BACHELOR
The reduction of iron oxide in a non-thermal hydrogen plasma a challenge to circular energy storage
Tennebroek, S.
Award date:
2019
Link to publication
Disclaimer
This document contains a student thesis (bachelor's or master's), as authored by a student at Eindhoven University of Technology. Student theses are made available in the TU/e repository upon obtaining the required degree. The grade received is not published on the document as presented in the repository. The required complexity or quality of research of student theses may vary by program, and the required minimum study period may vary in duration.
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
The reduction of iron oxide in a non-thermal hydrogen plasma
A challenge to circular energy storage
Sam Tennebroek
Supervisors:
Job Beckers, Boy van Minderhout, Bart Platier
Eindhoven, Saturday 16
thMarch, 2019
Circular energy storage is the missing link in the energy transition to using renewable energy instead of fossil fuels. Iron powder seems to be a promising candidate to provide a solution for this problem by serving as an energy carrier. In order to make this work, iron oxide powder has to be reduced to iron powder. When a non-thermal (radio frequency powered) hydrogen plasma is used to reduce iron oxide, hydroxyl radicals are formed, which can be detected by optical emission spectroscopy. It is investigated if the reduction process of iron oxide by a hydrogen plasma can be observed using this detection method.
First, the possible processes or phenomena which could influence the emission of light by the hydroxyl radicals are investigated, because these can (partly) account for its measured value. Pre- liminary results show that the flow of hydrogen does not influence the hydroxyl emission, while the pressure has a major influence on this, but can be corrected for accordingly by analysing the ratio between a nitrogen emission peak and a hydroxyl emission peak. Besides, water or air present in porous iron oxide particles can influence the measurement results. Water present in the vacuum chamber and very small leakages could be (small) causes as well. These causes could not be ruled out and should be considered when drawing conclusions.
Next, two types of iron oxide powders are placed in a hydrogen plasma and after that, the amount of hydroxyl particles is measured by analysing the relevant emission peaks. One measurement of a mixed iron powder (a mix of Fe2O3, Fe3O4 and FeO) with a power density of 0.32 W/cm3 indicated that the reduction process may have occurred, because of the remarkable changes in the emission spectrum around the emission peak of hydroxyl radicals while the rest of the spectrum stayed constant. Thereafter, two measurements with a higher power density, and of which one with pure magnetite powder provided insights in the reduction process through the ratios between nitrogen and hydroxyl emission peaks, possibly indicating the reduction process as well.
Contents iii
1 Introduction 1
2 Theory 4
2.1 Capacitively coupled plasma. . . 4
2.2 Paschen’s law . . . 6
2.3 Interactions between a hydrogen plasma and iron oxide . . . 8
2.3.1 Thermodynamics . . . 8
2.3.2 Optical emission . . . 11
3 Experimental set-up 14 3.1 Key components . . . 14
3.1.1 Vacuum chamber and discharge. . . 14
3.1.2 Photon detection . . . 17
3.1.3 Thermogravimetric analysis . . . 17
3.2 Calibrations . . . 18
3.2.1 Transmission of the quartz glass . . . 18
3.2.2 Dark current . . . 19
3.2.3 Background light . . . 20
3.3 Iron oxide powders . . . 20
4 Results and discussion 23 4.1 The Paschen curve . . . 23
4.2 A hydrogen plasma without iron oxide . . . 24
4.3 Pressure dependency . . . 25
4.4 Sub-question 1: other possible causes for changes of hydroxyl emission peaks . . . 28
4.4.1 Hydrogen flow . . . 28
4.4.2 Porous iron oxide powder . . . 29
4.4.3 Other causes . . . 30
4.5 Sub-question 2: The time evolution of the hydroxyl emission peaks . . . 31
4.5.1 Mixed iron oxide in a hydrogen plasma . . . 31
4.5.2 Increased plasma power . . . 34
4.5.3 Magnetite and increased plasma powder . . . 36
4.6 Recommendations . . . 40
4.7 Outlook . . . 41
5 Conclusion 43
Bibliography 45
Introduction
The human population is rapidly growing, increasing its demand for energy [1]. To sustain future generations, next to the current generations, it is paramount that this extra demand for energy is supplied without compromising future generations [2].
The current regime of energy is not sustainable. Fossil fuels, being hydrocarbons, are the major contributor to an increase in greenhouse gases (such as CO2) into the atmosphere, at a pace 200 times faster than typical for Earth’s pre-industrial carbon cycle. An increase in greenhouse gases and the resulting climate change has a disastrous impact on the quality of life on Earth for humans [3]. Also, fossil fuels release numerous pollutants that affect human health and natural ecosystems [4]. To mitigate global climate change, the energy system must transition away from the use of fossil sources. Instead, net zero carbon sources should be used [5]. To accelerate the transition to sustainable energy sources, commitments have been made internationally in the Paris Climate Agreement.
Beside the societal reason for stopping the use of fossil fuels, there is also an economic one. The fossil fuel reserves are limited and are increasingly more difficult to exploit and therefore becom- ing more expensive. Hence, a shift to renewable (naturally replenished) energy sources is necessary.
Most sustainable energy sources are inherently intermittent, causing excessive power fluctuations if used as primary energy. The sun shines only during the day and can be blocked by clouds, wind strength varies over the day and seasons, and the period of high-energy demand does usually not coincide with peak production. This causes unreliable energy supply and creates challenges in facilitating load balance to ensure power network stability [6]. Another viable solution is the use of energy storage [7].
There are several ways of storing energy, including mechanical (e.g. flywheel, pumped hydroelectric storage), electrical (e.g. capacitor), thermal (e.g. phase-change materials, aquifer), electrochem- ical (e.g. flow battery, rechargeable battery) or chemical (e.g. biofuels, hydrated salts, hydrogen, power to gas). However, most energy storage technologies currently lack the required storage dimensions (such as high energy density, large storage duration or a high degree of mobility) or are economically not feasible. Therefore, hydrocarbons still play an important role in our energy system.
For these reasons, it is necessary to develop an energy carrier that can be replenished with renew- able energy, possesses the required storage dimensions and when needed can be used for heating, electricity production and transportation. The ideal energy carrier should also be safe and be available in large quantities. More importantly, the ideal energy carrier should be green, non- polluting, and recyclable in an environmentally friendly way [8].
Metal powders are a promising, yet largely-overlooked, solution [8]. Using these metals as an energy carrier is (somewhat misleadingly) called “metal fuels”. In this report, there will be focused on iron powder as an energy carrier. The energy cycle shown in Figure 1.1 shows how energy can be stored in and gained from iron powder. The starting point in this cycle is a pile of iron oxide powder (rust). Energy from sustainable sources, like the sun or the wind, can be used to transform the rust powder into iron powder. This is called the reduction. The iron powder thereby contains the energy. This iron powder can be stored for prolonged periods of time or transported across the globe. To use the stored energy where-ever and whenever energy is needed, the iron powder can be combusted. In the combustion process, it is converted back into iron oxide powder, making iron powder a circular sustainable energy carrier.
Figure 1.1: The metal fuel cycle including three of the most promising applications.
