Hydrogen in

Intermetallic Compounds I
Electronic, Thermodynamic,
and Crystallographic Properties, Preparation
Edited by L. Schlapbach

With Contributions by
P. Fischer T.B. Flanagan R. Griessen M. Gupta
G. Hilscher W.A. Oates A. Percheron-Gu6gan
T. Riesterer L. Schlapbach J.-M. Welter
G. Wiesinger K. Yvon

With 118Figures and 12Tables

Springer-Verlag Berlin Heidelberg NewYork

London Paris Tokyo
Dr. Louis Schlapbach
L a b o r a t o r i u m ffir F e s t k 6 r p e r p h y s i k , E T H Zfirich, H 6 n g g e r b e r g ,
CH-8093 Ziirich, Switzerland and
Institut de Physique, Universit6 de F r i b o u r g ,
C H - 1 7 0 0 F r i b o u r g , Switzerland

ISBN 3-540-18333-7 Springer-Verlag Berlin Heidelberg New York

ISBN 0-387-18333-7 Springer-Verlag New York Berlin Heidelberg

Library of Congress Cataloging-in-Publication Data. Hydrogen in intermetallic compounds/edited by

L. Schlapbach; with contributions by P. Fischer... let al.]. p. cm. - (Topics in applied physics; v. 63)
Includes index. Contents: v. 1. Electronic, thermodynamic, and crystallographic properties, preparation.
1. Intermetallic compounds -- Hydrogen content. I. Schlapbach, L. (Louis), 1944-. II. Fischer, P. (Peter)
III. Series. QD171.H94 1988 546'.3-dc 19 87-32257
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,
reproduction on microfilms or in other ways, and storage in data banks. Duplication of this publication ,or
parts thereof is only permitted under the provisions of the German Copyright Law of September 9,1965, in
its version of June 24, 1985, and a copyright fee must always be paid. Violations fall under the prosecution
act of the German Copyright Law.
~2) Springer-Verlag Berlin Heidelberg 1988
Printed in Germany
The use of registered names, trademarks, etc. in this publication does not imply, even in the absence of a
specific statement, that such names are exempt from the relevant protective laws and regulations and
therefore free for general use.
Typesetting, printing and binding: Brtihlsche Universit/itsdruckerei, 6300 Giessen
2! 53/3150-5432!0
Preface

Phenomena related to the topic "Hydrogen in Metals" and the physics behind
them as they were known at the end of the seventies are reviewed in the books
Hydrogen in Metals, Vols. I and II, edited by G. Alefeld and J. V61kl (Vols. 28
and 29 of this series). These books, which treat mostly hydrogen in elemental
metals, are still very valuable and will continue to be important in the coming
years.
Since their publication, a lot of new results have been obtained, which
have improved both our knowledge and understanding considerably. The
work was stimulated by the extraordinary properties of hydrogen-storing
intermetallic compounds and by the energy crisis. The recent progress, which
has more often concerned intermetallic compounds and alloys than elemental
metals, has been reviewed in many good articles either at a technical, purely
scientific, or at a more popular level. Until now, however, the details of this
important field have never been drawn together and presented in the form of
a book.
Thus, the aim of the publication of the two volumes "Hydrogen in
Intermetallic Compounds" is to give a thorough description of the various
aspects of the topic in a series of chapters written by specialists in the field and
to review major progress on hydrogen in and on elemental metals. This volume
begins with a description of the preparation of intermetallics and their
hydrides. It contains further chapters on crystallographic, thermodynamic,
electronic and magnetic properties and on heat of formation models. Volume II
will be devoted to aspects of the kinetics and dynamics of hydrogen, to surface
phenomena and to applications and experimental techniques.
As the interaction of hydrogen with metals and alloys is, and will be, of
significant importance for basic research as well as for hydrogen energy
technology, fusion, catalysis, getters, electrochemical cells and for many more
applications, I hope that these two volumes will help many scientists to find the
information they are looking for, to spread the fascination which we the
authors already share, and to stimulate further work.
I should like to express my thanks to all the authors for their individual
contributions and for their willing and fruitful cooperation and to Angela
Lahee and to my wife Christine for careful reading of the manuscripts and
proofs.

