Master Thesis

(1)

Master Thesis

Improving Stability and Efficiency of

Earth-abundant Electrocatalysts for

Water Oxidation

Clemens Wunder

Master thesis, 60 hp Examiner:

(2)

(3)

Abstract

(4)

(5)

List of abbreviations

A Ampere, Unit for the current

CP Chrono potentiometry (Δt > 1 ms), Measures the stability CV Cyclic voltammogram, Measures the IV-Curve and the Activity

EDS Energy-dispersive X-ray spectroscopy, Elemental analysis, or chemical characterization

EIS Electrochemical impedance spectroscopy, Investigates the surface FRA Frequency Response Analyzer, Measures the internal resistance

(impedance)

HER Hydrogen Evolution Reaction, Production of H2

IR Internal Resistance, Resistance between working electrode and counter electrode

NRR Nitrogen Reduction Reaction, Production of NH3 OER Oxygen Evolution Reaction, Production of O2

RHE Reversible Hydrogen Electrode, Relates the potential to a standard hydrogen electrode

SEM Scanning Electron Microscopy, High quality picture of the surface V Volt, Unit for the potential

XRD X-ray Diffraction (Crystallography), Determines the atomic structure of a crystal

XPS X-ray photoelectron spectroscopy, Investigates the first surface layers

Author contribution

(6)

(7)

1. Introduction

Efficient electrochemical water oxidation is considered to be a key technology for allowing the storage of renewable electricity from sunlight or wind in fuels.[1][2]In this electrolytic process water is split into hydrogen and oxygen (Reaction 1).

𝐻2𝑂(𝑙) → 𝐻2(𝑔) + 1 2⁄ 𝑂2(𝑔) (1) This reaction can be separated into the cathodic half-cell reaction (hydrogen evolution reaction, HER) and the anodic half-cell reaction (oxygen evolution reaction, OER). Both half-cell reactions are shown for acidic conditions in Figure 1. The protons can travel through the electrolyte and the membrane separator, while the electrons that are generated on the anode via the OER travel to the cathode through an external circuit, where they are employed to reduce protons to molecular hydrogen, H2. An external potential is applied to drive the overall reaction since it is not spontaneous. The applied potential has to be at least 1.23 V in order to split the O-H bond. [3]

Figure 1: Schematic depiction of the electrochemical water splitting reaction under acidic conditions. The left part visualizes the oxygen evolution reaction leading to the production of molecular oxygen and protons, while the right part shows the reduction of protons to molecular hydrogen. The reaction is driven by an external potential.[3] The two compartments are separated by a proton exchange membrane to avoid mixing of the gases.

The best electrocatalysts for the acidic HER is platinum and the best catalyst for the acidic OER are IrO2 or RuO2. As these catalysts are precious and expansive, new materials for OER are currently under investigation. [4]

The investigated catalyst must fulfill two requirements. First, it should have a high activity. This means that it should split the water bond of 1.23 V with a minimal need of excessive energy (overpotential). One way to investigate the activity of a catalyst is to connect the cell to a potentiostat and measure a cyclic voltammogram (CV).[5] The applied potential vs Ag/AgCl is converted to the potential vs the reversible hydrogen electrode (RHE) according to Equation 2:

(10)

where I is the current, Rs the internal resistance and 0.222 V the standard potential of Ag/AgCl vs the standard hydrogen electrode (SHE). The factor 0.059 is calculated from R*T/F where R is the gas constant, T the absolute temperature (T=298K) and F the Faraday constant. The reference to the RHE is made because the hydrogen electrode is set as a zero-potential electrode in the literature. All electrode potentials are noted as the potential with reference to RHE, in order to allow cross-comparisons.

The energy loss (caused by various factors including the sluggish electrocatalytic performance, electrolyte resistance, electrical resistance in electrodes etc.) associated with practical electrocatalytic systems often results in the need to apply potentials greater than the water binding strength of 1.23V for water splitting reaction to occur. This excessive applied potential is denoted as overpotential needed to compensate the energy loss and will not result in the splitting of water, but e.g. lead to the transformation of part of the electric energy into thermal energy.[6] In this thesis, the overpotential is quoted with respect to the current density of 10 mA/cm2_(horizontal line, Figure 2) and should be as low as possible for a high activity. This specific current density is commonly regarded as a benchmarking figure of merit for solar-fuel device commercialization, because it corresponds to a solar-to-H2 conversion efficiency of roughly 10% when the electrochemical cell is coupled to a solar cell under 1 sun illumination.[7]

Figure 2: Change of current density with potential. The dashed red line marks the current density of 10 mA/cm2 used for benchmarking the catalysts.

The second requirement which must be fulfilled is that the catalyst should have a high stability. The stability describes how long the catalyst will remain stable within the given electrochemical surroundings (regarding the electrolyte pH and applied potential). The current density of 10 mA/cm2_{will be consistently upheld for hours} while adjusting the potential. An ideal sample would require the same potential for hours/days/years, but an actual catalyst will degrade or decay with time. This will in turn result in a higher resistance, therefore necessitating an even higher potential than before as expressed by Equation 3.

𝑈 = 𝑅 ∗ 𝐼 (3)

(11)

Figure 3: A schematic stability curve (CP) as a function of time vs the potential. While the sample degrades progressively over time, the point of instability is defined for benchmarking purposes as 0.5 V increase over the initial potential required to sustain water oxidation at a current density of 10 mA/cm2_{(red point).}

The impedances of the samples were investigated by electrochemical impedance spectroscopy (EIS). The resulting Nyquist plots can be used to determine the electrode resistance (Rs, uncompensated resistance between the working and the reference electrode), the catalyst-electrolyte charge transfer resistance Rct and the diffusion layer resistance R2 (Figure 4)[8]. The electrochemical equivalent circuit model used to fit the Nyquist plot was [Rs(RctQct)(R2Q2)], see insert in Figure 4.

Figure 4: The Nyquist plot shows the Cartesian coordinates, the real part of the impedance measurements on the X axis, and the imaginary part on the Y axis and is used to investigate the electrode resistance (Rs), the charge

transfer resistance (Rct) and the diffuse layer resistance (R2). Rct is specific for the catalyst, while Rs display the

electrolyte resistance and the geometric setup of the cell, and R2 shows the electrical conductivity of the catalyst

and its conductivity to the substrate. The parameters are extracted by fits (black line) to the data points (blue dots) using the equivalent circuit diagram shown in the inset. The fit consists of two semi-circles, where the respective width signifies the electrode resistance of the catalyst (Rct; orange) and of the diffuse layer resistance (R2; green).

The displacement from the Y-axis is due to the electrolyte resistance, marked with a blue double-sided arrow.

(12)

Aim of the Master Thesis

The aim of this thesis was to optimize the synthesis, composition and operation conditions of a catalysts for electrochemical water oxidation. The catalysts, combinations of Fe and Co, were evaluated in terms of activity and stability, and characterized regarding their composition and minerology. The performance of the best catalyst was compared to data reported in the literature.