The focus of this report lies on the reduction part of this energy cycle. Currently, iron ore is already being reduced on a large scale in blast furnaces in the industry. However, this smelting of iron ores (which is predominantly Fe2O3, iron oxide [9]) is performed using processes in which CO2 is emitted. The entire iron and steel industry accounts for 6.7% of the global anthropogenic carbon dioxide emissions [10]. It is therefore expected that more and more research will be done on ways on how to reduce iron oxide sustainably. It is already known that this can be carried out in several ways: the electrolysis of water could be used to produce hydrogen that can reduce iron oxide in a reactor with a temperature of about 480◦C and a pressure of 3.6 MPa [11]. Also, direct electrolysis of molten iron oxide is possible, in which energy will be used directly to split off an oxygen atom [12]. This report will focus on a more sophisticated way of reducing iron oxide; the abilities of hydrogen plasma to reduce iron oxide will be investigated.
One of the special and unique properties of the reduction with a hydrogen plasma is that it contains radicals (excited atoms or molecules) of hydrogen. These radicals have much higher reactivity than non-excited atoms and molecules, which enables them to react easier with iron oxide molecules.
After this reaction, hydroxyl (OH) radicals and iron will form.
This reduction method has already been researched in several articles [13,14, 15, 16]. However, the reduction of iron oxide powder with particle diameters of several tens of microns has never been researched before. Besides, much research done on the reduction of metal oxides use thermal plasmas (plasmas in which the temperature of the ions is equal to the temperature of the elec- trons). In the experiments conducted in this report, the focus is on non-thermal plasmas. Often, non-thermal plasmas can be operated at ambient temperatures, and can, therefore, be used to treat wounds, for example. In this report, the evolution of hydroxyl radicals has been observed by using optical emission spectroscopy (OES), which is a new method for monitoring the reduction process. Different from other methods, using OES has the advantage that the reduction process
can be monitored real-time and without having to take out or even touch the sample. This benefit can be especially important in large-scale applications.
Hence, the research question this report will try to answer can be formulated as:
Can the reduction of iron oxide in a non-thermal plasma be detected by analysing the spectral emission peaks of hydroxyl radicals?
This question is supported by two sub-questions:
What other processes next to the reduction of iron oxide can change the intensity of hydroxyl emission peaks?
How do the spectral emission peaks of hydroxyl change over time if iron oxide is present in a hydrogen plasma?
This investigation aims to find answers to the sub-questions and ultimately the main research question in collaboration with Philemon Koolen, who is also doing research in this field. He is investigating if the reduction of iron oxide takes place in a hydrogen plasma, by measuring the process with OES and comparing this to the reaction products being measured using X-ray dif- fraction and X-ray photoelectron spectroscopy. He does this research with three different types of iron oxide powder, and draws his conclusions from stationary OES-measurements in particular.
In this report however, the focus is maily on both studying the time-evolution of the emission spectra of the hydrogen plasma, and accounting for processes and phenomena which could disturb the OES-measurements and conducting relevant measurements accordingly. A few (calibration) measurements could be used both by Philemon Koolen and in this report. Therefore, it will be indicated in the caption of a figure of a measurement if a measurement conducted by Philemon Koolen is used. If no indication is given, the measurement is conducted by Sam Tennebroek. All analyses and conclusions drawn in this report are made individually as well.
In this report, the aspects of the theoretical background regarding the reduction of iron oxide in a hydrogen plasma will be explained, and hypotheses will be formulated along these explanations.
Next, the details of the experimental set-up will be elaborated used to conduct the experiments, and the results of relevant calibration measurements will be shown. Subsequently, the results of the experiments will be shown and discussed, leading to the conclusions and answers to the sub-questions and the research question of this report.
Theory
In this chapter, the theoretical background information which is relevant for understanding the behaviour of a hydrogen plasma, especially when it comes in contact with iron oxide powder, will be discussed. First, the method used to create the plasma which is used in the experiments (a capacitively coupled plasma) and its properties are explained. Next, the relation between pressure, distance between the electrodes and discharge voltage which is governed by Paschen’s law is shown. Finally, the expected (theoretical) interactions between the hydrogen plasma and iron oxide are discussed by elaborating on the thermodynamical processes and optical emission which occur during these interactions.
2.1 Capacitively coupled plasma
A plasma is a (partially) ionised gas, which is often called ‘the fourth state of matter’. There are several ways to create a plasma, but the most common method is to apply a voltage difference between two electrodes with gas in between. This section will elaborate on how a plasma can be created in such a system and what its behaviour is in this configuration.
A capacitively coupled plasma (CCP) consists of two electrodes of which one is grounded, and the other is connected to an AC power source, seeFigure 2.1. The term ‘capacitively’ originates from the fact that the resulting electrical circuit looks like that of a (dis)charging capacitor. A CCP is used a lot in the industry nowadays because it is simple to construct and has a high scalability [17]. This makes it one of the most common methods to create a plasma. In this report, a regular CCP with a radio-frequency (RF) power supply is used. This means that the voltage will alternate with a frequency in the radio-waves domain, which includes frequencies from 20 kHz up to 300 GHz. Usually, a frequency of 13.56 MHz is chosen, because certain bands in the radio-frequency domain are reserved for industrial, medical and scientific purposes, to avoid disruption of radio communication at the same frequency. A frequency of 13.56 MHz is located in such a band. Due to the electric field that will be created when the power supply is turned on, the gas atoms will be ionised, and electrons will be released. These electrons can then collide with other atoms to release secondary electrons that can ionise other gas atoms in their turn. If the voltage is suf- ficiently high (that is, above the so-called plasma’s breakdown voltage), this chain reaction of electrons that ionise their fellow atoms becomes so reactive that an abundant amount of electrons is created. Thereby, the (weakly) ionised gas between the electrodes becomes conductive, due to the self-sustaining gas ionisation caused by this chain reaction [18].
Figure 2.1: A typical CCP, powered by a RF voltage source. A plasma can be generated between the electrodes [18].
Even when the power supply has not been put on, the ions and electrons will still move due to thermal energy:
kBT =1
2mv2. (2.1)
The mass of ions is significantly higher than the mass of the electrons (mi>> me), while the tem- peratures of the electrons is equal or higher compared to the temperature of the ions (Te& Ti).
According to Equation 2.1, this means that when the gas is ionised, the thermal velocity of the electrons will be considerably higher than the thermal velocity of the ions. Therefore, electrons (especially at the boundary) will reach the electrodes faster than the ions, which results in a thin layer of positive ions directly next to the electrodes. These layers are called sheaths, and their formation happens in a very small time-scale. An electric field pointing away from the middle of the plasma is established in the sheaths, which forms a barrier for electrons to pass through: most of them are confined between the two electrodes, only those with an energy (or velocity) that is sufficiently high can penetrate through the sheath. The ions, however, are accelerated towards the electrodes, although at a much slower rate than the electrons [18]. Next, the sheath becomes more positive as ions are attracted to the sheath. In turn, the sheath will repel positive ions more and attract negative electrons, going through the whole process again. After several iterations, the plasma and its sheaths will reach a steady-state configuration in which the ion flux will be equal to the electron flux.
In a CCP with an RF power supply, the electrons move collectively, as an electron cloud under the changing electric field, as shown inFigure 2.2. Because the electrons are light, they will react to the instantaneous electric field, while the massive ions will only be influenced by the time-averaged electric field, due to their relatively high moment of inertia.
Figure 2.2: The electron cloud oscillates under the influence of the changing electric field in an CCP with RF power supply [18].
