Synthesis and Characterization of Bar- ium Titanate and Barium Indate-Zirconate Perovskite Oxyhydrides Master’s thesis in Nanotechnology Isac Johansson DEPARTMENT OF CHEMISTRY AND CHEMICAL ENGINEERING CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2025 www.chalmers.se www.chalmers.se Master’s thesis 2025 Synthesis and Characterization of Barium Titanate and Barium Indate-Zirconate Perovskite Oxyhydrides Isac Johansson Department of Chemistry and Chemical Engineering Division of Applied Chemistry Per-Anders Carlsson’s group Chalmers University of Technology Gothenburg, Sweden 2025 Synthesis and Characterization of Barium Titanate and Barium Indate-Zirconate Perovskite Oxyhydrides ISAC JOHANSSON © Isac Johansson, 2025. Examiner: Per-Anders Carlsson, Professor, Division of Applied Chemistry, Depart- ment of Chemistry and Chemical Engineering Supervisor: Rasmus Lavén, Post-Doctoral Researcher, Division of Applied Chem- istry, Department of Chemistry and Chemical Engineering Master’s Thesis 2025 Department of Chemistry and Chemical Engineering Division of Applied Chemistry Per-Anders Carlsson’s group Chalmers University of Technology SE-412 96 Gothenburg Telephone +46 31 772 1000 Cover: General structure of a perovskite oxyhydride of BaTiO3. Gothenburg, Sweden 2025 iv Synthesis and Characterization of Barium Titanate and Barium Indate-Zirconate Perovskite Oxyhydrides ISAC JOHANSSON Department of Chemistry and Chemical Engeering Chalmers University of Technology Abstract The global imperative to reduce CO2 emissions has driven interest in catalytic con- version technologies, particularly CO2 hydrogenation, which transforms CO2 into valuable chemicals. This reaction often relies on metallic nanoparticles supported on catalyst substrates, commonly metal oxides like Al2O3 or ZrO2. Perovskite oxides have emerged as promising alternatives due to their adjustable surface chemistry, thermal stability, and ability to host redox-active defect sites. Recent attention has turned to anion-adjusted perovskite materials, amongst them oxyhydrides, where oxygen anions are partially replaced by hydride ions. These modifications can en- hance catalytic performance and introduce properties such as hydride ion conductiv- ity and interesting electronic and magnetic properties. This project focused on the synthesis and structural analysis of reduced perovskite oxides of synthesised barium titanate (BaTiO3), nano-crystalline barium titanate and barium indate-zirconate (BaZr1−xInxO3− x 2 ). For BaTiO3, synthesis routes mainly investigated reduction with CaH2 enclosed in stainless steel capsules, filled with high purity argon. For BaZr1−xInxO3− x 2 , reduction by H2 gas annealing was investigated. Characterization heavily relied on powder X-ray diffraction (PXRD) and thermogravimetric analysis (TGA) measurements. Inelastic neutron scatterin (INS) was performed for a re- duced 50% indium substituted BaZr0.5In0.5O2.75. The study primarily investigated how synthesis parameters such as molar ratio of CaH2, temperature, and heating time affect reduction extent, anion composition, phase formation, impurity forma- tion and crystallinity. The CaH2 reduction of synthesized tetragonal BaTiO3 at 600◦ C for 48 hours yields reduced products with a cubic phase, accompanied by a colour change from white to dark blue or black. An increased molar ratio of CaH2 leads to a greater degree of reduction. Rietveld refinements indicate formation of a single phase in these reduced samples. In contrast, samples of nano-BaTiO3 subjected to the same reduction conditions exhibit a lower degree of reduction and show more pronounced two-phase indications. Higher molar CaH2 ratios result in the forma- tion of Ba2TiO4 impurities. These impurity phases can be reduced by decreasing the CaH2 ratio. For the nano-BaTiO3, a temperature decrease to 580◦ C doesn’t im- pact Ba2TiO4 amounts. Shortening the heating time to 24 hours leads to decreased amounts, at the expense of a lower reduction extent in the nano-BaTiO3 perovskite phase. Hydrogen annealing of BaZr1−xInxO3− x 2 with 50% indium substitution at 800◦ C for 24 hours and 70% substitution at 650◦ C for 20 hours give reduced per- ovskite oxides of barium indate-zirconate. The extent of reduction is comparable between the two compositions. Inelastic neutron scattering (INS) measurements on the 50% BaZr1−xInxO3−x/2 sample indicate minimal hydride incorporation. v Acknowledgements I would like to express a big thank you to Rasmus Lavén for for his tremendous help throughout these six months, and for giving me the opportunity to do this project. From experimental planning to result analysis and everything in between, Rasmus has been of huge help. I would also like to thank Guido J.L. de Reijer and Andreas Schaefer for giving assistance with several parts of the experimental synthesis and setup, and Per- Anders Carlsson for his general guidance, insightful discussions and giving me the opportunity to do this project. Thank you to Jeff Armstrong for performing INS measurements. I would like to thank the division of Applied Chemistry for providing the lab facilities necessary for this project, and for the people at the division for being so helpful and kind for these six months. This work was performed in part at the Chalmers Material Analysis Laboratory, CMAL, thank you. Experiments at the ISIS Neutron and Muon Source were supported by beamtime allocations through a Xpress access proposal 2590151. Isac Johansson, Gothenburg, June 2025 vii List of Acronyms BZO Barium indate-zirconate oxide BZOH Reduced barium indate-zirconate oxide CIF Crystallographic Information File INS Inelastic Neutron Scattering nH Molar ratio of H− to BaTiO3 PXRD Powder X-ray Diffraction TGA Thermogravimetric Analysis ix Contents List of Acronyms ix Nomenclature xi List of Figures xiii List of Tables xvii 1 Introduction 1 1.1 Aim of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Outline of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2 Perovskite oxyhydrides 5 2.1 General structure and properties . . . . . . . . . . . . . . . . . . . . . 5 2.2 Oxyhydride synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.3 BaTiO3 and BaTiO3−xHy . . . . . . . . . . . . . . . . . . . . . . . . 10 2.3.1 Structure and properties . . . . . . . . . . . . . . . . . . . . . 10 2.3.2 Previous synthesis reports . . . . . . . . . . . . . . . . . . . . 11 2.4 BaZr1−xInxO3− x 2 and BaZr1−xInxO3− x 2 −yHz : . . . . . . . . . . . . . . 13 3 Methods 15 3.1 Powder X-ray diffraction . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.1.1 Bragg’s law . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.1.2 Rietveld refinement and diffraction profile effects . . . . . . . . 16 3.2 Thermogravimetric analysis . . . . . . . . . . . . . . . . . . . . . . . 18 3.3 Inelastic neutron scattering . . . . . . . . . . . . . . . . . . . . . . . 18 4 Experimentals 21 4.1 Oxide Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 4.1.1 BaTiO3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 4.1.2 BaZr1−xInxO3− x 2 . . . . . . . . . . . . . . . . . . . . . . . . . 22 4.2 Oxyhydride Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 4.2.1 BaTiO3−xHy and nano-BaTiO3−xHy . . . . . . . . . . . . . . . 23 4.2.2 BaZr0.5In0.5O2.75−xHy and BaZr0.3In0.7O2.65−xHy . . . . . . . . 26 4.2.3 PXRD measurements . . . . . . . . . . . . . . . . . . . . . . . 27 4.2.4 TGA measurements . . . . . . . . . . . . . . . . . . . . . . . . 27 4.2.5 INS measurements . . . . . . . . . . . . . . . . . . . . . . . . 28 xi Contents 5 Results and discussion 29 5.1 Synthesis of BaTiO3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 5.2 Synthesis of BaTiO3−xHy . . . . . . . . . . . . . . . . . . . . . . . . . 30 5.3 Synthesis of nano-BaTiO3−xHy . . . . . . . . . . . . . . . . . . . . . . 41 5.4 Comparing BaTiO3−xHy and nano-BaTiO3−xHy . . . . . . . . . . . . 51 5.5 Synthesis of BaZr0.5In0.5O2.75 and BaZr0.5In0.5O2.75−xHy . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 5.6 Synthesis of BaZr0.3In0.7O2.65 and BaZr0.3In0.7O2.65−xHy . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 6 Conclusions 59 A Appendix 1 I xii List of Figures 1.1 a) Typical ABO3 perovskite structure. b) ABO3−xHy structure where a portion of the oxygen ions have been replaced with hydride ions. Green, light blue and red correspond to A, B and O respectively. Images generated using VESTA software [1]. . . . . . . . . . . . . . . 2 2.1 Top: Example of ABO3 perovskite structure. Green, light blue and red correspond to A site, B site and O respectively. Bottom: Vi- sualisation of twelve-fold and six-fold oxygen coordination around A and B respectively. Images were generated using VESTA software. . . 6 2.2 Top: ABO3−xHy structure where a portion of the oxygen atoms have been replaced with hydride ions. Bottom: Vacancy formation, where some oxygen has been removed from the perovskite structure. Images were generated using VESTA software. . . . . . . . . . . . . . . . . . 7 2.3 Left: Local coordination geometry around Ti in cubic BaTiO3. Right: Local coordination geometry around Ti in tetragonal BaTiO3. Bond lengths are shown for cubic and tetragonal phase respectively [1]. Images were generated using VESTA software. . . . . . . . . . . . . . 10 2.4 Left: Cubic perovskite structure of barium indate zirconate, with some oxygen vacancies as In3+ partially replaces the Zr4+. Right: Visualisation of partial indium substitution at zirconium atomic po- sitions. Images were generated using VESTA software. . . . . . . . . 13 2.5 Left: Hydride incorporation at oxygen 3d sites. Right: Hydride incorporation at interstitial [100] face centre position. Images were generated using VESTA software. . . . . . . . . . . . . . . . . . . . . 13 3.1 Constructive interference of X-rays. . . . . . . . . . . . . . . . . . . . 16 4.1 Glovebox with argon (99,999%, Linde), used for the mixing of CaH2 and BaTiO3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 4.2 Stainless steel capsule (after use) used for BaTiO3−xHy synthesis, shown both individually and placed inside the furnace. . . . . . . . . 24 5.1 Left: Rietveld plot of synthesised BaTiO3. Reference COD ID: 1507756, obtained from Crystallography Open Database. Right: Peak split at 2θ = 45.4o, displaying tetragonal phase of BaTiO3. Data was plotted in Matlab. . . . . . . . . . . . . . . . . . . . . . . . . . . 29 xiii List of Figures 5.2 Colours of products from reduction of synthesised BaTiO3 for differ- ent concentrations of CaH2, all heated at 600◦ C for 48 hours. From left to right, samples are reduced with molar ratios of CaH2 corre- sponding to nH = [2.0 ; 1.5 ; 1.0 ; 0.5]. . . . . . . . . . . . . . . . . . 30 5.3 PXRD pattern of products from reduction of synthesised BaTiO3 for different concentrations of CaH2, all heated at 600◦ C for 48 hours. Peak intensities are normalised to the most intense peak around 2θ = 31.5o. Measured in variable slit. Plotted in Matlab . . . . . . . . . 31 5.4 Zoomed images of [200] PXRD peak around 2θ = 45.4o. Plotted in Matlab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 5.5 Rietveld plots of products from reduction of synthesised BaTiO3 for different concentrations of CaH2. The blue crosses, green line, and turquoise line correspond to the observed data, calculated fit, and difference curve, respectively. Orange squares mark BaCO3 impurities (1.0 H, 2.0 H), green diamond symbols mark orthorhombic Ba2TiO4 impurities (2.0 H). The black cross likely corresponds to a monoclinic Ba2TiO4 phase, however this is unconfirmed. . . . . . . . . . . . . . . 33 5.6 Rietveld plot of 2.0 H sample for CaH2 reduction of synthesised BaTiO3 sample, with tetragonal BaTiO3 as phase reference. Ob- tained lattice parameters are shown in the Figure. . . . . . . . . . . . 34 5.7 Lattice parameter dependence on CaH2 molar ratio for CaH2 reduc- tion of synthesised BaTiO3. Linear fit was performed using polyfit function in Matlab. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 5.8 Left: Peak shape of 1.0 H CaH2 reduced BaTiO3 sample. Right: Peak shape of 2.0 H CaH2 reduced BaTiO3 sample. Red rectangles indicate areas of interest. . . . . . . . . . . . . . . . . . . . . . . . . . 