Zurich, January 1988 Louis Schlapbach

Contents

1. Introduction
By L. Schlapbach (With 2 Figures) . . . . . . . . . . . . . . . 1
1.1 M e t a l - H y d r o g e n Systems and Related Phenomena . . . . . . 1
1.2 Scope of These Two Volumes . . . . . . . . . . . . . . . 3
1.3 Recent Highlights and Outlook . . . . . . . . . . . . . . 6
References . . . . . . . . . . . . . . . . . . . . . . . . . 8

2. Preparation of Intermetallics and Hydrides

By A. Percheron-Gu6gan and J.-M. Welter
(With 30 Figures and 1 Table) . . . . . . . . . . . . . . . . 11
2,1 Introduction . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Thermodynamic Aspects o f Intermetallic Phases and
Their Synthesis . . . . . . . . . . . . . . . . . . . . . 12
2.3 Preparation of Intermetallic Phases: L a b o r a t o r y Scale . . . . 17
2.Yl The Starting Materials . . . . . . . . . . . . . . . 17
2.3.2 Melting Techniques . . . . . . . . . . . . . . . . . 19
2.3,3 Annealing Treatments . . . . . . . . . . . . . . . . 24
2.3.4 Single Crystal Growth Techniques . . . . . . . . . . 27
2.3,5 Noncrystalline Materials: Rapid Solidification
and Solid-State Reactions . . . . . . . . . . . . . . 28
2.3.6 Synthesis from the Gas Phase and Thin Films . . . . . 28
2,4 Characterization of Intermetallic Phases . . . . . . . . . . 29
2.5 Preparation of Intermetallic Phases: Industrial Scale . . . . . 32
2.6 Hydrogenation o f Intermetallic Phases . . . . . . . . . . . 37
2.7 Concluding Remarks . . . . . . . . . . . . . . . . . . . 40
References . . . . . . . . . . . . . . . . . . . . . . . . . 44

3. Thermodynamics of Intermetailic Compound-Hydrogen Systems

By T. B. Flanagan and W. A, Oates (With I4 Figures and 3 Tables). 49
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 49
3.2 Macroscopic Thermodynamics . . . . . . . . . . . . . . 50
3.2.1 Single and Two Condensed Phase Equilibria . . . . . . 50
a) Integral/Formation Properties . . . . . . . . . . . 50
b) Partial Molar Properties . . . . . . . . . . . . . 52
c) Plateau Properties . . . . . . . . . . . . . . . . 53
3.2.2 Three Condensed Phase Equilibria . . . . . . . . . . 56
VIII Contents

3.3 Experimental Aspects . . . . . . . . . . . . . . . . . . 57

3.3.1 p - r - T M e a s u r e m e n t s . . . . . . . . . . . . . . . . 57
3.3.2 C a l o r i m e t r i c M e a s u r e m e n t s . . . . . . . . . . . . . 58
a) R e a c t i o n C a l o r i m e t r y . . . . . . . . . . . . . . 58
b) T h e r m a l A n a l y s i s . . . . . . . . . . . . . . . . 58
c) H e a t C a p a c i t y . . . . . . . . . . . . . . . . . . 59
3.3.3 P r o b l e m s in O b t a i n i n g A c c u r a t e T h e r m o d y n a m i c D a t a 59
a) H y s t e r e s i s . . . . . . . . . . . . . . . . . . . 59
b) C h a r a c t e r i z a t i o n o f the S t a r t i n g I M C - A c t i v a t i o n
and Annealing . . . . . . . . . . . . . . . . . 60
c) S l o p i n g P l a t e a u x . . . . . . . . . . . . . . . . 61
d) D e c o m p o s i t i o n . . . . . . . . . . . . . . . . . 62
e) A m o r p h i z a t i o n . . . . . . . . . . . . . . . . . 62
3.4 R e s u l t s for R e p r e s e n t a t i v e S y s t e m s . . . . . . . . . . . . . 63
3.4.1 L a N i 5 a n d S o m e L a N i s - B a s e d S y s t e m s . . . . . . . . 63
3.4.2 T i F e . . . . . . . . . . . . . . . . . . . . . . . 64
3.4.3 ZrV2 a n d S o m e Z r M n z - B a s e d S y s t e m s . . . . . . . . 66
3.4.4 E r F % . . . . . . . . . . . . . . . . . . . . . . . 67
3.4.5 M g - B a s e d I M C s . . . . . . . . . . . . . . . . . . 68
3.5 M i c r o s c o p i c T h e r m o d y n a m i c s . . . . . . . . . . . . . . . 69
3.5.1 C o n f i g u r a t i o n a l M o d e l s . . . . . . . . . . . . . . . 69
a) Site a n d I n t e r a c t i o n E n e r g i e s . . . . . . . . . . . 69
b) H a r d C o r e , C o n s t a n t V0(j) M o d e l s . . . . . . . . . 71
c) M u l t i p l e P l a t e a u x M o d e l s . . . . . . . . . . . . . 73
3.5.2 N o n - C o n f i g u r a t i o n a l C o n t r i b u t i o n s . . . . . . . . . . 76
3.5.3 P h a s e T r a n s i t i o n s . . . . . . . . . . . . . . . . . 78
a) T h e ~ - ~ ' T r a n s i t i o n . . . . . . . . . . . . . . . 79
b) O r d e r / D i s o r d e r T r a n s i t i o n s . . . . . . . . . . . . 79
3.5.4 C o m p l e t e a n d P a r t i a l E q u i l i b r i u m . . . . . . . . . . 80
3.5.5 H y s t e r e s i s . . . . . . . . . . . . . . . . . . . . . 81
3.6 C o n c l u s i o n s . . . . . . . . . . . . . . . . . . . . . . 82
References . . . . . . . . . . . . . . . . . . . . . . . . . 83