By inserting different gasses between the electrodes, different plasmas can be created, each with unique properties. Because hydrogen is suitable for the reduction of iron oxide, the properties of a hydrogen plasma will be treated in the upcoming sections.
2.2 Paschen’s law
The discharge of a plasma is not a trivial phenomenon. As explained in the previous section, the energy of the electrons in the ionised gas has to be sufficiently high to ‘kick out’ electrons of other atoms to establish a chain reaction. This chain reaction then enables the plasma to discharge.
Various parameters play a role in the discharge process, of which the most important ones are the voltage difference between the two electrodes, the distance between the electrodes and the pressure of the gas.
Paschen’s law,
VB= Bpd
ln(Apd/ ln(1/γ)), (2.2)
describes a relation between the breakdown voltage (the voltage at which a discharge is estab- lished, VB) and the product of the distance between the electrodes d and the pressure p. γ is the probability that a secondary electron will be emitted when an electron is incident on an atom [19]. If the parameters contained by Equation 2.2are known, a graph can be plotted displaying Paschen’s law (sometimes this is done on a logarithmic scale). This graph is called a Paschen curve.
Figure 2.3: The Paschen curve of several different plasmas [20]. γ = 10−2 is used.
Establishing a discharge requires different voltages at different distances between the electrodes and pressure. If the pressure is too low, the electrons build up speed and become energetic, but cannot collide with many atoms simply because there are few. On the other hand, if the pressure is too high, the electrons cannot build up speed and collide too quickly with atoms, well before they reach a velocity (energy) sufficient to ionise an atom. The same goes for the distance between the electrodes. If this is too small, it is hard for the electrons to find an atom to collide with. In contrast, if the distance is too large, the electrons will have to go through a thicker layer of atoms, which results in a higher collision rate.
The Paschen curve has one global minimum (the minimum breakdown voltage), which indicates the lowest voltage difference needed to establish a discharge. Therefore, it is interesting to investigate where this minimum occurs. Differentiating with respect to pd yields
(pd)min= e ln(1/γ)
A (2.3)
and
(VB)min=eB ln(1/γ)
A . (2.4)
Literature suggests that for hydrogen the minimum occurs at about 2.55 Torr·cm and 320 V [20].
One Torr is about 133.3 Pa. This minimum indicates under what conditions the plasma can be ignited the easiest, that is, with the lowest breakdown voltage.
After discharge, a plasma will emit light, as will be described in sectionsubsection 2.3.2. Next, to information on the breakdown voltages at values of pd, the Paschen curve also provides insights on the dependence of the intensity on values of pd. This is because when a higher voltage than the breakdown voltage is used (at a particular value of pd), the extra energy that is stored in the electric field can be used to accelerate electrons even more. The emission of light, therefore, occurs more easily, and after a collision particles have a higher chance of having sufficient energy to collide with a second particle or to bringing the atoms they collide with to a higher excited
state. In particular, the generated information on the intensity by the Paschen curve can be useful in the explanation of the results of the conducted experiments.
2.3 Interactions between a hydrogen plasma and iron oxide
Up till this point, it is explained how a hydrogen plasma can be created, how it behaves in a CCP and what Paschen’s law tells about the breakdown voltage of a hydrogen plasma. In this section, the interactions with iron oxide powder are accounted for. Both the thermodynamic interactions and the optical emission phenomena are discussed, with the goal of lying a basis for the second sub-question (How do the spectral emission peaks of hydroxyl change over time if iron oxide is present in a hydrogen plasma?) of this report.
Several articles state that hydrogen plasma is suitable for the reduction of iron oxide under certain conditions [13,14,15,16]. The chemical reduction process of iron occurs in a series of steps [13]:
Fe2O3−−→ Fe3O4−−→ FeO −−→ Fe. (2.5) In this reduction process, the (molar) percentage of oxygen becomes less and less, until the iron oxide fully reduces in the last step. In the oxidation process, this will be the other way around:
Fe −−→ FeO −−→ Fe3O4−−→ Fe2O3. (2.6) If iron is combusted, first FeO will be produced. Very quickly, the FeO will disproportionate to Fe3O4 because FeO is thermodynamically unstable below temperatures of 575◦C [21]. Finally, Fe3O4 is converted to Fe2O3, but unlike the previous conversion, the kinetics of this reaction are rather slow [22].
There are different kinds of Fe2O3, each with their specific structure. α-Fe2O3 has a rhombo- hedral structure and occurs in iron ores (which are reduced in the industry). α-Fe2O3 exhibits weak ferromagnetism at room temperature and above, and is the most common form of Fe2O3. In comparison, the other two types of Fe2O3 are metastable. β-Fe2O3 has a body centred cubic structure and will transform to α-Fe2O3 when heated up to temperatures of 500◦C. γ-Fe2O3 has a cubic structure and is also converted to α-Fe2O3 at high temperatures [23,24].
If the breakdown voltage of Equation 2.2 has been reached in a hydrogen gas, a discharge will be established, and the hydrogen molecules will dissociate. The most important species [25] that then arise are monoatomic hydrogen (H), ionic hydrogen (H+, H2+, H3+) and rotationally and vibrationally exited hydrogen molecules (H2*) [26]. These are called hydrogen radicals, which can effectuate the reduction of Fe2O3:
2 Fe2O3+ 6 H/ 6 H+/ 3 H2+
/ 2 H3+
/ 3 H2∗−−→ 4 Fe + 6 OH∗. (2.7) In this reaction, it is assumed that Fe2O3 is the starting product. Radicals are often significantly more reactive than hydrogen molecules. The next sections will elaborate on this.
2.3.1 Thermodynamics
Traditional and sustainable reduction of iron oxide is currently done with hydrogen gas in the following way:
Fe2O3+ 3 H2−−→ 2 Fe + 3 H2O. (2.8) Because this process is endothermic, a lot of additional energy (e.g. heat) is needed. The Gibbs free energy (∆G) denotes the spontaneity of a chemical reaction. If the Gibbs free energy is negative, the reaction will be spontaneous. Therefore ∆Gtraditional > 0. However, the Gibbs
free energy of the conversion of molecular hydrogen into atomic hydrogen, hydrogen radicals and rotationally and vibrationally excited hydrogen is higher, which means that ∆Gradicals>> 0. If the Gibbs free energies ofEquation 2.8and the creation of radicals are combined, this results in a negative Gibbs free energy forEquation 2.7: ∆Gplasma= ∆Gtraditional− ∆Gradicals < 0. This also becomes clear from the Ellingham diagram shown inFigure 2.4. An Ellingham diagram gives the Gibbs free energy of several reactions as a function of the temperature. For the hydrogen molecules, atoms and radicals have to bond with the oxygen atoms from the iron oxide, it is relevant to check their corresponding lines in the Ellingham diagram. It can be seen that the reaction involving molecular hydrogen has the highest Gibbs free energy as compared to the other two red lines (H and H+). This indicates the advantage of using hydrogen plasma compared to hydrogen gas for the reduction of iron oxide.
Figure 2.4: An Ellingham diagram for several oxidation reactions [26]. The hydrogen- related lines are shown in red.
Both atomic hydrogen and ionic hydrogen are present in a hydrogen plasma. Zhang et al. investig- ated which of the hydrogen species that are present in the plasma are chemically the most active, which will, therefore, have a low Gibbs free energy. The results of this research are shown in Figure 2.5. It shows that the reduction ability from best to worst is H+> H2+> H3+ > H > H2.