36 5.9 Top: XRD patterns for CaH2 reduced 2.0 H of synthesised BaTiO3 samples before washing (blue curve) and after washing (red curve). Pink triangles, green diamonds, red circles and orange squares show presence of CaO, Ba2TiO4 (monoclinic + orthorhombic), Ti3O and BaCO3 respectively. Black crosses represent eliminated peaks in the XRD pattern. Bottom: Peaks for orthorhomic and monoclinic phase of Ba2TiO4 in unwashed 2.0 H sample. Images plotted in Matlab. . . 37 5.10 Comparison of Ba2TiO4 peak intensities around 28.5-30.5o for reduced synthesised BaTiO3 for different nH = [0.5 ; 1.0 ; 1.5 ; 2.0]. All peak intensities are normalised to the highest peak around 2θ = 31.5o, and the background level is set uniformly across all patterns. Images plotted in Matlab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 5.11 TGA data between 35-900◦ C for products of CaH2 reduction of BaTiO3 for 48 hours at 600◦ C for different molar ratios of CaH2. Black dotted lines correspond to approximate weight difference refer- ence lines before and after oxidation (likely not accurate as discussed below). Images plotted in Matlab. . . . . . . . . . . . . . . . . . . . . 39 xiv List of Figures 5.12 XRD patterns for 2.0 H sample of CaH2 reduced BaTiO3 before (blue) and after (red) TGA measurement. Black crosses represent uniden- tified phase(s), which peaks only appeared post-TGA. Once again orange squares, green diamonds and red circles represent BaCO3, Ba2TiO4 and Ti3O respectibely. Plotted in Matlab . . . . . . . . . . 40 5.13 PXRD patterns of products from reduction of nano-BaTiO3 for two different concentrations of CaH2, nH = 1.0 (blue) and nH = 2.0 (red). Orange squares and green diamond symbols represent BaCO3 and Ba2TiO4 impurities respectively. Peak intensities are normalised to the most intense peak around 2θ = 31.5o. Measured in variable slit. Plotted in Matlab. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 5.14 Showing of peak broadenings and absence of 2θ shift between the samples peak positions for XRD patterns from reduction of nano- BaTiO3 for two different concentrations of CaH2, nH = 1.0 (blue) and nH = 2.0 (red). Plotted in Matlab. . . . . . . . . . . . . . . . . . 42 5.15 Rietveld plots for refinements of nano-BaTiO3, performed with tetrag- onal reference (left) and cubic reference (right). The blue crosses, green line, and turquoise line correspond to the observed data, calcu- lated fit, and difference curve, respectively . . . . . . . . . . . . . . . 44 5.16 Rietveld plots for refinements of products from CaH2 reduction of nano-BaTiO3 at 600◦ C for 48 hours, with concentrations of 1.0 H (left) and 2.0 H (right). The blue crosses, green line, and turquoise line correspond to the observed data, calculated fit, and difference curve, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 5.17 PXRD patterns of products from CaH2 reduction of nano-BaTiO3 with molar ratio nH = 2.0 for different times and temperatures. Or- ange squares, green diamond symbols and purple upside-down trian- gles represent BaCO3, Ba2TiO4 and Ca(OH)2 impurities respectively. Peak intensities are normalised to the most intense peak around 2θ = 31.5o. Measured in variable slit. Plotted in Matlab. . . . . . . . . . 45 5.18 Comparison of Ba2TiO4 (green diamond symbol) peak intensities around 28.5-30o for reduced nano-BaTiO3 with molar ratio nH = 2.0 for different times and temperatures, before washing. Purple upside- down triangle correspond to Ca(OH)2 impurities. All peak intensities are normalised to the highest peak around 2θ = 31.7o, and the back- ground level is set uniformly across all patterns. Images plotted in Matlab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 5.19 Left: Peak shape of refined 1.0 H CaH2 reduced nano-BaTiO3 sam- ple, heated at 600◦ C for 48 hours. Right: Peak shape of refined 2.0 H CaH2 reduced nano-BaTiO3 sample, heated at 600◦ C for 48 hours. Red rectangles indicate areas of interest. . . . . . . . . . . . . . . . . 48 5.20 Top: Peak shape of refined 1.0 H CaH2 reduced nano-BaTiO3 sample, heated at 600◦ C for 48 hours. Bottom: Peak shape of refined 1.0 H CaH2 reduced nano-BaTiO3 sample, heated at 600◦ C for 48 hours using a smaller precursor amount. Red rectangles indicate areas of interest. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 xv List of Figures 5.21 TGA curves between 35-900◦ C for products of CaH2 reduction of nano-BaTiO3. Images plotted in Matlab. . . . . . . . . . . . . . . . . 50 5.22 Image of products from CaH2 reduction of synthesised (right) and nano-BaTiO3 (left), with molar ratio corresponding to 1.0 H for both samples. Samples heated at 600◦ C for 48 hours. . . . . . . . . . . . . 51 5.23 Left: TGA curves for samples of 1.0 H, heated at 600◦ C for 48 hours of CaH2 reduced nano-BaTiO3 (red) and synthesised BaTiO3 (blue). Right: TGA curves for samples of 2.0 H, heated at 600◦ C for 48 hours of CaH2 reduced nano-BaTiO3 (red) and synthesised BaTiO3 (blue). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 5.24 Left: Peak shape of refined 2.0 H CaH2 reduced synthesised BaTiO3 sample, heated at 600◦ C for 48 hours. Right: Peak shape of refined 2.0 H CaH2 reduced nano-BaTiO3 sample, heated at 600◦ C for 48 hours. Red rectangles indicate areas of interest. . . . . . . . . . . . . 53 5.25 PXRD pattern of synthesised 50% BZO, with vertical reference lines of BaZrO3 in red. COD ID: 1532743. Image extracted from EVA software. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 5.26 PXRD pattern of synthesised 50% BZO (blue) and H2 reduced 50% BZOH (red). BZOH sample heated at 800◦ C for 24 hours, with 8 ml/min H2 flow. Peak intensities are normalised to the highest intensity peak. Plotted in Matlab. . . . . . . . . . . . . . . . . . . . . 55 5.27 TGA curves between 35-900◦ C for 50% BZOH sample. Reported x values for BaZr0.5In0.5O2.75−xHy are shown, assuming vacancy forma- tion as well as hydride incorporation only. Plotted in Matlab. . . . . 55 5.28 Back-scattering INS spectrum for 50% BZOH sample (red). Back- ground measurement is shown in blue. Plotted in Matlab. . . . . . . 56 5.29 PXRD pattern of synthesised 70% BZO, with vertical reference lines of BaZrO3 in red. COD ID: 1532743. Image extracted from EVA software. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 5.30 PXRD pattern of synthesised 70% BZO (green) and H2 reduced 70% BZOH (orange). BZOH sample heated at 650◦ C for 20 hours, with 8 ml/min H2 flow. Peak intensities are normalised to the highest intensity peak. Plotted in Matlab. . . . . . . . . . . . . . . . . . . . . 57 5.31 TGA curves between 35-900◦ C for 70% BZOH sample. Reported x values for BaZr0.3In0.7O2.65−xHy are shown, assuming vacancy forma- tion as well as hydride incorporation only. Plotted in Matlab. . . . . 58 A.1 PXRD pattern of CaH2 reduced synthesised BaTiO3 in an alumina tube at 600oC for 48 hours. Plotted in Matlab. . . . . . . . . . . . . . I xvi List of Tables 2.1 Hydrogen processes during hydride reduction, adapted from [2]. . . . 8 2.2 Synthesis results reported from Nedumkandathil et al. for CaH2 re- duction of BaTiO3 heated at 600◦ C for 48 hours, table adapted from the original source [2]. . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.3 Synthesis results reported from Nedumkandathil et al. for CaH2 re- duction of BaTiO3 heated at 600◦ C with 1.2 H for different heating times, table adapted from the original source [2]. . . . . . . . . . . . . 12 4.1 Weighed precursor amounts for different molar ratios of synthesised BaTiO3 and CaH2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 4.2 Weighed precursor amounts for different molar ratios of nano-BaTiO3 and CaH2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 5.1 BaTiO3 lattice parameters for synthesised and reference. Reference COD ID: 1507756, obtained from Crystallography Open Database. . . 30 5.2 Refinement quality indicators for CaH2 reduced samples of synthe- sised BaTiO3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 5.3 BaTiO3−xHy lattice parameters before and after washing for different molar ratios nH , all heated for 48 hours at 600◦ C. BaTiO3 wasn’t washed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 5.4 Apparent x values for different molar ratios of CaH2, from products of CaH2 reduction of BaTiO3. Calculations of x were made considering only oxygen vacancies in the perovskite, as well as only hydride sub- stitution of oxygen. Lattice parameters were included for comparison. 41 5.5 Lattice parameters for CaH2 reduced nano-BaTiO3 for nH = [1.0 ; 2.0], heated at 600◦ C for 48 hours. Rietveld refinements were performed with a cubic BaTiO3 as reference for the oxide, which is discussed below. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 5.6 Comparison of lattice parameters and refinement quality indicators for Rietveld refinement of nano-BaTiO3 with cubic and tetragonal reference. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 5.7 Lattice parameters and refinement quality indicators for CaH2 re- duced nano-BaTiO3 for nH = [1.0 ; 2.0], heated at 600◦ C for 48 hours. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 5.8 Lattice parameters and refinement quality indicators of products from CaH2 reduction of nano-BaTiO3 with molar ratio nH = 2.0, for dif- ferent times and temperatures. . . . . . . . . . . . . . . . . . . . . . . 46 xvii List of Tables 5.9 Lattice parameters and refinement quality indicators of products from CaH2 reduction of nano-BaTiO3 with molar ratio nH = 2.0, for dif- ferent times and temperatures. . . . . . . . . . . . . . . . . . . . . . . 46 5.10 Lattice parameters for products of CaH2 reduction of synthesised and nano-BaTiO3 for different molar ratios of CaH2, heated at 600◦ C for 48 hours. No samples of 0.5 H and 1.5 H were synthesised for nano- BaTiO3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 5.11 Lattice parameters for CaH2 reduced nano-BaTiO3 for nH = 1.0 with different precursor amounts of nano-BaTiO3, heated at 600◦ C for 48 hours. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 xviii 1 Introduction The global goal of reducing CO2 emissions is widely recognized and has become a central focus across both environmental policy and scientific research. Part of the solution in reducing carbon emissions sustainably, is by converting CO2 into e.g biomass feedstock via CO2 hydrogenation. This process can be facilitated by the presence of catalytic metallic nanoparticles anchored onto catalyst support ma- terials. Commonly, these catalyst supports are metal oxides, such as aluminium oxide (Al2O3) or zirconium oxide (ZrO2) [3, 4]. Perovskite oxides have also been investigated in this context, for e.g. CO2 conversion to aromatics [5]. Metal ox- ides have shown to be effective as catalyst supports, due to their tunable surface properties, thermal stability, and ability to interact strongly with active metals [6]. Many oxides, such as CeO2 and TiO2, provide oxygen vacancies and defect sites that promote redox reactions and hydrogen spillover [7]. Additionally, their sta- bility under reaction conditions enables long-term operation without deactivation. Studies have shown that adjusting the anionic sites of these (oxide) catalyst sup- ports enhances their interaction with the nanoparticles, which in turn can improve overall catalytic performance [8]. One example of such anion-adjusted material is oxyhydrides, which show potential as catalyst support materials. These are formed by replacing some of the oxygen atoms (O2−) in oxides with hydride ions (H−). For ABO3 perovskite materials, which are the main materials of interest for this project, this occurs according to ABO3 → ABO3−xHy (1.1) This partial substitution has shown to potentially increase catalytic performance as well as give other interesting properties in e.g. hydride-ion conductivity and catalytic performance in other reactions such as ammonia synthesis [9, 10]. The materials also exhibit interesting electric and magnetic properties which can be tuned by the anionic substitution [9], for instance, several oxyhydride materials exhibit ferroelectric behaviour which in turn can be enhanced by anionic substitution [11]. 