2. Popular Scientific Summary including Social and Ethical

Aspects

2.1 Popular Scientific Summary

The idea of collecting energy from the sun and using it for our daily needs has been a dream of mankind for some time, but how can this be accomplished? Solar cells as well as wind and water turbines are already getting more and more common. However, for eventually replacing all fossil fuels one also needs to be able to store the renewable electricity they produce, because mankind has a need for an uninterrupted supply with energy and cannot wait for the sun to shine to be able to use electrically powered devises. Additionally, solar cells are most efficient near the equator due to the direct sun irradiation. Thus it would be advantages if solar energy could be stored and transported to where it is needed.[9]

(13)

2.2 Social and Ethical Aspects

The research done for this thesis was use-inspired and application-oriented research (Pasteur). The best possible outcome will contribute to improve our living conditions [15]. As a long term project, the OER will provide possibilities that allow the production of hydrogen and oxygen from water and solar energy (received through a solar cell), which can be used to power cars or any other electronic device, thus reducing the carbon emission on the Earth. Not only does this goal justify the research for this specific topic, but some adjustments were made in order to minimize the environmental burden. First, the experiments were all carried out using only small amounts of material. This attempt was used so that excessive waste formation was prevented. The not-avoidable waste was disposed following the rules and regulations. The risk for the researchers and other persons was kept as low as possible by working under a fume hood. One potential downside, however, is the general application of cobalt, the mining of which currently appears in parts of the world to be heavily reliant on child labor.[16][17]

3. Experimental

3.1 The Synthesis of the Catalysts

The catalysts were synthesized using spray-pyrolysis deposition on a hotplate. The precursors used were aqueous solutions of iron (III) chloride hexahydrate and/or cobalt (II) chloride hexahydrate. The total metal concentration was kept at 0.05 M (see Table 2 for precursor details). The spray nozzle was placed at 30 cm distance and tilted 45° from a titanium foil (substrate), which was heated on a hot plate at 250 °C or 450 °C. Compressed air at 0.5 bar was used as a carrier gas to spray the precursor solution onto the Ti substrate.

The sprayed samples were then placed in a sealed round-bottom flask and flushed with argon for 30 min before annealing for 30 min at 450°C under a static Ar environment. For P doping, the samples were annealed in the presence of sodium hypophosphite powder. A copper wire was glued to the back of the annealed sample using graphite conductive adhesive to establish an electrical contact for electrochemical measurement. 3.2 The Electrochemical Measurements

The electrochemical cell consisted of the catalyst, platinum coil and Ag/AgCl/1 M KCl as the working electrode, counter electrode and reference electrode, respectively. These electrodes were placed in contact with an aqueous electrolyte consisting of sulfuric acid or phosphate buffer solution, with pHs (pH 0.3 – pH 6) adjusted with sodium hydroxide or phosphoric acid.

(14)

The electrocatalytic current at the working electrode was measured at positively applied potential. In the used configuration (Figure 5), the OER occurs at the catalyst surface while the HER is catalyzed at the platinum surface. The geometric area of the catalyst exposed to the electrolyte was 0.2 cm2_.

Figure 5: The oxygen evolution reaction and the hydrogen evolution reaction in an acidic, wet chemical cell using platinum as the counter electrode and silver/silver chloride as the reference electrode.

3.3 Characterizations of the Catalysts

The samples were investigated by SEM, EDS, XPS and XRD. The mineralogical properties of the samples were examined using powder X-ray diffractometer (XRD) (PANalytical Xpert3; Cu Kα radiation; 45kV; 40mA).

The morphology was investigated by Scanning electron microscopy (SEM). The instrument was a Carl Zeiss Merlin field-emission scanning electron microscope FESEM (operating at 5 kV 120 pA), while the SEM images were taken with SmartSEM software.

The elemental mapping was carried out by energy-dispersive X-ray spectroscopy (EDS) on an Oxford Instruments energy dispersive X-ray spectrometer EDS (operating at 15 kV, 300 pA, dwell time for area analysis 30s, and map analysis 100us). The EDS data were taken with Aztec software. The samples were mounted onto an aluminum stub with carbon adhesive tape.

(15)

4. Results

4.1 The best Iron and Cobalt Combination for OER

The activities of different cobalt and iron compositions were investigated by measuring CV curves showing the generated current density plotted against the applied potential. The standard electrical potential for water oxidation is 1.23 V vs. RHE, but all known catalysts require a higher applied potential (i.e. overpotential) to overcome energy losses connected to, for example, catalytic charge transfer resistance and the charge transport across the catalyst-current collector interface.[3] Therefore, the best performing sample regarding the activity is the one closest to 1.23 V vs. RHE at bench-marking current density of 10 mA/cm2_.

Figure 6a shows the best polarization curves of the samples of two measurements prepared at an annealing temperature of 250°C. These samples will be denoted as F[y]C[10-y]-250, where F, C and y represent Fe, Co and the relative Fe content in 20% steps, respectively, while 250 specifies the annealing temperature. Both single-metal oxides of 100% FeOX (F10-250) (see Table 2 for synthesis conditions) and 100% CoOX (C10-250) produced similar levels of current densities within the applied potential region, which indicates a similar level of catalytic activity. Bimetallic oxides F[y]C[10-y]-250, on the other hand, generated higher current densities at given applied potentials, thus showing an improved catalytic activity as compared to F10-250 and C10-250 (the spraying temperature is further explained in the discussion section). Each catalyst was measured two times, while each measurement contains six CV curves. (3 from low potential to high potential and 3 the other way). All CV curves regarding the composition are shown in appendix G: Figure 36-Figure 38. The CV curves shifted to higher potentials over the 6 scans (Example of C4F6-250 in Figure 36) as the catalysts reached equilibrium with their electrolyte surrounding. Thus, the last scan was used for characterizing the activity, because it represents best the behavior of the catalyst at equilibrium.

(16)

Figure 6: (a) Best CV scans of different iron-cobalt catalysts at pH 0.3 (0.5 M H2SO4). The CV’s are displaying the

current density in mA/cm2_{as a function of the potential vs RHE, only the sixed scans (from low to high potential)}

for each sample are shown (all data are given in Appendix G (Figures 38-40). The scan rate was 10 mV/s. All catalysts were prepared and measured 2 times. (b) Average overpotentials with error bars derived from the two independently prepared (different days) catalyst samples.

The Tafel slope can be used to investigate the intrinsic catalytic activity of catalysts. The overpotential is plotted versus the current density J and shows the catalytic response with overpotential. The plot of log J vs η has a linear range (Figure 7), which

follows the equation:

log 𝐽 = 𝑎 +1 𝑏𝜂

The constant a displays the exchange current density, while the constant b displays the Tafel slope and can be used to interpret the rates of the mechanism for the whole reaction. A lower slope can be interpreted as a faster mechanism and a better charge transport. A good catalyst only needs a small increase of the overpotential in order to increase the catalytic current by a factor of 10. [7][18] The Tafel slope of C10-250 and F10-250 are 112 mV/dec and 115 mV/dec, while the slope of C4F6-250 is only 72 mV/dec.

Figure 7: The Tafel plots as a function of overpotential per current density of F10-250 (purple), C10-250 (red) and C4F6-250 (blue) in 0.5 M H2SO4 are shown. The measured data (dots) were fitted with a linear function (Tafel

slope) that is displayed in the corresponding color.

(17)

for the OER. The Rct is 18 Ω for C4F6-250, 20 Ω for C10-250 and 71 Ω for F10-250. Furthermore, F10-250 has another consecutive semicircle (R2) that is due to a poor charge transport (resistance of 40 Ω) in the bulk of the film.[8] C4F6-250 has the lowest resistance which agreed to the result from the Tafel plot. The Tafel slope and the Nyquist plot confirm that the combination of Fe and Co does produce an improved catalyst in comparison to both metals alone.

Figure 8: The Nyquist plots (Z´(Ω) vs -Z´´(Ω)) of C10-250 (blue), F10-250 (red) and C4F6-250 (orange) at 1.7 V at pH 2 are shown. The points are the measured data while the lines in the same color show the fit. The Rct is 18 Ω for C4F6-250, 20 Ω for C10-250 and 71 Ω for F10-250. F10-250 has a second semi-circle of 40 Ω.