Figure 2.5: An Ellingham diagram with the reactions of different hydrogen species [26].
Last but not least, a hydrogen plasma also contains rotationally and vibrationally excited mo- lecules. These species are essential for many reasons. The reduction ability of these ro-vibrationally excited molecules can be significantly higher than that of regular molecules, merely because they are at a higher energetic level. Over 95% [26] of the energy that is put in the plasma is transferred by electrons to excite hydrogen molecules. Additionally, H2 can remain excited for a relatively long time, as compared to some other molecules.
Furthermore, the Gibbs free energy, and therefore the reduction rate of the reaction depends on the temperature. The dependence of the degree of reduction on the temperature is reported in literature [13]. The results are shown inFigure 2.6.
Figure 2.6: Reduction rate of iron oxide in a hydrogen plasma at different temperatures.
In this figure, it can be seen that it takes more time to reduce iron oxide as the temperature becomes lower. Figure 2.6 clearly shows that substantial reduction takes place at 573 K and it is expected that this will also be the case for the selected working temperature of the experiments conducted in this report.
2.3.2 Optical emission
A hydrogen plasma is a mixture of neutral molecules, atoms, ions, ro-vibrationally excited atoms, and electrons. A plasma is very dynamic: electrons collide with all kinds of hydrogen species and excited atoms undergo relaxations and thereafter can become excited again. These interactions make that photons will be emitted. This happens for example when an electron is pushed in an outer electron shell of the ion (although for a short time), and the electron falls back to an inner shell. The energy of the photon (and therefore, the wavelength) changes according to
En2− En1= hν, (2.9)
where n1and n2are the principal quantum numbers, according to the shell the electron is in, with n1< n2. Rydberg’s formula,
1 λvac
= R 1 n21 − 1
n22
, (2.10)
in which λvac is the wavelength in vacuum as a result of the electron transition and R is the Rydberg constant (with a numerical value of approximately 1.097 · 107 m-1). Equation 2.10 is only valid for hydrogen, and it suggests a different series of emission wavelengths for different transitions. These series are the well-known Lyman, Balmer, and Paschen series. The Lyman series describes the wavelengths of the emitted photons for transitions from an arbitrary shell to the most inner shell (n1= 1 and n2= 2, 3, ..., ∞). The Balmer and Paschen series are constructed similarly, but they describe transitions to the second most inner shell (n1 = 2) and third most inner shell (n1 = 3) respectively. Table 2.1 gives some of the wavelengths corresponding to the three series.
Table 2.1: Wavelengths in vacuum of emitted photons in the Lyman, Balmer and Paschen series [27]. All values are in nanometers.
n2 Lyman (n1= 1) Balmer (n1= 2) Paschen (n1= 3)
2 121.6 - -
3 102.6 656.3 -
4 97.3 486.1 1875
5 95.0 434.0 1282
6 93.8 410.2 1094
∞ 91.2 364.6 820.4
In particular, the Balmer series is the only series that contains (four) wavelengths in the visible range. The spectral line corresponding to the first transition in the Balmer series is called H-α, the second one H-β, and so on. Noting that ν = c/λ and combining this with Equation 2.9, it can be derived that a decrease in wavelength corresponds with an increase in energy difference.
Hence, because the energy of the H-α transition has the smallest energy difference, this will be the spectral line with the highest intensity in the visible range.
In the emission spectrum of atomic hydrogen there will be sharp spectral lines at the wavelengths denotedTable 2.1(as with any other atom), seeFigure 2.7. A molecule can, unlike atoms, be in a rotationally or vibrationally excited state. Because of the symmetry of an atom, it cannot possess such properties. The vibrational states are more energetic, which is why molecules have different rotational states within a vibrational state [28]. For hydrogen molecules, these excited states can, for example, create the so-called Fulcher bands [29], which are also shown inFigure 2.7.
Figure 2.7: An emission spectrum of a hydrogen plasma [30]. Besides the Balmer spec- tral lines, the Fulcher bands originating from ro-vibrationally hydrogen molecules can be observed.
These bands are actually a collection of a large number of small peaks that lie very close to each other, since energy differences between different rotational and vibrational of molecular hydrogen (and every other molecule) states are minimal. When zooming in on the region where the band is located - as shown inFigure 2.8- the individual peaks of the ro-vibrational states become visible.
Figure 2.8: An emission spectrum of the Fulcher bands [31]. To make such a specific and accurate spectrum, a spectrometer with a very high spectral resolution is needed. The wavelength in nanometers is given on the horizontal scale.
AsEquation 2.8suggests, H2O is produced in the reduction process. This water enters the plasma when it is created, whereafter part of it will dissociate into hydroxyl radicals (OH*). As shown in Figure 2.9radicals has a clear spectral emission line at around 309 nm, although being wider than that of an atom. At around 285 nm, a smaller spectral emission line is located, which corresponds with another vibrational state of the hydroxyl radical. Because OH* is a molecule, it can have multiple ro-vibrational states. Therefore, as with hydrogen, the peak at 309 nm consists of a lot of small peaks as shown inFigure 2.10, which correspond with different rotational states.
Figure 2.9: An emission spectrum of hydroxyl radicals [32]. The blue line shows the measured spectrum, and the red line represents the spectrum which is obtained after the background correction has been applied.
Figure 2.10: An emission spectrum of the OH*(0-0) peak of Figure 2.9 [33]. It can be seen that such a peak actually consists of different rotational states. The general trend in this peak is depicted by the red line.
The spectrum inFigure 2.10contains multiple higher peaks, which can also be seen inFigure 2.9 as a kind of distortions in the main peak if a closer look is taken. In the zoomed spectrum with higher resolution, the first peak (as a result from a rotational transition, at about 306.5 nm) is high as compared to the spectrum measured below 306.5 nm. At the end of this spectrum, however, the intensity of the peaks decreases gradually. This sudden increase at 306.5 nm is due to the presence of forbidden quantum mechanical states of OH*. This peak shape is characteristic for hydroxyl radicals. Wavelengths longer than 319 nm are not measured inFigure 2.10. It is expected that these hydroxyl radical peaks will increase during the reduction of the iron oxide from the moment the plasma is discharged.
Experimental set-up
In this chapter, the most important components in the experimental set-up will be given and elucidated. Additionally, relevant calibration test results will be shown and a characterisation of the iron oxide powders (among which their composition) that are used in the experiments will be given.
3.1 Key components
This section will first describe the most relevant components that are used to create a hydrogen plasma and observing it. In addition, it will give an idea of the general overview of the experimental set-up. After, a more detailed description of the cavity (the container in which the plasma is created), the power supply and other measurement devices is given. Next, the working principles of an optical emission spectrometer are explained. Finally, a short note will be made on the reason why a thermogravimetric analyser is used in the experiments. The description of the experimental set-up will be useful especially regarding the first sub-question of this report (What other processes next to the reduction of iron oxide can change the intensity of hydroxyl emission peaks?).
3.1.1 Vacuum chamber and discharge
For the experiments with iron oxide inserted in a hydrogen plasma that are conducted in this report, the set-up inFigure 3.1is used.
Figure 3.1: The complete experimental set-up, including the devices that are used.