1 1. Introduction Figure 1.1: a) Typical ABO3 perovskite structure. b) ABO3−xHy structure where a portion of the oxygen ions have been replaced with hydride ions. Green, light blue and red correspond to A, B and O respectively. Images generated using VESTA software [1]. For perovskite oxyhydrides, A is typically a larger alkaline earth metal cation such as Ba2+ or Sr2+, whereas B is a smaller transition metal cation with variable oxidation states e.g. Ti or Co cations. The synthesis requires strongly reducing conditions, but these must be achieved without allowing the hydride, which is itself a powerful reducing agent, to transfer an electron to the metal cation [9]. Overly reducing conditions will hence not yield oxyhydrides, but can rather yield highly reduced metal oxides or the correspondent elemental metal phase. One of the most common ways of reducing the oxides in the synthesis is by metal hydride reduction. For example, CaH2 can be used as reducing agent. Oxyhydride formation via direct substitution can be written as BaTiO3 + x CaH2 → BaTiO3−xHx + x CaO + x 2 H2 (1.2) for BaTiO3 [2]. Recently, it has been shown that it is also possible to form oxy- hydrides via direct reduction with H2(g) [12]. For a indium-substituted barium- zirconate oxyhydride, this process can be described by BaZr0.5In(III)0.5O2.75 + 0.25 H2 → BaZr0.5In(II)0.5O2.25H0.5 (1.3) There are also other ways of synthesising oxyhydrides, such as high-pressure solid- state reactions or direct mechanochemical synthesis, starting from binary metal oxides and hydrides [13, 14]. 1.1 Aim of the thesis This project aims to synthesise selected perovskite oxyhydrides of BaTiO3 and BaZr1−xInxO3− x 2 and to conduct structural analysis of the samples using powder X-ray diffraction (PXRD) and thermogravimetric analysis (TGA). Inelastic neu- tron scattering (INS) is used for selected BaZr1−xInxO3− x 2 oxyhydride samples. The project explores the synthetic strategies for obtaining oxyhydride materials and how these strategies give rise to different properties, such as anion composition, vacancies and overall crystal structure. Molar ratio of CaH2 to BaTiO3, heating temperature and heating time are the main parameters that are investigated during synthesis. Of the two materials studied in this work, a greater emphasis is put on the oxyhydride 2 1. Introduction of BaTiO3. This comes naturally as it is more established in previous research and provides a more accessible starting point for method development. 1.2 Outline of the thesis The structure of the thesis is divided into six chapters. In chapter 2, a theoretical insight into perovskite oxyhydrides will be given, focusing on structural properties and syntheses procedures. Chapter 3 will give a brief and general description of the characterisation methods, as chapter 4 goes into specific detail about the exact experimental procedures. In chapter 5, all results will be presented and discussed, followed by a short conclusion of these results in chapter 6. 3 1. Introduction 4 2 Perovskite oxyhydrides This chapter aims to provide an understanding of the theoretical basis of the project, giving insight into oxide & oxyhydride perovskite materials, synthesis procedures and related concepts. 2.1 General structure and properties Oxyhydrides refer to a class of materials in which some of the oxygen (O2−) from an oxide has been replaced by hydride ions (H−), making it a mixed anion compound [9]. Examples of early synthesised oxyhydrides are e.g. Ba3AlO4H [15] and TiHxOy [16]. A very common type of oxyhydrides is transition metal oxyhydrides, where the cations consist of transition metals which normally have varying oxidations states. The varying oxidation states allow for hydride incorporation, since when substituting O2− for H− (or vacancy), the transition metal atom will be reduced to a lower oxidation state [17]. Hayward et al. were among the first to report a transition metal oxyhydride in 2002, LaSrCoO3H0.7 [18]. The material class of special interest in this report are perovskite transition metal oxyhydrides. ”Perovskite” refers to the specific crystal structure of the material, meaning the arrangement of the atoms in the crystal. The classic perovskite oxide nomenclature is ABO3, making the perovskite oxyhydride ABO3−xHy. A is typically a larger alkaline earth metal cation such as Ba2+ or Sr2+, whereas B is a smaller transition metal cation with variable oxidation states e.g. Ti or Co cations. Figure 2.1 displays a typical cubic ABO3 perovskite structure 5 2. Perovskite oxyhydrides Figure 2.1: Top: Example of ABO3 perovskite structure. Green, light blue and red correspond to A site, B site and O respectively. Bottom: Visualisation of twelve-fold and six-fold oxygen coordination around A and B respectively. Images were generated using VESTA software. The cubic perovskite unit cell is not uniquely defined, as it can be constructed in different ways, such as by placing either the A-site or B-site cation at the centre of the cell. For visualisation, the unit cell in Figure 2.1 is used as reference. Positioning the A cation at the centre places the B-site cations at the corners of the cubic unit cell. Each B-site cation is coordinated octahedrally by six oxygen atoms. As a result, each A-site cation is coordinated by twelve oxygen atoms from eight neighbouring BO6 octahedra. When reducing the perovskite, both hydride substitution and vacancy formation is possible. 6 2. Perovskite oxyhydrides Figure 2.2: Top: ABO3−xHy structure where a portion of the oxygen atoms have been replaced with hydride ions. Bottom: Vacancy formation, where some oxygen has been removed from the perovskite structure. Images were generated using VESTA software. In Figure 2.2, the oxyhydride and vacancy-filled perovskite phases are shown re- spectively. Oxygen vacancy formation induces a greater reduction of the transition metal cations, as the removal of O2− gives two excess electrons. In contrast, oxy- hydride formation results in a lower degree of reduction, as the introduction of H− partially compensates the charge imbalance. Although it is theoretically possible to form only one type, upon reduction of a perovskite oxide with H− available, it is likely that both oxygen vacancies and hydride incorporation occur. For BaTiO3, this has been confirmed [2]. The ABO3 perovskite oxides already exhibit interest- ing properties, such as ferroelectricity (BaTiO3) and ferromagnetism (SrRuO3) [19]. Depending on the specific atomic composition, they can be utilized in various areas, including catalytic nitric oxide (NO) conversion reactions and more [20]. Reduction of these perovskite oxides by hydride substitution or vacancy formation alters the electronic band structure of the material [21]. Thus by adjusting the anionic sites and hence the electronic structure, one can enhance existing (electronic, magnetic) properties or introduce new ones. For example, with H− and O2− having different charges, one can change the oxidation state of the metal atoms, adjust how electrons are arranged around the transition metal centers, and tune inter-cation couplings [21]. 7 2. Perovskite oxyhydrides 2.2 Oxyhydride synthesis Process Oxidation Side products Vacancy, 1e H− → H• + e 0.5(O2− + H2) Vacancy, 2e H− → H+ + 2e OH− Vacancy, H2 H• → H+ + e 0.5H2O Oxyhydride formation 2H− → H− + H• + e 0.5(O2− + H2) Table 2.1: Hydrogen processes during hydride reduction, adapted from [2]. There are some different ways of synthesising perovskite oxyhydrides. Generally, synthesis of oxyhydrides is a difficult procedure since one would expect the oxygen and hydrogen to form water under normal conditions. Very reducing conditions are required, making one of the most commonly used methods to reduce the oxide via topochemical metal hydride reduction. Metal hydride reduction is performed by solid state reaction at moderate temperatures. The oxide and metal hydride are mixed together using e.g. ball mill or mortar and pestle. This is normally done in a glove box with inert gas, since metal hydrides are reactive toward H2O and O2, causing them to be unstable in air. To avoid any air exposure, the oxide and metal hydride are enclosed in a sealed container, such as a vacuum sealed glass tube or sealed stainless steel capsule. Reaction occur at, for solid state reactions, moderate temperatures (around 400-600◦ C). The exact reaction mechanisms are not clear and obvious, however the reaction as such is, principally, about removing oxygen in the oxide and replacing them with hydride from the metal hydride. This is a process that has been explored for some time, where the first discovery of a mixed oxyhydride solid of LaHO was made already in 1982 [22]. For perovskite oxides, this synthesis process was first properly investigated and performed by Kobayashi et al. in 2012 [9]. It was discovered that, under strongly reducing conditions with CaH2 as reducing agent, one can substitute the O2− for H− in BaTiO3. Reduction of BaTiO3 using CaH2 is a reaction that will be heavily investigated in this project. This can take place in a number different of ways [2]. Vacancy formation - one electron process: BaTiO3 + x CaH2 → BaTiO3−x + x CaO + x H2 (2.1) Vacancy formation - two electron process: BaTiO3 + x CaH2 → BaTiO3−2x + x Ca(OH)2 (2.2) Vacancy formation, from H2: BaTiO3 + x H2 → BaTiO3−x + x H2O (2.3) Oxyhydride formation, direct substitution: BaTiO3 + x CaH2 → BaTiO3−xHx + x CaO + x 2 H2 (2.4) 8 2. Perovskite oxyhydrides Oxyhydride formation, via vacancy intermediate: BaTiO3−x + x 2 H2 → BaTiO3−xHx (2.5) For reduction of BaTiO3 with CaH2, the notation of nH will be used. This corre- sponds to BaTiO3 + nH, with H = 0.5 CaH2 (2.6) It refers to the molar ratio of H− to BaTiO3. The notation is used to be consistent with previous articles and to facilitate the comparison of reduction with other metal hydrides (e.g. LiH, NaAlH4), as it is mainly the hydride amount that is important rather then amount of metal hydride. An alternative and somewhat less complex method of synthesising (perovskite) oxyhydrides is by reduction with only H2 gas flow. In 2022, Toriumi et al. managed to reductively hydrogenate a 50 % indium- substituted barium-zirconate perovskite oxide to its reduced oxyhydride derivative via H2 annealing at 800◦ C [12]. The reaction is BaZr0.5In(III)0.5O2.75 + 0.25 H2 → BaZr0.5In(II)0.5O2.25H0.5 (2.7) The authors hypothesise that this process occurs in two consecutive reactions. First, In is reduced from +3 to +1 via BaZr0.5In(III)0.5O2.75 + 0.5 H2 → BaZr0.5In(I)0.5O2.25 + 0.5 H2O (2.8) followed by oxidative hydrogenation BaZr0.5In(I)0.5O2.25 + 0.5 H2 → BaZr0.5In(II)0.5O2.25H0.5 (2.9) This reaction gives the final product oxyhydride BaZr0.5In(II)0.5O2.25H0.5, in which indium is found to adopt a +2 oxidation state rather than a mixed-valence state of In(I) and In(III) [12]. This is somewhat unexpected, considering that indium has the valence electron configuration (4d105s25p1) and would therefore be expected to favour either the +1 or +3 oxidation states, corresponding to the loss of one or three electrons, respectively. The parent oxide of BaZr0.5In0.5O2.75 has been rigorously studied because of its proton conductivity in wet atmosphere (H2O) [23, 24] . One possible way for synthesising a mixed metal cation oxide with In and Zr cations by solid state reaction is BaCO3 + (1−x) ZrO2 + x 2 In2O3 → BaZr1−xInxO3− x 2 + CO2 (2.10) where x corresponds to the fraction of indium. As can be seen from the general reaction formula, In(III) incorporation inherently brings oxygen vacancy formation. For every two In(III) cations that are integrated in the oxide, one oxygen (O2−) must be removed to compensate for the charge imbalance. Consequently, the indium substituted zirconium oxide has natural oxygen vacancies in the perovskite structure which increase as the fraction of indium increases. Ahmed et al. have synthesised these oxides with 0 ≤ x ≤ 0.75 [24], meaning In substitution is possible over a large range. For an ideal 50 % (x=0.5) and 70 % (x=0.7) substitution of In, the reactions are 9 2. Perovskite oxyhydrides BaCO3 + 0.5 ZrO2 + 0.25 In2O3 → BaZr0.5In0.5O2.75 + CO2 (2.11) and BaCO3 + 0.3 ZrO2 + 0.35 In2O3 → BaZr0.3In0.7O2.65 + CO2 (2.12) respectively. 