4. Crystal and Magnetic Structures of Ternary Metal Hydrides:

A Comprehensive Review
By K. Y v o n a n d P. F i s c h e r ( W i t h 4 F i g u r e s a n d 3 Tables) . . . . . 87
4.1 I n t r o d u c t i o n . . . . . . . . . . . . . . . . . . . . . . 87
4.2 E x p e r i m e n t a l M e t h o d . . . . . . . . . . . . . . . . . . 89
4.2.1 S a m p l e P r e p a r a t i o n . . . . . . . . . . . . . . . . . 89
4.2.2 N e u t r o n D i f f r a c t i o n . . . . . . . . . . . . . . . . 90
4.2.3 S t r u c t u r e R e f i n e m e n t . . . . . . . . . . . . . . . . 91
4.3 S t r u c t u r a l R e s u l t s . . . . . . . . . . . . . . . . . . . . 93
4.3.1 M e t a l A t o m S u b s t r u c t u r e . . . . . . . . . . . . . . 93
4.3.2 D i s t r i b u t i o n o f H y d r o g e n A t o m s . . . . . . . . . . . 106
4.3.3 I n t e r a t o m i c D i s t a n c e s . . . . . . . . . . . . . . . . 109
 Contents IX

4.4 Stability Aspects . . . . . . . . . . . . . . . . . . . . 112

4.4.1 Interstitial Hole Size . . . . . . . . . . . . . . . . 113
4.4.2 Electrostatic Repulsion . . . . . . . . . . . . . . . 114
4.4.3 Electronic Factors . . . . . . . . . . . . . . . . . 115
4.4.4 Attractive Nearest-Neighbor Interactions . . . . . . . 116
4.5 Magnetic Ordering . . . . . . . . . . . . . . . . . . . 117
4.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . 118
4.A Appendix. Table: Structural D a t a for Ternary Hydrides Studied
by Neutron Diffraction . . . . . . . . . . . . . . . . . . 119
4.B Appendix. Table: Magnetic Structures . . . . . . . . . . . 130
References . . . . . . . . . . . . . . . . . . . . . . . . . 132
5. Electronic Properties
By M. G u p t a and L. Schlapbach (With 25 Figures and 2 Tables) . 139
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . 139
5.2 Theoretical and Experimental Methods: General F e a t u r e s . . . 140
5.2.1 Theoretical Methods . . . . . . . . . . . . . . . . 140
a) Stoichiometric Hydrides . . . . . . . . . . . . . 142
b) Nonstoichiometric Hydrides . . . . . . . . . . . . 146
c) H Impurity in Metals . . . . . . . . . . . . . . 146
d) Calculation of Other Observables . . . . . . . . . 147
5.2.2 Experimental Methods . . . . . . . . . . . . . . . 148
5.3 Results for Binary Hydrides . . . . . . . . . . . . . . . 150
5.3.1 Hydrides of Pd and Ni . . . . . . . . . . . . . . . 150
a) Energy Bands and Densities of States . . . . . . . 150
b) Fermi Surfaces (FS) . . . . . . . . . . . . . . . 160
c) Study of Other Observables . . . . . . . . . . . . 160
5.3.2 Hydrides of Ti, Zr, and H f . . . . . . . . . . . . . 162
5.3.3 Hydrides of Nb, V, and Ta . . . . . . . . . . . . . 166
5.3.4 Hydrides of Rare Earth Metals
Including Sc, Y, and La . . . . . . . . . . . . . . . 168
5.3.5 Hydrides o f Ca and Mg . . . . . . . . . . . . . . . 176
5.3.6 Hydrides of Actinide Metals . . . . . . . . . . . . . 176
5.4 Results on Ternary Hydrides . . . . . . . . . . . . . . . 177
5.4.1 Hydrides of A B s C o m p o u n d s (Haucke Compounds) . 177
5.4.2 Hydrides of FeTi . . . . . . . . . . . . . . . . . . 185
5.4.3 MgzNi and Isoelectronic C o m p o u n d s and Their Hydrides. 193
5.4.4 The A B z Laves Phase, and Related C o m p o u n d s
