Structural Basis for Hydrogen Interaction in Selected Metal Hydrides

Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1234

JONAS ÅNGSTRÖM

ACTA UNIVERSITATIS UPSALIENSIS ISSN 1651-6214 ISBN 978-91-554-9187-1 UPPSALA urn:nbn:se:uu:diva-245046 2015 Dissertation presented at Uppsala University to be publicly examined in Häggsalen, Ångströmlaboratoriet Lägerhyddsvägen 1, Uppsala, Friday, 24 April 2015 at 09:15 for the degree of Doctor of Philosophy. The examination will be conducted in English. Faculty examiner: Fermin Cuevas (East Paris Institute of Chemistry and Materials Science, CNRS).

Abstract Ångström, J. 2015. Structural Basis for Hydrogen Interaction in Selected Metal Hydrides. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1234. 74 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-554-9187-1.

Metal hydrides have existing and potential uses in many applications such as in batteries, for hydrogen storage and for heat storage. New metal hydrides and a better understanding of the behaviour of known metal hydrides may prove crucial in the realisation or further development of these applications. The aims of the work described in this thesis have been to characterise new metal hydrides, investigate how the properties of known metal hydrides can be improved and understand how their structure influences these properties. Metal hydrides, in most cases synthesised via high-temperature techniques, were structurally characterised using X-ray powder diffraction, X-ray single crystal diffraction and neutron powder diffraction and their thermodynamic and kinetic properties by in-situ X-ray powder diffraction, thermal desorption spectroscopy and pressure-composition-temperature measurements. The investigations showed that: the storage capacity of the hexagonal Laves phase Sc(Al1- xNix)2 decreases with increasing Al content. There is a significant decrease in the stability of the hydrides and faster reaction kinetics when Zr content is increased in the cubic Laves phase Sc1- xZrx(Co1-xNix)2. Nb4M0.9Si1.1 (M=Co, Ni) form very stable interstitial hydrides which have very slow sorption kinetics. MgH2 mixed with 10 mol% ScH2 reaches full activation after only one cycle at 673 K while it takes at least four cycles at 593 K. LnGa (Ln=Nd, Gd) absorb hydrogen in two steps, it is very likely that the first step is interstitial solution of hydride ions into Ln4 tetrahedra and the second step places hydrogen atoms in Ln3Ga tetrahedra. The nature of the Ga-H bond is still unclear.

Jonas Ångström, Department of Chemistry - Ångström, Box 523, Uppsala University, SE-75120 Uppsala, Sweden.

ISSN 1651-6214 ISBN 978-91-554-9187-1 urn:nbn:se:uu:diva-245046 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-245046) “Jag känner mig däremot såsom transformist nästan förpliktigad antaga endast ett enkelt ämne, ur vilket de andra genom förtunning, förtätning, kopulation, korsning och så vidare uppståt, och detta utan att vilja nämna ur-ämnet vid namn, kanske inte en gång vid namnet väte.” August Strindberg (Antibarbarus, 1905)

List of papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I A. Sobkowiak, J. Ångström, T. K. Nielsen, Y. Cerenius, T. R. Jensen and M. Sahlberg. “Hydrogen absorption and desorption properties of a novel ScNiAl alloy” Applied Physics A 104 1 (2011) 235-238 DOI: 10.1007/s00339-010-6116-z

II M. Sahlberg, J. Ångström, C. Zlotea, P. Beran, M. Latroche and C. P. Gómez. “Structure and hydrogen storage properties of the hexagonal Laves phase Sc(Al1−x Nix)2” Journal of Solid State Chemistry 196 (2012) 132-137 DOI: 10.1016/j.jssc.2012.06.002

III J. Ångström, R. Johansson, L. H. Rude, C. Gundlach, R. H. Scheicher, R. Ahuja, O. Eriksson, T. R. Jensen and M. Sahlberg. “Hydrogen storage properties of the pseudo binary Laves phase (Sc1−xZrx)(Co1−yNiy)2 system” International Journal of Hydrogen Energy 38 23 (2013) 9772-9778 DOI: 10.1016/j.ijhydene.2013.05.053

IV D. Korablov, J. Ångström, M. B. Ley, M. Sahlberg, F. Bessenbacher and T. R. Jensen. “Activation effects during release and uptake of MgH2” International Journal of Hydrogen Energy 39 18 (2014) 9888-9892 DOI: 10.1016/j.ijhydene.2014.02.112

V J. Ångström, C. Zlotea, M. Latroche and M. Sahlberg. “Hydrogen-sorption properties of Nb4M0.9Si1.1 (M=Co,Ni) hydrides” International Journal of Hydrogen Energy 40 6 (2015) 2692-2697 DOI: 10.1016/j.ijhydene.2014.12.031

VI J. Ångström, R. Johansson, O. Balmes, M. H. Sørby, R. H. Scheicher, U. Häussermann and M. Sahlberg. “Hydrogen absorption in rare-earth-gallide Zintl-phases LnGa (Ln=Nd, Gd)” manuscript

Reprints were made with permission from the publishers. The author also contributed to the following published and/or accepted scholarly works which are not included in this thesis:

i S. Kamali, N. Shahmiri, J. J. S. A. Garitaonandia, J. Ångström, M. Sahlberg, T. Ericsson and L. Hägerström. “Effect of mixing tool on magnetic properties of hematite nanoparticles prepared by sol–gel method” Thin Solid Films 534 (2013) 260-264 DOI: 10.1016/j.tsf.2013.03.009

ii S. Frykstrand, C. Strietzel, J. Forsgren, J. Ångström, V. Potin and M. Strømme. “Synthesis, electron microscopy and X-ray characterization of oxymagnesite, MgO2·2MgCO3, formed from amorphous magnesium carbonate” CrystEngComm 16 (2014) 10837-10844 DOI: 10.1039/C4CE- 01641F

iii V. Höglin, J. Ångström, M. S. Andersson, O. Balmes, P. Nordblad and M. Sahlberg, “Sample cell for in-ﬁeld X-ray diffraction experiments” Results in Physics 5 (2015) 53-54 DOI: 10.1016/j.rinp.2015.01.006

iv J. Ångström, M. Sahlberg and R. Berger, “Real-time in-situ monitoring of the topotactic transformation of TlCu3Se2 into TlCu2Se2” Journal of Alloys and Compounds accepted (2015) DOI: 10.1016/j.jallcom.2015.02- .180 My contributions to the papers

The authors contribution to the papers included in this thesis:

I I made some of the syntheses, took part in the conception of the project and did the in-situ diffraction experiments together with the ﬁrst author. I was involved in the writing and discussions about the manuscript. I approved the ﬁnal proof of the paper.

II I planned the work together with the main author and did most of the experimental work except the structural reﬁnements and PCT measurements. I wrote parts of the manuscript and was involved in all of the discussions. I approved the ﬁnal proof of the paper.

III I planned and executed all the experimental work and the treatment and interpretation of the resulting data. The theoretical calculations were performed by the second author. I wrote most of the manuscript and was involved in all discussions. I approved the ﬁnal proof of the paper.

IV I executed most of the in-situ experiments, took part in the treatment in the data and wrote parts of the manuscript. I approved the ﬁnal proof of the paper.

V I planned all and executed most of the experimental work and the treatment and interpretation of the resulting data. I wrote most of the manuscript and approved the ﬁnal proof of the paper.

VI I planed and executed most of the experimental work. The theoretical calculations were performed by the second author. I did the majority of the treatment of the experimental data. I wrote most of the manuscript and approved the version appended to this thesis.

Contents

1 Introduction ...... 13 1.1 Examples of Metal Hydrides ...... 13 1.1.1 Ionic Metal Hydrides ...... 13 1.1.2 Metallic Metal Hydrides ...... 15 1.1.3 Covalent Metal Hydrides ...... 17 1.1.4 Zintl Phase Hydrides ...... 17 1.2 Existing and Potential Applications of Metal Hydrides ...... 17 1.2.1 Ni–MH Batteries ...... 17 1.2.2 Hydrogen storage ...... 18 1.2.3 Thermal Storage ...... 19 1.2.4 Other Applications and Effects of Hydrogen in Metals 20

2 Aims ...... 21

3 Methods ...... 23 3.1 Synthesis ...... 23 3.1.1 Arc Melting ...... 23 3.1.2 Heat Treatment ...... 23 3.1.3 Hydrogenation/Deuteration ...... 24 3.2 Diffraction ...... 24 3.3 Characterisation by Diffraction Techniques ...... 28 3.3.1 X-ray Single Crystal Diffraction ...... 28 3.3.2 X-ray Powder Diffraction ...... 28 3.3.3 Neutron Powder Diffraction ...... 29 3.3.4 Determination of Lattice Parameters ...... 30 3.3.5 Full Pattern Reﬁnement Using the Rietveld Method .... 30 3.4 Thermodynamic and Kinetic Behaviour of Metal Hydrides ...... 31 3.4.1 Thermodynamics of Metal Hydrides ...... 31 3.4.2 Kinetics ...... 33 3.4.3 Density Functional Theory ...... 34

4 Results and Discussion ...... 37 4.1 Laves Phases Containing Sc ...... 38 4.1.1 The Hexagonal Laves Phase Sc(Al1−xNix)2 ...... 38 4.1.2 The Cubic Laves Phase (Sc1−xZrx)(Co1−yNiy)2 ...... 43 4.2 Activation Effects in MgH2 Mixed with ScH2 ...... 47 4.3 Nb4M0.9Si1.1 (M=Ni, Co) Hydrides ...... 49 4.4 Hydrogen Absorption in LnGa (Ln=Gd, Nd) ...... 54 4.5 Ti–Zr–Ni–Co approximants ...... 59 5 Conclusions and Outlook ...... 60

6 Sammanfattning på Svenska ...... 62

7 Acknowledgements ...... 65 Abbreviations

A list of the abbreviations used in this thesis.

abbreviation meaning

BCC Body Centered Cubic CCD Charge-Coupled Device CSP Concentrated Solar Power DFT Density Functional Theory FWHM Full Width at Half Maximum ICSD Inorganic Crystal Structure Database Ni–MH Ni-Metal Hydride NPD Neutron Powder Diffraction PCT Pressure Composition Temperature TDS Thermal Desorption Spectroscopy wt% weight % of hydrogen XRPD X-Ray Powder Diffraction XRSCD X-Ray Single Crystal Diffraction

1. Introduction

Metals and hydrogen have a long and intimate history: hydrogen gas was arguably first observed by Robert Boyle, when he dissolved steel filings in acid in 1671.[1] Dissolution of iron in sulphuric acid was also used to fill the balloon Örnen, in the disastrous Arctic expedition in 1897 led by Salomon A. Andrée.[2] The interaction between metals and hydrogen is, however, far from limited to the production of hydrogen gas when metals are oxidised in acidic solutions; that metals absorb hydrogen has been known since the middle of the 19th century [3]. Since then, some kind of metal hydride has been reported for a majority of the metals1; in these compounds hydrogen atoms form ionic, metallic or covalent bonds with the metal atoms (figure 1.1). The existing and potential uses of metal hydrides are mainly energy related and include: as rechargeable batteries [5] anodes, as materials for a safer and more compact hydrogen storage [6] and as materials for thermal storage in concentrated solar power [7]. A deepened understanding of how the properties of known hydrides can be enhanced as well as the discovery of new metal hydrides could prove crucial for the realisation or improvement of these and other applications.

1.1 Examples of Metal Hydrides 1.1.1 Ionic Metal Hydrides In some sense the ionic metal hydrides are the “true” metal hydrides since they actually contain the hydride ion. The hydrides of the stable group I elements are all ionic and can be seen as analogues of the halides; all crystallise in the NaCl-type structure (figure 1.2a); there is indirect evidence of a transition to the denser CsCl-structure for all but NaH and LiH under very high static pressure (figure 1.2b).[8] Most hydrides of the stable group II elements are ionic, the exception is BeH2 which has a BeH4-network structure [9]; all but one of the ionic group II hydrides crystallise in the PbCl2-type structure [10, 11, 12] while MgH2 crystallises in the rutile-type structure [13] (figure 1.2c and d). Typical for group II ionic hydrides are that they are very stable (e.g. enthalpies of forma- −1 −1 tion are 73 kJ mol for MgH2 [14] and 174 mol for CaH2 [15]).

1The Inorganic Crystal Structure Database (ICSD) contains structures for binary metal hydrides (or deuterides) for 75% of the metallic elements (excluding the transuranic elements) [4].

13 M+H-

ionic bond. met. bond. + cov. bond.

H2 M MHx

M-H Figure 1.1. Metal hydrides exhibits all three types of chemical bonding: ionic, metallic and covalent.

a) b)

c) d)

Figure 1.2. The crystal structures of the group I and II ionic hydrides. Metal ions are shown as blue spheres and hydride ions as red spheres. a) NaCl-type structure b) CsCl-type structure. c) PbCl2-type structure d) Rutile-type structure.

14 MgH2 2 MgH2 is arguably the most investigated hydride for hydrogen storage because of the high weight percent of hydrogen in the compound (7.66 wt%), low cost and the abundance of Mg in the earth’s crust. The aforementioned stability of MgH2 and a tendency to oxidise to MgO are, however, large obstacles for some applications. Additionally, the kinetics of hydrogen desorption in the absence of additives combined with ball milling are rather sluggish.[16]

1.1.2 Metallic Metal Hydrides Binary metallic metal hydrides are mainly formed by the transition metals; however, there is a general (but not universal) trend of decreasing affinity for hydrogen with increasing group number for the transition metals in the peri- odic table. In these compounds H atoms are located in interstitial sites between the much larger metal atoms. In an interstitial hydride, where the M–H bond is metallic, the stoichiometry is not limited by requirements of charge neutrality or by the formation covalent bonds. This means that the composition can vary. When the affinity for hydrogen is “just right” hydrogen can easily and quickly be cycled in and out of a material at moderate pressures and temperatures. PdHx [3] is an example of such a Goldilocks hydride, but alloying a metal with high hydrogen affinity with one with low hydrogen affinity can achieve a similar result.

ABx Interstitial Hydrides Many interstitial hydrides are binary, or pseudo-binary, combinations of a metal (or metals) with a large metallic radius (rm), A, and a metal (or metals) with a smaller rm, B. Typical combinations are AB5, AB2 and AB. The high weight of transition-metal atoms in comparison to the H atom makes the gravimetric density of hydrogen rather low (<2 wt%); however, the storage capacity per volume is often high, e.g. the volumetric density of hydrogen(deuterium) is higher in LaNi5D6.7 [17] than in D(s) [18] or H2(l) [19]. Additionally, the properties can be tuned to ﬁt an application by changing the composition.