A hydrogen plasma is created in a cavity which is placed in the vacuum chamber. The cavity consist of a top and a bottom electrode of which the latter is grounded, the details are provided in Figure 3.2. The top electrode is connected to a RF power supply which is used to establish a discharge of a plasma. Several gases, like argon and hydrogen, can be connected to the set-up.
These gases flow from the flow meter at the input through the vacuum chamber with the cavity to the vacuum pump. A valve is situated just before the vacuum pump so that the pressure can be varied. Because this is a manual operation, this will result in a average relative uncertainty in the pressure of approximately 20%. Additionally, two manometers (one is for measuring pressures below 1 mbar, the other for pressures above 1 mbar) are connected to the vacuum chamber. Light which is emitted by the plasma in the cavity can be detected by a spectrometer placed on the outside.
Figure 3.2: The cavity in the vacuum chamber of the experimental set-up.
A more detailed overview of the cylindrical cavity is shown inFigure 3.2. The cavity is 70 mm in diameter and 40 mm in height. Inside of its bottom electrode a thermocouple can be placed, so the temperature in the cavity can be measured. This is important because specific isolating materials (peek in this case) are used to isolate the top electrode from the wall of the cavity and this peek cannot withstand high temperatures (> 150◦C). A wire mesh functions as a window in the cavity, passing the light emitted by the plasma, which can in turn be detected by a spectrometer. When iron oxide powder is placed on top of the bottom electrode, the power supply which is connected to the top electrode can be switched on. Then, when a proper gas flow is set, a plasma can be created, and the interactions with the iron oxide will start.
As explained the hydrogen plasma within the cavity is powered by an RF power source in the experiments. For this purpose, a signal generator on 13.56 MHz and a variable peak-to-peak voltage of 0 - 600 mVpp is used. The signal is sent to the amplifier, which enlarges the peak- to-peak voltage of the input signal. Next, the amplified signal enters a so-called matchbox. The matchbox enables maximum power transfer between its input and its output. These devices enable the plasma to reach a power of 24 W. Because the cavity is a confined volume in which the plasma is present, it makes more sense to calculate the power density. Accounting for the dimensions of the cavity, and the assumption that the plasma will occupy half of the space in the cavity, the power density inside the plasma can reach values of 0.32 W/cm3. As a comparison, Sabat et al.
managed to reduce iron oxide with a plasma power density of 4.5 W/cm3, which is 14 times as much. Moreover, a microwave plasma was used during their experiments, which usually results in a higher electron number density [34]. In the experiments conducted in this report an RF plasma will be used. Sabat et al. confirmed that for the plasma they worked with there is a strong positive correlation between power density and reduction ability of the hydrogen plasma [13, 35]. Also, a higher power density makes that the electron density ne in the plasma will increase, which is favourable for the reduction process, because there are more electrons to excite other (hydrogen) atoms or molecules, which is also stated in [36]. This makes it interesting to try higher power densities. However, there should be accounted for differences between an RF and a microwave plasma. Compared to in an RF plasma, electrons travel shorter distances before they switch direction in a microwave plasma due to the higher frequency. This suppresses the surface charging of the electrodes and the cavity, which can lead to different behaviour in RF plasmas.
3.1.2 Photon detection
For the detection of the emitted wavelengths and their respective intensities, an AvaSpec-2045TEC is used together with the suitable software (Avasoft 8.0). This type of spectrometer is equipped with an internal cooling element, which reduces the noise significantly. Moreover, this spectrometer can measure emission spectra in a broad range; between 200 and 1100 nm. This range will however be narrowed because the light has to propagate through the quartz glass to be measured in this experimental set-up (further explained in subsection 3.2.1). With the help of an optical fiber, this spectrometer is aimed at the wire mesh of the cavity, with the quartz glass still in between.
This way, the emitted light from the plasma can be analysed externally. A spectrometer contains a row of small detectors, which can measure the intensity of a specific frequency component.
These frequency components are generated by splitting up the incoming light. An AvaSpec-2048 spectrometer (of which the working components are shown inFigure 3.3) is used in the experiments.
Figure 3.3: Schematic overview of the inside of the AvaSpec-2045TEC spectrometer [37].
A mirror first collimates the incoming divergent beam of light and then split up in frequency components by the grating. The focusing mirror then aims the light at an array of tiny detectors.
This array of pixels detects the photons that come into the spectrometer. For each photon, this array gives a so-called count, making use of the photoelectric effect by charging small elements which this array can detect. Because some elements are charged up spontaneously (this is called the dark current Id, expressed in electrons per pixel per second), and there is a readout noise Nread, the number of false counts that are measured although no photon has reached the detector Nf is given by
Nf = Idt + Nread (3.1)
in which t is the integration time. This integration time is the measuring time, a specific time interval in which all incident photons are measured. The spectrometer has a function to average over certain integration times when desired. The dark current should always be taken into account when observing spectra.
3.1.3 Thermogravimetric analysis
Thermogravimetric analysis (TGA) is used to study physical phenomena such as adsorptions and desorptions of samples and phase transitions. A TGA system can measure the mass of a sample (which has a mass usually in the order of 10 mg) very accurately while regulating the temperature of the gas around the sample. A schematic view of a typical TGA system is given inFigure 3.4.
Figure 3.4: Schematic view of a typical TGA system. Courtesy: Creative Biostructure.
The gas in the chamber around the sample can be chosen, according to the preferences regarding the sample. In the experiments conducted in this report, a TGA analysis will be used to check if water is present in a sample. If the sample is dried for a while at a certain temperature, the mass of the sample can decrease. In that case, the decrease can indicate the evaporation of water. If the mass is equal to the mass of the sample before the analysis, the presence of water can probably be ruled out. This analysis will contribute to the answer on the second sub-question of this report;
on finding other causes for hydroxyl peaks.
3.2 Calibrations
This section contains the results of three different calibrations: the transmission of the quartz glass (which is situated between the plasma and the spectrometer), dark current of the spectrometer and background light. These calibrations are used in the Results and discussion to correct the measurements if necessary.
3.2.1 Transmission of the quartz glass
Light emitted by the excited species in the hydrogen plasma goes through the wire mesh and the quartz glass (seeFigure 3.2). This quartz glass can attenuate the incident light beam for different wavelengths. Therefore the glass is examined on the wavelengths it reflects and transmits, and to what extent. An Ar-Hg calibration lamp is used to investigate the transmission of the glass; the results are given inFigure 3.5.
Figure 3.5: The transmission of the light through quartz glass for different wavelengths.
Measurement conducted by both Philemon Koolen and Sam Tennebroek.
During every measurement of the intensity of the light, this transmission has to be taken into account. Also, the wavelength of the spectrometer is calibrated using an Ar-Hg calibration lamp as well. The known characteristic spectral emission lines are matched with the right wavelengths on the spectrometer to achieve this.
3.2.2 Dark current
The dark current has to be measured, so there can be accounted for these false counts in the measurements. If the detector is covered, no photons from the environment or a light source will reach it, and therefore this method is suitable to measure the dark current Id. The average number of total counts with a covered detector is measured as a function of the integration time of the spectrometer; the results are displayed inFigure 3.6.
Figure 3.6: The amount of false counts as a function of the integration time of the spectrometer.
This is the linear relation suggested byEquation 3.1. The data points are manually determined averages over the wavelength, which is why there may deviate from the fit. From the equation, it can be seen that the slope of the fit depicts the dark current Id and the offset depicts the read-out noise Nread. Their numerical values are given by Id= (1.8 ± 0.1) · 102 electrons/pixel/s and Nread= 39 ± 6 electrons.
