant [55]–[57].

The third configuration, shown in Figure 9c, is called an inductively coupled plasma (ICP) and uses an inductive coil to cause ionisation. The high degree of ionisation is generated through inductive coupling. When a radio-frequency alternating current (A.C.) is sent through the coil, a time varying magnetic field of the same frequency is generated [58]. The changing magnetic field will then produce electric currents in the conductive entity present inside the coil [59]. In a gas, this manifests in the acceleration of the free electrons [60], that will collide numerous times with the other species inside the gas and transfer their energy [61]. When a free electron with enough energy collides with an electron of a neutral atom, that bound electron will vacate the atom and the atom will be left ionized. Once this process of ionisation occurs in a high enough degree, the plasma state is achieved [58]. Because of the low number of free electrons in an unionised gas, a form of ignition, by a tesla coil for example, is required to initialize the formation of the plasma [62]. To prevent the plasma from reaching the walls of the torch in an ICP, a sheath gas is sent along the wall to shield it [62]. The sheath gas is composed of a gas that is difficult to ionise, like H2, and is applied at higher velocities as to minimize heat transfer to the torch wall [63]. The central gas serves as the plasma gas. Applications of ICP’s are excessively found in quantitative chemical analysis. Different atomic spectroscopy methods and mass spectrometry can employ the plasma as ionization source [60], [61], [64]. The plasma is further used for preparation and coating of metal powders [61], [63], [65].

(40)

16 An alternative configuration to the inductive coil plasma (not shown in figure 1) is the capacitively coupled plasma. This configuration works with capacitive coupling instead of inductive coupling but has very similar properties. Furthermore, its principles are comparable to the dielectric barrier discharge plasma discussed further in chapter 2.2.2, where the non-thermal plasma generating techniques are summarised. The capacitively coupled plasma technique will therefore not be discussed further in this chapter.

The last configuration shown in Figure 9d is a combination of the previously mentioned techniques. Both a high temperature arc between two electrodes and an inductive coil are used to maintain ionization. This enables the application of higher flow velocities compared to the inductive coil configuration and a higher control of the plasma than with arc configurations [54], [66]. The use of this configuration is still limited because of its complicated nature, but the combined advantage of both techniques shows high potential [54].

In a thermal plasma, the temperatures of the electrons and heavier particles have about the same temperature, or they are in local thermal equilibrium (LTE). This indicates that, when observing one of the species, the temperature is the same as its neighbouring species. A complete thermal equilibrium is not achieved because of significant energy losses in the form of electromagnetic radiation. Nevertheless, there are a couple of useful laws, like the Maxwell and Boltzmann distributions, still satisfied under LTE conditions. These laws can be used to characterise the plasmas [52], [67].

2.2.2. Non-thermal techniques

The high temperatures accumulated in thermal plasmas are not desirable for many applications, for example when heat sensitive materials are treated. Plasmas can also be maintained at lower temperatures, even at room temperature, and are referred to as non-thermal [52]. Under these low temperature conditions, the heavier particles, which determine the observable temperature, move much slower than in a thermal plasma. Non-thermal plasma is generated through low temperature electrical discharges between two electrodes, with ionization of the gas between the electrodes as a result. The same principle of ionization is employed in every configuration, which is called the electron avalanche (Figure 13), where accelerated electrons are responsible for ionisation [68]–[70]. Note how, in principle, ICPs are generated in the same way, but they are classified as thermal plasma. The difference has to do with the degree of ionisation, which is much higher in ICPs, resulting in much faster movement of the heavier species. There are three major groups of non-thermal plasma generating techniques: the glow discharge, corona discharge and dielectric barrier discharge. The same gasses can be used as plasma gas as in the thermal techniques.

A glow discharge plasma is generated by applying a direct current to two electrodes positioned at opposite ends of a gas filled tube. The gas in the tube is ionized through the electron avalanche phenomenon (more thoroughly described in chapter 2.2.3), where electrons are produced at the cathode through secondary emission by cations [69], also referred to as sputtering. Different zones in the plasma can be visually distinguished [69], [71], [72]. A schematic rendition of the different zones is shown in Figure 10.

(41)

17

Figure 10 Glow discharge plasma [73]

Formation of the different zones can be explained when going over the path an electron follows when it is produced at the cathode. The Aston Dark Space is the result of the initial low energy of the electrons right after sputtering, where the energy is not high enough for ionisation to occur. When the electrons reach the Cathode Glow zone, their energy is sufficient for ionisation, producing visual radiation in the process. Further rise in the energy of the electrons results in ionisation of atoms without the emission of visible light in the Cathode Dark Space. Upon reaching the Negative Glow zone, the number of electrons has exponentially risen to such an amount that electron energy and ionisation rate is decreasing, resulting in a very intense emission of visible light. Further from the cathode, the electron energy is further decreased so that the emission of visible light is stopped in the Faraday Space. The electron density decreases and the electrical field gets stronger to form a uniform quasi-neutral plasma in the Positive Column. An increase in the electrical field at the end of the Positive Column towards the anode results in a spike in the emission of visible light, the Anode Glow. Close to the anode, the electron density reduces so that emission of visible light stops in the Anode Dark Space [69], [71], [72].

A glow discharge plasma is mainly used in the chemical analysis of solids, liquids and gasses. Atomic absorption and atomic emission spectroscopy, as well as mass spectrometry, are methods in which the plasma can be used to ionise the sample for detection. Further applications are modification and cleaning of metallic surfaces. Since the Positive Column zone in the glow discharge is the most uniform and stable, this part is most frequently used in the aforementioned applications [71].