AB Interstitial Hydrides AB or BCC3-type compounds crystallise in the W-type structure, or sometimes its more ordered cousin the CsCl-type structure4 (ﬁgure 1.2b). To form a substitution alloy with the W-type structure the atoms cannot differ much in size so these compounds are often made up of two (or more) 3d transition metals, where TiFeHx [20] is a classic example.

2see applications on page 19 3Body Centred Cubic 4Which does not have BCC symmetry!

15 a) b)

c) d)

Figure 1.3. Laves phases and the Friuaf polyhedron (green lines). The A atom is shown in a lighter shade of grey than the B atom. a) The Friauf polyhedron: one of the four AB3 tetrahedra is shown in black. b) The tetrahedral A2B2-interstitial sites are shown in white on the hexagonal surfaces of the Friauf polyhedron. c) The cubic Laves phase MgCu2 with Friauf polyhedron. d) The hexagonal Laves phase MgZn2 with Friauf polyhedron.

AB2 Interstitial Hydrides The crystal structure of the AB2-compounds (also known as Laves phases) can be constructed from Friauf-polyhedra stacked in different ways.[21] The polyhedron itself can be seen as four tetrahedra formed by the central A atom in the polyhedron and three B atoms (figure 1.3a). Interstitial hydrogen is often located in tetrahedral A2B2 voids on the faces of the Friauf-polyhedron (figure 1.3b). The two most common types of AB2 compounds are the cubic MgCu2-type (C15) and hexagonal MgZn2-type (C14) (figure 1.3c and d). The difference in size of the A and B atoms are often larger than in the AB- hydrides. A typical example of an AB2 hydride is Zr(1−y)Tiy(Cr(1−x)Nix)2Hz. [22]

AB5 Interstitial Hydrides The AB5-type interstitial hydrides crystallise in the CaCu5-type structure and can be formed by Y and a large number of lanthanides (A) combined with pe- riod four transition metals (B).[23] The structure can be rationalised as stacking of polyhedra built from AB3 tetrahedra in a similar way as in the Laves phases. The archetypical examples of AB5-hydrides are LaNi5Hx and its many variants.[24]

16 1.1.3 Covalent Metal Hydrides Metals in the p-block form covalent hydrides that are typically unstable; one example is the methane analogue stannae (SnH4).[25]

Complex Hydrides Complex hydrides are formed by positive metal ions of group I or II elements and a negatively charged complex ion; these complex ions consist of one or many p-block atoms covalently bound to H atoms with structures re- sembling hydrocarbons. Much effort has gone into studying some of these hydrides [26] because of their high gravimetric hydrogen densities, e.g. NaAlH4 with 7.5 wt% and LiBH4 with 18.5 wt%. Complicated decomposition reactions and poor reversibility have, so far, made them unsuitable for most applications.

1.1.4 Zintl Phase Hydrides Hydrides of Zintl phases are interesting since they are somewhere on the boarder between all the previously described metal hydrides. Zintl phases are formed by combining electropositive (also known as an active) alkali, alkaline earth, or rare-earth metals with electronegative p-block metals.[27] The crystal structures of these compounds can be rationalised as a cation of the active metal and a complex ion of the electronegative metal (where the complex ion is bound to fulﬁl the octet rule). Hydrides of Zintl phases are often either hydridic or complex.[28] In hydridic hydrides, hydride ions (H−) are located in interstitial sites coordinated by the active metal ions and, since the formal charge of the hydride ion is -1, this requires oxidation of the complex ion to preserve charge neutrality. In complex hydrides of Zintl phases, hydrogen atoms bind covalently to the complex ion, forming a complex hydride.

1.2 Existing and Potential Applications of Metal Hydrides Metal hydrides are used in a number of niche and mainstream applications and have been suggested for use in a number of other applications (ﬁgure 1.2). A couple of examples are provided below.

1.2.1 Ni–MH Batteries The main application of metal hydrides today is probably in Nickel-Metal Hy- dride (Ni–MH) batteries where they are used as anode materials (ﬁgure 1.2a).

17 a) b)

- Ni-MH

c) d)

Figure 1.4. Existing and potential applications of metal hydrides a) Ni–MH batteries. b) Hydrogen storage in submarines and other applications were wt% is not critical. c) Hydrogen storage for cars and other light mobile applications. d) Heat storage in CSP plants.

In a typical battery, an interstitial metal hydride (often a pseudo-binary version of LaNi5Hx) anode is combined with a NiOOH/Ni(OH)2 cathode in a KOH solution. Ni–MH batteries have lost a large part of the mobile energy storage market to Li-ion batteries during the past decade; however, they are still used in a number of applications.

1.2.2 Hydrogen storage

H2 is an attractive energy carrier because of its high gravimetric energy density and its complete lack of CO2 emissions when it is oxidised in a combustion engine or fuel cell. An alternative storage technique to high pressure tanks has been sought to (depending on the application) improve safety, system gravimetric density and/or volumetric density. Metal hydrides often require activation by a couple of hydrogenation/dehydrogenation cycles or heating in H2, before before they reach the reaction rates required for hydrogen storage applications. This process, the activation effect, is not well deﬁned but probably involves changes in the microstructure of the hydride [29] and reduction or cracking of the surface oxide. In applications where weight, and maybe also cost, is of lower impor- tance hydrogen storage in interstitial metal hydrides is attractive because of their high volumetric storage capacity and safety. One example is submarines such as the Type 212 U-boats operated by the German and Italian navies that

18 uses GfE hydralloy C [26] (a commercial AB2-type interstitial hydride material [30]) in combination with liquid oxygen storage to power the propulsions system during quiet operation (ﬁgure 1.2b).

Light Mobile Applications Finding a metal hydride suitable for hydrogen storage in light mobile applications (e.g. cars) has long been the holy grail of metal hydride research (figure 1.2c).[6] Since light weight is paramount in such an application the main focus has been on light metal hydrides which contain more than one hydrogen per metal, such as Mg-based or complex metal hydrides. It should be noted, however, that no metal hydride currently surpasses carbon fibre reinforced high pressure tanks in all relevant aspects.[26] The first commercially produced fuel-cell automobiles have recently en- tered, or will soon enter, the market. One example is Toyota Mirai [31] which will be launched in Japan, western Europe and the USA [32] this year. While these vehicles use the high pressure storage option, this may increase the in- terest in research into metal hydrides as a stand-alone storage alternative or in a combined high-pressure/metal-hydride storage solution. Another example where metal hydrides are used for hydrogen production (in this case in a non-reversible way), is the portable charger POWERTREKK 2.0 by myFC. [33] This device uses a mixture of sodium silicide, NaBH4 and Al which is mixed with water to produce H2 for a small fuel cell.

1.2.3 Thermal Storage The absorption of hydrogen in a metal is often a very exothermic process, especially for the ionic hydrides (see page 13); this can be exploited to store heat for Concentrated Solar Power (CSP) plants (ﬁgure 1.2d).[7] A possible set-up is that a very stable hydride, e.g. CaH2, is heated by CSP during the day forming Ca and H2. This increases the pressure enough so that a more labile 5 hydride , e.g. FeTi, absorbs the H2 and this absorption allows more hydrogen to be released from the stable hydride and so on:

Day : CaH2 + TiFe + heat → Ca + TiFeH2 (1.1) During the night the stable hydride is cooled down, causing it to absorb hydrogen which releases heat. The lowered pressure causes the labile hydride to release H2 which the stable hydride absorbs and so on:

Night : Ca + TiFeH2 → CaH2 + TiFe + heat (1.2) The heat released by the formation of the stable hydride can be used to produce electricity during the night.

5which is not heated but in contact with the gas released from the stable hydride

19 1.2.4 Other Applications and Effects of Hydrogen in Metals Here is a short list of additional applications/interesting properties of metal hydrides which will not be described in detail: production of high purity H2(g) by diffusion through a metallic foil [34]; increased critical temperature for superconductors after hydrogenation [35] and using the difference in optical properties between a metal and its metal hydride to produce a switchable mir- ror.[36]

20 2. Aims

The aims of this thesis were to characterise new metal hydrides and to study how the properties of known metal hydrides can be improved by changing the chemical composition and crystal structure of these compounds via substitution for and addition of elements in the materials. Relating structural characteristics to properties was an important part of this process.

Laves Phases Containing Sc AB2 or Laves phase compounds can be used in hydrogen storage and in Ni– MH batteries. The aim in this projects was to chart the extension of hexagonal and cubic Laves phases in the Sc(Al1−xNix)2 and (Sc1−xZrx)(Co1−yZry)2 systems and study how the composition and crystal structure affect the hydrogen sorption properties.

Activation In MgH2 Catalysed With ScH2 MgH2, one of the most studied potential hydrogen storage materials is ham- pered by, among other things, slow kinetics. The aim of this project was to investigate the activation effect during cycling of a 90mol% MgH2 mixed with 10mol% ScH2, a possible catalyst.

Nb4MSi (M=Co, Ni) Hydrides Nb4MSi (M=Co, Ni) have recently been shown to be superconductive and interstitial hydrogen has been known to increase the critical temperature of some superconductors. The aim of this project was to investigate hydrogen absorption/desorption of the Nb4MSi (M=Co, Ni) hydrides and possible order between the M/Si sites to facilitate further studies of the superconductive properties of these hydrides.

LnGa (Nd, Gd) Hydrides REGa compounds order magnetically at low temperatures and could potentially be used in magnetocaloric refrigeration; additionally, they are Zintl phases which can form different kinds of metal–hydrogen bonds. The aim of this project was to determine the crystal structure of the LnGa (Ln=Nd, Gd) hydrides and investigate how hydrogen is introduced into the structure as well as the nature of the metal-hydrogen bonds; this could facilitate the understanding of the magnetic properties of the REGa hydrides.

21 Ti–Zr–Co–Ni Approximants Ti–Zr–Ni quasicrystals and their related approximants have been shown to absorb large amounts of hydrogen. The aim of this project was to synthesise Ti–Zr–Co–Ni approximants and hydrogenate them.

Techniques The tools used to meet these aims were: high temperature synthesis techniques to produce samples; XRPD, NPD and XRSCD for characterisation of crystal structures as well as phase analysis and in-situ XRPD, TDS and PCT measurements for characterisation of thermodynamic and kinetic properties.

22 3. Methods

3.1 Synthesis Syntheses of samples for all but paper IV were done using the high temperature techniques described below (ﬁgure 3.1). The samples for paper IV were produced using ball milling.

3.1.1 Arc Melting An electric arc is a continuous high voltage/low current discharge between two electrodes where the gas between them is broken down into ions (a plasma which carries the current). Reasonably conductive materials can be welded or melted using such an arc. Most of the samples in this thesis were produced using an arc furnace which consisted of a water-cooled Cu hearth (electrode one) on which the sample was placed and a W rod (electrode two) both encased in an airtight chamber (figure 3.1a). To minimise the partial pressure of O2 and other reactive gases the chamber was flushed by high purity Ar and a Ti “getter” was melted prior to melting the sample. The samples were re-melted a number of times and flipped between each melting to improve homogeneity.

3.1.2 Heat Treatment The poor temperature control and high cooling rate of arc melting often neces- sitated heat treatment to improve the crystallinity and phase homogeneity of the samples. For most samples the treatment was carried out in fused silica ampoules which were evacuated during at least 0.5 h using an oil-vacuum pump. The ampoules were then sealed and placed in a pit furnace and heated to the desired temperature (figure 3.1b). If higher temperatures than 1373 K were required1 a tube furnace with SiC heating elements was used (figure 3.1c); the samples were placed in an alumina boat and heated to the desired temperature under flowing high purity Ar.

1Heat treatment in ampoules is limited by the crystallisation of the silica and the maximum temperature of the pit furnaces.

23 a) b)

c) d)

Figure 3.1. The high temperature techniques used. a) Arc melting. b) Heat treatment of samples in evacuated fused silica ampoules in a pit furnace or c) Heat treatment in an alumina ship under ﬂowing Ar(g) in a tube furnace. d) Hydrogenation or deuteration in an alumina crucible placed in a steel autoclave under H2 or D2 pressure in a pit furnace.

3.1.3 Hydrogenation/Deuteration Powdered samples were hydrogenated or deuterated in an alumina “finger” crucible which in turn was placed in a steel autoclave (figure 3.1d). The autoclave was flushed three times with the reactive gas followed by an activation of the material by heating it in the reactive gas (H2/D2 pressures of >5 MPa). Fi- nally the temperature was lowered down to room temperature (∼293-303 K).

3.2 Diffraction Using the diffraction phenomenon is one of the most effective ways of determining the crystal structure of and identifying crystalline compounds. The techniques rely on ﬁnding some process that scatters incoming waves2 elastically; if there is any kind of order in the material (short range or long range) the dispersed waves can interact in constructive and destructive ways and the patterns they form can be recorded. X-rays, which are easily produced in a laboratory, behave in this way and are therefore extensively used in diffraction experiments. Neutrons and electrons are also possible to use when accelerated to sufﬁcient velocities, in part because of the wave/particle duality.

2with wavelengths λ close to length scale of what is to be studied

24 -1b1+2b2+0b3 ks 2θ k

Figure 3.2. A two dimensional slice of the Edwald’s sphere with the b3 vector orthogonal to the ﬁgure and the reciprocal lattice points are at the intersects of the grey grid.

Diffraction Conditions An ideal crystal has an infinite array of atoms which have long range order; this order can often be described using translational symmetry with a three dimensional unit cell repeated infinitely in all directions, but there are excep- tions: e.g. quasi-crystals and incommensurate structures. Crystals with translational symmetry can be described using an infinite lattice of points with identical surroundings and by defining a unit cell with the lattice vectors a1, a2 and a3 (equivalent to lattice parameters a, b, c ,α, β and γ). It is convenient to convert these real-space vectors to reciprocal ones (b1, −1 b2 and b3), where kbnk = kank and bn is orthogonal to the two a=6 n vectors. The incident radiation is described by the vector k, where kkk=1/λ. If the incident wave is scattered elastically the vector describing the scattered wave vector ks must have the same length, i.e. kkk=kksk=1/λ. For the scattered waves to interact constructively the difference between the two vectors must be a linear combination of integers (the indices h, k and l) times the reciprocal lattice vectors, i.e. k − ks = hb1 + kb2 + lb3. This process is nicely illustrated by the Edwald’s sphere, see figure 3.2. In powder diffraction it is usually easier to think of these conditions as Bragg’s law where a reflection occurs at an angle θhkl if:

−1 2khb1 + kb2 + lb3k sinθhkl = 2dsinθhkl = λ (3.1) where d is the distance between crystallographic planes.