3.2.3 Background light
Finally, the space in which the experiments take place will not be completely dark. There has to be accounted for background light: photons that do not originate from the excited species in the plasma, but the environment. This background light (including the dark current) is measured in Figure 3.7.
Figure 3.7: A characterisation of the background light as a function of wavelength at an integration time of 600 ms averaged 49 times.
It can be seen that the mean number of counts inFigure 3.7is approximately equal to 120. Because the intensity is measured in 600 ms, the amount of measured counts per second is equal to 200.
This is very close to the measured value of the dark current Id. Therefore, the background light has a minor influence on the measurements compared to the dark current.
3.3 Iron oxide powders
This section will give a characterisation of the iron powders which are used in the experiments and give an explanation why these powders are used.
In the conducted experiments, two types of iron oxide powders are used:
1. ’Mixed’ iron oxide: this iron oxide powder is obtained from the combustion of iron powder in a cyclonic burner, designed by Tim Spee [38]. This kind of burner can establish a self- propagating iron flame, with a temperature of about 800◦C. Because not all iron particles
have reacted completely, according toEquation 2.6, a mixture of different types of iron oxide are present in this powder. The XRD analysis of this powder shown in Figure 3.8supports this statement. This powder is chosen because it is an actual powder which could be used in the future to store energy in. This iron powder is the result from an actual combustion, which makes it a realistic example with regard to the future in which iron powder may serve as an energy carrier.
The particle diameters are not measured but are estimated to be approximately equal to the diameter of 30 µm of the iron powder which was used.
2. Magnetite: this powder can also be denoted by iron(II,III) oxide or Fe3O4. This powder is bought on Sigma Aldrich, has a purity of 95% and consists of particles with a diameter of approximately 5 µm [39]. This type of iron oxide is an intermediate type of iron oxide which can be a result of incomplete combustion of iron oxide, as shown in Equation 2.6.
This powder is chosen because its composition is significantly better defined compared to the mixed iron powder. Also, the particle size is still in the range of what would be realistic if iron powder will serve as an energy carrier.
These particles are significantly smaller than the powders (or pellets) used in other experiments.
Sabat et al. used iron oxide pellets of approximately 500 µm in size [13] and in other experiments pellets in the form of disks with a diameter of 40 mm and a height of 3 to 9 mm [35]. Smaller particles will have a larger specific area (when assuming the mass of the sample is the same) that will be in contact with the plasma and therefore the hydrogen radicals compared to larger particles. This will positively influence the reduction process. No research has been done on the effect of different particle sizes of iron oxide on their reduction in a hydrogen plasma, which is why it cannot be ruled out there are other differences between small and large particles. However, many experiments report a positive correlation between a smaller particle size and higher reaction rate [40,41].
Figure 3.8: An XRD measurement of the mixed iron oxide powder. It is clear that this powder consists of several types of iron oxide. Courtesy: Martijn Weijers.
Using this equipment the experiments as described in the next section can be conducted. Both the pressure and the flow of the hydrogen gas in the cavity is varied to investigate their influence on the intensity of the light emitted by the plasma. The measurements on the intensity of the emitted light of the hydrogen plasma when iron oxide is present are conducted using different gap widths (between the electrodes) and different voltages in order to create different power densities.
Furthermore, the intensity is always measured over a specified period so that the reduction process can be monitored accurately.
Results and discussion
In the previous part of this report, the theory describing plasma behaviour and optical emission and the experimental set-up including calibration measurements are discussed. This chapter begins with a characterisation of a hydrogen plasma without iron oxide in order to create a reference point for the measurement in which iron oxide is included. For this purpose, the results of the measured Paschen curve of the hydrogen plasma, a full spectrum and the dependency of the spectrum on the pressure are analysed and discussed. When it is clear what the reference point is, first several other causes for hydroxyl emission peaks to change (except for the reduction of iron oxide of course) are discussed in order to answer the first sub-question. Thereafter, using the analysis of the other causes, the results of the experiments in which iron oxide was included are discussed to answer the second sub-question of this report. In this final part, experiments with different power densities and different powders are conducted.
4.1 The Paschen curve
As discussed in section 2.2the breakdown voltage of a certain plasma is governed by Paschen’s law. In this report, this is done for a hydrogen plasma. The breakdown voltage is plotted against the product of the pressure and the distance between the two electrodes inFigure 4.1.
Figure 4.1: The experimentally measured Paschen curve for a hydrogen plasma. Meas- urement conducted by both Philemon Koolen and Sam Tennebroek.
It can be seen that when pd decreases (starting from the minimum) there is an abrupt change in breakdown voltage, whereas that when pd increases there is a slow rise in breakdown voltage. This is according to the behaviour of Paschen curves in general. Equation 2.2is used to fit the measure- ments (actually, an adapted version of this equation - in which pd is replaced by
pd+ln(γ) A
−ln(γ) - is used to avoid fitting problems with the negative part of the function). The numerical valueA of the experimentally determined minimum is VB = (1.5 ± 0.1) · 102 mVpp at (pd)min= 3.4 ± 0.1 mbar·cm.
Literature suggests that this minimum should occur at about 2.55 ± 0.05 Torr·cm, which is equal to 3.40 ± 0.07 mbar·cm [20], with a breakdown voltage of (3.2 ± 0.2) · 102 V. The measured pd matches the one from the literature very well. However, the breakdown voltages cannot be compared because the measured voltage is from the signal generator (hence expressed in mVpp), whereas the voltage from the literature is the one which, in this set-up, would be measured by the Octiv (a device which measures voltage and power). SeeFigure 3.1. Nevertheless, the Paschen curve will still provide useful information for next sections.
4.2 A hydrogen plasma without iron oxide
Before any experiments on the reduction of iron oxide in a hydrogen plasma can be done, the emission spectrum of the hydrogen plasma which is used in the experiments has to be analysed first. The emission spectrum on the left in Figure 4.2 is what was measured when a hydrogen plasma was made in the cavity. The spectrum is expected to look like the spectrum shown in Figure 2.7 and Figure 2.9 when zooming in on the region around 300 nm. However, if these spectra are compared, it can be seen that there are several additional emission peaks which cannot be explained by the emission of hydrogen. InFigure 4.3the emission spectrum of both atomic and molecular nitrogen is shown. As treated in subsection 2.3.2atoms account for the characteristic peaks (in the case for nitrogen located at high wavelengths), and molecules account for bands (in the case for nitrogen located at lower wavelengths). Molecular nitrogen can be in different stages of excited states, two of them are called the First positive series (or system) and the Second positive series (or system). Especially the Second positive series overlaps with the hydrogen spectrum in the measured range. A detailed view of the Second positive series is given in the spectrum on the right inFigure 4.3. The characteristic emission peaks of molecular nitrogen match with the observed peaks inFigure 4.2.
Figure 4.2: The measured spectrum of the hydrogen plasma (without iron oxide) before and after the helium leakage test and stopping the leak. The red rectangle denotes the Second positive series of molecular nitrogen. Measurement conducted by both Philemon Koolen and Sam Tennebroek.
Figure 4.3: Left: The spectrum of both atomic and molecular nitrogen [42]. Two signi- ficant bands in this wavelength range are the First and Second positive series. Right: A zoom-in of the Second positive series [43].