and Their Hydrides . . . . . . . . . . . . . . . . . 201
5.4.5 A m o r p h o u s Alloys and Their Hydrides . . . . . . . . 205
5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . 207
References . . . . . . . . . . . . . . . . . . . . . . . . . 209
6. Heat of Formation Models
By R. Griessen and T. Riesterer (With 23 Figures and 3 Tables) . . 219
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . 219
X Contents

6.2 Thermodynamics o f Hydrogen Absorption in Metals . . . . . 222

6.3 Empirical Correlations . . . . . . . . . . . . . . . . . . 231
6.4 Semi-empirical Models . . . . . . . . . . . . . . . . . . 234
6.4.1 General Properties of Semi-empirical Models . . . . . . 235
a) Binary Alloys . . . . . . . . . . . . . . . . . . 235
b) Binary Metal Hydrides . . . . . . . . . . . . . . 237
c) Ternary Metal Hydrides . . . . . . . . . . . . . 239
d) Local or Site-Dependent Enthalpies . . . . . . . . 241
6.4.2 The Atomic-Cell Model . . . . . . . . . . . . . . . 244
a) Intermetallic C o m p o u n d s and Alloys . . . . . . . . 244
b) Binary Metal Hydrides . . . . . . . . . . . . . . 245
c) Ternary Metal Hydrides . . . . . . . . . . . . . 247
d) Quaternary Metal Hydrides . . . . . . . . . . . . 248
6.4.3 The Band-Structure Model . . . . . . . . . . . . . 249
a) Binary Metal Hydrides . . . . . . . . . . . . . . 249
b) Ternary Metal Hydrides . . . . . . . . . . . . . 255
c) Volume Effects . . . . . . . . . . . . . . . . . 260
6.4.4 The Local Band-Structure Model . . . . . . . . . . . 262
6.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . 265
6.6 Tables . . . . . . . . . . . . . . . . . . . . . . . . . 266
References . . . . . . . . . . . . . . . . . . . . . . . . . 279
Additional References . . . . : . . . . . . . . . . . . . . . 284

7. Magnetic Properties, Mi~ssbauer Effect and Superconductivity

By G. Wiesinger and G. Hilscher (With 20 Figures) . . . . . . . 285
7.1 Introduction and Scope . . . . . . . . . . . . . . . . . 285
7.2 Theoretical Background . . . . . . . . . . . . . . . . . 286
7.2.1 Theoretical Concepts in Magnetism . . . . . . . . . . 286
7.2.2 Physical Concepts of M6ssbauer Spectroscopy . . . . . 291
a) Introductory Remarks . . . . . . . . . . . . . . 291
b) Experimental Techniques . . . . . . . . . . . . . 292
c) Hyperfine Interactions . . . . . . . . . . . . . . 293
7.3 Magnetism and M6ssbauer S p e c t r o s c o p y - Experimental
Results and Discussion . . . . . . . . . . . . . . . . . . 295
7.3.1 General Remarks . . . . . . . . . . . . . . . . . . 295
7.3.2 Transition-Metal-Rich R-TM C o m p o u n d s . . . . . . . 296
a) R2TM17 . . . . . . . . . . . . . . . . . . . . 296
b) R T M 5 . . . . . . . . . . . . . . . . . . . . . 296
c) R2TM7 (Space G r o u p P63/mmc No. 194) . . . . . . 301
d) R2FelgB-Type C o m p o u n d s . . . . . . . . . . . . 302
e) R6TM23 ( T M = M n , Fe) . . . . . . . . . . . . . 303
f) RTM3 ( T M = F e , Co, Ni) . . . . . . . . . . . . . 307
7.3.3 Laves-Phase-Type C o m p o u n d s . . . . . . . . . . . . 309
a) Mn Compounds . . . . . . . . . . . . . . . . . 309
b) Fe C o m p o u n d s . . . . . . . . . . . . . . . . . 312
 Contents XI