2.3 BaTiO3 and BaTiO3−xHy 2.3.1 Structure and properties As mentioned, Kobayashi et al. managed to synthesise the first perovskite oxyhy- dride of BaTiO3 in 2012 [9]. These BaTiO3 perovskites can form cubic or tetragonal crystal structure depending on temperature, meaning that in the tetragonal phase one lattice parameter is different from the two other, as opposed to the cubic phase where all three lattice parameters are equal. At room temperature, BaTiO3 is tetragonal. In tetragonal BaTiO3, one lattice parameter is marginally larger then the other two. Figure 2.3: Left: Local coordination geometry around Ti in cubic BaTiO3. Right: Local coordination geometry around Ti in tetragonal BaTiO3. Bond lengths are shown for cubic and tetragonal phase respectively [1]. Images were generated using VESTA software. This effect is illustrated in Figure 2.3. In the tetragonal phase, the Ti ion is displaced along the c-axis, leading to an asymmetry in its position and breaking inversion symmetry [11]. This displacement alters the Ti–O bond lengths and the O–Ti–O bond angles, resulting in an overall distortion of the lattice parameters compared to the undistorted cubic phase of BaTiO3. It is this structural non-centrosymmetric displacement of Ti4+ in the tetragonal BaTiO3 structure that gives the material a permanent dipole moment and hence spontaneous electric polarisation, causing it to be ferroelectric [11]. As mentioned, BaTiO3 exhibits different structural per- ovskite phases depending on temperature. It’s rhombohedral at low temperatures, orthorhombic around 183 K, normally tetragonal near 278 K, and eventually tran- sitions to a cubic phase once heated above 396 K [11]. In the oxide, O2− coordinate octahedrally around the Ti(IV) cations. The oxyhydride form of BaTiO3, that is BaTiO3−xHy, has a mix of O2− and H− ions in the octahedral environment around 10 2. Perovskite oxyhydrides the Ti ions. As a result of the change in charge balance, Ti exist in a mixed oxidation state of Ti(III/IV) [9]. Ti(III) has one valence d-electron (3d1) as opposed to Ti(IV) who has an empty 3d shell (3d0). This extra electron increases the ionic radius of the Ti ion leading to an expansion of the perovskite structure, increasing the lattice parameters. This changes the materials long-range average crystal symmetry, and causes the oxyhydride material to be of cubic phase at all temperatures, as opposed to its parent oxide [11][25]. The lattice expands not only as a result of Ti cation radius, but also because the H− radius is larger then the O2−, and the Ti-anion bond gets weaker [26]. Inclusion of the electron changes the colour of the material from white to blue. There is some controversy whether this colour change is due to the extra electron creating a polaron (localised electrons) or whether it’s delocalised in the electronic band. Schrader et al. suggested the formation of polarons [27], while more recently, Granhed et al. found that the extra electron introduces delocalised electronic states in the band gap structure [28]. 2.3.2 Previous synthesis reports Metal hydride reduction of BaTiO3 was first performed by Kobyashi et al. where they reported the formation of oxyhydride BaTiO2.4H0.6 [9]. However, synthesis conditions were explored much more extensively by Nedumkandathil et al. in 2018 [2]. The article from 2018 thoroughly explored BaTiO3 reduction with regards to use of different metal hydrides in different molar ratios as well as different heating times for samples with constant molar ratio. Nedumkandathil et al. found several interesting results with regards to reduction with different molar ratio of CaH2. The following results are collected from [2], used as a basis for comparison nH Product phase/fraction (w%) Lattice parameters (Å) x from TG 0 (BaTiO3) tetragonal a = 3.9964(1), c = 4.0310(1) 0 0.2 tetragonal a = 3.9971(1), c = 4.0260(1) 0.03 0.6 cubic 4.0051(6) 0.10 1.2 cubic-I/89(1) 4.0096(2) 0.24 cubic-II/11(1) 4.0219 1.8 cubic-I/89(1) 4.0138(1) 0.34 cubic-II/11(1) 4.0288 Table 2.2: Synthesis results reported from Nedumkandathil et al. for CaH2 re- duction of BaTiO3 heated at 600◦ C for 48 hours, table adapted from the original source [2]. Based on x (for BaTiO3−xHy) and the lattice parameters increasing for higher CaH2 molar ratio, they find that the extent of reduction increases as nH increases. This trend is consistent for nH = [0.6 ; 1.2 ; 1.8], however for nH = 0.2 they still see a tetragonal phase and attain x = 0.03 from TGA, indicating little reduction. For 1.2 H and 1.8 H they identify the formation of two separate cubic phases. One majority phase (89 %) with smaller lattice parameter, and one minority phase (11 %) with a larger lattice parameter. This differs from the 0.6 H sample with only one cubic phase, seemingly corresponding to the majority phase of the samples with higher 11 2. Perovskite oxyhydrides CaH2 concentration. They also report a deepening of the colour from blue to dark blue/black, for increasing nH [2]. This is in agreement with reports from Kobyashi et al. [9]. With regards to heating time, they studied a sample of 1.2 H for 1, 2, 4 and 7 days respectively. Heating time (days) Product phase/fraction (w%) Lattice parameters (Å) x from TG 1 cubic-I/91(2) 4.0093(2) 0.24 cubic-II/9(2) 4.0219(1) 2 cubic-I/87(2) 4.0079(1) 0.24 cubic-II/13(2) 4.0205(1) 4 cubic-I/89(2) 4.0094(1) 0.26 cubic-II/11(2) 4.0221(1) 7 cubic-I/70(1) 4.0173(1) 0.51 cubic-II/28(1) 4.0275(1) Ti3O/2(1) Table 2.3: Synthesis results reported from Nedumkandathil et al. for CaH2 re- duction of BaTiO3 heated at 600◦ C with 1.2 H for different heating times, table adapted from the original source [2]. For all heating times of the 1.2 H samples, a mixture of two phases of cubic BaTiO3 oxyhydride is reported. The 1, 2- and 4 day experiment yield similar results, with approximately similar weight% distribution between the phases, lattice parameters and x values. For the 7-day experiment, the results show a much higher extent of reduction. The reports of a Ti3O phase suggest onset of decomposition of BaTiO3 [2]. 12 2. Perovskite oxyhydrides 2.4 BaZr1−xInxO3−x 2 and BaZr1−xInxO3−x 2 −yHz : For the oxide BaZr1−xInxO3− x 2 structure, Ahmed et al. found that all samples, for 0 < x < 0.75, possessed a cubic symmetry in the perovskite structure [24]. Figure 2.4: Left: Cubic perovskite structure of barium indate zirconate, with some oxygen vacancies as In3+ partially replaces the Zr4+. Right: Visualisation of partial indium substitution at zirconium atomic positions. Images were generated using VESTA software. As seen in Figure 2.4, the substitution of Zr4+ for In3+ cations brings oxygen vacan- cies. For every two In3+ ions that are incorporated, one oxygen vacancy is formed. These already existing oxygen vacancies should facilitate hydride ion incorporation. Ahmed et. al. also found that the size of the lattice parameters increase linearly with increased indium concentration, which comes as a result of the larger ionic radius of In3+ compared to Zr4+ [24]. Recently, in 2022, Toriumi et al. presented a oxyhydride of a 50% indium substituted zirconium oxide (BaZr0.5In(II)0.5O2.25H0.5) [12]. They could demonstrate good H− ion conductivity and also explored the crystallographic sites at which hydride ions were incorporated into the structure. Using neutron diffraction, they found that hydride could be incorporated into oxygen vacancy sites (3d) but also at interstitial sites close to [100] face centre sites (demonstrated below). Figure 2.5: Left: Hydride incorporation at oxygen 3d sites. Right: Hydride incorporation at interstitial [100] face centre position. Images were generated using VESTA software. Thanks to their high H− ion conductivity and relatively simple synthesis procedure, this barium indate-zirconate perovskite oxyhydride show potential for use in ceramic electrolysis cells and membrane reactors for processes like ammonia synthesis, CO2 hydrogenation, and methane conversion [12]. 13 2. Perovskite oxyhydrides 14 3 Methods To characterize the oxyhydrides, powder X-ray diffraction (PXRD) and thermogravi- metric analysis (TGA) was used. This chapter aims to provide theoretical insight into these characterization techniques, explaining what they are and how they work. Inelastic neutron scattering (INS) measurements was conducted on selected barium indate-zirconate samples, and is hence briefly described in this section. 3.1 Powder X-ray diffraction In chemical synthesis there is an evident need of characterising your samples. This characterisation is performed for many reasons, but mainly to find out whether the synthesis has been successfully completed or not. For solid state chemistry synthesis, specifically if your materials are crystalline, powder X-ray diffraction (PXRD) is perhaps the most important and necessary technique. It allows one to determine information regarding the crystal structure such as phases (single, multi) and lattice parameters. 3.1.1 Bragg’s law The theoretical basis of PXRD lays in the fact that electron densities around atoms in a material can diffract electromagnetic waves (EM), such as X-rays, that pene- trate through the sample. For an amorphous (no long-range order) material, this diffraction will be disordered/destructive. If the sample material atoms are of an longe range ordered and symmetric nature, that is crystalline, the diffraction of EM waves can occur constructively. Constructive interference can only occur if the dis- tance between two parallel, direction-specific atomic planes is exactly equal to an integer multiple of the wavelength of the X-ray [29]. In order to visualise why this is true see Figure 3.1. 15 3. Methods Figure 3.1: Constructive interference of X-rays. Shall the X-rays interact constructively, they must reach the detector at the same point across the wave, meaning the waves are parallel once they reach the detec- tor. For this to happen, the extra distance that the more penetrative wave travels needs to be an integer multiple of the wavelength. The extra distance can be deter- mined using simple trigonometry relations utilizing the interplanar spacing d and the incident angle of the X-ray θ. This reasoning ultimately leads to Bragg’s law 2d sin(θ) = nλ (3.1) Where 2d sin(θ) corresponds to the extra distance travelled [29]. 3.1.2 Rietveld refinement and diffraction profile effects A powerful method to extensively analyse obtained diffraction data is the Rietveld refinement method. It can be used for both neutron diffraction and powder diffrac- tion, where the focus will be on PXRD applications. The method is based on an iterative fitting of calculated diffraction patterns based on structural and instrumen- tal parameters, to experimental diffraction data [30]. By utilizing a reasonably good initial approximation of some parameters, such as unit cell dimensions and atom co- ordinates, one can refine these parameters (+ other parameters) to calculate a better fit to the experimental data, and hence obtain more accurate crystallographic in- formation for their samples. When used correctly, with good initial guesses and high quality data, it can give reliable information about e.g. lattice parameters, phase quantities, crystal sizes and micro-strain. The goal for a Rietveld refinement is to minimize the difference between the observed diffraction pattern and the calcu- lated pattern generated from a structural model, also known as the difference curve. Rietveld refinement utilizes a non-linear least square method to do this, where a function M is defined as M = ∑ i Wi(Iobs i − Icalc i )2 (3.2) where Iobs i and Icalc i correspond to intensity data for the experimental and calculated profiles respetively, and Wi is a statistical weight factor [31]. The Icalc i is itself defined 16 3. Methods as a function of pn. Icalc i = f(p1, p2, ...pn) (3.3) Icalc(pn) is non-linear and depend on transcendental functions (sin, cos etc.) and can hence not be solved for directly [31]. One has to Taylor expand Icalc(pn) and then solve a set of matrix equations iteratively to minimise the function. The exact mathematical derivations are not of focus here, but can be found in e.g. [32]. Lattice parameters, crystal phases and crystallite size are some examples of parameters that are refined. Differences in size of lattice parameters primarily affects diffraction peak positions while, for example, crystallite size affects the peak shape. Smaller crystal- lites generally result in broader diffraction peaks due to increased uncertainty in the position of lattice planes. This broadening arises because a fewer number of repeat- ing planes in the individual crystals limits the extent of