The presence of nitrogen molecules in the plasma could be evidence for the presence of leakage.
Therefore a helium leakage test was performed on the cavity, and several leakages were found and closed. After this, the emission spectrum of a hydrogen plasma was measured again, see the spectrum at the right inFigure 4.2. It can be seen that the amount of nitrogen in the plasma has decreased significantly, leaving a purer hydrogen plasma.
In the remainder of this chapter, an answer on the sub-questions will be sought, using the inform- ation gathered up till this point. First, the influence of the pressure of the hydrogen gas inside the cavity is investigated, because this will be used in the analysis of the subsequent measurements on iron oxide. Next, an overview of other the possible causes for changes in hydroxyl emission peaks is given. Finally, the measurement results of iron oxide in a hydrogen plasma will be discussed.
4.3 Pressure dependency
Since it is hard to set the pressure to a constant value due to the valve which has to be manu- ally operated (as described insubsection 3.1.1), and can only be done with approximately 20%, the dependency of the pressure inside the cavity on the spectrum and its intensity is measured.
Situations can occur in which the valve is set in a way that the pressure seems constant, varies slowly, without the experimenter noticing. If the pressure is too high, there will be more hydrogen molecules to excite. However, the electrons in the plasma will not be able to build up enough speed to excite hydrogen molecules, giving them a smaller chance to excite the hydrogen molecules.
On the other hand, if the pressure is low, electrons will build up more than enough speed, but due to the low amount of hydrogen molecules, they will not excite them frequently enough. This would lead to the hypothesis that one maximum will occur in the intensity. To find out what the influence of the pressure is on the intensity of the emitted light from the plasma, measurements of spectra of a hydrogen plasma without iron oxide are made at different pressures at a gap width of 23 ± 1 mm.
Figure 4.4: The emission spectrum of a hydrogen plasma in the region between 300 and 325 nm at different pressures. The pressure is given in mbar all with a relative uncertainty of 10%.
If the spectra inFigure 4.4are followed according to increasing pressure, it becomes clear that the overall intensity attains a maximum at p = 0.60 mbar, then decreases again, increases attains a second maximum at p = 1.65 mbar, and finally decreases again. If the intensity at the location of the highest point of the hydroxyl emission peak (310 nm) is plotted as a function of the pressure, the result is a graph like inFigure 4.5.
Figure 4.5: A plot of the intensity of the hydrogen plasma as function of pd at 310 nm.
Measurements at other wavelengths result in the same shape of the graph.
Two maxima appear in this figure, which is not in line with the hypothesis. No similar observa-
tions have been made before in literature. It could be that the the electrodes and/or the cavity become charged and therefore repel or attract electrons.
All in all, the pressure can influence the intensity substantially. Therefore, it is interesting to look at the ratio of the hydroxyl emission peak at 310 nm and the nitrogen emission peak at 315 nm, rather than the absolute value of the intensity of a peak, because that one is probably heavily influenced by pressure. This ratio should stay constant when no iron oxide is present and vary during the reduction process because nitrogen is not involved in this process, but hydroxyl radicals are. The ratios of the peaks for different values of pd are given in Table 4.1. In this table the ratio between the nitrogen and hydroxyl emission peaks (IN2/IOH ·) is given. Both the ratios with respect to (w.r.t.) zero and the baseline are calculated. The baseline is determined by taking the mean value of the ’flat’ parts of the spectrum at the left and the right of the spectrum. This is illustrated inFigure 4.4. Sharp peaks are peak intensities measured at one specific wavelength only (so one measurement), mean peaks are averaged with two measurements in the close neighborhood of the peak for more accurate results.
Table 4.1: The ratios of the nitrogen and hydroxyl emission peaks.
pd (mbar·cm) 0.41 0.92 1.4 1.6 1.9 2.6 3.8 5.6
ratio sharp peaks w.r.t. zero 1.4 1.3 1.3 1.2 1.3 1.2 1.2 1.2 ratio mean peaks w.r.t. zero 1.4 1.3 1.2 1.2 1.2 1.2 1.2 1.2 ratio sharp peaks w.r.t. baseline 2.2 2.4 2.6 2.7 2.9 3.3 3.7 3.7 ratio mean peaks w.r.t. baseline 2.1 2.1 2.2 2.4 2.5 2.6 2.7 3.0
Figure 4.6: The ratios of the nitrogen and hydroxyl emission peaks in a hydrogen plasma containing no iron powder as a function of pd.
The ratios of the nitrogen and hydroxyl peaks with respect to zero stay constant with an accuracy of approximately 14%. At low pressures, the ratio is higher, which can be possibly explained by the more significant differences in pressure between the environment (p = 1 atm) and the pressure inside the cavity (p = 0.18 mbar at pd = 0.41 mbar·cm). Tiny leaks in the cavity can suck air, and
therefore nitrogen inside, which will happen faster with more significant differences in pressure, so at low pressures inside the cavity. The increasing ratio for the peaks with respect to the baseline is difficult to explain. In the next sections, these values will be compared to other measured values in a hydrogen plasma with iron oxide, possibly providing useful insights. The pressure can account for overall intensity in- and decreases of the spectrum and for small differences in the ratio of the nitrogen and hydroxyl emission peaks with respect to zero. In conclusion, the ratio IN2/IOH ·
depends much less on p than IOH · does, which is an observation that will improve the accuracy of the measurement result in the experiments with iron oxide.
4.4 Sub-question 1: other possible causes for changes of hydroxyl emission peaks
This section will analyse if the used hydrogen flow, porosity of the iron oxide powder, the presence of water in the cavity or small leaks could influence hydroxyl emission peaks. Conclusions on the influence of these processes will be supported by measurement results. In this way, the first sub-question of this report can be answered adequately.
4.4.1 Hydrogen flow
The hydrogen flow which is set can influence the spectrum as well. A part of the spectrum is measured for different flows of hydrogen gas; the pressure is kept as constant as possible in this experiment. The result is give inFigure 4.7.
Figure 4.7: The measured spectra for different flows of hydrogen. The pressure is kept constant. The flow is depicted in sccm (standard cubic centimetres per minute, that is, at 1 atm and 273 K), with an uncertainty of 0.5 sccm for each flow. Left: Enlarged spectrum between 300 and 325 nm. Right: Wide spectrum.
Apart from the spectrum at a flow of 8 sccm (which is probably an incorrect measurement) the baselines of the spectra match very well. If a closer look is taken to the region containing the hydroxyl peak (around 310 nm), it can be seen that the nitrogen peak at 315 nm is strongly influenced by the flow of hydrogen. This may have to do with small leakages which are still present in the cavity through which nitrogen can leak to the inside. The pressure is kept constant.
When the flow of hydrogen is low, the flow of nitrogen leaking inside the cavity is relatively large compared to the flow of hydrogen. For high flows of hydrogen, the nitrogen leakage flow is relatively small.
In the experiments, the flow can be set digitally and will stay substantially more constant than the pressure (approximately 10% versus 100% relative fluctuations). Therefore, in the experiments, it is assumed that the flow of hydrogen is constant and that it will not change much in the measured spectra. In the experiments, a hydrogen flow of 10 sccm is used, unless indicated otherwise.