c) Fe-Containing Pseudobinaries . . . . . . . . . . . 315

d) C o m p o u n d s of Co, Ni and Others . . . . . . . . . 317
7.3.4 TiFe and Related CsC1-Type C o m p o u n d s (Space G r o u p
Pm3m, No. 221) . . . . . . . . . . . . . . . . . . 318
7.3.5 C o m p o u n d s of Low Transition Metal Content . . . . . 323
7.3.6 A m o r p h o u s Alloys . . . . . . . . . . . . . . . . . . 325
7.4 Superconductivity . . . . . . . . . . . . . . . . . . . . 329
References . . . . . . . . . . . . . . . . . . . . . . . . . 332
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . 343
Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . 345
Contributors

Fischer, Peter
Labor ffir Neutronenstreuung, ETH Zfirich,
CH-5303 Wfirenlingen, Switzerland

Flanagan, Ted B.
Chemistry Department, University of Vermont,
Burlington, VT 05405, USA

Griessen, Ronald
Natuurkundig Laboratorium, Vrije Universiteit, De Boelelaan 1081,
NL-1081 HV Amsterdam, The Netherlands

Gupta, Mich61e
Institut des Sciences des Materiaux, U.A. 446 du C.N.R.S.,
Bgtt. 415, Universit6 Paris-Sud, F-91405 Orsay, France

Hilscher, Gerfried
Institut ffir Experimentalphysik, TU Wien,
A-1040 Wien, Austria

Oates, W. Alan
Department of Chemical and Materials Engineering,
The University of Newcastle, Shortland, N.S.W. 2308, Australia

Percheron-Gu6gan, Annick
Laboratoire de Chimie M6tallurgie des Terres Rares, C.N.R.S.,
1, Place A. Briand, F-92190 Meudon, France

Riesterer, Thomas
Laboratorium ffir Festk6rperphysik, ETH Zfirich,
CH-8093 Zfirich, Switzerland
XIV Contributors

Schlapbach, Louis
Laboratorium fiir Festk6rperphysik, ETH Ziirich,
CH-8093 Z/irich, Switzerland and
Institut de Physique, Universit6 de Fribourg,
CH-1700 Fribourg, Switzerland

Welter, Jean-Marie
Institut ffir Festk6rperforschung der Kernforschungsanlage Jfilich GmbH,
Postfach 1913, D-5170 Jtilich, Fed. Rep. of Germany
Permanent address: Tr6fimetaux, Centre de Recherche, B.P. 11,
F-60590 Serifontaine, France

Wiesinger, G/inter
Institut fiir Experimentalphysik, TU Wien,
A-1040 Wien, Austria

Yvon, Klaus
Laboratoire de Cristallographie aux Rayons X, Universit6 de Gen6ve,
24, Quai E. Ansermet, CH-1211 Gen6ve 4, Switzerland
1. Introduction

Louis Schlapbach

With 2 Figures

This short introduction to the more general topic of hydrogen in and on metals
starts with a simple description of metal-hydrogen systems and of related
phenomena. Then the relevance of research on metal-hydrogen systems for
solid state science and for applications will be emphasized, the discovery of
hydrides of intermetallics and their differences to binary hydrides will be
sketched and references to standard literature will be given. Finally, in the third
part, Sect. 1.3, some highlights of the topics treated in this volume will be
outlined.