constructive interference between scattered X-rays [33], as described by the Scherrer equation τ = Kλ βcos(θ) (3.4) where the important thing to note is that if the mean size of the crystalline domains (τ) decrease, the full-width half maximum (β) increase [34]. K is a dimensionless shape factor. Aside from contributions originating from actual relevant structural characteristics, the need for such a refinement method also comes from sample and instrument parameters that give unwanted contributions to the diffraction and detection of X- rays, and hence affects the appearance of the diffraction data. These parameters can impact background contributions, give shifts in peak positions and affect the line profile (the shape of the peaks). Background contributions, for XRD patterns with well-resolved peaks, are normally estimated by either linear interpolation be- tween selected points between peaks or by using empirical/semi-empirical functions to model the contributions [35]. For more complex patterns, with e.g. significant overlap between peaks, the background estimation becomes less straight forward and may require more advanced modelling approaches. Shifting in peak positions (2θ) mostly originates from sample parameters such as surface roughness or sample dis- placement. Sample displacement refers to whether the sample position, in relation to the incident X-ray beam and detector, is displaced vertically. Sample displacement correction is described differently depending on the geometry of the instrument. The standard geometry used in most PXRD instruments is the Bragg–Brentano geometry, for such instruments the sample displacement can be described with ∆2θ = −2s cos(θ) R (3.5) where s is the displacement of the sample and R is the radius of the goniometer circle [35]. Optimally, the diffraction peaks in PXRD would appear as δ functions. However, instrumental and sample parameters cause peak broadening. Effects from instru- mental parameters generally give Gaussian line profile contributions, while sample effects inflicts a Lorentzian line profile [31]. 17 3. Methods G(∆2θ, Γ) = √ 4ln2 πΓ2 e −4ln2(∆2θ)2 Γ2 (3.6) L(∆2θ, γ) = 2 πγ · 1 1 + (2·∆2θ γ )2 (3.7) where Γ and γ are related to the peaks full-width half-maximum (FWHM) and half-width at half-maximum (HWHM) [31, 35]. These effects are best described using peak shape functions (PSF), where two of the most common ones is Voigt and Pseudo-Voigt. Voigt PSFs are defined as the convolution of G(∆2θ, Γ) and L(∆2θ, γ) V (∆2θ, Γ, γ) = G(∆2θ, Γ) ⊛ L(∆2θ, γ) (3.8) while Pseudo-Voigt PSFs is represented as a linear combination of the Gaussian and Lorentzian contribution. Vp(∆2θ, Γ, γ) = ηL(∆2θ, γ) + (1 − η)G(∆2θ, Γ) (3.9) In the pseudo-Voigt function, η is the mixing parameter that determines the relative contribution of the Lorentzian and Gaussian components. It can take values between 0 and 1, where η = 0 corresponds to a purely Gaussian peak shape and η = 1 corresponds to a purely Lorentzian peak shape. 3.2 Thermogravimetric analysis Thermal analysis consist of a wide range of methods in which the properties of a material is studied as a function of temperature. One such analytic technique which will be utilized is Thermogravimetric analysis (TGA). TGA is utilized in order to de- termine how the mass of a substance varies with temperature. The sample is placed in a furnace and subjected to progressive temperature changes (normally around 20◦ C/min) while the mass is simultaneously measured [36]. Consequently the raw data received can be used to plot the sample substance mass as a function of tempera- ture, which in turn can be used to study e.g. phase transitions or decompositions of the material. For this project, TGA measurements will be of importance in order to study the degree of reduction for the synthesised oxyhydride samples. By heat- ing the reduced samples in air, reoxidation of the hydride- or vacancy-containing perovskite structure occurs, allowing quantification of the extent of reduction. The mass gain in reaction BaTiO3−x + O2 → BaTiO3 is monitored to determine this. 3.3 Inelastic neutron scattering By directing a beam of neutrons toward a solid state sample, letting them interact inelastically with the atom nuclei, one can obtain information about the dynamic 18 3. Methods properties of the material, such as atomic/molecular vibration modes [37]. This characterisation technique is called inelastic neutron scattering (INS). When neu- trons hit sample nuclei, there needs to be an exchange in both momentum (Q̄) and energy (E) in order for an neutron scattering event to be inelastic. This means that ∆E = Ei − Ef ̸= 0 (3.10) ∆Q̄ = k̄i − k̄f ̸= 0 (3.11) where Ei and Ef denotes neutron energy before and after the scattering event, and k̄i and k̄f correspond to the neutron wave vector before and after the scattering event. This is different from neutron diffraction where the neutrons scatter elastically (∆E = 0) and only momentum transfer occur. Most importantly for this project, INS is sensitive to hydrogen following that hydrogen has a large incoherent neutron scattering cross-section [38]. Hydrogen nuclei scatter neutrons strongly, allowing for hydrogen to be detected using INS. This is in contrast to PXRD which relies on electron density, making hydrogen very difficult to detect. 19 3. Methods 20 4 Experimentals This chapter will address and describe the experimental procedures, including all steps of the syntheses as well as the characterization measurements. 4.1 Oxide Synthesis In order to proceed with the synthesis of actual oxyhydrides, the precursor oxides had to be synthesised. This was performed via solid state reaction of crystalline powders, specific for each oxide material type. For barium titanate, this was done according to BaCO3 + TiO2 → BaTiO3 + CO2 (4.1) For barium zirconate-indate BaCO3 + (1−x) ZrO2 + x 2 In2O3 → BaZr1−xInxO3− x 2 + CO2 (4.2) Where x is the amount of successfully substituted indium. During the mixing pro- cess, a few millilitres of high-purity (>99.5%) ethanol were added. The addition of liquid ethanol significantly improves the mixing of the solid powders, ensuring sufficient contact between the precursor grains and promoting a uniform mixture. This uniformity should result in a higher reaction yield and a more homogeneous oxide upon heating the mixture. The synthesis includes heating the powders in alumina crucibles using a box furnace. This procedure includes pressing the powder into a pellet and utilizing the remaining powder to cover the top and bottom side of the pellet in the crucibles. Pellet pressing compacts the powder, increasing proximity of grains and subsequently bettering the conditions for forming the desired oxide. It was performed using a pellet press and belonging die set, forming 20 mm diameter pellets. Cover powder is used as extra safety, to reduce any risk of contaminating the pellet. After heating, only the pellet itself is used as sample. 4.1.1 BaTiO3 Amounts of reactants were chosen to form approximately 5g of barium titanate product, where 4.2313 g and 1.7125 g of BaCO3 (>99%, Sigma-Aldrich) and TiO2 (> 99%, Sigma-Aldrich, Anatase -325 mesh) was weighed respectively to attain equal molar proportions of BaCO3 and TiO2 (nBa ≈ nT i). The precursors were mixed together and grinded using an agate mortar and pestle. 21 4. Experimentals The grinding process was carried out for approximately 15-20 minutes with around 5 ml of ethanol (>99.5%) added in the beginning. An additional 3 ml ethanol was added in the middle of the mixing once the original 5 ml had evaporated. The powder mixture was heated in a furnace at 900◦ C for 8 hours. The powder mixture was then ground again for the same duration with same amounts of ethanol, where about 70-80 % of the content was pressed into a pellet and the remaining powder was used to cover the bottom and top of the pellet. The crucible was subsequently heated for 48 hours at 1200◦ C, optimally forming the desired product, BaTiO3. The cover powder was disposed, and the pellet of BaTiO3 was crushed into a fine powder for XRD measurements and oxide reduction with CaH2. Before placed in the glovebox, the powder was heated at 500◦ C to discharge any adsorbed water content. 4.1.2 BaZr1−xInxO3− x 2 For 50% In substituted barium-zirconate, amounts of reactants were chosen to form approximately 15 g of barium zirconate-indate product. 9.014 g, 2.8147 g and 3.1703 g of BaCO3 (>99%, Sigma-Aldrich ), ZrO2 (99%, Sigma-Aldrich) and In2O3 (Alfa Aesar, 99.9%) was weighed respectively to give molar proportions of barium, zirco- nium and indium according to nBa ≈ 0.5 nZr + 0.5 nIn. Using agate mortar and pestle, the three precursor powders were grinded and mixed together. Roughly 5 ml of ethanol (>99.5%) was added and the mix was grinded for 10 minutes. Then, a smaller amount of ethanol was added (2-3 ml) and the mix was grinded for an additional 10 minutes. The powder mix was transferred to two Al2O3 (alumina) crucibles and heated in a furnace for 8 hours at 1000◦ C. After heating, the powder was once again mortled in the same fashion as previously mentioned (10 + 10 minute grinding and addition of ethanol). Around 70% of the powder was then pressed in to a pellet and placed in a alumina crucible. The remainder of the powder was placed at the bottom and top of the pellet (with pellet in-between), in order to avoid contamination/unwanted diffusion issues. The crucible, containing pellet and cover powder, was then placed in the furnace for a second time, this time for 72 hours at 1200◦ C. After heating, the pellet and powder was crushed. The intent was to only crush the pellet and re-use the same cover powder for the third heating, however this was not viable as the powder had ”morphed” with the pellet. Consequently the entire content of the crucible (pellet + powder) was mortled together with a few ml of ethanol. A pellet was made once again and powder was placed at the bottom and top of the crucible. The crucible was placed in the furnace for the third time, this time for 48 hours at 1325◦ C. Upon removal from the furnace, the cover powder could be separated from the pellet using sandpaper, allowing only the pellet to be crushed. The pellet was crushed into a fine powder for XRD measurements and oxyhydride synthesis. The oxide for 70% substituted barium zirconate was synthesised in the same exact fashion, using the same precursors to give molar proportions of barium, zirconium and indium according to nBa ≈ 0.3 nZr + 0.7 nIn. 22 4. Experimentals 4.2 Oxyhydride Synthesis In total, three ways of synthesising an oxyhydride was tried. The most frequent and most examined was that of the BaTiO3 reduction with metal hydride CaH2, con- sequently heating it in an enclosed stainless steel capsule filled with argon. Metal hydride reduction was performed with both synthesised (microcrystalline) and in- dustrially manufactured nanocrystalline BaTiO3, purchased from Sigma-Aldrich. For synthesised BaTiO3, it was also attempted to conduct metal hydride reduction with CaH2 under simultaneous (reducing) H2 gas flow. For BaZr1−xInxO3− x 2 the oxide was heated in H2 flow, using the hydrogen gas as reducing agent to form the oxyhydride. 4.2.1 BaTiO3−xHy and nano-BaTiO3−xHy A series of measurements with the synthesised BaTiO3 and CaH2 (Sigma-Aldrich, ≥ 97.0%) was made with regards to different molar ratios, more specifically nH = [0.5; 1.0; 1.5; 2.0]. An arbitrary mass of barium titanate was decided for each sample, and the correct stoichiometric amount of calcium hydride was calculated based on that. nH / Precursor amount BaTiO3 (g) CaH2 (g) 0.5 0.4999 0.02256 1.0 0.4005 0.0362 1.5 0.3006 0.0406 2.0 0.2995 0.0535 Table 4.1: Weighed precursor amounts for different molar ratios of synthesised BaTiO3 and CaH2. All synthesis steps using CaH2 were conducted in a glove box filled with argon (O2 < 0.1 ppm, H2O = 0.6 ppm ), to avoid exposure to air and water (see Figure 4.1). Stoichiometric amounts (see table 4.1) of both precursors were separately weighed and subsequently crushed and mixed using agate mortar and pestle. The grinding took place for a longer time compared to the oxide synthesis, about 30-40 minutes, without any ethanol. 