25 The Structure Factor The intensity of a reﬂection is ideally proportional to the absolute value of the 3 structure factor Fhkl:

n j j sinθhkl j sinθhkl 2πi(hx j+ky j+lz j) Fhkl = ∑ g t ( ) f ( )e (3.2) j=1 λ λ where n is the total number of atoms; θhkl and λ are the angle and wavelength, j j sinθhkl j sinθhkl respectively; g , t ( λ ) and f ( λ ) is the occupancy, the displacement and atomic scattering factor of the jth atom and x j, y j and z j are the fractional th j sinθhkl coordinates of the j atom. f ( λ ) is very dependent on what type of radiation is used in the experiment.

Scattering of X-rays (Thompson Scattering) One way that X-rays can interact with matter is that the electrons start to undu- late “on” the electro-magnetic wave. This makes each electron emit a spheri- cal wave with the same wavelength as the incident radiation, a process called Thompson scattering. Thompson scattering relies on the very low weight of the electrons, so the nuclei (which are also charged particles) do not scatter X-rays in this way. Since the electrons scatter X-rays the amount one atom scatters is proportional to the number of electrons. This makes it hard (but not impossible) to “see” the lightest atoms if they are surrounded by heav- ier atoms (such as in a metal hydride); it is also hard to distinguish between atoms or ions with equal, or close to equal, number of electrons (e.g. in NaF, ﬁgure 3.3a).

Scattering of Neutrons Unlike X-rays, neutrons mainly interact with the nucleus of an atom rather than its electrons (ﬁgure 3.3b). When a neutron hits a nucleus the neutron can be: elastically scattered coherently (i.e. with the same phase as the other neutrons and is therefore interesting for diffraction experiments); elastically scattered in-coherently (i.e. not with the same phase as the other neutrons); scattered in-elastically by the nucleus (i.e. some energy is transferred to or from the nucleus) or absorbed by the nucleus. Unlike X-rays, with a linear relationship between elastically scattered photons and atomic number, the likelihood that isotopes scatter neutrons elastically and coherently follow an apparent random pattern; this makes it possible to distinguish some atoms with similar numbers of electrons, e.g. in the powder diffraction pattern of NaF (ﬁg- ure 3.3c). Additionally, since the neutron has a spin it is scattered by unpaired electrons and can thus be used to probe the magnetic ordering of a material.

3In a real experiment this is complicated by a number of factors, see equation 3.3 on page 30.

26 a) b)

18e- 18e- 18e- 9.58 3.71 9.58

18e- 18e- 18e- 3.71 9.58 3.71

- - 18e- 18e 18e 9.58 3.71 9.58

c ) X - r a y ) t i n u

b n r a ( / I

1 5 3 5 5 5 7 5 9 5 1 1 5 2 θ/ d e g r e e Figure 3.3. The difference between the way NaF scatters X-rays and neutrons. a) What the X-ray “sees” (i.e. the electrons of the ions), b) What the neutron “sees” (i.e. the nuclei and their bound scattering lengths, b, in fm). c) Simulated X-ray and neutron powder diffraction patterns (λ = 1.54056 Å).

27 3.3 Characterisation by Diffraction Techniques

“I suppose it is tempting, if the only tool you have is a hammer, to treat every problem as if it were a nail.”

Abraham Maslow (The Psychology of Science, 1966)

3.3.1 X-ray Single Crystal Diffraction The X-ray Single Crystal Diffraction (XRSCD) experiment was performed on a single crystal mounted on a cactus spine with epoxy glue in a Bruker D8 single crystal diffractometer using MoKα radiation and a Charged-Coupled Device (CCD) detector (ﬁgure 3.4a).

3.3.2 X-ray Powder Diffraction Most X-ray Powder Diffraction (XRPD) experiments were performed on either of two Bruker D8 Advance diffractometers: one using monochromatic CuKα1 radiation and a Våntec position sensitive detector and the other using CuKα radiation and an energy dispersive Lynxeye XE detector; both diffractometers were in the Bragg-Brentano set-up4 (figure 3.4c). Powdered samples were applied to a flat single-crystal silicon wafer5 either by dusting powder trough a very fine sieve or by dispersing the powder in ethanol which was evaporated on the wafer.

In-situ X-ray Powder Diffraction All in-situ Powder X-ray Diffraction (in-situ XRPD) experiments were performed at the I711 beam-line [37] on the MAXII synchrotron of the MAXIV laboratory in Lund, Sweden. The diffraction patterns were recorded in the Debye-Scherrer set-up6 on a large area Titan or Mar165 CCD detector, using a sample cell designed to withstand high pressures and temperatures [38] (ﬁg- ure 3.4b). Sample–detector distances and wavelengths were determined using LaB6 (in many cases NIST660b [39]). The short wavelength (λ ≈ 1 Å), large detector area and short sample– detector distance allowed measurements of rather short d-values without mov- ing the detector; this combined with the high X-ray ﬂux provided by the synchrotron allowed for very short exposure times (1-30 s) for each diffractogram. The recorded 2-dimensional images were reduced to 1-dimensional diffractograms by integrating the intensities in the Fit2D program[40].

4i.e. the incident angle and detector angle are identical. 5oriented in such a way as to produce no diffraction pattern 6i.e. transmission geometry.

28 a) b)

θ θ

c) d)

2θ

Figure 3.4. The diffraction techniques used in this thesis. a) X-ray Single Crystal Diffraction. b) X-ray Powder Diffraction in the Bragg–Brentano set-up. c) In-situ X-ray Powder Diffraction in the Debye–Scherrer set-up using synchrotron radiation. d) Neutron Powder Diffraction diffraction using a monochromatic reactor source.

Powdered samples were placed in fused silica capillaries7 with inner di- ameters of 0.5 mm. The capillaries were placed in single-crystal sapphire tubes8 which were mounted into the sample holder using graphite ferules. Af- ter ﬂushing the system with Ar three times and leak checking, the samples were subjected to different H2 pressures (<15 MPa) or dynamic vacuum at temperatures ranging from 303 to 823 K.

3.3.3 Neutron Powder Diffraction The Neutron Powder Diffraction (NPD) experiments were performed using monochromatic neutron radiation on powdered samples in vanadium tubes on the MEREDIT instrument [41] at the Nuclear Physics Institute in Rez, Czech Republic using λ=1.46 Å radiation or on the PUS diffractometer [42] at the Institute for Energy technology in Kjeller, Norway using λ=1.55 Å radiation (ﬁgure 3.4d).

7to reduce absorption 8oriented in such a way as to produce no diffraction pattern

29 3.3.4 Determination of Lattice Parameters The lattice parameters of a crystalline phase can be determined from a powder diffraction pattern if the Bravais lattice is known. This is done by a least squares ﬁt of expected peak positions from Bragg’s law9 to the observed peak positions. The lattice parameters determined in this manner presented in this thesis were obtained using the UnitCell [43] and CHECKCELL [44] programs.

3.3.5 Full Pattern Refinement Using the Rietveld Method Full pattern refinement is a powerful technique to determine the crystal structure of a crystalline material that is approximately known or the phase content of a sample from powder data. In this technique a calculated pattern is refined to fit the experimental diffractogram by changing structural and profile parameters using the least squares method. The technique was developed and first used by H. M. Rietveld [45] and is commonly called Rietveld refinement. Ideally, diffraction peaks should be located at angles 2θ determined by Bragg’s law9; however, this angle is affected by a number of instrumental imperfections such as absorption and sample displacement. The integrated 10 intensity (Ihkl) of a peak depends on the structure factor (Fhkl), but also a number of other parameters and is calculated as:

2 Ihkl = K phklLθ Pθ Aθ ThklEhkl|Fhkl| (3.3) were K is a constant known as the scale factor which roughly accounts for the amount of compound, measurement time and incident radiation flux; phkl is the multiplicity of the specific reflection; Lθ , Pθ and Aθ are multipliers which take the geometry of the experiment, partial polarisation of the electromag- netic wave and absorption of incidence and diffracted beam as well as porosity into account; Thkl is the preferred orientation factor and Ehkl in an extinction multiplier (which is usually not important for small crystals). The peak shape is often calculated as a linear interpolation of a Gaussian and Lorentzian peak function, called a pseudo Voigt peak function, where the Full Width at Half Maximum (FWHM) depends on θ: p FWHM = Utan2θ +Vtanθ +W (3.4) where U,V and W are constants. The structural determinations and phase analyses done in the work presented in this thesis were performed with the FullProf [46] or JANA2006 [47] programs. Interstitial D atoms were located by combining NPD and XRPD data using Fourier difference maps in JANA2006. Standard deviations are given in parenthesis after all determined values.

9equation 3.1, page 25 10equation 3.2, page 26

30 3.4 Thermodynamic and Kinetic Behaviour of Metal Hydrides The thermodynamic and kinetic properties of a metal hydride are crucial for most applications. Methods for determining the heat of formation and activation energies of hydrogen desorption are described below.

3.4.1 Thermodynamics of Metal Hydrides In a ﬁrst order phase transition where hydrogen dissociates from a metal hydride: 1 1 1 MH (s) M(s) + H (g) (3.5) x x x 2 2 the equilibrium constant is calculated as11:

1 1 1 −◦ 1 {M(s)} x {H2(g)} 2 1 x · (PH dis/P ) 2 1 2 −◦ 2 K = 1 ≈ 1 = (PH2dis/P ) (3.6) {MH(s)} x 1 x where P−◦ =1 bar=0.1 MPa. If K is inserted into the van ’t Hoff equation a relationship between the heat of formation, ∆H−◦ and the equilibrium pressure is obtained: 1 ∆H−◦ ∆S−◦ lnK = ln(P /P−◦ ) = − (3.7) 2 H2dis RT R where R is the ideal gas constant and T is the temperature. The entropy of formation ∆S−◦ is almost constant for all metal hydrides and depends mostly of the entropy loss in the transition from gas to solid phase. Equation 3.7 predicts that the phase transition will take place at a specific pressure for each temperature. In a solid solution of hydrogen in metal on the other hand, the metal hydride is not a pure solid and its activity is not equal to unity; therefore, the pressure at which a certain amount of hydrogen is desorbed is dependent on both hydrogen concentration and temperature. In a real metal hydride both behaviours are often observed. When the partial H2 pressure is increased at a fixed temperature, hydrogen is first dissolved in the metal (called the α phase) where the concentration will be a function of pressure; at some pressure a phase transition to a more ordered hydride phase (the β phase) occurs which leads to a large increase in hydrogen concentration without a large change in pressure, i.e. a plateau (figure 3.5). Just as in the α phase more hydrogen can be dissolved in the β phase, and the concentration will be a function of pressure. If a continual phase transition from α to β phase is structurally possible a first order phase transition may occur above a critical temperature (Tc).

11Using the deﬁnition that the activity of a pure solid is unity and assuming that hydrogen is an ideal gas.

31 Tc

T3 α α+β β T2

log(P/arb. unit) log(P/arb. T1

H/M Figure 3.5. Pressure-Composition-Temperature (PCT) diagram for a system containing the solid solution α phase and ordered hydride β phase. T1

Pressure Composition Temperature Measurements By measuring a series of Pressure Composition Temperature (PCT) curves at different temperatures, ∆H−◦ of a hydride can be determined by plotting equilibrium pressures of the plateaus against the temperatures in a van ‘t Hoff plot (equation 3.7). This value is not only important to solid gas applications but also for electrochemical systems, e.g. batteries, since the standard potential depends on ∆H−◦ . The PCT curves presented in this thesis were measured using a Setaram PCTPro–2000 automatic Sieverts apparatus (figure 3.6a). The sample was placed in a stainless steel autoclave and heated to a fixed temperature and the H2(g) pressure was stepwise increased by filling a large or small reservoir and then opening a set of valves to the sample. After equilibration the amount of hydrogen absorbed was calculated by the drop in pressure and the equilibrium pressure was noted. The samples were activated by cycling a number of times before the PCT curves were recorded.

Predicting the Stability of a Hydride A number of factors have been suggested to correlate with and/or determine the stabilities of metal hydrides [48]: the size of interstitial voids [49], the unit cell volume [24], the difference between ∆H of the inter-metallic compound and the binary hydrides of the metals that it consists of [50] and the difference between the Fermi energy and the centre of the lowest band of the host metal [51]. It should be noted that the number of valence electrons, hydrogen afﬁnity and atomic radii are correlated. With the ever increasing computational power available, and therefore decreasing cost of calculations, computational methods such as Density Functional Theory (DFT) are more and more viable and seems to yield reasonable predictions for a number of hydrides [52].

32 a) b)

s l

Figure 3.6. Solid gas techniques used for characterisation. a) Schematic view of an automatic Sieverts apparatus used to record PCT curves. b) The sample in high vacuum during the TDS measurements.

3.4.2 Kinetics The Kissinger equation [53], originally developed for differential thermal analysis, can be used to determine the activation energy (Ea) of a reaction. It is obtained from the Arrhenius equation and the rate equation: dx k(T) = Ae−Ea/RT and = k(t) f (x) (3.8) dt where k(T) is the rate constant, A the pre-exponent factor, R the ideal gas constant, T the temperature, t time and f (x) a function of the amount of reactant x (which describes the kinetic model). The combination of the two equations yields the general rate equation: dx dx = A f (x)e−Ea/RT or f (x) = A−1eEa/RT (3.9) dt dt differentiating the function by time yields:

d dx d f (x) dx E dT dt = e−Ea/RT ( + f (x) ) (3.10) dt dx dt RT 2 dt inserting f(x) from equation 3.9 yields:

d dx dx d f (x) E dT dt = e−Ea/RT ( + eEa/RT ) (3.11) dt dt dx ART 2 dt 33 dT If the temperature is increased linearly ( dt = β) then, at some temperature dx d dt Tmax, the maximum reaction rate will be reached, which implies dt = 0:

d f (xmax) Ea/RTmax Ea β 0 = + e 2 (3.12) dx AR Tmax Finally taking the natural logarithm of the result yields the general Kissinger equation[54]:

β d f (xmax) AR −1 ln( 2 ) = ln( ) + ln( ) − Ea · (RTmax) (3.13) Tmax dx Ea

d f (xmax) −1 If ln( dx ) is assumed to be independent of (RTmax) (or rather β), which 12 is a good approximation in many cases ,Ea can be obtained by determin- β ing the Tmax at a number of β and doing a linear regression of ln( 2 ) and Tmax −1 −(RTmax) . Thermal Desorption Spectroscopy (TDS) is a method used to study the desorption of a volatile compound from a material during heating where the partial pressure of the desorbed species is monitored by a mass-spectrometer. If the sample is heated linearly and the maximum partial pressure is assumed to indicate the maximal reaction rate a series of TDS measurements may be used to determine Ea via the Kissinger equation. The TDS data presented in this thesis were obtained by linear heating samples placed in a stainless steel sample holder in high vacuum (ﬁgure 3.6a). The sample temperature was monitored by a thermocouple in close proximity to the sample holder and the partial pressure of H2 (or D2) was monitored using a Microvision Plus residual gas analyser/mass spectrometer.