1.1 Metal-Hydrogen Systems and Related Phenomena

A metal-hydrogen system consists of a metal (possibly containing dissolved
hydrogen), hydrogen in the gaseous or in a condensed phase, and of an interface
in between. The metal can be a high purity single crystal, a crystalline ordered
or disordered or amorphous single-phase alloy or a multiphase alloy of
technical purity. Hydrogen, and also its isotopes deuterium and tritium, can be
present as a clean molecular gas, a gas mixture, dissolved in an organic or
aequous liquid, as hydrogen atoms, or as ionized atoms in a hydrogen plasma
or for hydrogen implantation. The interface may also be a clean crystalline,
ordered or disordered or amorphous metallic surface, it may be oxidized,
contaminated or really dirty. Some examples are
- H2 gas and a Pd single crystal with the (H-induced) reconstructed (110)
surface as interface,
- a LaNi 5 powder electrode in an aequous K O H solution as electrolyte,
- H~-containing coal gas in a carbon steel pipeline.
A simplified and frequently used (one-dimensional) model to describe H2
gas and a hydrogen dissolving metal is shown in Fig. 1.1. A hydrogen molecule
approaching the metal can be dissociated at the interface, adsorbed at
appropriate surface and near surface sites and dissolved at interstitial sites of
the host metal. If the local hydrogen concentration exceeds a certain limit, a
hydride phase precipitates.
The thermodynamic aspects of hydride formation (gaseous hydrogen) can
be described by pressure-composition isotherms (pcT curves, Fig. 1.2). The host
metal dissolves some hydrogen as a solid solution (s-phase). As the He pressure,
2 L. Schlapbach

H2 GAS INTERFACE METAL Fig. 1.1. Simplified model of the dis-

sociation of molecular hydrogen at an
interface and of the solution of hydro-
% gen atoms in the bulk. The potential
energy indicates from left to right a
shallow minimum for physisorbed Hz,

o0 a deep minimum for chemisorbed dis-

sociated H, a rather deep minimum for
near surface hydrogen and periodic
minima for H dissolved on interstitial
O sites of the host metal, separated from
each other by diffusion barriers

Peq ('I"2)
//

.
\\/%
\/ . . . .

\\

CH
~ H

liT
JR

Fig. 1.2. Pressure-composition isotherms for the solid solution of hydrogen (s-phase) and
hydride formation (fl-phase). The region of coexistence of the two phases is characterized by
the flat plateau at the equilibrium pressure pea(T) and ends at the critical temperature To.The
enthalpy of hydride formation AH is obtained from the variation of the equilibrium pressure
(on a logarithmic scale!) with temperature in a van't Holt plot

and herewith the concentration cn of dissolved H, is increased, the H - H

interaction becomes locally important and nucleation and growth of the
hydride phase (/~) start. Whilst the two phases ~ and/~ coexist, the isotherms
show a plateau, the length of which determines how much H2 can be stored
reversibly with small pressure variations. In the pure/%phase the H 2 pressure
rises steeply with the concentration. At higher H2 pressure further plateaux and
further hydride phases m a y be formed. The two-phase region ends in a critical
point T~. The plateau pressure pea(T), strongly depends on temperature. F r o m
the slope of a so-called van't Hoff plot of the plateau pressure (on a logarithmic
scale) versus T - 1 , the heat of hydride formation A H can be evaluated (see
Chap. 3).
 Introduction 3

The solution of hydrogen in the bulk at interstitial sites of the host metal and
the formation of metal hydride phases affect the host metal properties and
induce a variety of phenomena:
Crystal Structure. The lattice expands, the crystal structure changes mostly
with a reduction of symmetry, a H-subtattice is formed, order-disorder
transitions of the H-sublattice occur, lattice defects are formed, they interact
and diffuse, strain fields grow, and even non crystalline and probably quasi-
crystalline materials can be formed.
Phase Diagram. The hydrogen metal and hydrogen-hydrogen interaction
(nearest neighbours, next nearest neighbours .... ) are the origin of the occur-
rence of a variety of phases as a function of temperature and hydrogen
concentration and phase transitions. Hydrogen in metals can often be treated
as a lattice gas.
Electronic and Magnetic Properties. The lattice expansion, the additional
electron and proton strongly perturb electrons and phonons. Metal-semi-
conductor transitions, loss and appearance of magnetic moments, of magnetic
order, and also of superconductivity can be observed.
Cohesion. The change of the phonon spectra and of the electronic properties
deteriorates the mechanical, particularly elastic properties and leads to
decohesion, embrittlement, and even to disintegration into powder.
Dynamics of Hydrogen. The hydrogen atoms and their isotopes (D, T, g+)
vibrate about their equilibrium positions, perform local motions and long-
range diffusion. The high mobility allows a redistribution of hydrogen atoms
and the formation of concentration gradients under the influence of mechanical
deformation (strain field), temperature gradient (thermotransport), electric and
magnetic fields (electromigration, proton Hall effect). Quantum diffusion
occurs down to the 10 K range.
Surface. The sorption of hydrogen covers adsorption and desorption, surface
migration, surface reconstruction and surface segregation, chemical reduction,
surface hydride formation and embrittlement.
It is evident that these effects and phenomena are at the same time exciting
for basic research and important for technical applications and all steps in
between.