23 4. Experimentals Figure 4.1: Glovebox with argon (99,999%, Linde), used for the mixing of CaH2 and BaTiO3. After grinding, the powder mix was placed directly into a metal capsule using a spat- ula and enclosed tightly at both ends by Swagelok caps. In the glovebox, wrenches were used to strongly close the caps, hence ensuring that the sample is surrounded by argon with very low amounts of air entering the capsule. The capsule containing the sample was then transferred to a furnace to be heated. Figure 4.2: Stainless steel capsule (after use) used for BaTiO3−xHy synthesis, shown both individually and placed inside the furnace. All samples in the series were heated at 600◦ C for 48 hours. For the first two samples, nH = [1.0 ; 1.5], the entire metal capsule was placed inside a box furnace. However this method was abandoned since the swageloks were extremely difficult to open after heating, most likely as a result of the metal expanding at high temperatures. This problem was avoided for samples with nH = [0.5 ; 2.0] by using a clam shell type tube furnace with open ends, along with longer metal capsules. Thus allowing heating of one side of the capsule (containing the powder), leaving the other end outside the furnace at room temperature. Consequently, the swageloks were much easier to open. 24 4. Experimentals Generally, the different furnace setups are not believed to affect the results signif- icantly. Temperature and time were closely monitored to try to assure as equal conditions as possible. However, a smaller metal capsule does theoretically lead to higher H2 pressure in the container, which could potentially affect the reducing conditions. For subsequent syntheses, the tube furnace setup was used. At this stage, an unwanted phase of barium orthotitanate (Ba2TiO4), and highly reduced Ti3O was detected. In an attempt to remove this impurity, an additional synthesis was performed for the synthesised BaTiO3 where the powder was placed in a aluminium oxide tube inside the metal capsule. This was done to investigate whether the contact between the steel and powder caused this impurity, by e.g. Ti diffusing through the metal. Synthesis in alumina tube was performed with nH = 2.0, mBaT iO3 = 0.3000 g and mCaH2 = 0.0547 g The same synthesis procedure was conducted for samples of industrially manufac- tured nano-crystalline (< 100 nm) BaTiO3 (≥ 99%, Sigma-Aldrich). Samples with nH = [1.0 ; 2.0] were grinded and heated at 600◦ C for 48 hours. Based on trial filtration of the synthesised BaTiO3 (described later in this section), an increased amount of precursor material was used to attain more sample product in later fil- tration steps. nH / Precursor amount nano-BaTiO3 (g) CaH2 (g) 1.0 1.0007 0.0903 2.0 1.0005 0.1810 Table 4.2: Weighed precursor amounts for different molar ratios of nano-BaTiO3 and CaH2. The Ba2TiO4 phase was also apparent for these nano-BaTiO3−xHy samples. In a further attempt to remove the impurity, a sample with nH = 2.0 was heated at a lower temperature of 580◦ C to lower reducing conditions. This was carried out with mBaT iO3 = 1.0003 g and mCaH2 = 0.1809 g. Also, the same sample type was heated for only 24 hours instead of 48 to try and remove the impurity. In order to see if the increased powder amounts for the nano-BaTiO3 samples had an impact on the synthesis results, a sample was prepared using similar amounts of precursor to those of the synthesised BaTiO3 samples. This was mainly with regards to a suspicion of a two-phase formation, indicated by a slight asymmetry in the XRD peak line profiles. There was an attempt to perform CaH2 metal reduction of the synthesised BaTiO3 samples in H2 flow. The powder mix was placed in an Al2O3 boat inside the glovebox, enclosing it in a small box filled with argon in order to minimise air exposure during transfer of the sample. A separate tube furnace was used for this, which allowed for flow of hydrogen gas. While argon was flowing, the sample was quickly transferred from the small argon- filled box and placed in the furnace (with argon flowing simultaneously). Argon flow was continued at 25 ml/min in an attempt to purge all oxygen before increasing temperature and turning on H2. After five days of purging, argon flow was lowered to 10 ml/min and H2 flow was started at 8 ml/min. These conditions were held for 2 hours. Argon flow was then stopped, while H2 was continued at 8 ml/min and 25 4. Experimentals heating started. The sample was heated for 48 hours at 600◦ C, same as most of the samples heated in metal capsules. Reduction of BaTiO3 with CaH2 typically produces substantial amounts of calcium oxide (CaO) in the reaction, hence the samples of oxyhydride BaTiO3−xHy had to be washed and filtered in order to discharge any remaining CaO. For washing, each nano-BaTiO3−xHy sample was added to a beaker with 50 ml, 0.1 M solution of ammonium chloride (NH4Cl, >99.5%, Riedel-de Haën) in methanol (>99.9%). For the synthesised BaTiO3−xHy, 25 ml 0.1 M NH4Cl solution was used due to the smaller amount of powder in those samples. The solution was stirred for two hours, using magnetic stirrers. The samples were vacuum filtered using a Buchner flask and funnel, along with a vacuum pump. After filtration, the powder was left on the filter paper to dry for a long duration (15-20 hours). NH4Cl reacts with CaO according to NH4Cl + CaO → NH3 + CaCl2 + H2O (4.3) Forming CaCl2, which is highly soluble in methanol and can consequently be filtered away after washing with pure methanol [2]. 4.2.2 BaZr0.5In0.5O2.75−xHy and BaZr0.3In0.7O2.65−xHy Both the 50 % and 70% barium indate-zirconate oxyhydrides were synthesised in a very similar fashion. For both BaZr0.5In0.5O2.75 and BaZr0.3In0.7O2.65, the sample powder was placed and spread evenly onto a Al2O3 crucible ”boat”. The crucible was then placed in a quartz tube in a tube furnace. In the gas flow synthesis, Ar (99.999%, Linde) and H2 gas (99.999%, Linde) were used. For BaZr0.5In0.5O2.75, gas flow experiments were conducted according to the following flowchart: Ar flow was set at 20 ml/min for 24 hours at 600◦ C. Ar flow was then lowered to 10 ml/min and H2 flow was initiated at 8 ml/min, while starting a temperature increase to 800◦ C. After 1 hour, when 800◦ C set temperature was reached, Ar was stopped completely and H2 flow was continued at 8 ml/min. Heating was stopped 24 hours later, with H2 flow still on. Once temperature had reached circa 150◦ C, H2 was switched off and Ar was turned on at 15 ml/min. With room temperature being reached, Ar was turned off and sample was removed from the furnace. For BaZr0.3In0.7O2.65, gas flow experiments were conducted according to the following flowchart: Heating was initiated in Ar at 25 ml/min. 2 hours later, when temperature had reached 650◦ C, Ar was lowered to 10 ml/min and H2 was initiated at 5 ml/min. Ar flow was then stopped, as H2 was increased to 8 ml/min. Heating was stopped 20 hours later. With room temperature being reached, H2 was stopped and Ar was turned on at 25 ml/min to purge after which the sample was removed from the furnace. The samples was exposed to argon gas during the heating in order to dehydrate the sample, as OH− ions incorporates into position of oxygen vacancies [39]. 26 4. Experimentals 4.2.3 PXRD measurements PXRD measurements were conducted using a Bruker D8 Discover diffractometer with a Cu Kα radiation source [40]. All powder samples were measured in variable slit on a zero background, monocrystalline silicon holder, in the range of 2θ = [10- 90◦]. The obtained powder diffractograms were evaluated using EVA software and GSAS-II [41, 42]. EVA was used for general analysis and to give initial insights about the synthesis results. Phase analysis of the PXRD diffractograms was performed in the software, and was used to confirm/strongly indicate whether the synthesis was successful and to identify impurity phases in the samples, comparing the obtained peaks to a crystallography database in the software. GSAS-II software was used for Rietveld refinement analysis in order to analyse the PXRD diffractograms in greater detail. The refinement was performed using theoretical/calculated structures obtained from CIF files and standard instrumental files for a Cu Kα source instrument. Crystallographic information was calculated for the samples, mainly for phase identification and to determine lattice parameters. In the refinement, the background was described as a weighted sum of Chebyshev polynomials. An auto-background was computed to set fixed points in the back- ground, then the background (described by the sum of Chebyshev polynomials) was refined to fit the set background points. Generally, 12-14 polynomials with dif- ferent weight coefficients were used to accurately describe the background. Aside from the background, all refinements were performed with sample displacement, sur- face roughness, unit cell parameters, crystallite size, micro-strain and atomic mean square displacement (U) as refinement parameters. Pseudo-Voigt PSFs were used to describe the peak shapes. From the refinement, only the values for lattice parameters are of interest. Param- eters such as crystallite size and strain were refined to improve the line profile fit, but the resulting values are not physically meaningful. For them to be meaningful one would need pre-determined instrument parameters from the specific PXRD in- strument, which describe the Gaussian and Lorentzian instrumental contributions to the line profile. The refinement performed in this project used a standard Cu Kα instrument file. 4.2.4 TGA measurements Before TGA measurements, all CaH2 reduced BaTiO3 samples were vacuum-dried for 7 hours at 50◦ to discharge any remaining methanol from the washing process. The instrument for TGA measurements was a Mettler TGA/DSC 3+ [43]. 70 µl aluminium oxide crucibles were used, containing around 10-20 mg of sample powder. Experiments were run in air in a temperature range from 30-900◦ C, with a ramp rate of 10 ◦ C/min. An empty crucible was measured and subtracted from all obtained TGA sample curves, to compensate for background noise originating from the crucible. Hence all crucibles were assumed to give approximately the same background contribution. 27 4. Experimentals 4.2.5 INS measurements The sample of BaZr0.5In0.5O2.75−xHy was sent to the ISIS Neutron and Muon Source, located at the Rutherford Appleton Laboratory in Oxfordshire, United Kingdom, for inelastic neutron scattering (INS) measurements using the TOSCA spectrometer. Measurements were carried out on both the sample and an empty reference to deter- mine background noise. Further details regarding the facilities and the instrument can be found from [44, 45]. 28 5 Results and discussion There are various interesting comparisons and analyses to be made. This chapter will present the results of the syntheses, mainly on the basis of XRD and TGA measurements. 5.1 Synthesis of BaTiO3 In order to confirm the formation of oxide BaTiO3 in the synthesis and also the phase (tetragonal/cubic), one can simply perform a Rietveld refinement and evaluate how well the calculated pattern fits the experimental diffraction data. Figure 5.1: Left: Rietveld plot of synthesised BaTiO3. Reference COD ID: 1507756, obtained from Crystallography Open Database. Right: Peak split at 2θ = 45.4o, displaying tetragonal phase of BaTiO3. Data was plotted in Matlab. As depicted in the Rietveld plot to the left in Figure 5.1, the experimental pattern match up well with the reference of tetragonal BaTiO3. This is mainly visualised by the turquoise residual, displaying the difference in intensity values of the experimen- tal and reference pattern. The synthesised BaTiO3 is in tetragonal phase, which is shown in the right panel Figure 5.1. This corresponds to the {200} peak. In tetrag- onal BaTiO3, as a result of c > b = a, the (002) direction will have a slightly larger lattice parameter, causing the intensity corresponding to that direction to shift to 29 5. Results and discussion larger 2θ. This causes the peak split where around 1/3 of the peak intensity shifts. For a cubic phase this peak would be more symmetric. By further utilizing the Rietveld analysis, the lattice parameters can be compared. Lattice parameter (Å) BaTiO3 synthesised a = 3.9984(1), c = 4.0257(1) BaTiO3 reference a = 3.9999, c = 4.0170 Table 5.1: BaTiO3 lattice parameters for synthesised and reference. Reference COD ID: 1507756, obtained from Crystallography Open Database. Table 5.1 shows that the c lattice parameter is larger for the synthesised BaTiO3 compared to the reference, while the a lattice parameter is marginally smaller. While it appears that the synthesised oxide is slightly more distorted to tetragonal phase, overall the lattice parameters are very comparable. The reference originates from an article which exploring synthetic methods for BaTiO3, however the exact synthesis conditions corresponding to the reference phase used in this analysis are not specified [46]. 