A corona discharge plasma is generated through the corona effect. When a conductive material, like a metal, is charged highly enough, it will start attracting or rejecting electrons close to the surface of the material, depending on the charge. The shape of the conductive material is important for the effectiveness of these attracting/rejecting forces, as the surface charge is more potent in needle-like structures. The accelerated electrons will, when having a high enough energy, ionize the molecules surrounding it through the electron avalanche process. The plasma is only formed very close to the charged surface and is hardly visible [68], [71]. A sketch of this phenomenon is shown in Figure 11. Electrons produced in natural ionization events initiate the ionization process. The electrons around the negatively charged surface accelerate towards the grounded electrode, while the produced cations will be attracted to the discharge electrode. Collision of electrons with electronegative molecules, like O2, produces anions. These negatively charged ions will accelerate away from the negatively charged electrode, just like the electrons [68], [71]. Plasma reactor configurations consist of a collection of grids with pins on which a high voltage is applied. The pins will each provide corona discharges causing the ionization of the gas sent through the grids [69].

(42)

18

Figure 11 Negative D.C. corona discharge; figure is not to scale, the plasma region takes only a small part of the interelectrode spacing [68]

Corona discharge plasma is applied in ozone production reactors, for the surface treatment of polymers and fabrics, and as electrostatic precipitators [68]. The latter application is a method to remove unwanted volatiles from an air stream. Corona discharge plasma is also used for the removal of unwanted charges on the outside of aircrafts to prevent uncontrolled electrical discharges [71]. A dielectric barrier discharge (DBD) plasma is one of the more technologically relevant non-thermal plasmas at the present because of its reproducibility [74]. It works similar to a glow discharge, but the electrodes are much closer to each other and the space between the two electrodes is partly filled with a dielectric (electrically insulating) material. A high A.C. voltage initiates ionization of the gas in between the two electrodes instead of the D.C. used in a glow discharge [75]. Different configurations are shown in Figure 12.

Both planar and cylindrical configurations exist. Configurations where the electrode is covered by the dielectric barrier are less susceptible to corrosion and erosion. However, these configurations need higher voltages because of the drop in voltage caused by the barrier [74]. Because of the similarity in build to a capacitor, it cannot be operated with a direct current. An alternating current is required to prevent saturation of the capacitor, after which the electrical flow would stop [71], [74], [75].

(43)

19 The dielectric barrier, which can be made of glass, quartz, ceramics, rubbers or other non-conductive materials [71], [74], is the key element in this plasma generating technique that provides the properties that make it unique. It provides for an electrical separation between the two electrodes, so that many micro discharges occur between the electrodes and the dielectric barrier instead of a single arc discharge [74], [76]. Every time the polarity of the electrodes changes because of the applied A.C. voltage, a discharge occurs. Because of the slow movement of the ions between the two electrodes relative to the electrons, which is a property of all non-thermal plasmas, the micro discharges follow the same path every time a discharge occurs. The plasma is therefore not completely uniform and is not in thermal equilibrium [69].

A uniform plasma can still be generated with a dielectric barrier discharge configuration under specific conditions. The purity and composition of the plasma gas are the most important factors for achieving this, as well as the frequency of the applied voltage. When high gas ionization is achieved in these conditions, the plasma generation is more a product of the electron avalanche than the occurrence of discharges. This creates a uniform plasma over the whole volume between the electrodes. Maintaining this uniformity can prove to be challenging however, and for many applications it is not necessary [69], [74], [75].

Applications of the DBD plasma are ozone generation, destruction of pollutants, surface treatment, as lamps or displays, synthesis of certain molecules, decontamination of tools and the treatment of liquids. It is also used in industrial high power CO2 lasers [52], [69], [71], [74], [75].

2.2.3. Plasma chemistry

A diverse array of mechanisms occurring in plasma influence the reactivity of the molecules. Both chemical and physical processes play an important role in plasma reactions. Some of the more important chemical mechanisms will be described in this subchapter in order to give an idea of the complex nature of plasma reactions. A summary of the different mechanisms is shown in Table 1. The most important phenomenon happening in a plasma, which is already mentioned multiple times in previous subchapters, is the ionisation of the molecules. In thermal plasmas, a high degree of ionization is achieved because of the way the plasma is induced. This gives rise to a high temperature and vice versa [55]. In fact, in 1920, Saha M. derived an equation that directly relates the degree of ionization with the temperature of a gas in thermal equilibrium, named the Saha Ionization Equation. For a monoatomic gas where only lower energy excitations are observed, the equation is shown in equation (11) [67], [70], [77]: 𝑥+_{∙ 𝑒}− 𝑥 = 𝑍+ 𝑍 (2𝜋𝑚𝑒 𝑘𝐵𝑇 ℎ2 ) 3 2_{∙ 𝑒}− 𝑈 𝑘𝐵𝑇 (11)

Where the left-hand side is the equilibrium constant for the ionisation reaction of element 𝑥, 𝑍 and 𝑍+ are the partition functions of the neutral and ionized particle respectively, 𝑚𝑒 is the mass of an electron, 𝑇 is the temperature, 𝐾𝐵 and ℎ are the Boltzmann and Planck constant respectively and 𝑈 is the ionization potential. This equation is derived using the combination of different thermodynamic laws, statistical mechanics and quantum mechanical assumptions. As this is beyond the scope of this thesis, no further attention will be given to this formula, but the complete derivation can be viewed in the referenced paper [77].