4.2 The Mineralogical Properties

The mineralogical compositions of F10-250, C2F8-250, C4F6-250, C6F4-250, C8F2-250 and C10-C8F2-250 were investigated by XRD (Figure 9) with freshly prepared and washed samples. All samples show characteristic peaks of the titanium substrate around 38°, 40° and 53°. The first two samples of 0% cobalt (F10-250) and 20% cobalt (C2F8-250) show the characteristic peaks of the hematite structure (24°, 33°, 35°, 41°, 50°, 53°, 63° and 64°)[19], while the samples of 40-80% cobalt (C4F6-250, C6F4-250 and C8F2-250) show the characteristic peaks of the maghemite structure (30°, 35°, 44°, 57° and 63°).[20] The 100% cobalt sample (C10-250) shows the peaks from Co(II) oxide.[21] No peaks of cobalt-compounds were found for samples other than the one with 100% cobalt, thus suggesting that Co is either present as an amorphous Co(II) phase or is included in the vacancies of the Fe(III) oxide crystal structure as a substitutional aliovalent Co(II) doping.[22]

(18)

peaks for maghemite [24] were detected and only small deviations were seen due to the Co doping.

Figure 9: The XRD spectra of F10-250, C2F8-250, C4F6-250, C6F4-250, C8F2-250 and C10-250 are shown and the peaks are matched with hematite (H), maghemite (M), cobalt oxide (C) and the blank titanium substrate (*).

4.3 Elemental Distribution

The XRD suggested that C4F6-250 builds a maghemite Fe3+_{oxide crystal structure} with a Co substitutional aliovalent Co(II) doping. A SEM/EDS scan (Figure 10) was used to investigate the elemental distribution of C4F6-250 to check if a second amorphous Co-oxide region that was not involved in the doping could be present. The freshly sprayed samples F10-250, C2F8-250, C4F6-250, C6F4-250, C8F2-250 and C10-250 were investigated by EDS/SEM (Appendix B, Figure 21-Figure 26) after rinsing them with water. The Fe ion concentration was equally distributed over most of the substrate, but Figure 10a shows two spots that contain nearly no Fe. At the same time, these spots had the highest Co concentration (Figure 10b). The O concentration is mainly homogenous all over the substrate, but also visibly reduced in the Co-rich areas (Figure 10c). The same trend can be seen for C2F8-250 and C6F4-250 (shown in appendix C: Figure 27). The SEM investigation (Figure 10d) showed that the “spots” were three dimensional on top of the substrate in form of drops. The XPS analyzes (Appendix M: Table 5) of a second set of sprayed, but unwashed samples confirmed the Co-Cl on top of the surface (atomic concentration (AC) of 14%). The difference between these two is only that the Co-Fe ratio was nearly 1:1, which is closer to the sprayed ratio of 40 to 60%. This can be used to verify that the washing step removes most, but not all Co chloride from the surface.

(19)

concentrated solution contains different Co complexes like [Co(H2O)6]2+_, [CoCl(H2O)5]+_{, [CoCl2(H2O)4], [CoCl3(H2O)]}-_{and [CoCl4]}2-_{.[28] The pH of the} solution of the Fe and/or Co chloride salt in water was not adjusted and resulted in an acidic pH between 5.6 (100% Co) and 1.7 (100% Fe) (Appendix S: Table 6), since the positive charged Fe3+_{can bind OH}-_{ions. These ions are generated from water, which} results in the release of protons into the solution and the resulting change in pH. The Fe and Co chloride precursor mixture was sprayed onto the heated Ti surface to form the oxide film. As discussed, due to dehydration of the water drops during the deposition process we do not know which exact species reaches the Ti-surface. For the further discussion we thus use the known properties of the starting salts as approximation. FeCl3 ·6H2O begins the oxidation process around 250 °C, while CoCl2·6H2O begins the oxidation process above 400 °C.[29][30] The required high temperature for CoCl2·6H2O implies that the Co precursor tends to stay as the chloride form when the precursor aerosol dehydrates on the surface of the substrate, while the Fe complex oxidizes. The thermal treatment at 250 °C was done before the measurement in the acidic electrolyte or the measurement by SEM/EDS, so that there may be some Co chloride which did not form oxides and thus is easily rinsed off thereafter with water. This is in line with the result that the atomic ratio of 40% Co and 60% Fe for this sample was not achieved but only 20% Co were measured during the elemental analysis (SEM/EDS) of the whole sample (Appendix V: Figure 55). The samples of the XPS scan were not washed before the measurement and contained the sprayed ratio (Table 5), revealing that the washing procedure removed Co chloride from the surface. The same trend could be seen for 60 and 80% Co samples.

Figure 10: The C4F6-250 sample was investigated by EDS/SEM and the mapping of the a) detected Fe atoms, b) detected Co atoms and c) detected O atoms are shown. d) The SEM image of the C4F6-250 is shown.

4.4 The Influence of the Electrolyte pH

(20)

from 0.3 to 6. The pH dependencies of C2F8-250 and C6F4-250 (20% and 60% cobalt samples) are shown in Figure 27 (Appendix C), which confirms the trend that the overpotential is increasing with pH. In order to investigate this behavior, a more detailed investigation of the surface of the catalyst in the electrolyte was required. The layer between the bulk aquatic electrolyte and the solid Fe phase can be assumed to be a double layer with O or OH groups on the surface. The characteristics of this layer can be used to describe the transfer of ions (charge carriers, e.g. protons) between the electrolyte and the surface. The characteristics of this layer is strongly pH dependent [31] and can be displayed by looking at the Nyquist plot of C4F6-250 (Figure 11c) and the resulting uncompensated resistance (Rs) (Figure 11d). The uncompensated resistance displays the resistance between the working electrode and the reference electrode (the resistance of the electrolyte). A lower pH implies an increase in charge carries, which increases the layer capacitances while lowering the resistance.

Each resistance was measured in two attempts with 4 curves each. The uncompensated resistance increased with pH from 15 Ω at pH 0.3 up to 60 Ω at pH 6. This increase in the electrochemical/electrolyte resistance (Figure 11d) leads to a decrease in activity (Figure 11b) since a higher resistance in the electrolyte means that a higher potential is required to achieve a current density of 10 mA/cm2_._{This can be explained by the} relation U(Potential)=R(Resistance)*I(Current), which shows that the potential must increase if the resistance increases (as long as the current remains constant).

Figure 11: (a) CV curves (6th_{scan) and (b) overpotentials of C4F6-250 (2 measurements each) at different pHs (0.3}

(21)

4.5 The Catalytic Stability

The best activity was achieved using C4F6-250 (similar to C6F4-250 and C8F2-250) at pH 0.3 with 0.5 M H2SO4, but a good catalyst must as well offer a high stability. The stability was investigated by measuring a chrono potentiometry (CP) curve of C4F6-250 at different pH values between 0.3 and 6 (Figure 12a). This measurement was only performed once due to the long measurement times. The stability at pH 0.3 was 2 h 50 min increasing toward 5 h 10 min at pH 1, reaching a peak at pH 2 with an overall stability of 14 h and 30 min. The stability of 5 h 30 min at pH 3 was decreasing slightly towards pH 6 (2 h 30 min). Before we look at the reason why pH 2 was the best condition for the stability of C4F6-250, the highest stability of C4F6-250 at pH 2 was compared with the stability of C10-250 and F10-250 at pH 2 (Figure 12b). The stability of C10-250 was only 3 h 30 min, while the stability of F10-250 (15 h 50 min) and C4F6-250 (14 h 30 min) were on a similar, but slightly higher level. The difference between C4F6-250/F10-250 and C10-250 can be explained by the varying Co content since the stability of F10-250 confirms that a higher Fe content is correlating to a higher stability. This can be used to explain why higher Co contents resulted in a lower stability, i.e. in a reduced time where 10 mA/cm2_{can be achieved. The activity of} C4F6-250, C10-250 and F10-250 at pH 2 are shown in Appendix N: Figure 48c and the overpotential is shown in Appendix N: Figure 48d. The overpotential of C4F6-250 was the lowest one with 670 mV, while F10-250 and C10-250 (750-780 eV) were less active, following the same trend as at pH 0.3 (Figure 6).