5.2 Synthesis of BaTiO3−xHy The result analysis of the BaTiO3−xHy oxyhydride will rely heavily on determining the extent to which the BaTiO3 oxide has been reduced. By analyzing the PXRD spectra, including Rietveld refinement analysis, alongside the TGA data, strong indications can be obtained regarding the degree of reduction. As stated in the theory section, the lattice parameter of the perovskite is a strong reflection of the extent of reduction for BaTiO3. A larger lattice parameter indicates a more reduced sample, this will be important for the analysis. All samples presented in this section showed significant colour change from white to dark blue/black, suggesting reduction to some extent. Figure 5.2: Colours of products from reduction of synthesised BaTiO3 for different concentrations of CaH2, all heated at 600◦ C for 48 hours. From left to right, samples are reduced with molar ratios of CaH2 corresponding to nH = [2.0 ; 1.5 ; 1.0 ; 0.5]. Figure 5.2 shows a gradient-like colour change, where the colour of the reduced products go from black/dark blue to a lighter blue when CaH2 concentration de- creases. As mentioned in the experimental section, one series of CaH2 reduction of 30 5. Results and discussion synthesised BaTiO3 was made with different molar ratios of CaH2. PXRD pattern for each BaTiO3 sample heated at 600◦ C for 48 hours, can be shown in Figure 5.3 Figure 5.3: PXRD pattern of products from reduction of synthesised BaTiO3 for different concentrations of CaH2, all heated at 600◦ C for 48 hours. Peak intensities are normalised to the most intense peak around 2θ = 31.5o. Measured in variable slit. Plotted in Matlab It is not straightforward to deduce conclusions just looking at Figure 5.3 directly. However, zooming in on one of the peaks reveals some information. Figure 5.4: Zoomed images of [200] PXRD peak around 2θ = 45.4o. Plotted in Matlab Looking at Figure 5.4, a general indication can be noted regarding the degree of reduction. Comparing the oxide (black) with the reduced samples (various colors) in the plot, all nH samples have a shift toward lower 2θ, indicating an increase in lattice parameter consistent with a higher degree of reduction [2]. Samples with higher concentrations of CaH2 exhibit a more pronounced shift toward lower 2θ angles, indicating a greater extent of reduction. This is not a direct conclusion, 31 5. Results and discussion as there are multiple other sample and instrumental parameters that can affect the shift in 2θ. In order to more accurately determine the extent of reduction, one needs to use Rietveld refinement to determine the lattice parameters of the samples. By looking at the line profiles in Figure 5.4, there is a clear difference between the nH samples and the parent oxide. As previously discussed and visualised in Figure 5.1, the oxide shows a tetragonal line profile. None of the reduced samples exhibits this however, as they show a more sharp and symmetrical line profile, characteristic for a cubic phase. To analyse this further, below are shown the Rietveld plots for all nH = [0.5 ; 1.0 ; 1.5 ; 2.0] samples, using a cubic BaTiO3 phase as reference. 32 5. Results and discussion Figure 5.5: Rietveld plots of products from reduction of synthesised BaTiO3 for different concentrations of CaH2. The blue crosses, green line, and turquoise line correspond to the observed data, calculated fit, and difference curve, respectively. Orange squares mark BaCO3 impurities (1.0 H, 2.0 H), green diamond symbols mark orthorhombic Ba2TiO4 impurities (2.0 H). The black cross likely corresponds to a monoclinic Ba2TiO4 phase, however this is unconfirmed. As can be observed directly from the difference curves in the Rietveld plots, they show good agreement between the observed and calculated patterns. There is min- imal deviation between the observed data and the calculated curve, indicating a good fit. This can be further analysed by comparing refinement quality indicators, Rwp and χ2. 33 5. Results and discussion nH Rwp χ2 0.5 4.451 % 2.56 1.0 4.923 % 3.28 1.5 3.863 % 1.58 2.0 5.144 % 2.14 Table 5.2: Refinement quality indicators for CaH2 reduced samples of synthesised BaTiO3 Values of Rwp vary between 3.863% and 5.144% while χ2 values vary from 1.58 to 3.28, showing reasonably good fits for all samples. One Rietveld refinement was performed for the 2.0 H sample, using a tetragonal phase BaTiO3 as reference. Figure 5.6: Rietveld plot of 2.0 H sample for CaH2 reduction of synthesised BaTiO3 sample, with tetragonal BaTiO3 as phase reference. Obtained lattice parameters are shown in the Figure. As can be seen from Figure 5.6, the refinement generates lattice parameters of a = 4.03276 Å, c = 4.03558 Å, further suggesting the formation of a cubic phase. The values are close to each other, and also close to the lattice parameter of a = 4.03376 Å for the 2.0 H sample obtained from refinement with a cubic reference, reported in table 5.3. The lattice parameters before and after washing, obtained from Rietveld analysis, can be shown in Table 5.3 34 5. Results and discussion nH Lattice parameter(s) (Å), pre-wash Lattice parameter(s) (Å), post-wash 0 (BaTiO3) a = 3.9984(1), c = 4.0257(1) 3.9984(1), c = 4.0257(1) 0.5 4.00891(7) 4.00880(5) 1.0 4.0075(1) 4.00705(7) 1.5 4.0262(2) 4.0269(1) 2.0 4.0333(4) 4.03376(9) Table 5.3: BaTiO3−xHy lattice parameters before and after washing for different molar ratios nH , all heated for 48 hours at 600◦ C. BaTiO3 wasn’t washed. In agreement with previous research, the lattice parameter increase for higher con- centrations of CaH2 [2]. This is likely due to higher concentrations of CaH2 leading to greater reduction of the oxides, which increases the abundance of Ti3+. Since Ti3+ has a larger ionic radius than Ti4+, this results in an expansion of the lattice parameter. The trend is relatively consistent, with the exception of the nH = 1.0 sample having a smaller lattice parameter then for nH = 0.5. Minor differences be- tween the values before and after washing can be observed, however these are small and considered negligible. The trend of increasing lattice parameters for higher concentrations of CaH2 can be visualised when plotting lattice parameter against nH , Figure 5.7: Lattice parameter dependence on CaH2 molar ratio for CaH2 reduction of synthesised BaTiO3. Linear fit was performed using polyfit function in Matlab. Looking at Figure 5.7, there is a linear trend if one disregards the 1.0 H sample. However using three data points while omitting one is not definitive. It is likely that for appreciably smaller or larger molar ratios, the data would deviate from this trend. It remains unclear as to why the 1.0 H sample has a smaller lattice parameter than the 0.5 H sample. This result is not expected as higher concentration of metal hydride should reduce BaTiO3 more extensively, as reported from [2]. Nedumkan- dathil et al. also reports formation of two separate cubic phases, one minority phase 35 5. Results and discussion with a slightly larger lattice parameter than that of the majority phase. Two-phase formation should give asymmetry in the peak shape, as two phases of different lat- tice parameter size will generate two peaks (overlapping). Below is shown the peak shape of the peak around 2θ = 56o for nH = [1.0 ; 2.0]. Figure 5.8: Left: Peak shape of 1.0 H CaH2 reduced BaTiO3 sample. Right: Peak shape of 2.0 H CaH2 reduced BaTiO3 sample. Red rectangles indicate areas of interest. When performing Rietveld refinement using only one cubic phase as reference, a two-phase formation should become apparent by an asymmetry in the peak shape. In Figure 5.8, the peak shape area of interest is shown with red rectangles. Neither 1.0 H or 2.0 H show any clear suggestions of two phases, as the calculated curve fits the line profile of the experimental data well. However, the 1.0 H sample does show a slight tendency toward what would be expected for multi-phase samples. The calculated fit (green) slightly underestimates the intensity on the left side of the experimental data (blue crosses), compared to the 2.0 H sample where the profile shows near complete symmetry with regards to this. That said, the difference is too small to attribute confidently to the presence of two phases and could instead be explained by uncertainties in the refinement process, or by instrumental and sample effects. Nedumkandathil et al. report much more apparent peak broadening from the additional phase [2]. If an additional phase is present, its weight fraction is likely small. Peaks for samples of 0.5 H and 1.5 H show similar peak shapes to those presented in Figure 5.8. By comparing lattice parameters obtained in table 5.3 with the ones reported from Nedumkandathil et al. in table 2.2, the results suggest that the values obtained are comparative to those of the minority (cubic-II) phase. All reduced BaTiO3 samples were washed with 0.1M NH4Cl methanol solution to remove CaO. Before washing, for samples of nH = [0.5 ; 1.0 ; 1.5 ; 2.0], only CaO and Ba2TiO4 were identified as impurities by XRD. Also Ti3O was identified for the 2.0 H sample. The Ba2TiO4 is mainly in orthorhombic phase, while one low intensity peak around 2θ = 30.3o seems to correspond to the monoclinic phase. For visualisation, XRD patterns for the 2.0 H sample (sample with most intense impurity peaks) before and after washing are shown below. 36 5. Results and discussion Figure 5.9: Top: XRD patterns for CaH2 reduced 2.0 H of synthesised BaTiO3 samples before washing (blue curve) and after washing (red curve). Pink triangles, green diamonds, red circles and orange squares show presence of CaO, Ba2TiO4 (monoclinic + orthorhombic), Ti3O and BaCO3 respectively. Black crosses repre- sent eliminated peaks in the XRD pattern. Bottom: Peaks for orthorhomic and monoclinic phase of Ba2TiO4 in unwashed 2.0 H sample. Images plotted in Matlab. Figure 5.9 clearly shows an effective removal of CaO, as all CaO diffraction peaks are eliminated after washing. This goes for all samples nH = [0.5 ; 1.0 ; 1.5 ; 2.0]. Amounts of Ba2TiO4 (barium orthotitanate) are heavily reduced for all samples, even though some small amounts remain (mainly for the 2.0 H sample). After washing, visible BaCO3 peaks appear for 1.0 H and 2.0 H samples, while 0.5 H and 1.5 H do not show clear signs of BaCO3 formation. Considering Ba2TiO4 amounts are reduced and BaCO3 appears, some type of process where Ba2TiO4 is converted to BaCO3 is viable. Felgner et al. report of such a process in room temperature with CO2, in the presence of atmospheric moisture [47]. Ba2TiO4 + H2O → Ba(OH)2 + BaTiO3 (5.1) Ba(OH)2 + CO2 → BaCO3 (5.2) Marks et al. claim this reaction is possible for both monoclinic and orthorhombic Ba2TiO4 [48]. However results from Felgner et al. show that the orthorhombic phase 37 5. Results and discussion undergoes this transformation to a greater extent [47]. From Figure 5.9, the two peak intensities around 29o of the orthorhombic phase are heavily reduced after washing, while the 30.5o peak of the monoclinic phase remains largely unchanged. These findings are therefore consistent with those reported by Felgner et al [47]. Felgner et al. examined this reaction in air saturated with water vapour for 65 hours, which could provide more favourable reaction conditions. In contrast, it seems unlikely that the same extent of reaction would occur under ambient air during a two-hour methanol washing step. Also, there is no sign of formation of an additional BaTiO3 phase, which the reaction formula suggest. Amounts of Ba2TiO4 impurity seem to significantly increase for higher concentra- tions of CaH2. Figure 5.10: Comparison of Ba2TiO4 peak intensities around 28.5-30.5o for reduced synthesised BaTiO3 for different nH = [0.5 ; 1.0 ; 1.5 ; 2.0]. All peak intensities are normalised to the highest peak around 2θ = 31.5o, and the background level is set uniformly across all patterns. Images plotted in Matlab Samples of 0.5 H and 1.0 H show noticeably less amounts compared to samples of 1.5 H and 2.0 H, as seen in Figure 5.10. This trend is expected, as more strongly reducing conditions are likely to favour the formation of Ba2TiO4. It is likely that the Ti gets over-reduced, leaves the perovskite structure and forms e.g. Ti3O which formation is confirmed by the peak in Figure 5.9 (marked with red circle). Amounts of Ba2TiO4 before washing is not consistent with amounts of BaCO3 formed after washing for the samples of different nH . For example, the 1.5 H sample has significantly more intense diffraction peaks from Ba2TiO4 (seen in Figure 5.10) than the 1.0 H sample, while BaCO3 impurity can only be seen for the 1.0 H sample and not for 1.5 H. To draw more conclusions concerning the extent of reduction of synthesised BaTiO3 reduced with different molar ratios of CaH2, TGA measurements are highly relevant. Below is shown TGA curves for respective sample with nH = [0.5 ; 1.0 ; 1.5 ; 2.0]. 