3.4.3 Density Functional Theory Density Functional Theory (DFT) is a theoretical method that can be used to calculate properties of a material using quantum physics, independent of experimental results. These properties could be obtained by solving the many electron time independent Shrödinger equation directly; however, every electron added to the system adds four dimensions to the problem, so a system with many atoms is very computationally expensive. In DFT the electrons are replaced by an electron density dependent on three coordinates, the location in physical space. DFT in this pure form approximates the nuclear–electron interactions well but is not as good at, for example, approximating electron–electron interactions; a number of corrections are used to compensate for these shortcom- ings. The DFT data presented in this thesis were calculated using The Vienna

12but not for e.g. Avrami-Erofev and diffusion kinetics[54]

34 Ab initio Simulation Package [55] and the The Perdew–Burke–Ernzerhof ap- proach [56] was used to compensate for the approximations.

4. Results and Discussion

“Chemists are always right, but for the wrong reasons.” Olle Eriksson

The results presented in this thesis are divided into four parts: hydrides of Laves phases containing Sc (papers I, II and III), MgH2 mixed with ScH2 (paper IV), Nb4MSi (M=Ni, Co) hydrides (paper V) and NdGa and GdGa hydrides (paper VI) (ﬁgure 4.1). In addition there is a ﬁfth part dealing with the attempted synthesises of Ti–Zr–Ni–Co quasicrystal approximants.

Figure 4.1. The crystal structures of the hydrides or hydride forming compounds studied in this thesis. First row: A hexagonal Laves phase and a cubic Laves phase. Second row: MgH2, Nb4MSi and NdGaD1.51.

37 4.1 Laves Phases Containing Sc Papers I, II and III concern Laves phases containing Sc. Laves phase, or AB2-type, compounds can be used in Ni–MH batteries and as hydrogen storage materials. Paper I and II deals with the hexagonal Laves phase formed in the Sc(Al1−xNix)2 system and Paper III concerns the cubic Laves phase (Sc1−xZrx)(Co1−yNiy)2 system. Both of these systems are pseudo binary, where Sc and Zr act as the larger A atom and Co, Ni and Al act as the smaller B atom. Although Sc may not be a viable raw material for most applications (because of its high cost) it is interesting from a fundamental point of view, since it is the lightest d-block metal as well as the lightest and smallest rare earth metal. The structural relationship between the hexagonal and cubic Laves phases is discussed in the introduction on page 16.

4.1.1 The Hexagonal Laves Phase Sc(Al1−xNix)2 Phase Analysis and Structure Determination A hexagonal Laves phase with the MgZn2-type structure was observed in as cast Sc(Al1−xNix)2 samples while 0.25≤x≤0.75 and single phase samples were obtained when 0.36≤x≤0.65 (figure 4.2 and 4.3a). At the extremes x=0.00 and 0.90≤x≤1.00 the samples contained only a cubic Laves phase. A linear decrease in lattice parameters was observed as a function of x (i.e. Ni content) which is not surprising since the Al atom is much larger than the Ni atom[19] (table 4.1 and figure 4.3b and c); the fact that lattice parameters kept changing even at compositions where multiphase samples were obtained suggested that the hexagonal Laves phase had a larger homogeneity area at elevated temperatures. Structure determination using XRSCD and NPD con- firmed that Sc(Al0.50Ni0.50)2 crystallises in the MgZn2-type structure where Ni and Al are close to randomly distributed between the two B sites available in the structure.

Interstitial Hydrogen Absorption Lattice expansion was seen when samples with 0.36≤x≤0.67 were hydrogenated at 573 K under 2 MPa H2 pressure and the change in lattice parameters was larger the higher the Ni content (table 4.1). Piesl [57] found that the expansion of the unit cell volume as a function of hydrogen content was sur- prisingly constant for many metals (∼2.9 Å3/H atom). Using this relationship the stoichiometry of the interstitial hydrides was approximated to vary from Sc(Al0.64Ni0.36)2H0 to Sc(Al0.33Ni0.67)2H0.76 (table 4.1). Determination of the structure of Sc(Al0.33Ni0.67)2D0.39 from NPD data revealed that the compound absorbs deuterium into interstitial tetrahedral A2B2 sites (i.e. coordinated by two Sc and two Ni/Al atoms). A series of PCT measurements determined that the maximal storage capacity of Sc(Al0.33Ni0.67)2 was 1.95 H/(form. unit) at 353 K.

38 X - r a y

x = 1 . 0 0

x = 0 . 9 0

x = 0 . 7 5

x = 0 . 7 0

x = 0 . 6 7

x = 0 . 6 5 ) t i n u

. x = 0 . 6 0 b r a ( / I x = 0 . 5 0

x = 0 . 4 0

x = 0 . 3 6

x = 0 . 3 0

x = 0 . 2 5

x = 0 . 2 2

x = 0 . 0 0

1 5 2 5 3 5 4 5 5 5 6 5 7 5 8 5 2 θ/ d e g r e e

Figure 4.2. XRPD patterns of as-cast Sc(Alx−1Nix)2-samples. Expected peak positions of the hexagonal Laves phase is marked with vertical bars.

39 Table 4.1. Lattice parameters of the hexagonal Laves phase of the Sc(Al1−xNix)2 system and their interstitial hydrides with hydrogen content approximated from unit cell volume expansion.

Sample a/Å c/Å ahyd/Å chyd/Å H/form. unit

Sc(Al0.25Ni0.75)2 5.0809(3) 7.8605(7) Sc(Al0.30Ni0.70)2 5.0728(2) 8.0268(5) Sc(Al0.33Ni0.67)2 5.0838(2) 8.0032(7) 5.1613(5) 8.1471(7) 0.76 Sc(Al0.35Ni0.65)2 5.0943(2) 8.0288(4) Sc(Al0.40Ni0.60)2 5.1171(5) 8.067(1) 5.1487(2) 8.1679(4) 0.4 Sc(Al0.50Ni0.50)2 5.1434(1) 8.1821(2) 5.1695(2) 8.3000(6) 0.4 Sc(Al0.60Ni0.40)2 5.1749(3) 8.309(1) 5.1783(3) 8.3459(5) 0.1 Sc(Al0.64Ni0.36)2 5.1908(9) 8.3629(1) 5.1859(4) 8.3558(6) (-0.05) Sc(Al0.69Ni0.31)2 5.2051(4) 8.4039(9) Sc(Al0.75Ni0.25)2 5.2324(3) 8.5262(8)

Structural Reason for Maximal Interstitial Storage Capacity The decreasing amount of hydrogen absorbed as a function of Al content is probably caused by Al blocking interstitial sites, since H atoms avoids occu- pying voids in close proximity to p-block elements [58]. Supposing a random Al/Ni distribution the likelihood that none of the B sites is occupied by an Al 1 atom decreases with Al content in a given A2B2 site , e.g. it is 42% when x=0.65, 25% at x=0.5. and 13% at x=0.65.

Interstitial Hydrogen Cycling and Desorption In-situ XRPD experiments showed that there was a continuous unit cell volume expansion and contraction when Sc(Al0.50Ni0.50)2 was cycled between dynamic vacuum and 50 MPa H2(g) at 453 K; the hydrogenation/dehydrogenation process was reversible over multiple cycles. The gradual change of the unit cell volume suggests that a solid solution rather than a second order phase transition occurs. Desorption curves obtained from TDS of the interstitial Sc(Al0.33Ni0.67)2Hx show one distinct bell-shaped peak (ﬁgure 4.4a). Kissinger analysis (ﬁg- 2 −1 ure 4.4b) of the desorption data gives Ea = 51.7 kJ mol which is close to −1 that of other interstitial hydrides such as LaNi5 which has Ea = 40 kJ mol [59].

Decomposition to ScH2 and NiAl Both ex- and in-situ PXRD experiments showed that Sc(Al0.50Ni0.50)2 decom- poses according to:

Sc(Al0.50Ni0.50)2 + H2 ScH2 + NiAl (4.1)

1or any other type of void since there are no interstitial sites surrounded by only A atoms in the Laves phase structures. 2 −1 Ea is incorrectly calculated as 4.6 kJ mol in paper II.

40 a ) S c 0 1 0 0

2 5 7 5

l % A S % 5 0 c 5 0

7 5 2 5

1 0 0 A l 0 N i 0 2 5 5 0 7 5 1 0 0 % N i

b ) c ) 8 . 6 5 . 2 5

8 . 4 5 . 2 0 Å Å

/ / 8 . 2 c a 5 . 1 5

8 . 0 5 . 1 0

7 . 8 0 . 0 0 . 2 0 . 4 0 . 6 0 . 8 1 . 0 0 . 0 0 . 2 0 . 4 0 . 6 0 . 8 1 . 0 x i n S c ( A l N i ) x i n S c ( A l N i ) 1 - x x 2 1 - x x 2

Figure 4.3. a) Composition map for Sc–Al–Ni with reported phases marked as large open circles, synthesised single phase samples marked with ﬁlled circles and multiphase samples marked as half-ﬁlled circles. b) Lattice parameter a of the hexagonal Laves phase as a function of composition. c) Lattice parameter c of the hexagonal Laves phase as a function of composition.

41 a ) b ) 1 . 0

- 1 4 - 1 0 . 8 E a = 5 1 . 7 k J m o l ) ) 1 t -

i 2 s

n R = 0 . 9 9 7

1 u -

0 . 6 . K / b 2

r - 1 5 - a T (

/ 0 . 4 2 β H ( n P l 0 . 2 - 1 6 0 . 0 3 5 0 4 0 0 4 5 0 5 0 0 - 3 . 2 - 3 . 0 - 2 . 8 - 2 . 6 T / K - R - 1 T - 1 / m o l J - 1 1 0 - 4

Figure 4.4. Result of TDS of Sc(Al0.33Ni0.67)2Hx. a) Partial pressure of H2 as a function of temperature for a number of ramp rates (β). Peaks are shifted towards higher temperature with higher β. b) Ea calculated using the Kissinger equation.

when heated to 823 K under 5 MPa H2 pressure (ﬁgure 4.5). The peaks of the ScH2 and NiAl phases (with CaF2-type and CsCl-type structure respectively) were much broader than those of the mother compound (indicating a much smaller average size of, and/or strain in, the crystals); TDS of the resulting ScH2+NiAl sample showed one large and one small bell- shaped peak; Kissinger analysis of the larger peak gave an activation energy −1 of 182 kJ mol . Sc(Al0.50Ni0.50)2 was reformed indicating that this reaction may be reversible over multiple cycles.

X - r a y

i n t e r s t i t i a l a b s . 5 0 ~ 4 7 5 K n a

c 1 0 0 s d e c o m p o s i t i o n 8 2 3 K

1 5 0

1 5 2 0 2 5 3 0 3 5 2 θ/ d e g r e e Figure 4.5. Densitometric view of diffractograms recorded in situ during heating of Sc(Al0.50Ni0.50)2 under 5 MPa H2 pressure. The white line at ∼scan 120 is caused by missing scans.

42 4.1.2 The Cubic Laves Phase (Sc1−xZrx)(Co1−yNiy)2 Phase analysis All (Sc1−xZrx)(Co1−yNiy)2 samples contained a cubic Laves phase and most samples were single phase (ﬁgure 4.6). The two samples, Sc0.75Zr0.25CoNi and ScNi2, which were not completely phase pure, contained small amounts of a phase with the CsCl-type structure. Lattice parameters increased with increasing Zr content while the samples without Zr had very similarly sized unit cells (table 4.2 and ﬁgure 4.7); these lattice parameters were in good agreement the ones reported in the literature [60, 61].

Absorption and Desorption of Interstitial Hydrogen Multiple cycles of interstitial hydrogen absorption and desorption were observed for all samples with a Zr content equal to or less than Sc0.50Zr0.50CoNi, when cycled at 303 or 333 K under 15 MPa H2/dynamic vacuum monitored using in-situ XRPD (ﬁgure 4.8a). A gradual shift of peak positions of the pris- tine materials, followed by a decrease in peak intensities at the expense of a growing hydride phase during hydrogen absorption indicated a solid solution of hydrogen in the α-phase followed by a phase transition to the β phase. During desorption this process was reversed. The amount of absorbed hydrogen, approximated from unit cell expansion using Peisl’s relationship [57], increased as a function of the Co and Sc content (table 4.2). For the samples without Zr this is in good agreement with the results of Yoshida and Akiba [62] who reported a maximum storage capac- −1 ity of 1.03 H/M for ScCo2 (∆H=30 kJ mol ) and 0.65 H/M for ScNi2(∆H= −1 16 kJ mol ) under 4MPa H2 pressure at 313 K. There is, to my knowledge, no literature data to compare the Zr containing samples to; however, hydrogen only forms a solid solution up to 0.4 H/M with ZrCo2 even under the extreme H2 pressure of 80 MPa at 298 K![63] It is reasonable to assume that the hy-

Table 4.2. Lattice parameters determined from ex- and in-situ PXRD and ab-inito calculations of the (Sc1−xZrx)(Co1−yNiy)2 cubic Laves phases and their hydrides. Hy- drogen content per metal atom was approximated using Peisl’s relationship [57] and, in parenthesis, estimated expansion from DFT.