Figure 12: The stability curves of a) C4F6-250 at pH 0.3 to 6 and b) of C4F6-250, C10-250 and F10-250 at pH 2 measured in H2SO4 and NaOH.

(22)

stability at first, but it reduces the stability with increasing pH. This behavior can be explained by looking at the phase transition at a higher oxidative potential (red line). The amount of excess oxidative potential (the overpotential) for the switch to the unstable FeO42- _{phase decreases with increasing pH, thus reducing the overall stability.} This behavior is in line with the observation, which shows a generally decreasing stability and increasing solubility associated with increasing pH and applied potential. It can be concluded that Co doping improves the activity, but is ineffective in improving the stability.[32][33][34]

Figure 13: The Pourbaix diagram of (a) cobalt and (b) Fe showing the stable phases in dependence of the applied E(V) and the pH. It is noted that the Pourbaix diagram of Fe does not distinguish between the mineral forms of Fe2O3, hematite and maghemite, and thus is a guide only. The diagrams were constructed for a low metal concentration, i.e., 10 nM, which is chosen owing to the low catalyst loading and a likely gradual dissolution of the catalyst.[35] They are used as an indicator of equilibrium of the catalyst in aqueous system. Diagrams obtained from references[32][33][34].

4.6 Changing the Spraying Temperature and the Electrolyte

(23)

Figure 14: (a) Activity, (b) overpotential, (c) CP curve and (d) stability of C4F6-450 at pH 2, 4 and 6 adjusted with 0.5 M NaOH in 1M monosodium /disodium phosphate.

(24)

Figure 15: The stability of C4F6-450 under fluctuating potential displayed as different CV curves (1.2 to 2.4 V uncompensated potential) at pH 6 in 3 h steps over a time period of 24 h.

4.7 The Surface Chemical Properties

(25)

4.8 Introducing/Replacing Metal Oxide with Metal Phosphate

The synthetic conditions for C4F6-450 were further modified by changing the Fe/Co oxides to Fe/Co phosphate with the aim to further improve the activity and/or stability in the phosphoric electrolyte[38][39][40]. The oxide samples were transformed into phosphate samples via phosphidation using PH3 gas, which was produced by heating up sodium hypophosphite to 450 °C while annealing. The impact of different amounts of sodium hypophosphite on the activity of C4F6-450 were investigated. Therefore, the samples are denoted as C4F6-450/XX to indicate the amount of added sodium hypophosphite to the flask before annealing, ensuring that C4F6-450/03 contains 40% Co, 60% Fe and 0.3 g of sodium hypophosphite. The CV curves of C4F6-450 doped with different amounts of hypophosphite are shown in Appendix P: Figure 50. The 6th scans are shown in Figure 16a and the calculated overpotential is shown in Figure 16b. The sample C4F6-450/03 had the lowest overpotential of the hypophosphite samples with 600 mV, while both, the sample with more hypophosphite C4F6-450/04 (overpotential of 620 mV) and the sample with less hypophosphite (overpotential of 690 mV), resulted in a higher overpotential (no repeat measurements). However, all these samples showed an equal or if at all slightly worse result as compared to C4F6-450, which was not treated with hypophosphite. C4F6-450 had an average overpotential of 590 mV with the better measurement (pH 2).

Figure 16: (a) CV scans and (b) overpotential of C4F6-450, C4F6-450/02, C4F6-450/03 and C4F6-450/04. The spraying was performed at 450 °C and the electrolyte pH during the measurement was adjusted to pH 2.0 using 0.5 M H3PO4 and 0.5 M NaOH. The differently treated phosphate samples were only measured once.

4.9 The Stability of the Metal Phosphates

(26)

Figure 17: (a) Stability as a function of the amount of hypophosphite in the flask before annealing (pH 2, 450 °C). (b) The Tafel plot of C4F6-450 (110 mV/dec) and C4F6-450/03 (127 mV/dec) at pH 2.

4.10 The Chemical Composition of the Phosphate Samples

The chemical composition of C4F6-450/02, C4F6-450/03 and C4F6-450/04 were investigated by XRD (Figure 18). The titanium substrate shows two specific peaks around 38° and 40°, which were visible on C4F6-450/02 and C4F6-450/04, but not on C4F6-450 and 4F6-450/03. The absence or different intensities of these peaks find their explanation in the different thicknesses of the catalyst layers, which are a consequence of incomplete control of the synthetic conditions. The dominant phase for all samples was maghemite, which was visible at 30°, 36° and 43°. The samples C4F6-450/03 showed additional FeP peaks at 33°, 37°, 46°, 47° and 49°.[41] The sample treated with 0.3 g sodium hypophosphite (C4F6-450/03, light blue line, Figure 18) resulted in the highest activity (Figure 16) and stability (Figure 17) and is the only one displaying these FeP peaks. The SEM investigation of 450/02, 450/03 and C4F6-450/04 (Appendix L: Figure 45Figure 47) showed that Co, Fe and P are distributed homogeneously over the surface, but only C4F6-450/03 showed the FeP peaks in XRD. The SEM scan (Appendix L: Table 4) showed that C4F6-450/03 contains a higher atomic ratio of 23.8% P and 56.5% O compared to C4F6-450/02 and C4F6-450/04. It can be asssumed that only C4F6-450/03 contained phosphate on the surface while C4F6-450/02 and C4F6-450/04 had amorphous phosphate on the surface.

(27)

The XPS scan of the surface (Appendix M: Table 5) confirmed the SEM result, which suggested that the surface of C4F6-450/03 is covered with bonded phosphate. The P 2p3/2 scan

resulted in an atomic concentration of 8.35% at a binding energy of 133.1 eV, which is typical for phosphate[42]. Nearly half of the surface (46.3%) was covered with metal oxides/hydroxides and another 7.72% of the atomic concentration on the surface were phosphate bounded oxygen. This can be seen at a binding energy of 533.1 eV, which could also fit C-OH bonds but these would also be visible in the C 1s spectra at 287.1 eV, where nothing was detected for C4F6-450/03. The carbon content from contaminations was present with a total of 26.25%. This value is lower than for C4F6-450, since the atomic concentrations are relative and the surface contains mainly phosphate. Furthermore, the carbon contamination came from the surrounding during the spraying process while the catalyst was heated. Comparing C4F6-450 and C4F6-450/03, the core Fe3+_2p

3/2 peak of phosphate and/or

maghemite is located at a higher binding energy, which indicates that the Fe sites of the phosphate sample exhibit a more positive partial charge. It was assumed that this appears to adversely affect the OER activity by binding with the OH- _{(dissociated from water molecules)}

too strongly, so that it hinders the desorption of O2. The FePO4 peak of C4F6-450/03 comes

from the phosphidation process where phosphate is produced, while C4F6 -450 could contain these peaks due to contamination or contact to the phosphate electrolyte.[43]