38 5. Results and discussion Figure 5.11: TGA data between 35-900◦ C for products of CaH2 reduction of BaTiO3 for 48 hours at 600◦ C for different molar ratios of CaH2. Black dotted lines correspond to approximate weight difference reference lines before and after oxidation (likely not accurate as discussed below). Images plotted in Matlab. Generally, the TGA curves in Figure 5.11 show deviation from expected thermal be- haviour of the samples. There is a significant weight decrease for all samples around 35-300◦ C, followed by an expected weight increase caused by oxidation of the re- duced BaTiO3 around 350-600◦ C (different temperature ranges for different CaH2 concentrations). An additional weight decrease is seen toward the higher tempera- ture range, near 600◦ C, which seems to stop around 800◦ C. Some of the mass loss at lower temperatures could be attributed to surface water or possibly remaining methanol evaporating, even though they were vacuum dried before measurement. Any mass loss after oxidation at temperatures over 500-600◦ C is completely unex- pected for BaTiO3, and is likely caused by some decomposition process originating from impurities in the sample. The only impurities confirmed from XRD are BaCO3 and Ba2TiO4, where BaCO3 decomposes into BaO and releases CO2 according to BaCO3 → BaO + CO2 (5.3) however this decomposition normally occurs at temperatures around 900-1000◦ C [49]. Also, PXRD diffraction peaks from BaCO3 are not found for all samples, which further suggest that the cause for the mass decrease at high temperatures is not due to BaCO3 decomposition. To gain further knowledge into the features of the TGA 39 5. Results and discussion curves, the 2.0 H sample was measured in PXRD after the TGA measurement. Figure 5.12: XRD patterns for 2.0 H sample of CaH2 reduced BaTiO3 before (blue) and after (red) TGA measurement. Black crosses represent unidentified phase(s), which peaks only appeared post-TGA. Once again orange squares, green diamonds and red circles represent BaCO3, Ba2TiO4 and Ti3O respectibely. Plotted in Matlab It is clear from Figure 5.12, that unknown phase(s) are formed during the TGA mea- surement, suggested by the emergence of new diffraction peaks (shown by the black crosses). This heavily indicates a decomposition process which could explain mass- decreases in the TGA curves. Noticeably, diffraction peaks from BaCO3 and Ti3O disappear after the measurement, suggesting that these phases undergo some sort of decomposition/reaction. However, the formed impurity phase has not been iden- tified. Multiple Ba–Ti–O phases were considered and compared against databases in EVA software, in spite of this, none of them matched well with the unidentified peaks observed in the post-TGA XRD pattern. It also remains possible that some amorphous impurity, undetectable by XRD, may contribute to the observed mass- decreasing processes. This is especially relevant considering BaCO3, Ba2TiO4 and Ti3O are not identifiable for all molar ratios of CaH2. From the wider and more asymmetric peaks in the post-TGA XRD pattern (red), it can be confirmed that the reduced perovskite BaTiO3−xHy is oxidised back to its parent tetragonal BaTiO3 oxide. The two (or more) mass-decreasing processes cause the quantification of x (in BaTiO3−xHy) to be difficult and unreliable, as they likely overlap with the mass- increasing oxidation step. Yet, x was calculated based on the black-dashed lines in Figure 5.11. 40 5. Results and discussion nH Lattice parameter (Å) x from TGA (only vacancies / only hydride ions) 0.5 4.00880(5) 0.106 / 0.113 1.0 4.00705(7) 0.159 / 0.170 1.5 4.0269(1) 0.125 / 0.133 2.0 4.03376(9) 0.463 / 0.494 Table 5.4: Apparent x values for different molar ratios of CaH2, from products of CaH2 reduction of BaTiO3. Calculations of x were made considering only oxygen vacancies in the perovskite, as well as only hydride substitution of oxygen. Lattice parameters were included for comparison. As expected, values of x are not consistent with determined lattice parameters of the samples. In general, a larger lattice parameter is expected to correlate with a higher x value, as both serve as indicators of the extent of reduction, as reported by Nedumkandathil et al. in Table 2.2. For example, the lattice parameter for the 1.5 H sample of 4.0269(1) Å should bring a higher x value compared to the 1.0 H sample with a lattice parameter of 4.00705(7) Å. This is not observed for the obtained values in Table 5.4, where x is higher for 1.0 H (x ≈ 0.16) compared to 1.5 H (x ≈ 0.13). As mentioned, the deviating and non-optimal TGA curves give unreliable values of x, causing the lattice parameter to be a better indicator of the degree of reduction. Worth noting however, is the significant reduction of the 2.0 H sample, reflected by high values of both lattice parameter size and x. Samples from attempts of reduction with CaH2 under H2 flow showed no colour change, or any indications of reduction from PXRD measurements. Likely, this is due to air not being purged away efficiently in the tube furnace, causing CaH2 to react with oxygen gas and water in the air instead of reducing BaTiO3. 5.3 Synthesis of nano-BaTiO3−xHy CaH2 reduction using manufactured nano-BaTiO3 (cubic phase) was performed to study the impact of BaTiO3 crystallite size in the reduction, while also briefly in- vestigating some effects of heating time and temperature, partially in relation to the Ba2TiO4 impurity. All samples presented in this section showed notable colour change from white to dark blue/black, suggesting reduction to some extent. Firstly, similar to results for synthesised BaTiO3 reduction with CaH2, two samples with molar ratio nH = [1.0 ; 2.0] were synthesised. PXRD patterns of these, along with the parent nano-BaTiO3 oxide is shown below. 41 5. Results and discussion Figure 5.13: PXRD patterns of products from reduction of nano-BaTiO3 for two different concentrations of CaH2, nH = 1.0 (blue) and nH = 2.0 (red). Orange squares and green diamond symbols represent BaCO3 and Ba2TiO4 impurities re- spectively. Peak intensities are normalised to the most intense peak around 2θ = 31.5o. Measured in variable slit. Plotted in Matlab. In general there is little difference in the XRD patterns in Figure 5.13, where peak shapes and peak positions are similar for the reduced samples and the oxide. Notable is the BaCO3 impurity for the nano-BaTiO3 oxide, which seem to disappear for the reduced, washed samples of 1.0 H and 2.0 H. From Figure 5.13, quite significant peak broadening can be observed for the perovskite peaks, which is expected from the small crystallite size of nano-BaTiO3. By zooming in on the peaks more closely, this can be examined more closely. Figure 5.14: Showing of peak broadenings and absence of 2θ shift between the samples peak positions for XRD patterns from reduction of nano-BaTiO3 for two different concentrations of CaH2, nH = 1.0 (blue) and nH = 2.0 (red). Plotted in Matlab. 42 5. Results and discussion As opposed to for CaH2 reduced samples of synthesised BaTiO3, no apparent 2θ shift trend can be noted from Figure 5.14, suggesting less degree of reduction. To the right in the figure, a zoomed image of the peak around 56.3o displays the wide peaks obtained from XRD. As mentioned, this is expected due to smaller crystallite size results in fewer number of repeating planes in the crystals, limiting the extent of constructive interference between scattered X-rays [33]. Peak shapes for the samples show little asymmetry, indicating cubic phases for the parent oxide and the reduced samples. Additionally, the label of the industrially manufactured nano- BaTiO3 stated it was in a cubic phase. Initial Rietveld refinements of the samples generated following values of lattice parameters (performed with a cubic reference for all samples, including the oxide). nH Lattice parameter (Å) 0 (nano-BaTiO3) 4.0088(1) 1.0 4.0081(1) 2.0 4.0114(1) Table 5.5: Lattice parameters for CaH2 reduced nano-BaTiO3 for nH = [1.0 ; 2.0], heated at 600◦ C for 48 hours. Rietveld refinements were performed with a cubic BaTiO3 as reference for the oxide, which is discussed below. The results from Table 5.3 raised questions. It is unlikely that the 1.0 H reduced sample would have a smaller lattice parameter then that of the unreduced oxide. All reports suggest that a reduction of BaTiO3 should yield a larger lattice parameter, and if the oxide is not reduced at all the lattice parameter should remain unchanged [9, 2]. Hence, a refinement was performed for the nano-BaTiO3 using a tetragonal BaTiO3 phase as reference. Results from this refinement, as well as values for the original cubic refinement are shown in the table below. Phase of nano-BaTiO3 Lattice parameter(s) (Å) Rwp χ2 Cubic 4.0088(1) 5.759 % 2.85 Tetragonal a = 4.0049(1), c = 4.0180(1) 4.700 % 1.90 Table 5.6: Comparison of lattice parameters and refinement quality indicators for Rietveld refinement of nano-BaTiO3 with cubic and tetragonal reference. Table 5.6 strongly indicates that the precursor nano-BaTiO3 is in tetragonal phase with a = 4.00493 Å and c = 4.01801 Å. This is shown by the improved refinement quality indicators Rwp and χ2, when performing the refinement with a tetragonal reference. The reason for this not showing in the line profile is likely due to the peak broadening caused by the crystallite size, as the asymmetry is not visible. Rietveld plots are shown for the oxide and samples of nH = [1.0 ; 2.0] in following figures. 43 5. Results and discussion Figure 5.15: Rietveld plots for refinements of nano-BaTiO3, performed with tetrag- onal reference (left) and cubic reference (right). The blue crosses, green line, and turquoise line correspond to the observed data, calculated fit, and difference curve, respectively Confirmed from the difference curve (turquoise) in Figure 5.15, tetragonal phase reference generates a better fit. Figure 5.16: Rietveld plots for refinements of products from CaH2 reduction of nano-BaTiO3 at 600◦ C for 48 hours, with concentrations of 1.0 H (left) and 2.0 H (right). The blue crosses, green line, and turquoise line correspond to the observed data, calculated fit, and difference curve, respectively. Obtained lattice parameters from the correct tetragonal refinement in Figure 5.15, as well as for the reduced samples in Figure 5.16 are shown below. nH Lattice parameter(s) (Å) Rwp χ2 0 (nano-BaTiO3) a = 4.0049(1), c = 4.0180(1) 4.700 % 1.90 1.0 4.0081(1) 5.724 % 3.94 2.0 4.0114(1) 4.519 % 2.08 Table 5.7: Lattice parameters and refinement quality indicators for CaH2 reduced nano-BaTiO3 for nH = [1.0 ; 2.0], heated at 600◦ C for 48 hours. An expected higher lattice parameter for the 2.0 H sample compared to the 1.0 H sample can be seen in Table 5.7. When comparing to previous synthesis results 44 5. Results and discussion from from Nedumkandathil et al., the lattice parameter (for 2.0 H primarily), seem to match that of the majority phase reported from [2]. Effects of CaH2 molar ratio on degree of reduction was explored quite extensively in the result section of synthesised BaTiO3, and is hence not be of focus in this sect