Sample aex−situ/Å acalc./Å ain−situ/Å ahyd.in−situ/Å H/M ZrCoNi 6.9550(1) 6.90 Sc0.25Zr0.75CoNi 6.9479(5) 6.89 Sc0.50Zr0.50CoNi 6.9400(3) 6.88 6.9388(2) 7.1983(4) 0.56 (0.60) Sc0.75Zr0.25CoNi 6.9364(3) 6.87 6.9371(3) 7.2795(4) 0.75 (0.80) ScNi2 6.9213(2) 6.88 6.9185(7) 7.2660(6) 0.75 (0.93) ScCoNi 6.9263(2) 6.86 6.9257(4) 7.3036(4) 0.82 (1.01) ScCo2 6.9276(2) 6.82 6.9260(5) 7.378(1) 1.00 (1.23)

43 X - r a y

Z r C o N i

S c 0 . 2 5 Z r 0 . 7 5 C o N i

S c 0 . 5 0 Z r 0 . 5 0 C o N i ) t i n u . b r a ( / I

S c 0 . 7 5 Z r 0 . 2 5 C o N i

S c N i 2

S c C o N i

S c C o 2 2 5 3 5 4 5 5 5 6 5 7 5 8 5 2 θ/ d e g r e e

Figure 4.6. Experimental (red dots) and calculated (black lines) diffraction patterns of the (Sc1−xZrx)(Co1−yNiy)2-samples and their corresponding difference curves (blue lines). Expected peak positions are indicated as vertical bars.

44 a ) b ) 6 . 9 5 6 . 9 2

6 . 9 2 6 . 8 9 Å Å / / a a 6 . 8 9 6 . 8 6

6 . 8 3 6 . 8 6

0 . 0 0 0 . 2 5 0 . 5 0 0 . 7 5 1 . 0 0 0 . 0 0 0 . 2 5 0 . 5 0 0 . 7 5 1 . 0 0 y i n S c C o N i x i n S c 1 - x Z r x C o N i 1 - y y

Figure 4.7. Values for lattice parameter a in (Sc1−xZrx)(Co1−yNiy)2 as a function composition, determined from XRPD (red squares) and calculated using DFT (black dots). Lines are added as aid to the eye. a) As a function of x (i.e amount of Zr). b) As a function of y (i.e amount of Ni).

a ) 0 b ) 0 X - r a y

1 0 0 0 1 0 0 0 s s

/ t / s / t t 2 0 0 0 2 0 0 0

3 0 0 0 3 0 0 0

1 0 2 0 3 0 4 0 0 5 0 1 0 0 2 θ/ d e g r e e w t % o f p h a s e

Figure 4.8. a) Densitometric view diffraction patterns recorded in situ as hydrogen is cycled in and out of Sc0.50Zr0.50CoNi. The change in H2 pressure from 15 MPa to dynamic vacuum, or vice-versa, caused an almost instant shift of the peaks indicating hydrogen absorption/desorption. b) wt% of the α (green line) and β phase (black line) obtained from sequential reﬁnement of the in-situ diffractograms.

45 dride of ZrCoNi is very unstable3. A gradual decrease in hydride stabilities from ScCoNi to ZrCoNi is a plausible explanation for the reduction in hydrogen absorption between the two compositions. Sequential refinement made it possible to follow the amount of the α and β phases as a function of time (figure 4.8b); even though no kinetic parameters were extracted from these data it was clear that the samples containing Zr had much higher reaction rates4, especially during desorption. A possible explanation for this is that Zr has lower affinity for hydrogen than Sc [64] which could lead to faster diffusion.

Comparison with Ab-initio Calculations The ab-initio calculations slightly underestimated the size of the unit cell (table 4.2 and ﬁgure 4.7); the difference was close to the thermal expansion from 0 to 300 K expected for Laves phases based on the linear expansion parameters calculated by Mayer et. al [65]. Additionally, the ab-initio calculations predicted that the unit cell volume would expand by ∼2.72 Å3/H atom introduced into the unit cell for the compounds containing Zr and by ∼2.36 Å3/H atom for the ones without. Using these expansions to calculate the amount of absorbed hydrogen yielded the same trends as using the 2.9Å3/H atom expansion (table 4.2).

3 If it behaves like the ScCo2/ScNi2 system it is probably even slightly less stable than ZrCo2. 4This is the reason why the samples without Zr were cycled at 333 K rather than 303 K.

46 4.2 Activation Effects in MgH2 Mixed with ScH2 As mentioned in the introduction (page 15) the low cost and high wt% of hydrogen in MgH2 have made it very interesting for some applications, especially hydrogen storage. However, its thermodynamic stability, sensitivity to oxygen and slow kinetics present major obstacles to its use in these applications. One way to improve the kinetics is to add a catalyst to the material, another way is to reduce the grain size (e.g. by ball-milling the material); these methods are often combined. Paper IV concerns the activation effect during cycling of MgH2 ball-milled together with the possible catalyst ScH2 investigated by in-situ XRPD.

Phase Analysis A sample preprepared via ball-milling of MgH2 and ScCl3 formed MgCl2 and ScH2, while the ball-milled MgH2 (90 mol%) and ScH2 (10 mol%) showed no sign of reaction between the hydrides after milling. For this reason, the second ball milled mixture was selected for in-situ experiments.

In-situ Cycling Multiple transitions between Mg and MgH2 were observed using in-situ XRPD when three different MgH2/ ScH2 samples were cycled at 593 K, 673 K and 723 K between 10 MPa or 15 MPa H2 and dynamic vacuum (ﬁgure 4.9a, b and c). No side reactions were observed except for formation of MgO.

Reaction Rates and Activation Effect Sequential refinements using the Rietveld method yielded the wt% of the con- stituent phases which made it possible to follow the reactions as a function of scan number(or time). As expected the reaction rates (the slope at 50% con- version) increased with increasing temperature and cycle number; however, at 723 K the reactions were as fast as or faster than the exposure time of one diffractogram, this meant that the reaction rates could not be determined for this temperature. The activation effect was seen as a linear increase in relative reaction rate as a function of cycle number in the 593 K experiment, while the increase was only observed from the first to second cycle in the 673 K experiment (figure 4.9d). The results suggest that activation of the material takes multiple cycles at 593 K and 10 MPa but only one cycle at 673 K and 15 MPa. Since the publication of Paper IV a thorough study by Luo et al. [16] on the effect of ScH2 addition to MgH2 has shown that these mixtures have a reduced activation energy compared to pure MgH2. The best ball-milled ScH2/MgH2 mixture had a Ea of dehydrogenation to just above half that of pure untreated MgH2.

47 a ) b ) X - r a y X - r a y 1 0 0 d 1 5 0 d 1 2 0 0 d 2 1 0 0 n n

a a d 2 c c

s 3 0 0 s d 3 1 5 0 4 0 0 d 3

2 0 0 5 0 0 d 4 d 4

1 5 2 5 3 5 4 5 1 5 2 5 3 5 4 5 2 θ/ d e g r e e 2 θ/ d e g r e e c ) d ) X - r a y 2 . 0 0 5 0 d 1 1 . 7 5 1 0 0 n d 2 1 r a / n c 1 . 5 0 r s 1 5 0 d 3 1 . 2 5 2 0 0 d 4 6 7 3 K 1 . 0 0 2 5 0 5 9 3 K 1 5 2 5 3 5 4 5 1 2 3 4 2 θ/ d e g r e e d e h y d . c y c l e Figure 4.9. Densitometric view of the diffractograms recorded during hydrogenation/de-hydrogenation cycling of MgH2 (90 mol%) and ScH2 (10 mol%), between a high and a low H2 pressure at isothermal conditions, and the relative reaction rates for each dehydrogenation. a) Cycling at 593 K with high pressure 10 MPa. b) Cycling at 673 K with high pressure 15 MPa. c) Cycling at 723 K with high pressure 15 MPa. d) Relative reaction rate r/r1 compared to the ﬁrst cycle for the desorption for at 593 K and 673 K as a function of cycle number.

48 4.3 Nb4M0.9Si1.1 (M=Ni, Co) Hydrides

The structures of Nb4MSi (M=Fe, Ni or Co) were ﬁrst reported by Gladu- chevsky and Ku’zma [66] in 1965 and the deuterides of the Co and Ni variants by Vennström and Andersson [67] in 2004. Both studies agree on the structure type of the materials (Al2Cu, ﬁgure 4.10) and on the location of the Nb atom (in the Al site); however, while the earlier study suggested that the M and Si atoms were ordered into two different sites, the latter found them randomly distributed between the two sites. The more recent study also found that H(D) atoms are located in two types of Nb4-tetrahedral voids in the Nb4MSi deuterides. Paper V concerns the hydrides of Nb4M0.9Si1.1 (M=Ni, Co). Recently Ryu et al.[68] found that Nb4MSi (M=Fe, Ni or Co) are all superconducting below 6 K. In this paper it was is also claimed that their samples consisted of two phases with ordered and disordered M/Si sites and that only the ordered phase was superconductive.

Analysis of Hydride (Deuteride) Structure If the metal atoms are removed from the Nb4MSiDx structure perpendicular hexagon “chains” of partially occupied D-atom sites appear (figure 4.10b); each hexagon is made up of five H sites: four sites of its own and two that connects it to (and shares with) the two adjacent hexagons. These sites are only partially occupied (about 1.3/5) with a much higher occupation of the interconnecting sites than the four other sites in the chain. Westlake[69, 48] found that a minimum hole radius of 0.4 Å and H–H distance of 2.1 Å can be used to predict site occupancy and maximum storage capacity in many interstitial metal hydrides. The shortest inter “chain” distance is 2.6 Å, so only distances in the “chains" need to be considered. Filling the two sites closest to one of the two linking sites in each hexagon (in the same direction) yields a shortest H–H distance of 2.3 Å and a stoichiometry of Nb4MSiH4. Filling the four sites surrounding every second linking site and every other linking site yields a shortest H–H distance of 2.1 Å and a stoichiometry of Nb4MSiH6. The hole radii, calculated as the shortest Nb-hole distance minus the shortest Nb-Nb distance, is at least 0.48Å for both types of sites. As previously mentioned hydrogen avoids sites close to p-block elements in interstitial hydrides.[58] The interconnecting H site in the hexagons is surrounded by four equidistant M/Si sites. The non-interconnecting H sites have two M/Si sites at about the same distance as the interconnecting H site but two M/Si sites ∼0.5 Å closer than that (figure 4.10c and d). The blocking effect of the p-block elements would mean that none of the non-interconnecting sites could be occupied if the M/Si sites were ordered, since one of the two closest M/Si sites would always be occupied by a Si atom. Assuming a random distribution of M and Si atoms between the two sites 25% of these sites would be available for hydrogen occupation.

49 a) b) x x x x

x x x x

c) d)

Figure 4.10. The structure of Nb4MSi and some structural details. Nb is shown as light grey spheres and M/Si as dark grey spheres. D atom sites are shown as white spheres with the interconnecting sites market with an x. a) The Nb4MSi with the coordination polyhedra of the two D sites shown with green lines. b) “Chains” of D atom sites ﬁlled with blue D atoms in two ways. c) The four closest M/Si sites to the interconnecting site d) The four closest M/Si sites to the non-interconnecting sites in the hexagons.

50 a ) b ) X - r a y X - r a y ) ) s s t t i i n n u u

. . b b r r a a ( ( / / I I

1 5 3 0 4 5 6 0 7 5 1 5 3 0 4 5 6 0 7 5 2 θ/ d e g r e e 2 θ/ d e g r e e

Figure 4.11. Result of full pattern reﬁnements of XRPD patterns of heat treated Nb4M0.9Si1.1 samples. Experimental pattern, calculated pattern, difference curves and expected peak positions are shown as red dots, a black line, a blue line and vertical bars, respectively. a) Nb4Ni0.9Si1.1 b) Nb4Co0.9Si1.1.

Phase Analysis The Nb4Ni0.9Si1.1 and Nb4Co0.9Si1.1 samples contained a majority phase with the CuAl2-type structure and disordered M/Si sites after heat-treatment (ﬁg- ure 4.11). No phase with ordered M/Si positions was observed in any of the samples produced, even though a number of compositions, heat treatment temperatures and times, with and without quenching, were performed. Further investigation of the superconductivity of the hydrides of Nb4MSi suggest that some Nb4MSi samples were composed of two Nb4MSi phases but both had disordered M/Si sites.[70] In addition to the aforementioned Nb4MSi phase a few faint peaks were found in the sample indicating the presence of other phases. The strongest extra peaks in the Nb4Ni0.9Si1.1 sample could be explained by a phase with W-type structure. The unit cell volume was slightly smaller than that of pure Nb suggesting that it was a Nb-rich alloy with Si and/or Ni.

Kinetics TDS showed two distinct desorption peaks for some ramp rates (β) for the Nb4Ni0.9Si1.1Hx sample and all β for the Nb4Co0.9Si1.1Hx sample (figure 4.12a). The rather noisy signal complicated the Kissinger analysis, which had 2 −1 low R values, which gave Ea=79 kJ mol for Nb4Ni0.9Si1.1Hx and 94 and −1 170 kJ mol for Nb4Co0.9Si1.1Hx (figure 4.12b). Slow diffusion caused by the high affinity for hydrogen of Nb and the “chains” of available H sites may explain the slow kinetics in both samples. However, the assumption that the kinetic model can be ignored in Kissinger analysis does not work well for diffusion kinetics [54] and the apparent Ea

51 a ) b ) C o P e a k 2 N i 1 . 0 - 1 3 - 1 E = 7 9 k J m o l - 1 E a = 1 7 0 k J m o l a 0 . 8 R 2 = 0 . 9 0 R 2 = 0 . 9 0 ) 1

- - 1 4 x s a 1 - m 2 K H 0 . 6 /

2 P - / 2

T - 1 5 H β

P C o P e a k 1 0 . 4 ( n

l - 1 E a = 9 4 k J m o l - 1 6 2 0 . 2 R = 0 . 8 6

0 . 0 - 1 7 3 5 0 5 0 0 6 5 0 8 0 0 1 4 1 6 1 8 2 0 T / K T - 1 / ( K - 1 * 1 0 - 4 )

Figure 4.12. a) TDS curves for desorption of Nb4Ni0.9Si1.1Hx (black squares) and −1 Nb4Co0.9Si1.1Hx (white circles) β ≈ 0.04 K s . b) Kissinger analysis to determine activation energies.

0 . 1 a P

M / P 0 . 0 1

1 E - 3 Nb4Co0.9Si1.1 623 K

Nb4Co0.9Si1.1 573 K

Nb4Ni0.9Si1.1 573 K

Nb4Ni0.9Si1.1 573 K Nb Ni Si 373 K 1 E - 4 4 0.9 1.1 0 . 0 0 0 . 2 5 0 . 5 0 0 . 7 5 1 . 0 0 1 . 2 5 1 . 5 0 1 . 7 5 2 . 0 0 2 . 2 5 H / f o r m . u n i t .