The XPS analyzes of O 1s revealed that the Fe-O on the surface is not completely gone after transferring the oxide sample into the phosphate sample, but the amount of Fe-O connection on the surface (around 530 eV) was reduced. Mainly OH/FePO4 (531.5 eV)[44] and H2O (533

eV) were found on the surface. These findings are in accordance with the previous result that the surface is mainly build out of Fe-P and FePO4 compounds. The comparison of the O 1s

spectra of C4F6-450 and C4F6-450/03 showed that the Fe-O peak of the phosphate sample shifted to a lower binding energy. The partial charge at the O site of the phosphate sam ple is more negative compared to that of the oxide sample. This adversely affects the OER activity by acting as an electrically repulsive shield to the O of water molecules, thus hindering the dissociative water adsorption process (primary step for OER) on the sample surface.[31] The more negative partial charge at the O site of the phosphate sample attracts and reacts with the free H+_{ions (from the electrolyte). This leads to more severe corrosion as compared to the}

oxide sample, which explains the decrease in the stability for the phosphate samples compared to the oxide samples. The P 2p spectrum of C4F6-450/03 (Appendix E: Figure 29) showed that P2p1/2 and P2p3/2 are visible between 131 and 135 eV, confirming the phosphate on the surface.

The combination of XPS and XRD showed that the catalyst surface is mainly covered with phosphate, which is most likely amorphous as planned, since the peaks in the XRD are rather small and broad than intense and sharp.

4.11 Phosphate Samples – The Impact of the pH

(28)

Figure 19a: (a) Activity, (b) overpotential and (d) stability of C4F6-450/03 as a function of pH adjusted with 0.5 M NaOH in 1M monosodium /disodium phosphate. (c) The Pourbaix diagram of phosphate at a low metal concentration, i.e., 10 nM to display the behavior in solution.[32][33][34].

The stability of C4F6-450/03 was investigated (Figure 20) and is displayed in Figure 20b. The stability was increasing with increasing pH from 50 min (pH 1) to 31 h and 50 min (pH 6). The increase in pH increases the stability, since less protons are in the electrolyte which can attack the phosphate on the surface. A lower pH means that more protons can bind with phosphate, which then detaches from the surface and dissolves into the electrolyte.[35][36]

(29)

5. Discussion

The original idea of this work was to investigate the production of ammonia from gaseous nitrogen and a liquid proton source. The mechanism to produce ammonia consists of the nitrogen reduction reaction (NRR) and the oxygen evolution reaction (OER). Unfortunately, the study of the NRR could not be adapted to a work of this scale, which is why the focus was shifted on the OER side instead. In this work, different combination of iron and cobalt were investigated for their ability to catalyze the OER. They were tested at two different spraying conditions and at different pHs between 0.3 and 6.

The synthesis was carried out at two different temperatures. The substrate temperature influences the physical properties of the deposited catalysts, including the catalyst loading, mineralogy, and overall quality (e.g. catalyst adhesion to the Ti foil). In this study, a minimal substrate temperature of 250 °C was chosen to be above the temperature for the decomposition of FeCl3. _{6H2O (120-180 °C for first-stage} dehydration) [29] and CoCl2.6H2O (130 °C for first-stage dehydration) [30] and to compensate for heat loss to the surrounding. This temperature was above the temperature of oxide formation for Fe chloride but under the one from Co chloride (400 °C). In later part of the work, a higher substrate temperature of 450 °C was chosen leading to an improved film quality (film adhesion to substrate, homogeneity) and the resultant enhanced catalytic performance (Figure 14, Appendix I, Figure 41). The best working composition in terms of stability and activity was achieved with a spraying content of 40 to 80% cobalt and 60 to 20% iron for all three samples because Co-doping induces the mineralogical phase transformation from hematite in the pure and lightly-doped Fe oxide samples, to maghemite in the highly doped (40-80% Co) Fe oxide. The actual composition for these samples was 20% Co and 80% Fe, since most of Co was present as chloride, which was washed away before the measurement (SEM). The enhanced intrinsic OER activity of Co-doped Fe oxide samples as compared to that of the pure Fe oxide is due to (1) the crystal polymorph effect where previous studies have provided evidence that maghemite shows increased OER activity compared to hematite[18] and (2) because of the co-doping effect where Co improves (i) the intrinsic catalytic activity and (ii) the electron transport within the film (as seen in EIS data)[31]. While Co-doping improves the activity, it neither deteriorates nor improves the stability as compared to pure Fe oxide. The catalysts sprayed at 250 °C had the highest average activity (average overpotentials between 570 mV and 630 mV) at pH 0.3 and the highest stability (14 h 20 min) at pH 2.

(30)

and Co was investigated, but did not result in any improvement on neither the activity nor the stability (Appendix T: Figure 53), which is why the data were not analyzed in this work.

The catalysts C4F6-250 and C4F6-450 will be set into context to check how good they are, therefore the overpotentials of C4F6-250 at pH 0.3 (570 mV) and C4F6-450 at pH 2 (560 mV) were compared with the literature. First, Mondschein et al [47] investigated a cheap Co based catalyst which will be compared with the catalysts from this work. Mondschein investigated a nanostructured film of Co3O4 on fluorine-doped tin oxide in 0.5 M H2SO4 which had an overpotential of 570 mV under acidic conditions and was stable for over 12h. The nanostructured Co3O4 had a lower activity than C4F6-450 (560 mV) but was better than C4F6-250 (670 mV) and the C10-250 sample (780 mV, same electrolyte and pH). The impact of a nanostructure was not investigated during this work, but five recently published OER catalysts (Table 1) are nanostructured in form of nanowires, nanosheets or other nanostructures, which makes them of interest for future studies. The pros and cons of nanostructures are that they are not as fast and as easy to synthesize as the spraying method used in this work, but the surface area is increased which results in a higher activity. The Ir nanowire and the nitrogen-doped carbon nanosheet are more active than C4F6-450, but Ir is a noble metal and therefore more expansive and not as suitable for mass production as C4F6-450. The nitrogen-doped carbon nanosheet has the advantage of the high surface area of the nanosheet, which C4F6-450 does not have. A future attempt could be to investigate if the synthesis of C4F6-450 on a nanosheet could lead to an increase in the activity. The investigated electrolyte in this thesis was either H2SO4 or H3PO4, the impact of HClO4 as a potential electrolyte was not part of these studies.

Table 1: Literature comparison of recently published OER catalysts under acidic conditions

Ca ta lyst Electrolyte Method Overpotentia l Sta bility Source Ir na nowires (ultra thin) 0.1 M HClO4 Wet chemica l method 390 mV Sta ble for 11+ h [48] Ma ghemite/Hematite 0.5 M H2SO4 Spra y pyrolysis 650 mV 24 h [18]

Nitrogen-doped ca rbon na nostructures

0.1 M HClO4

Wet impregna tion technique

390 mV 100

cycles

[49] Co-MoS2 na nosheet 0.5 M H2SO4 Hydrotherma l

method 540 mV Sta ble for 11+ h [50] Ag-doped Co3O4 na nowires 0.5 M H2SO4 Hydrotherma l method 680 mV 10 h [51]

6. Conclusions and Outlook

(31)

The investigation of non-noble metals for the oxygen evolution reaction is a promising path that can lead to a combination of metals which are cheap, stable and suitable for mass production with a high activity in order to use and store the energy of the sun or other renewable electricity from other sources. The combination of Fe and Co is promising, but must be investigated in the alkaline region as well to see if the activity and stability could be refined any further. Another possibility to increase the activity and stability would be to investigate the impact of the substrate or alternatively the fabrication of a nanostructure to increase the surface. Lastly, the addition of yet another metal to the composition may lead to further improvements regarding catalyst stability and activity.