4.4.2 Porous iron oxide powder
Another process that could contribute to the creation of hydroxyl radicals is the presence of water in porous iron oxide powder. Porous iron powder is often obtained after the combustion of iron particles [44]. After combustion, and during storage the iron oxide powder comes in contact with the humid air in the environment. The water in the air could be absorbed by the porous iron oxide powder, which in turn can dissociate to hydroxyl radicals when present in the hydrogen plasma.
To determine if there is water present in the porous iron oxide particles, an experiment involving a TGA (as described insubsection 3.1.3) is conducted. A sample of iron oxide is placed in the small bucket which is continuously weighed and conserved in a nitrogen environment with a relative humidity of 0 at a temperature of 26.4 ± 0.1◦C for 4 hours. If water is present in the porous iron powder, this will slowly evaporate, and the total sample mass will decrease. The results of this experiment are given inFigure 4.8.
Figure 4.8: The evolution of the mass of the sample of iron oxide as a function of time.
The graph on the left displays the results of the magnetite, the graph on the right displays the results of the mixed iron powder.
It can be seen that the mass of the magnetite powder decreases by approximately 15 µg. Com- pared to the mass of the sample, which is around 11.7 mg, this is quite low. Using these masses and the molecular weight of water and magnetite, it can be calculated that this decrease in mass corresponds with 5 · 1017 water molecules. The sample contains 3 · 1019 molecules of magnetite.
Therefore, there are 60 times more magnetite molecules than water molecules. However, also low concentrations of water can change the plasma characteristics.
The measurements on the mixed powder have more remarkable results: its mass increases over time, which is hard to explain. This could be caused by the bucket in which the sample is con- tained oscillates. Thus, the possible contamination of the powder with water has to be considered when this powder is used.
Another issue that is possible is the presence of air (in particular nitrogen) in the porous iron oxide. If the spectrum on the right in Figure 4.2(in which the cavity only contains a hydrogen plasma) is compared withFigure 4.9(in which the cavity contains both the hydrogen plasma and iron oxide), it can be seen that the nitrogen emission peaks (as inFigure 4.3) are slightly higher in Figure 4.9, relative to the hydrogen emission peaks. This indicates that there probably is a small amount of nitrogen gas present in the porous iron oxide.
Figure 4.9: The full emission spectrum of a hydrogen plasma with the mixed iron powder.
4.4.3 Other causes
Apart from the fact that the pressure, hydrogen flow and porous iron oxide particles can cause changes in hydroxyl emission peaks, there are some other causes to which a change can be assigned:
1. The vacuum chamber always contains a little bit of water, because it is opened regularly to take the cavity out (to observe or replace the iron powder for example) and humid air can enter the vacuum chamber. It is also aerated in this way. This water can then dissociate to hydroxyl radicals after the discharge of the hydrogen plasma, and therefore cause an increase in hydroxyl emission peaks, independent of the cavity or the sample it contains. This can be prevented by heating the vacuum chamber to over 100◦C, so the water evaporates and leaves the chamber. However, this is a task which costs much effort to do and will not evaporate all of the water. This measurement is not taken before conducting the measurements in this report.
2. Even after the helium leakage test, there could be (very) small leakages present in the vacuum chamber. Air could dissipate through these small holes. Thus, nitrogen, oxygen and possibly even a small amount of water can enter the vacuum chamber. Nitrogen in the chamber could cause the nitrogen emission peaks to rise and could disturb the ratios between a nitrogen and a hydroxyl emission peak in the next sections. Oxygen could recombine with the hydrogen radicals in the plasma, and could potentially form hydroxyl radicals, which would influence the height of the hydroxyl emission peak, which would rise as well.
In conclusion, the hydrogren flow used in the experiments stays constant, and is assumed not to influence the hydroxyl emission peaks. The number of water molecules in the porous iron oxide is 60 times less than the number of iron oxide molecules. However, this small amount could still influence the plasma characteristics and therefore its emission peaks. Also, the presence of water in the cavity could not be ruled out. Small leaks may also still be present, although before the experiment several leaks have been stopped.
4.5 Sub-question 2: The time evolution of the hydroxyl emission peaks
This section focusses on answering the second sub-question in this report. Three experiments and measurements are conducted in order to achieve this. First, a measurement with a relatively low power density on the time-evolution of they hydroxyl emission peaks with the mixed iron oxide is done. Next, the time-evolution is analysed another two times using the ratio between a fairly stable nitrogen emission peak and a hydroxyl emission peak, which is just very little dependent on the pressure. One of these measurements is conducted using a higher power density and the other is conducted using pure magnetite powder and an even higher powder density. Finally, using tables and plots of the ratios conclusions about the reduction process are drawn.
4.5.1 Mixed iron oxide in a hydrogen plasma
After the leak had been stopped, the experimental set-up is expected to be suitable to observe the reduction of iron oxide powder. The mixed iron oxide powder (as explained insection 3.3) is placed in the cavity with hydrogen gas with a pressure of 0.40 mbar. After discharge, the power density inside the cavity is 0.32 W/cm3. An emission spectrum is saved every five minutes from the moment of discharge, the results of a part of the spectrum between 300 and 500 nm are shown inFigure 4.10.
Figure 4.10: A part of the spectrum of the emission spectrum of a hydrogen plasma con- taining the mixed iron oxide. Every spectrum is made on a specific time after discharge, as indicated in the figure. The time depicted is in minutes. Measurement conducted by both Philemon Koolen and Sam Tennebroek.
The first thing that stands out in this figure is that the spectra from t = 5 min and later all match each other approximately in a large part of the spectrum. The spectrum from just after
the moment of discharge at t = 0 min, however, lies significantly lower. As has been made clear insection 4.3this has to do with the pressure, which can lower the total spectrum. Probably the pressure was still being tuned manually using the valve, and constant pressure was established only after a few minutes (but less than 5 minutes). Because the rest of the spectra approximately match, it is assumed that the pressure was approximately constant during the measurement of those spectra.
Additionally, the part of the spectrum at the very left of this spectrum, between 300 and 325 nm is remarkable, because there seems to be a variation there. The spectra measured after 5 minutes do not match each other very well in that region. Figure 4.11shows a close-up of the variant part of the spectrum.
Figure 4.11: Enlarged figure of the part of the varying spectra. The time depicted is in minutes. The red line refers to the shape of the hydroxyl peak as suggested by literature as can be seen in Figure 2.7. A hydrogen flow of 10 sccm is used. The mass of the sample is 3 g and the energy density is 0.32 W/cm3. Measurement conducted by both Philemon Koolen and Sam Tennebroek.
According toFigure 2.10the OH* (0-0) peak, which is also the most intense one, is visible between 306 and 319 nm approximately and lies in the same region as the gradually increasing spectra.
Therefore, the variation indicates a change in the hydroxyl emission peaks in the spectra. The characteristic shape that is indicated in Figure 2.10 is also depicted in Figure 4.11. Moreover, these peaks increase in intensity as time passes, which indicates that more hydroxyl radicals are formed during the experiment. This could be caused by the presence of iron oxide in the cavity.
It can also be seen that at first, the hydroxyl emission peak increases, although after t = 100 min it reaches a steady state. In the experiment, a flow of hydrogen gas is present in the cavity, which can influence the reduction behaviour of iron oxide. A possible explanation for this stagnation is that directly after discharge the hydrogen gas tries to reduce the iron oxide. However, it takes