Figure 4.13. PCT curves for Nb4M0.9Si1.1 (M=Co, Ni) samples.

52 may be underestimated. The overlapping nature and “noisyness” of the peaks made it impossible to determine the correct kinetic model for this system.

Thermodynamics Regardless of the correct values of Ea, absorption and desorption of hydrogen in the Nb4M0.9Si1.1 samples were very slow which complicated the PCT- measurements (ﬁgure 4.13). Even after a few cycles the equilibration at each point was very slow; the equilibration time was ﬁxed to 10 h so the curves were only reliable at elevated temperatures. Even though ∆H could not be extracted from this data it was clear that the hydrides were very stable and the close to linear appearance of the PCT curves at 573 K suggest that this temperature is above Tc.

53 4.4 Hydrogen Absorption in LnGa (Ln=Gd, Nd) Most REGa-compounds (where RE is a rare earth element) form typical Zintl phases with the CrB-type structure.[71, 72] The active metals are present as tetrahedra of RE3+ ions and the more electronegative Ga forms poly-anionic chains of Ga3−-species (ﬁgure 4.14a). Hydrogen could potentially be introduced into this structure in two ways (page 17) [28]: either as interstitial hydride ions (H−) in RE3+ tetrahedra causing the oxidation of the poly-anion or by covalent bonding to the Ga chains, forming a new poly-anion. The REGa compounds order magnetically at low temperatures and could potentially be used in magnetic refrigeration.[73] The magnetic properties of their hydrides have not been investigated and knowing the structure and the nature of M–H bonds in these compounds could help in understanding their magnetic properties. Paper VI concerns the structures and hydrogen sorption properties of the hydrides of LnGa (Ln=Gd, Nd), especially the nature of the M–H bonds.

Determination of the Structure of a NdGa Deuteride 5 The structure of NdGaD1.51 was determined from PXRD and PND patterns of a deuterated NdGa-sample (figure 4.14b, figure 4.15 and table 4.3). D atoms were found in two tetrahedral sites: one fully occupied Nd4 site and one just off the centre of a ∼1/4 occupied-Nd3Ga site, with a Ga–D distance of 1.80 Å. The fractional coordinates of the metal atoms were about the same as in the non-deuterated material and the unit cell volume was also almost unaffected; however the a axis was significantly shortened and the c axis significantly elon- gated. In addition to the main NdGaD1.51 phase the sample contained weak peaks identified as small amounts of the GaNd3, GaNd2 and NdD2 phases.

Desorption of D2 from NdGaD1.51 TDS curves of the NdGaD1.51 sample used in the structure determination showed two clear peaks indicating that hydrogen(deuterium) is desorbed in two dis- −1 crete steps (ﬁgure 4.16a). Kissinger analysis gave Ea of 115 kJ mol and

5 GdGaDx cannot be studied in this way because of the extreme neutron absorption of Gd.

Table 4.3. Crystallographic data for NdGaD1.51 at ∼298 K. Space Group a/Å b/Å c/Å V/Å3 Cmcm 4.1093(3) 12.2531(9) 4.1644(3) 209.70(3) Atom Wykoff pos. x y z B/Å2 occ. Ga 4c 0 0.0581(4) 1/4 1 1 Nd 4c 0 0.3465(2) 1/4 1 1 D1 4c 1/2 0.2484(6) 1/4 2.3(1) 1 D2 8g 0.434(3) 0.041(1) 1/4 0.84(4) 0.26(1)

54 a) b)

b c

Figure 4.14. The structure of REGa and a REGa deuteride. a) REGa: RE ion are shown as blue spheres and Ga species as red spheres. The Ga chains are shown by black lines. b) NdGaD1.51: deuteride ions are shown as large red spheres and the deuterium atoms as small white spheres.

156 kJ mol−1 for the ﬁrst and second peak respectively (ﬁgure 4.16b). Post- mortem PXRD showed that the NdGa phase had reformed after desorption suggesting that the compound can be cycled multiple times.

Hydrogen Absorption of GdGa Two significant peak position shifts were observed using in-situ XRPD when GdGa was heated linearly to and cooled from 773 K under 2 MPa H2 pressure (figure 4.17a); sequential refinement revealed that these shifts, the first (larger one) during heating and the second (smaller one) during cooling, were caused by contraction of the a axis and elongation of the c axis (figure 4.17a). Just as in the structure of NdGaD1.51 the b axis and the unit cell volume were almost unaffected.

Comparison with Results from DFT The inter-atomic distances of NdGaH and NdGaH2 calculated using DFT were in good agreement with the experimental values for NdGaD1.51. Hydrogen occupation of the tetrahedral Nd4 voids was energetically favourable compared to the site closer to the Ga chain in NdGaH . The calculated NdGaH2 structure was a slightly distorted version of the NdGaD1.51 structure with all of the Nd3Ga voids filled on one side of the Ga chain. Filling the tetrahedral Gd4 void first and then the sites adjacent to the Ga chain to the composition GdGaH1.66 reproduced the experimental results remarkably well (figure 4.17).

55 a ) X - r a y ) t i n u

. b r a ( / I

2 0 3 5 5 0 6 5 8 0 2 θ/ d e g r e e b ) n ) t i n u

. b r a ( / I

2 5 5 0 7 5 1 0 0 1 2 5 2 θ/ d e g r e e

Figure 4.15. Diffractograms used to determine the structure of NdGaD1.51. Experi- mental data are shown as red rings, the calculated pattern as a black line, the difference curve as a blue line and expected peak positions of identiﬁed phases vertical bars. a) XRPD pattern. b) NPD pattern.

a ) b ) 1 . 0 P e a k 1

) - 1 )

1 - 1 5 1 1 5 k J m o l - x s

a 2

1 m R = 0 . 9 6 - 2 P e a k 2 D K

( - 1 P / / 2 2 1 5 6 k J m o l -

D 0 . 5

T 2

P - 1 6 R = 0 . 9 1 β ( n l

- 1 7 0 . 0 4 0 0 6 0 0 8 0 0 1 0 0 0 - 2 . 2 - 2 . 0 - 1 . 8 - 1 . 6 T / K - R - 1 T - 1 / ( m o l J - 1 1 0 - 4 )

Figure 4.16. a) TDS curves of NdGaD1.51 for a number of ramp rates (β). Increasing β shifted peaks to higher temperatures. b) Kissinger analysis of the two peaks.

56 a ) 0 . 0 X - r a y 3 0 3 K

0 . 5

h 7 7 3 K / t 1 . 0

1 . 5 2 9 4 K 1 5 2 0 2 5 3 0 3 5 4 0 2 θ/ d e g r e e b ) 3 0 3 K 7 7 3 K 2 9 4 K

b / b 0

e 1 0 5 u l a v

t r a t 1 0 0 s

f c / c 0 o

% 9 5

a / a 0 0 . 0 0 0 . 2 5 0 . 5 0 0 . 7 5 1 . 0 0 1 . 2 5 1 . 5 0 1 . 7 5 t / h

Figure 4.17. a) Densitometric view of the in-situ PXRD patterns recorded during heating and cooling of a GdGa sample under 2 MPa H2 pressure. b) Relative change in lattice parameters as a function of time is shown as red lines. Calculated changes from ﬁrst ﬁlling the Gd4-tetrahedral site and then the site adjacent to the Ga chains is shown as dashed lines.

57 Analysis of the Structure of and Bonding in LnGaHx Results for both GdGa and NdGa indicate that hydrogen will ﬁrst occupy the Ln4 site and then the Ln3Ga site. The ﬁrst step can easily be explained by the current understanding of Zintl-phase hydrides. Hydride ions go into the Ln4 voids, their formal charge requires that the Ga chain is oxidised from Ga3− to Ga2− to keep charge neutrality; this could be accomplished by the formation of a π-bond along the chain. A possible second step would be the formation of a covalent bond to the H atom in the Ln3Ga site, the Ga–D distance(1.80 Å) in NdGaD1.51 is a bit longer than the covalent Nd–D bonds reported in the literature [74, 75]. The value of the electronic localisation function obtained from DFT between Ga and H in NdGaH2 was ∼0.37 suggesting a weak covalent interaction. The nature of Ga–H interaction in the second step of hydrogenation is thus not yet known and requires further investigations.

58 4.5 Ti–Zr–Ni–Co approximants The quasicrystals and approximants in the Ti–Zr–Ni system have been shown to absorb and desorb large amounts of hydrogen interstitially.[76, 77] These compounds have local structures very similar to those of the Laves phases and, as seen in the results presented in this thesis, it has been possible to substitute Co for Ni because of their almost identical metallic radius. A number of samples with compositions Ti50Zr35(Ni1−xCox)15 (x=0, 0.17, 0.5, 0.83, 1) were produced. A Ti50Zr35Ni12.5Fe2.5 sample was also produced. While the Ti50Zr35Ni15 approximant was easily reproduced no sign of an approximant phase was found in any of the other samples even though a number of heat treatment regimes were performed. A quasicrystal phase with composition Ti53Zr27Co20 synthesised via melt- spinning has been reported [78]; it is possible that fast cooling rates, such as in melt-spinning or suction-casting can produce Ti50Zr35(Ni1−xCox)15 approximants.

59 5. Conclusions and Outlook

Sc-Based Laves-Phase Hydrides As cast Sc(Al1−xNix)2 samples only contained the hexagonal MgZn2-type Laves phase when 0.36≤x≤0.65. Ni and Al were randomly distributed between the B sites in the x=0.5 sample. Hydrogenation either lead to interstitial solution in tetrahedral A2B2 sites or decomposition into ScH2 and NiAl. Both reactions were reversible. The amount of interstitial hydrogen that could be stored in the material was inversely correlated with the Al content; this is probably caused by Al proximity to available interstitial sites as H atoms often avoid interstitial sites close to p-block elements. All the samples in the cubic Laves phase (Sc1−xZrx)(Co1−yNiy)2 system that have been studied were single, or close to single, phase. An inverse corre- lation between hydrogen absorption and the content of Zr was observed when hydrogenation/de-hyrogenation cycles were performed. This was attributed to a decreased stability of the hydride affecting the plateau pressure. Zr content improved hydrogen sorption kinetics that may be caused by faster hydrogen diffusion.

Activation in MgH2 with ScH2 Activation of MgH2 mixed with 10mol % ScH2 was observed during hydrogenation/de-hydrogenation cycling. No reaction between the two hydrides could be observed. A plateau of desorption rate seems to have been reached after the second cycle at 673 K while the rate was still increasing signiﬁcantly between the third and fourth cycle at 593 K.

Nb4MSi (M=Co, Ni) Hydrides The hydrides of Nb4MSi (M=Co, Ni) were shown to be very stable and have very slow hydrogen sorption kinetics. Stoichiometry of the hydrides where probably limited by the proximity of most of the available interstitial sites to M/Si sites which may be occupied by Si (a p-block element). No phase with ordered M/Si sites was observed.

LnGa (Ln=Nd, Gd) Hydrides Hydrogen was absorbed/desorbed by LnGa (Ln=Nd, Gd) in two steps. After the second step H(D) atoms were located in fully occupied Nd4 tetrahedra and ∼1/4 occupied Nd3Ga tetrahedra in the NdGaD1.51 phase. Occupation of the Ln4 tetrahedra was probably the ﬁrst step, this required an oxidation of the Ga chains present in the structure possibly by a formation of a π bond. The nature of the Gd–D bond formed in the second step is still unclear.

60 Ti–Zr–Ni–Co Approximants No approximant phases were observed in the Ti50Zr35(Ni1−xCox)15 system unless x=0.

Future Outlook Measurements of the superconductivity of the Nb4MSi (M=Co, Ni) hydrides were beyond the scope of this thesis but will be presented soon; the same goes for the magnetic properties of the LnGa (Ln=Nd, Gd) hydrides. In the latter case there are still many unanswered questions regarding the nature of the second hydrogenation step that will hopefully be explained in the not to distant future.

61 6. Sammanfattning på Svenska

Produktion och lagring av förnybar energi ställer stora krav på utvecklingen av nya material. Metallhydrider är en stor grupp av föreningar med väldigt varierande egenskaper som används, eller skulle kunna användas, i ett flertal energirelaterade tillämpningar, såsom: uppladdningsbara batterier, vätelagring och värmelagring för termisk solkraft (se figur 6.1). Metallhydriderna kan de- las in i tre grupper utifrån vilken typ av kemisk bindning som dominerar i dem: metalliska-, joniska- och kovalenta metallhydrider (se figur 6.2). I de joniska metallhydriderna bildar negativa hydridjoner (H−) och positiva metalljoner (Mn+) mycket stabila salter som strukturellt liknar halidernas1 salter med samma metaller. MgH2 är en jonisk hydrid med en mycket hög viktandel väte (7.66%), vilket gör den intressant som vätelagringsmaterial för lätta tilläpningar som vätgasdrivna bilar (figur 6.1a). Två mycket stora problem för användningen av denna hydrid är att den är för stabil för att avge väte vid rumstemperatur och att absorptionen och avgivandet av väte sker mycket långsamt. I den här avhandlingen visas att blandningen nio delar MgH2 och en del ScH2 uppnår maximal reaktionshastighet efter en absorption/desoptionscykel vid 673 K. De metalliska metallhydriderna, där väteatomerna sitter i hålrum mellan de mycket större metallatomerna, har en relativt låg viktprocent väte (∼2%). Trots detta är de de metallhydrider som används i störst usträckning, till ex- empel i batterier och i vätelagring i ubåtar (figur 6.1b och c). De metalliska metallhydridernas styrkor är den extrema volymetriska tätheten av väte (i vis- sa fall högre än i fast och flytande väte) och att deras stabilitet, reaktionshastighet och elektrokemiska egenskaper är relativt lätta att kontrollera genom att ändra den kemiska sammansättningen. Kombinationen av en stabil jonisk metallhydrid och en instabil metallisk metallhydrid skulle kunna användas för att lagra värme i termisk solkraft (figur 6.1d). I den här avhandlingen visas att ökad aluminiumhalt i Sc(Al1−xNix)2 mins- kar mängden väte som vill lösa sig i föreningen och att ökad mängd zirkoni- um i (Sc1−xZrx)(Co1−yNiy)2 destabiliserar metallhydriden, vilket gör att extrema tryck krävs för väteabsorption. Dessutom visas att föreningarna Nb4MSi (M=Co, Ni) bildar mycket stabila interstitiella hydrider med mycket långsam kinetik.