Acknowledgement

I am grateful that I was able to study the topic of the oxygen evolution reaction in the research group of Johannes Messinger at the university of Umeå. A special thanks to Wai Ling for supporting and helping me with all the practical work in the lab. I want to thank Léon and Lisa for proofreading my thesis and their support. It was a nice year investigating the topic of the ammonia production and the water oxidation. Even though I had to drop the topic of the ammonia production out of gaseous nitrogen due to the time shortage, it was interesting and helpful to see how a new research topic is set up from scratch and then experiencing the subsequent development of the project. Finally, I want to thank my friends and family for supporting me during all this time.

References

[1] H. Da u, C. Limberg, T. Reier, M. Risch, S. Rogga n, a nd P. Stra sser, “The Mecha nism of Wa ter Oxidation: From Electrolysis via Homogeneous to Biological Catalysis,” ChemCatChem, vol. 2, no. 7, pp. 724–761, 2010, doi: 10.1002/cctc.201000126.

[2] R. Schlögl, “The role of chemistry in the energy challenge,” ChemSusChem, vol. 3, no. 2, pp. 209–222, 2010, doi: 10.1002/cssc.200900183.

[3] I. Roger, M. A. Shipman, and M. D. Symes, “Earth-abundant catalysts for electrochemical and photoelectrochemical water splitting,” Nat. Rev. Chem., vol. 1, 2017, doi: 10.1038/s41570-016-0003.

[4] S. Kim, M. Cho, and Y. Lee, “ Iridium Oxide Dendrite as a Highly Efficient Dual Electro -Ca ta lyst for Wa ter Splitting a nd Sensing of H 2 O 2 ,” J. Electrochem. Soc., vol. 164, no. 5, pp. B3029–B3035, 2017, doi: 10.1149/2.0061705jes.

[5] S. Chen, T. Takata, and K. Domen, “Particulate photocatalysts for overall water splitting,” Nat.

Rev. Mater., vol. 2, pp. 1–17, 2017, doi: 10.1038/na trevmats.2017.50.

[6] N. Elgrishi, K. J. Rountree, B. D. McCa rthy, E. S. Rountree, T. T. Eisen ha rt, a nd J. L. Dempsey, “A Practical Beginner’s Guide to Cyclic Voltammetry,” J. Chem. Educ., vol. 95, no. 2, pp. 197– 206, 2018, doi: 10.1021/a cs.jchemed.7b00361.

[7] M. G. Wa lter et al., “Sola r wa ter splitting cells,” Chem. Rev., vol. 110, no. 11, pp. 6446–6473, 2010, doi: 10.1021/cr1002326.

[8] B. A. Mei, O. Munteshari, J. Lau, B. Dunn, and L. Pilon, “Physical Interpretations of Nyquist Plots for EDLC Electrodes a nd Devices,” J. Phys. Chem. C, vol. 122, no. 1, pp. 194–206, 2018, doi: 10.1021/a cs.jpcc.7b10582.

[9] L. Zhuang, L. Ge, Y. Yang, M. Li, Y. Jia, and X. Yao, “Ultrathin Iron-Cobalt Oxide Nanosheets with Abundant Oxygen Vacancies for the Oxygen Evolution Reaction,” Adv. Mater., vol. 29, no. 1606793, pp. 1–7, 2017, doi: 10.1002/a dma.201606793.

[10] E. A. Fletcher a nd R. L. Moen, “Hydrogen a nd oxygen from wa ter,” Science (80-. )., vol. 197, no. 4308, pp. 1050–1056, 1977, doi: 10.1126/science.197.4308.1050.

[11] D. W. Keith a nd A. E. Fa rrell, “Rethinking Hydrogen Ca rs,” Science (80-. )., vol. 301, pp. 315– 317, 2003, doi: 10.1016/S0301-4215(02)00298-7.14.

[12] G. Brumfiel, “Hydrogen cars fuel debate on basic research,” Nature, vol. 422, no. 6928, p. 104, 2003, doi: 10.1038/422104a.

(32)

thermometry,” Appl. Phys. B Photophysics Laser Chem., vol. 51, no. 1, pp. 59–60, 1990, doi: 10.1007/BF00332326.

[14] H. Yoneyama, H. Sakamoto, and H. Tamura, “A Photo-electochemical cell with production of hydrogen a nd oxygen by a cell rea ction,” Electrochim. Acta, vol. 20, no. 5, pp. 341–345, 1975, doi: 10.1016/0013-4686(75)90016-X.

[15] J. Kovac, “Ethics in Science: The Unique Consequences of Chemistry,” Account. Res., vol. 22, no. 6, pp. 312–329, 2015, doi: 10.1080/08989621.2015.1047709.

[16] F. Benjamin; B. Krause; S. Raul, “Artisanal mining, livelihoods, and child labor in the cobalt supply chain of the Democratic Republic of Congo,” Cent. Eff. Glob. Action, 2017.

[17] U. Chohan, “Blockchain and the Extractive Industries #2: Diamonds Case Study,” SSRN

Electron. J., 2018, doi: 10.2139/ssrn.3141883.

[18] W. L. Kwong, C. C. Lee, A. Shchukarev, E. Björn, and J. Messinger, “High -performance iron (III) oxide electrocatalyst for water oxidation in strongly acidic media,” J. Catal., vol. 365, pp. 29–35, 2018, doi: 10.1016/j.jca t.2018.06.018.

[19] R. L. Blake, R. E. HessevicK, T. Zoltai, and L. W. Finger, “Refinement of the Hematite Structure,” Am. Mineral., vol. 51, pp. 123–129, 1966, doi: 10.1016.2010.01.025.

[20] B. A. Wechsler, D. H. Lindsley, a nd C. T. Prewitt, “Crysta l structure a nd ca tion distribution in tita noma gnetites (Fe3-xTixO4).,” Am. Mineral., vol. 69, no. 7–8, pp. 754–770, 1984.

[21] M. B. B. R. T. Downs, K. L. Bartelmehs, G. V. Gibbs, “Interactive software for calculating and displa ying X-ra y or neutron powder diffra ctometer pa tterns of crysta lline ma teria ls.,” Am.

Mineral., vol. 78, no. 9–10, pp. 104–107, 1993.

[22] Y. Feng, J. Wu, Q. Chi, W. Li, Y. Yu, and W. Fei, “Defects and Aliovalent Doping Engineerin g in Electroceramics,” Chem. Rev., 2020, doi: 10.1021/acs.chemrev.9b00507.

[23] W. L. Kwong, C. C. Lee, A. Shchukarev, and J. Messinger, “Cobalt-doped hematite thin films for electrocatalytic water oxidation in highly acidic media,” Chem. Commun., vol. 55, no. 34, pp. 5017–5020, 2019, doi: 10.1039/c9cc01369e.

[24] Z. Hua ng et al., “All-room-tempera ture solution-processed new na nocomposites ba sed hole tra nsport la yer from synthesis to film forma tion for high -performa nce orga nic sola r cells towa rds ultima te energy-efficient fa brica tion,” Nano Energy, vol. 47, no. Februa ry 2018, pp. 26–34, 2018, doi: 10.1016/j.na noen.2018.02.019.

[25] U. Schwertma nn, R. M. Cornell, U. Schwertmann The Iron Oxides. 2003.

[26] I. Persson, “Ferric Chloride Complexes in Aqueous Solution: An EXAFS Study,” J. Solution

Chem., vol. 47, no. 5, pp. 797–805, 2018, doi: 10.1007/s10953-018-0756-6.

[27] M. D. Lind, “Crystal structure of ferric chloride hexahydrate,” J. Chem. Phys., vol. 47, no. 3, pp. 990–993, 1967, doi: 10.1063/1.1712067.

[28] I. suna ga wa K. Wa izumi, H. Ma suda , H. Ohta ki, K. Tsuka moto, “In Situ Observa tions of the Pha se Tra nsition a mong Coba lt(II) Dichloride Hydra tes a nd Crysta l Structures of the Tetra - and Hexahydrate,” Chem. Soc. Japan, vol. 63, no. 12, pp. 3426–3433, 1990.

[29] S. B. Ka nungo a nd S. K. Mishra , “Therma l Dehydra tion a nd Decompostion of FeCl3 xH2O,” J.

Therm. Anal., vol. 46, pp. 1487–1500, 1996.

[30] S. K. Mishra and S. B. Kanungo, “Thermal dehydration and decomposition of cobalt chloride hydrate (CoCl2·xH2O),” J. Therm. Anal., vol. 38, no. 11, pp. 2437–2454, 1992, doi: 10.1007/BF01974622.

[31] K. Shimizu and J. F. Boily, “Electrochemical properties and relaxation times of the hematite/water interface,” Langmuir, vol. 30, no. 31, pp. 9591–9598, 2014, doi: 10.1021/la 501669a.

[32] Y. Surendra na th, D. A. Lutterma n, Y. Liu, a nd D. G. Nocera , “Nuclea tion, growth, a nd repa ir of a coba lt-ba sed oxygen evolving ca ta lyst,” J. Am. Chem. Soc., vol. 134, no. 14, pp. 6326–6336, 2012, doi: 10.1021/ja 3000084.

[33] G. Ong, Shyue Ping a nd Wa ng, Lei a nd Ka ng, Byoungwoo a nd Ceder, “Li-Fe-P-O2 Pha se Diagram from First Principles Calculations,” Chem. Mater., vol. 20, no. 5, pp. 1798–1807, 2008, doi: 10.1021/cm702327g.

[34] G. Ja in, Anubha v a nd Ha utier, Geoffroy a nd Ong, Shyue a nd Moore, Cha rles a nd Fischer, Christopher a nd Persson, Kristin a nd Ceder, “Forma tion entha lpies by mixing GGA a nd GGA + U calculations,” Phys. Rev. B, vol. 84, no. 4, p. 045115, 2011, doi: 10.1103/PhysRevB.84.045115.

[35] K. A. Persson, B. Waldwick, P. Lazic, and G. Ceder, “Prediction of solid-aqueous equilibria: Scheme to combine first-principles ca lcula tions of solids with experimenta l a queous sta tes,”

Phys. Rev. B - Condens. Matter Mater. Phys., vol. 85, no. 23, pp. 1–12, 2012, doi:

10.1103/PhysRevB.85.235438.

[36] A. Li et al., “Sta ble Potentia l Windows for Long‐Term Electroca ta lysis by Ma nga nese Oxides Under Acidic Conditions,” Angew. Chemie, vol. 131, no. 15, pp. 5108–5112, 2019, doi: 10.1002/a nge.201813361.

(33)

Characterization of Iron Oxide Nanoparticles,” Appl. Surf. Sci., vol. 405, no. February 2018, pp. 337–343, 2017, doi: 10.1016/j.a psusc.2017.02.002.

[38] P. T. Ba ba r et al., “SILAR deposited iron phospha te a s a bifunctional electroca ta lyst for efficient water splitting,” J. Colloid Interface Sci., vol. 534, pp. 350–356, 2019, doi: 10.1016/j.jcis.2018.09.015.

[39] Y. Li and C. Zhao, “Iron-Doped Nickel Phosphate as Synergistic Electrocatalyst for Water Oxidation,” Chem. Mater., vol. 28, no. 16, pp. 5659–5666, 2016, doi: 10.1021/a cs.chemmater.6b01522.

[40] J. Qi et al., “Modula tion of crysta l wa ter in coba lt phospha te for promoted wa ter oxida tion,”

Chem. Commun., vol. 56, no. 33, pp. 4575–4578, 2020, doi: 10.1039/d0cc01023e.

[41] R. Wyckoff, Crystal structures, vol. 1. 1964.

[42] R. Franke, T. Chassé, P. Streubel, and A. Meisel, “Auger parameters and relaxation energies of phosphorus in solid compounds,” J. Electron Spectros. Relat. Phenomena, vol. 56, no. 4, pp. 381–388, 1991, doi: 10.1016/0368-2048(91)85035-R.

[43] Y. Ya o et al., “Iron phosphide enca psula ted in P-doped gra phitic ca rbon a s efficient a nd sta ble electrocatalyst for hydrogen and oxygen evolution reactions,” Nanoscale, vol. 10, no. 45, pp. 21327–21334, 2018, doi: 10.1039/c8nr06752j.

[44] Q. Xie et al., “Effective Adsorption a nd Remova l of Phospha te from Aqueous Solutions and Eutrophic Wa ter by Fe-ba sed MOFs of MIL-101,” Sci. Rep., vol. 7, no. 1, p. 3316, Dec. 2017, doi: 10.1038/s41598-017-03526-x.

[45] P. H. Struthers and D. H. Sieling, “Effect of organic anions on phosphate precipitation by iron and aluminum as influenced by ph,” Soil Science, vol. 69, no. 3. pp. 205–213, 1950, doi: 10.1097/00010694-195003000-00005.

[46] I. B. A. Razaq, “Effect of pH and exchangeable metals on phosphate adsorptio n by soils,”

Retrosp. Theses Diss., vol. 9170, p. 265, 1989.

[47] J. S. Mondschein et al., “Crysta lline Coba lt Oxide Films for Susta ined Electroca ta lytic Oxygen Evolution under Strongly Acidic Conditions,” Chem. Mater., vol. 29, no. 3, pp. 950–957, 2017, doi: 10.1021/a cs.chemmater.6b02879.

[48] L. Fu, F. Yang, G. Cheng, and W. Luo, “Ultrathin Ir nanowires as high -performance electrocatalysts for efficient water splitting in acidic media,” Nanoscale, vol. 10, no. 4, pp. 1892– 1897, 2018, doi: 10.1039/c7nr09377b.

[49] K. Ma mta ni et al., “Insights into oxygen reduction rea ction (ORR) a nd oxygen evolution rea ction (OER) a ctive sites for nitrogen-doped ca rbon na nostructures (CNx) in a cidic media,”

Appl. Catal. B Environ., vol. 220, pp. 88–97, 2018, doi: 10.1016/j.a pcatb.2017.07.086.

[50] Q. Xiong et al., “One-step synthesis of coba lt-doped MoS2 na nosheets a s bifunctional electrocatalysts for overall water splitting under both acidic and alkaline conditions,” Chem.

Commun., vol. 54, no. 31, pp. 3859–3862, 2018, doi: 10.1039/c8cc00766g.

[51] K. L. Ya n et al., “Mesoporous Ag-doped Co3O4 na nowire a rra ys supported on FTO a s efficient electrocatalysts for oxygen evolution reaction in acidic media,” Renew. Energy, vol. 119, pp. 54–61, 2018, doi: 10.1016/j.renene.2017.12.003.