1F, Cl, Br och I

62 a) b)

- Ni-MH

c) d)

Figur 6.1. Realiserade och tänkbara tillämpningar för metallhydrider. a) En tänk- bar framtida tillämpning är vätelagring i vätgasdrivna bilar. b) Det största använ- dingsområdet för metallhydrider är som elektrodmaterial i uppladdningsbara nickel- metallhydridbatterier. c) Metallhydrider används för att lagra väte för att driva ubåtar. d) En tänkbar tillämpning för metallhydrider är värmelagring i termisk solkraft så att de kan producera el även på natten.

M+H-

jonbindning met.bindn. + kov. bindn.

H2 M MHx

M-H Figur 6.2. Metall-väteföreningar (metallhydrider) kan bindas samman av jon-, metall- eller kovalenta bindningar och detta har stor inverkan på föreningens egenskaper.

63 a) b)

b c

Figur 6.3. a) Två enhetsceller av NdGa där neodymjoner (Nd3+) kan ses som blå sfärer och galliummolekyljonkedjorna av Ga3− som röda sfärer bundna med svarta streck. b) En enhetscell av NdGaD1.51: hydridjoner (egentligen deuteridjoner) ses som stora röda sfärer mellan de blå neodymjonerna och väteatomer (deuteriumatomer) som små vita sfärer.

Zintlfaser ligger i gränslandet mellan föreningarna med kovalent-, jonisk- och metallbidning. Deras strukturer kan beskrivas som positiva metallkatjo- ner av en aktiv metall Man+ och molekylanjoner av en elektronegativ metall m− Mpx , där Mp försöker få åtta elektroner i sitt yttersta elektronskal genom sin jonladdning och sina kovalenta bindningar. Väte kan absorberas på två sätt i strukturen: antingen en hydridjon (H−) i hålrum mellan de positiva metall- n+ m− katjonerna (Ma ) eller kovalent bundna till molekylkatjonen (Mpx ). I den här avhandlingen visas att LnGa (Ln=Gd, Nd), som byggs upp av Ln3+-joner och Ga3−-kedjor (ﬁgur 6.3a), absorberar väte i två steg: först absorberas väte i hål mellan Ln3+-jonerna vilket kräver att Ga3− kedjorna oxide- ras till Ga2− för att laddningsbalansen ska behållas. Detta kräver att det bildas en extra bindning i Ga-kedjan för att Ga fortfarande har åtta elektroner i sitt yttersta elektronskal. I det andra steget absorberas väte i hålrum nära Ga-kedjan. Bindningen mellan Ga och H är fortfarande oklar.

64 7. Acknowledgements

“Don’t worry, everybody gets paid.” Leif Hammarström

First of all I would like to thank all my supervisors for your support and your guidance during my time as a PhD student. I am extra thankful to Martin Sahlberg for always introducing me to new people and instigating collabora- tions, Yvonne Andersson for the meticulous proofreading and Björgvin Hjör- varsson for always asking the (for a chemist) surprising questions. Secondly I would like to thank the troglodytes: First of all I would like to say I am sorry, and you are welcome, to Viktor “Abobo” Höglin for my back- seat reﬁnement. I am also thankful to Girma Gebresenbut for teaching me about quasicrystals and Ethiopian beer brands. I would also like to thank Jo- han Cedervall, Ocean Fang and Pedro Berastegui for livening up the basement. Cesar Pay Gómez deserves a special thanks for all his help with crystallography. I would also like to acknowledge the very talented summer and degree project workers Tom Faske, Maciek Kaplan and Petter Tammela for their all their help. I am also very grateful to Adrian Renne for proof-reading parts of the thesis on very short notice. The collaborators outside the UU department of chemistry who helped make this thesis possible are also acknowledged: ﬁrst of all the group of Tor- ben Jensen at Aarhus University (Morten Ley, Line Holdt Rude, Dmytro Ko- rablov, Thomas Kollin Nielsen, Flemming Besenbacher et al.) for helping me to get started with in-situ diffraction. I am also hugely indebted to Claudia Zlotea and Michel Latroche at CNRS for your help with PCT-measurements and for the opportunity to work in your lab. I am also thankful to, the ever enthusiastic, materials theory group at UU (Olle Eriksson, Rajeev Ahuja and Ralph Scheicher with a special shout-out to my beard + plaid shirt + glasses brother Robert Johansson). I am grateful to Ulrich Häussermann at Stock- holms University for handing me a GdGa sample to measure at MAX-lab which started the very interesting LnGa project. Finally I would like to thank the beam line scientists and other personnel at MAX-lab, the nuclear physics institute in the Czech republic and institute for energy technology in Norway (Yngve Cerenius, Cartsten Gunlach, Olivier Balmes, Dörthe Haase, Premysl Beran, Magnus Sørby et al.) for all their help. I am very thankful to all current and former members of the department of chemistry at Ångström for making it a great place to work, but a number

65 of people deserves special mention: Rolf Berger for teaching me why lattice energy is unfortunately named. Adam “kanske” Sobkowiak, Viktor “Eazy-E” Renman and Rickard “Sheldon” Eriksson who I like to remind to always re- member, newer forget I711. Ulf Jansson for directing my application for summer work to Yvonne and Martin which, eventually, led to this thesis. Brewing partners Mats Tydén and Gabriel Oltean in Rackarbergets Bathroom Brewery. Mikael Ottosson for discussion about beer and help with diffraction. Fang Mao for being the best possible ofﬁce-mate. Crossword wizards Sara Munk- tell and Jill Sundberg. Linus von Fiandt and Tim Nordh for coffee from their coffee machine. Fredrik “Mannen på bilden” Lindgren for having the vege- tarian blood discussion with me three times. Tomas Edvinsson for elevating the lunch-room discussions to philosophy from time to time. Mats Boman and Erik Lewin for introducing me the secrets of Glögg-making. Leif Ham- marström for providing the quote above at the end of a meeting describing the Byzantine economy of the department. The administrative and technical staff: especially Katarina “Tatti” Israelsson and Eva Larsson for help with many administrative problems, Anders Lund for building and ﬁxing innumer- able things remarkably fast. I would like to apologise to Leif Nyholm for all the where/were and plural/singular errors...

A very special thanks goes to David Rehnlund for always having my back, you are a true friend!

I thank my family for your support, especially my parents for inspiring me to pursue a scientiﬁc career.

Last but not least I would like to thank my wife Sara Frykstrand Ångström for your love and support, especially during these last few weeks before handing in the thesis.

66 Bibliography

[1] R. Boyle. “Experiment III: Of the lasting of the ﬂame of a metalline sub- stance in the same vaccum”. In: The Works of the Honourable Robert Boyle. In six volumes. To which is preﬁxed The life of the author. Vol. 6. New Experiments, touching the Relations betwixt Flame and Air. And about Explosions. Printed for J. Riverton et al. in London, 1772, pp. 575– 576. [2] G. V. E. Svedenborg. “Ballongen och dess konstruktion”. Swedish. In: Med Örnen mot polen: Andrées polarexpidition år 1897. Albert Boniers Förlag, Stockholm, 1930, pp. 48–61. URL: http://runeberg.org/ medornen/0059.html. [3] T. Graham. “On the Absorption and Dialytic Separation of Gases by Colloid Septa”. In: Philosophical Transactions of the Royal Society of London 156 (1866), pp. 399–439. DOI: 10.1098/rstl.1866.0018. [4] ICSD. Inorganic Crystal Structure Database. online database. 2014. URL: http://icsd.fiz-karlsruhe.de/ (visited on 12/10/2014). [5] Y. Liu, Y. Cao, L. Huang, M. Gao, and H. Pan. “Rare earth–Mg–Ni- based hydrogen storage alloys as negative electrode materials for Ni/MH batteries”. In: Journal of Alloys and Compounds 509.3 (2011), pp. 675 –686. DOI: 10.1016/j.jallcom.2010.08.157. [6] L. Schlapbach and A. Züttel. “Hydrogen-storage for mobile applications”. In: Nature 414 (2001), pp. 1456–1463. DOI: 10.1038/35104634. [7] D. Harries, M. Paskevicius, D. Sheppard, T. Price, and C. Buckley. “Con- centrating Solar Thermal Heat Storage Using Metal Hydrides”. In: Pro- ceedings of the IEEE 100.2 (2012), pp. 539–549. DOI: 10.1109/JPROC. 2011.2158509. [8] E. G. Ponyatovskii and I.O. Bashkin. “New Phase Transitions in Hy- drides of the I-A, III-A, and IV-A Group Metals”. In: Zeitschrift für Physikalische Chemie 146 (1985), pp. 137–157. DOI: 10.1524/zpch. 1985.146.2.137. [9] G. S. Smith, Q. C. Johnson, D. K. Smith, D. E. Cox, R. L. Snyder, R.- S. Zhou, and A. Zalkin. “The crystal and molecular structure of beryl- lium hydride”. In: Solid State Communications 67.5 (1988), pp. 491– 494. DOI: 10.1016/0038-1098(84)90168-6.

67 [10] J. Bergsma and B. O. Loopstra. “The crystal structure of calcium hydride”. In: Acta Crystallographica 15.1 (1962), pp. 92–93. DOI: 10 . 1107/S0365110X62000225. [11] N. E. Brese, M. O’Keeffe, and R. B. Von Dreele. “Synthesis and Crys- tal Structure of SrD2 and SrND and Bond Valence Parameters for Hy- drides”. In: Journal of Solid State Chemistry 88.2 (1990), pp. 571–576. DOI: 10.1016/0022-4596(90)90255-V. [12] W. Bronger, S. Chi-Chien, and P. Müller. “Die Kristallstruktur von Bari- umhydrid, ermittelt über Neutronenbeugungsexperimente an BaD2”. Ger- man. In: Zeitschrift für anorganische und allgemeine Chemie 545.2 (1987), pp. 69–74. DOI: 10.1002/zaac.19875450208. [13] T. Noritake, S. Towata, M. Aoki, Y. Seno, Y. Hirose, E. Nishibori, M. Takata, and M. Sakata. “Charge density measurement in MgH2 by synchrotron X-ray diffraction”. In: Journal of Alloys and Compounds 356- 357.0 (2003). Proceedings of the Eighth International Symposium on Metal-Hydrogen Systems, Fundamentals and Applications (MH2002), pp. 84 –86. DOI: 10.1016/S0925-8388(03)00104-X. [14] J. F. Stampfer, C. E. Holley, and J. F. Suttle. “The Magnesium-Hydrogen System1-3”. In: Journal of the American Chemical Society 82.14 (1960), pp. 3504–3508. DOI: 10.1021/ja01499a006. [15] R. W. Curtis and P. Chiotti. “Thermodynamic Properties of Calcium Hy- dride”. In: The Journal of Physical Chemistry 67.5 (1963), pp. 1061– 1065. DOI: 10.1021/j100799a027. [16] X. Luo, D. M. Grant, and G. S. Walker. “Catalytic effect of nano-sized ScH2 on the hydrogen storage of mechanically milled MgH2”. In: Jour- nal of Alloys and Compounds 622 (2015), pp. 842 –850. DOI: 10.1016/ j.jallcom.2014.10.161. [17] C. Lartigue, A. Le Bail, and A. Percheron-Guegan. “A new study of the structure of LaNi5D6.7 using a modiﬁed Rietveld method for the reﬁne- ment of neutron powder diffraction data”. In: Journal of the Less Com- mon Metals 129 (1987), pp. 65–76. DOI: 10.1016/0022- 5088(87) 90034-8. [18] A. F. Schuch and R. L. Mills. “Crystal Structure of Deuterium at Low Temperatures”. In: Physical Review Letters 16 (14 1966), pp. 616–618. DOI: 10.1103/PhysRevLett.16.616. [19] G. Aylward and T. Findlay. “Properties of the elements”. In: SI-Chemical Data. 6th Edition. Milton: John Wiley and Sons Australia Ltd, 2008. Chap. 4, pp. 5–13. [20] J. J. Reilly and R. H. Wiswall. “Formation and Properties of Iron Tita- nium Hydride”. In: Inorganic Chemistry 13.1 (1974), pp. 218–222. DOI: 10.1021/ic50131a042.

68 [21] Y. Komura and K. Tokunaga. “Structural Studies of Stacking Variants in Mg-base Friauf–Laves phases”. In: Acta Crystallographica Section B 36.7 (1980), pp. 1548–1554. DOI: 10.1107/S0567740880006565.

[22] M. Bououdina, H. Enoki, and E. Akiba. “The investigation of the Zr1−yTiy (Cr1−xNix)2–H2 system 0.0≤y≤1.0 and 0.0≤x≤1.0 Phase composition analysis and thermodynamic properties”. In: Journal of Alloys and Com- pounds 281.2 (1998), pp. 290 –300. DOI: 10.1016/S0925-8388(98) 00792-0. [23] J. H. Wernick and S. Geller. “Transition Element–Rare Rarth Com- pounds with Cu5Ca Structure”. In: Acta Crystallographica 12.9 (1959), pp. 662–665. DOI: 10.1107/S0365110X59001955. [24] J. C. Achard, A. Percheron-Guegan, H. Diaz, F. Briancourt, and F. De- many. “Hydrures ternaries de terres rares. Application au stockage de l’hydrogène”. French. In: Precedings of the 2nd international congress on hydrogen storage. Paris: Pergamon Press, 1977, 1E 12. [25] I. J. Maley, D. H. Brown, R. M. Ibberson, and C. R. Pulham. “Solid- state structures of the covalent hydrides germane and stannane”. In: Acta Crystallographica Section B 64.3 (2008), pp. 312–317. DOI: 10.1107/ S0108768108010379. [26] U. Eberle, M. Felderhoff, and F. Schüth. “Chemical and Physical Solu- tions for Hydrogen Storage”. In: Angewandte Chemie International Edi- tion 48.36 (2009), pp. 6608–6630. DOI: 10.1002/anie.200806293. [27] G. J. Miller, M. W. Schmidt, F. Wang, and T.-S. You. “Quantitative Advances in the Zintl–Klemm Formalism”. In: Zintl Phases. Ed. by Thomas F. Fässler. Vol. 139. Structure and Bonding. Springer Berlin Heidelberg, 2011, pp. 1–55. ISBN: 978-3-642-21149-2. DOI: 10.1007/ 430_2010_24. [28] U. Häussermann, V. F. Kranak, and K. Puhakainen. “Hydrogenous Zintl Phases: Interstitial Versus Polyanionic Hydrides”. In: Zintl Phases. Ed. by Thomas F. Fässler. Vol. 139. Structure and Bonding. Springer Berlin Heidelberg, 2011, pp. 143–161. ISBN: 978-3-642-21149-2. DOI: 10 . 1007/430_2010_20. [29] A. Léon, E.J. Knystautas, J. Huot, and R. Schulz. “Inﬂuence of the evaporation rate and the evaporation mode on the hydrogen sorption kinetics of air-exposed magnesium ﬁlms”. In: Thin Solid Films 496.2 (2006), pp. 683 –687. DOI: 10.1016/j.tsf.2005.08.227. [30] GfE Metalle und Materialen GmbH. Hydralloy C. data sheet. 2006. [31] Toyota Motor Europe. Toyota Ushers InThe Future With Launch Of ‘Mi- rai’ Fuel Cell Sedan. press release. 2015. URL: http://newsroom.to yota.eu/pressrelease/4124//toyota-ushers-future-launch- mirai-fuel-cell-sedan (visited on 03/10/2015).