(34)

Appendix

A The Sample Preparation

The weight at 250 °C was limited to 0.05 M in total. The input and the volume were quadrupled for the 450 °C samples to achieve the same loading.

Table 2: The sample preparation for 250°C and 450 °C

Sample Temperature [°C] Volume [mL] Iron [g] Cobalt [g] 0% Cobalt 100% Iron 250 100 1.351 0 20% Cobalt 80% iron 250 100 1.081 0.238 40% Cobalt 60% Iron 250 100 0.811 0.476 60% Cobalt 40% Iron 250 100 0.540 0.713 80% Cobalt 20% Iron 250 100 0.270 0.951 100% Cobalt 0% Iron 250 100 0 1.189 0% Cobalt 100% Iron 450 400 5.404 0 40% Cobalt 60% Iron 450 400 3.242 1.903 100% Cobalt 0% Iron 450 400 0 4.757

The electrolytes for the phosphate samples were prepared as shown in Table 3. The pH was adjusted with 0.5 M H3PO4 and 0.5 M NaOH if needed.

Table 3: Weight in for 50 mL electrolyte for different pHs

(35)

B SEM Pictures of the Sample

Figure 21: SEM picture of F10-250.

(36)

Figure 23: SEM mapping of C4F6-250.

(37)

Figure 25: SEM mapping of C8F2-250.

(38)

C The Overpotential of C2F8-250, C4F6-250 and C6F4-250

Figure 27: The overpotential of C2F8-250, C4F6-250 and C6F4-250. The trend that the overpotential was increasing with pH was seen through all samples. The local minima at pH 0.3 using C4F6-250 could be confirmed.

D Different Ionic Strengths for the Phosphate and Oxide Samples at 450 °C

(39)

E The XPS Measurements

The atomic concentrations are shown in Table 5.

Figure 29: The XPS a) P 2p3/2 spectrum of C4F6-450/03 and b) S 2p3/2 spectrum of C4F6-450.

(40)

Figure 31: The XPS C 1s spectrum of a) C4F6-250, b) C4F6-450 and c) C4F6-450/03.

(41)

Figure 33: The XPS Fe 2p3/2 spectrum of a) C4F6-250, b) C4F6-450 and c) C4F6-450/03.

(42)

F Stability Curves of C4F6-450 at pH 2,4 and 6

Figure 35: The CP curves of C4F6-450 at pH 2 in a 1M Monosodium/disodium phosphate electrolyte.

G All CV Curves of the 250°C Samples at pH 0.3

Figure 36: a) All 6 CV curves and b) the resulting overpotential of C4F6-250 at pH 0.3. The last curve of the 6. scan was used. The same concept was used for all CV curves shown in this thesis.

(43)

Figure 37: The measured CV curves of a) F10-250, b) C2F8-250, c) C4F6-250, d) C6F4-250, e) C8F2-250 and f) C10-250 at pH 0.3. The two measurements of C10-250 were measured from the same sample to see the degradation and do not represent a second set of measurements.

(44)

Figure 38: The overpotential of the 6th_{CV curves of F10-250, C2F8-250, C4F6-250, C6F4-250, C8F2-250 and}

C10-250. All 6th curves are shown in a), while b) shows the average between the curves for the same catalyst together with the error range.

H All CV Curves of C4F6-250 at Different pHs

(45)

Figure 39: The measured CV curves of C4F6-250 in H2SO4 at a) pH 0.3, b) pH 1, c) pH 2, d) pH 3, e) pH 4, f) Ph 5

and g) pH 6.

Figure 40: The overpotential of the 6th CV curves of C4F6-250 at different pHs. All 6th curves are shown in a), while b) shows the average between the curves for the same pH together with the error range.

I The CV Curves of C4F6-250 and C4F6-450 at pH 2

(46)

(47)

J The Maghemite and Hematite Structure

Figure 42: The a) hexagonally close-packed structure of hematite and the b) cubic close-packed structure of maghemite.[52]

K The XRD and SEM Spectra of C4F6-250 and C4F6-450

(48)

Figure 44: The SEM mapping of C4F6-450 of a) Fe, b) Co, c) O, d) Cl, e) C and f) the SEM picture.

e _f

c d

(49)

L The SEM Spectra of C4F6-450/02, C4F6-450/03 and C4F6-450/04

Figure 45: The SEM mapping of C4F6-450/02 of a) Fe, b) Co, c) O, d) Cl, e) P and f) the SEM picture.

e f

c d

(50)

(51)

Figure 47: The SEM mapping of C4F6-450/04 of a) Fe, b) Co, c) O, d) Cl, e) P and f) the SEM picture.

Table 4: The atomic concentration (SEM) of C4F6-450/02, C4F6-450/03 and C4F6-450/04

C4F6-450/02 C4F6-450/03 C4F6-450/04

Element Atomic concentration [%]

(52)

M The Atomic Concentrations (XPS) of C4F6-250, C4F6-450 and C4F6-450/03 The XPS graphs are shown in Appendix E.

Table 5: The XPS analyzes of the different atomic concentrations of C4F6-250 and C4F6-450. Binding

energy [eV] Name

(53)

N The Activity of C4F6-250, C10-250 and F10-250 at pH 2

Figure 48: The a) CV curve and b) overpotential of C4F6-250, C10-250 and F10-250 at pH 2 measured in H2SO4 and

NaOH.

O The CV Curves of C4F6-450 at pH 2, 4 and 6

Figure 49: The CV curve of C4F6-450 at a) pH 2,b) pH 4 and c) pH6 in a phosphoric electrolyte. The 6th_{curve and}

(54)

P The CV Curves of C4F6-450/02, C4F6-450/03 and C4F6-450/04 at pH 2

Figure 50: The CV curves of a) C4F6-450/02, b) C4F6-450/03 and c) C4F6-450/04. All of this samples were measured one time. The CV curves of e) the 6th_{scans were converted into e) the overpotential together with C4F6-450. The}

(55)

Q The CV Curves of C4F6-450/03 at pH 1 to 6

Figure 51: The CV curves of C4F6-450/03 at a) pH 1, b) pH 2, c) pH 3, d) pH 4, e) pH 5 and f) pH 6 in 0.5 M H3PO4.

(56)

R The Nyquist Plot of C4F6-450/03 at pH 1 to 6

Figure 52: The a) Nyquist plot of C4F6-450/03 at pH 1 to 6 and b) the resulting electrolyte resistance.

S The pH of Co and Fe Chloride in Solution

Table 6: The pH of different Co and Fe chloride salt compositions in water.

Sample Test 1 Test 2 Test 3 Average

(57)

T The Nickel Doped Catalysts

Figure 53: (a) CV curves, (b) activity of C2F8-250, C4F6-250 and C6F4-250 with 10% nickel substitution.(c) CP curve and (d) stability of C4F6-250 and C4F6-250 -10% Nickel substitution.

U The Solubility of Fe and the Eh-pH Diagram

(58)

V The Achieved Co Content by SEM and XPS

(59)

Master Thesis

Master Thesis

Improving Stability and Efficiency of

Earth-abundant Electrocatalysts for

Water Oxidation

Clemens Wunder

Abstract

List of abbreviations

Author contribution

Table of contents

1. Introduction

2. Popular Scientific Summary including Social and Ethical

Aspects

3. Experimental

4. Results

5. Discussion

6. Conclusions and Outlook

Acknowledgement

References

Appendix