69 [32] Toyota USA Newsroom. Toyota Mirai Hydrogen Fuel Cell Vehicle Makes East Coast Debut. press release. 2015. URL: http : / / pressroom . toyota . com / releases / toyota + mirai + washingtondc + auto + show+east+coast.htm (visited on 03/10/2015). [33] myFC. Powertrekk - Technical Information. web page. 2014. URL: ht tp : / / www . powertrekk . com / pages / powertrekk - about - the - charger (visited on 03/10/2015). [34] J. B. Hunter. US2773561 A: Silver-palladium film for separation and purification of hydrogen. U.S. Patent. 1955. URL: https://www.googl e.com/patents/US2773561. [35] J. E. Schirber and C. J. M. Northrup. “Concentration dependence of the superconducting transition temperature in PdHx and PdDx”. In: Physical Rewiew B 10.9 (1974). DOI: 10.1103/PhysRevB.10.3818. [36] T. J. Richardson, J. L. Slack, R. D. Armitage, R. Kostecki, B. Farangis, and M. D. Rubin. “Switchable mirrors based on nickel–magnesium films”. In: Applied Physics Letters 78.20 (2001). DOI: 10.1063/1.1371959. [37] Y. Cerenius, K. Ståhl, L. A. Svensson, T. Ursby, Å. Oskarsson, J. Al- bertsson, and A. Liljas. “The crystallography beamline I711 at MAXII”. In: Journal of Synchrotron Radiation 7.4 (2000), pp. 203–208. DOI: 10. 1107/S0909049500005331. [38] T. R. Jensen, T. K. Nielsen, Y. Filinchuk, J.-E. Jørgensen, Y. Cerenius, E. M. Gray, and C. J. Webb. “Versatile in situ powder X-ray diffraction cells for solid–gas investigations”. In: Journal of Applied Crystallogra- phy 43.6 (2010), pp. 1456–1463. DOI: 10.1107/S0021889810038148. [39] D. L. Kaiser and R. L. Watters. Standard Reference Material R 606b. National Institute of Standards & Technology Certificate. 2010. URL: https://www-s.nist.gov/srmors/view_cert.cfm?srm=660B. [40] A. P. Hammersley, S. O. Svensson, M. Hanfland, A. N. Fitch, and D. Häusermann. “Two-dimensional detector software: From real detector to idealised image or two-theta scan”. In: High Pressure Research 14.4- 6 (1996), pp. 235–248. DOI: 10.1080/08957959608201408. [41] P. Beran. MEREDIT - Neutron Powder Diffractometer. web page. 2015. URL: http://neutron.ujf.cas.cz/meredit (visited on 03/10/2015). [42] B. C. Hauback, H. Fjellvåg, O. Steinsvoll, K. Johansson, O. T. Buset, and J. Jørgensen. “The High Resolution Powder Neutron Diffractometer PUS at the JEEP II Reactor at Kjeller in Norway”. In: Journal of Neutron Research 8.3 (2000), pp. 215–232. DOI: 10.1080/10238160008200055. [43] BeN Systems. UnitCell version 1. Comuter Program. 2000. [44] J. Laguier and B. Bochu. CHECKCELL. computer program. 2000. URL: http://www.ccp14.ac.uk/tutorial/lmgp/.

70 [45] H. M. Rietveld. “A Proﬁle Reﬁnement Method for Nuclear and Mag- netic Structures”. In: Journal of Applied Crystallography 2.2 (1969), pp. 65–71. DOI: 10.1107/S0021889869006558. [46] J. Rodriguez-Caravajal. “Recent developments of the program FULL- PROF”. In: Comission on Powder Diffraction Newsletter 26 (2001), pp. 12–19. URL: http://www.iucr.org/resources/commissions/ powder-diffraction/newsletter. [47] V. Petricek, M. Dusek, and L. Palatinus. JANA2006–Crystallographic Computational System for Standard and Modulated Structures. computer program. 2006. URL: http://jana.fzu.cz/. [48] D. G. Westlake. “Hydrides of intermetallic compounds: A review of stabilities, stoichiometries and preferred hydrogen sites”. In: Journal of The Less-Common Metals 91.1 (1983), pp. 1–20. DOI: 10 . 1016 / 0022 - 5088(83)90091-7. [49] C. B. Magee. “Structures and stabilities of the group IIIa dihydrides”. In: Journal of the Less Common Metals 72.2 (1980), pp. 273 –290. DOI: 10.1016/0022-5088(80)90147-2. [50] P. C. P. Bouten and A. R. Miedema. “On the heats of formation of the binary hydrides of transition metals”. In: Journal of the Less Common Metals 71.1 (1980), pp. 147 –160. DOI: 10 . 1016 / 0022 - 5088(80 ) 90110-1. [51] R. Griessen and A. Driessen. “Heat of formation and band structure of binary and ternary metal hydrides”. In: Physical Review B 30 (8 1984), pp. 4372–4381. DOI: 10.1103/PhysRevB.30.4372. [52] L. G. Hector, J. F. Herbst, W. Wolf, P. Saxe, and G. Kresse. “Ab Ini- tio thermodynamic and elastic properties of alkaline-earth metals and their hydrides”. In: Physical Review B 76.1 (2007), p. 014121. DOI: 10.1103/PhysRevB.76.014121. [53] H. E. Kissinger. “Reaction Kinetics in Differential Thermal Analysis”. In: Analytical Chemistry 29.11 (1957), pp. 1702–1706. DOI: 10.1021/ ac60131a045. [54] J. P. Elder. “The general applicability of the Kissinger equation in thermal analysis”. English. In: Journal of Thermal Analysis 30.3 (1985), pp. 657–669. DOI: 10.1007/BF01913612. [55] G. Kresse, M. Marman, and J. Furthmüller. VASP the guide. manual. URL: http://cms.mpi.univie.ac.at/vasp/vasp.pdf. [56] J. P. Perdew, K. Burke, and M. Ernzerhof. “Generalized Gradient Approx- imation Made Simple”. In: Physical Review Letters 77 (1996), pp. 3865– 3868. DOI: 10.1103/PhysRevLett.77.3865.

71 [57] H. Peisl. Lattice Strains due to Hydrogen in Metals. Vol. 28. Topics in Applied Physics. Springer Berlin, 1978, pp. 53–74. ISBN: 978-3-540- 08705-2. DOI: 10.1007/3540087052_42. [58] S. Rundqvist, R. Tellgren, and Y. Andersson. “Hydrogen and deuterium in transition metal-p element compounds: Crystal chemical aspects of interstitial solid solubility and hydride phase formation”. In: Journal of the Less Common Metals 101.0 (1984). Proceedings of the Interna- tional Conference on Hydrogen in Metals, pp. 145 –168. DOI: 10.1016/ 0022-5088(84)90092-4.

[59] A. J. Goudy and R. A. Wallingford. “Desorption kinetics of LaNi5, LaNi4.7Al0.3, (CFM)Ni5 and MNi4.5Al0.5 hydrides”. In: Journal of the Less Common Metals 99.2 (1984), pp. 249 –256. DOI: 10.1016/0022- 5088(84)90222-4. [60] E. W. Collings, R. D. Smith, and R. G. Lecander. “Magnetic susceptibil- ity studies of the cubic laves phases Sc(Ni1−xCox)2”. In: Journal of the Less Common Metals 18.3 (1969), pp. 251–266. DOI: 10.1016/0022- 5088(69)90163-5. [61] J. M. D. B. Oliveira and I. R. Harris. “Valency compensation in the Laves system, Ce(Co1−xNix)2”. In: Journal of Materials Science 18.12 (1983), pp. 3649–3660. DOI: 10.1007/BF00540737. [62] M. Yoshida and E. Akiba. “Hydrogen absorbing properties of ScM2 Laves phase alloys (M = Fe, Co and Ni)”. In: Journal of Alloys and Compounds 226.1-2 (1995), pp. 75–80. DOI: 10.1016/0925-8388(95) 01611-2. [63] S. M. Filipek, V. Paul-Boncour, N. Kuriyama, N. Takeichi, H. Tanaka, Liu R.-S, R. Wierzbicki, R. Sato, and H.-T. Kuo. “Hydrides of Laves phases intermetallic compounds synthesized under high hydrogen pressure”. In: Solid State Ionics 181.5-7 (2010), pp. 306–310. DOI: 10 . 1016/j.ssi.2010.01.009. [64] L. C. Beavis. “Characteristics of some binary transition metal hydrides”. In: Journal of the Less Common Metals 19.4 (1969), pp. 315–328. DOI: 10.1016/0022-5088(69)90177-5. [65] B. Mayer, H. Anton, E. Bott, M. Methfessel, J. Sticht, J. Harris, and P. C. Schmidt. “Ab-initio calculation of the elastic constants and thermal expansion coefﬁcients of Laves phases”. In: Intermetallics 11.1 (2003), pp. 23 –32. DOI: 10.1016/S0966-9795(02)00127-9.

[66] E. I. Gladyshevskii and Y. B. Kuz’ma. “The compounds Nb4FeSi, Nb4- CoSi, Nb4NiSi and their crystal structures”. In: Journal of Structural Chemistry 6.1 (1965), pp. 60–63. DOI: 10.1007/BF00743870.

72 [67] M. Vennström and Y. Andersson. “Hydrogen absorption in Nb4CoSi and Nb4NiSi”. In: Journal of Alloys and Compounds 364.1-2 (2004), pp. 141–145. DOI: 10.1016/S0925-8388(03)00539-5. [68] G. Ryu, S. W. Kim, S. Matsuishi, H. Kawaji, and H. Hosono. “Supercon- ductivity in Nb4MSi (M=Ni, Co, and Fe) with a quasi-two-dimensional Nb network”. In: Physical Review B 84.22 (2011), p. 224518. DOI: 10. 1103/PhysRevB.84.224518. [69] D. G. Westlake. “Stoichiometries and interstitial site occupation in the hydrides of ZrNi and other isostructural intermetallic compounds”. In: Journal of the Less-Common Metals 75.2 (1980), pp. 177–185. DOI: 10.1016/0022-5088(80)90115-0. [70] J. Ångsröm, T. Faske, and M. Sahlberg. unpublished material. 2015. [71] O. Schob and E. Parthé. “AB compounds with Sc, Y and rare earth metals. I. Scandium and yttrium compounds with CrB and CsCl structure”. In: Acta Crystallographica 19.2 (1965), pp. 214–224. DOI: 10.1107/ S0365110X65003134. [72] A. E. Dwight, J. W. Downey, and R. A. Conner Jnr. “Equiatomic compounds of Y and the lanthanide elements with Ga”. In: Acta Crystallo- graphica 23.5 (1967), pp. 860–862. DOI: 10.1107/S0365110X67003834. [73] R. A. Susilo, J. M. Cadogan, D. H. Ryan, N. R. Lee-Hone, R. Cobas, and S. Muñoz Pérez. “Spin-reorientation in GdGa”. English. In: Hyperﬁne Interactions 226.1-3 (2014), pp. 257–266. DOI: 10 . 1007 / s10751 - 013-0924-4. [74] H. Fahlquist, D. Noréus, S. Callear, W. I. F. David, and B. C. Hauback. 5− “Two New Cluster Ions Ga[GaH3]4 with a Neopenthane Structure in n− Rb8Ga5H15 and [GaH]n in Rbn(GaH2)n, Represent a New Class of Compunds with Direct Ga-Ga Bonds Mimiking Common Hydrocar- bons”. In: Journal of the American Chemical Society 133.37 (2011), pp. 14574–14577. DOI: 10.1021/ja2067687. [75] T. Björling, D. Noréus, and U. Häussermann. “Polyanionic Hydrides from Polar Intermetallics AeE2 (Ae = Ca, Sr, Ba; E = Al, Ga, In)”. In: Journal of the American Chemical Society 128.3 (2006), pp. 817–824. DOI: 10.1021/ja054456y. [76] J. Y. Kim, R. Hennig, V. T. Huett, P. C. Gibbons, and K. F. Kelton. “Hydrogen absorption in Ti–Zr–Ni quasicrystals and 1/1 approximants”. In: Journal of Alloys and Compounds 404-406.0 (2005). Proceedings of the 9th International Symposium on Metal-Hydrogen Systems, Funda- mentals and Applications (MH2004) The International Symposium on Metal-Hydrogen Systems, Fundamentals and Applications, {MH2004}, pp. 388 –391. DOI: 10.1016/j.jallcom.2005.02.089.

73 [77] R. G. Hennig, E. H. Majzoub, and K. F. Kelton. “Location and energy of interstitial hydrogen in the 1/1 approximant W-TiZrNi of the icosahedral TiZrNi quasicrystal: Rietveld reﬁnement of x-ray and neutron diffraction data and density-functional calculations”. In: Physical Re- view B 73.18 (2006), p. 184205. DOI: 10.1103/PhysRevB.73.184205. [78] W.J. Kim and K. F. Kelton. “Icosahedral-phase formation and stability in Ti–Zr–Co alloys”. In: Philosophical Magazine Letters 74.6 (1996), pp. 439–448. DOI: 10.1080/095008396179977.

Acta Universitatis Upsaliensis Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1234 Editor: The Dean of the Faculty of Science and Technology

A doctoral dissertation from the Faculty of Science and Technology, Uppsala University, is usually a summary of a number of papers. A few copies of the complete dissertation are kept at major Swedish research libraries, while the summary alone is distributed internationally through the series Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology. (Prior to January, 2005, the series was published under the title “Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology”.)

ACTA UNIVERSITATIS UPSALIENSIS Distribution: publications.uu.se UPPSALA urn:nbn:se:uu:diva-245046 2015