1.2 Scope of These Two Volumes

Since T. Graham's observation of the ability of Pd to absorb large amounts of
hydrogen in 1866, i.e. almost a hundred years after the discovery of hydrogen by
Lavoisier in 1783, who had followed Cavendish's work on "factitious air",
4 L. Schlapbach

research on hydrogen in and on metals has seen various motivations and many
ups and downs. In addition to the continuous interest of physicists and chemists
in metal hydrides as an interesting class of materials as well as in the hydrogen
adsorption as a prototype step of a catalyzed reaction, and of metallurgists
fighting against hydrogen embrittlement, metal-hydrogen systems frequently
became "runners". This was the case e.g. after the recognition of their potential
application as moderators in nuclear reactors or after the discovery of the
astonishing hydrogen sorption properties ofintermetallie compounds and their
use for the reversible storage of hydrogen as an energy carrier.
A no less important part of the motivation is, however, related to basic
research and to the fact that many key phenomena of solid state science can be
studied in metal-hydrogen systems: magnetism and superconductivity, heavy
electron behaviour, alloy formation, superlattices, fraetals, metal-
semiconductor transitions in the bulk and at the surface, surface reconstruc-
tion and surface segregation, phase diagrams, order-disorder transitions,
quantum diffusion, formation of disordered and non-crystalline materials and
many more.
The results of the world wide research efforts on metal-hydrogen systems
are published in a variety of journals covering physics, chemistry, metallurgy,
and engineering. A large part of the results concerning metal hydrides are
regularly presented at the "International Symposium on the Properties and
Applications of Metal Hydrides" and at the "International Conference on
Hydrogen in Metals", whose proceedings are published as special issues of the
Journal of Less-Common Metals [1.1] and Zeitschrift ffir physikalische Chemie
[1.2]. The two conferences will merge in 1988. Hydrogen-energy related
papers are often presented at the World Hydrogen Energy Conference [1.3] or
published in the International Journal of Hydrogen Energy.
Two NATO Conferences were dedicated to the electronic structure of
hydrogen metals [1.4] and to the effect of disorder and amorphicity on metal-
hydrogen systems [1.5]. Energy storage in metal hydrides was reviewed with
emphasis on automotive applications [1.6] and on energy technology in
general [1.7].
Among the further review papers written within the last few years some
introductory papers at a popular level should be mentioned [1.8], as well as
reviews which focus on particular materials: Zr based alloys and ABE-Laves
phase alloys [ 1.9]; the physicochemistry of intermetallic compounds and their
hydrides [1.10]; atomistic and electronic approaches to hydrogen in bcc metals
[1.11]; and finally two rather extended reviews focussed on thermodynamic
and electronic properties of hydrides formed from intermetallic compounds of
rare earth and transition metals [1.12], which were finished at the end of 1981
and 1983, respectively.
As is evidenced by the title these two books on Hydrogen in Intermetallic
Compounds, Vols. I and II, give a rather complete description of our knowledge
and understanding of hydrogen in and on intermetallic compounds of the
major families. Furthermore, some important results on hydrogen in and on
 Introduction 5

elemental metals and in some amorphous alloys are given in order that the
description of the state of the art at the end of 1986 is as complete as possible.
Intermetallic compounds are, as will be described in detail in Chap. 2 on
preparation, a special case of ordered single-phase alloys, characterized by
stoichiometric (or near stoichiometric) concentration ratios of the two metallic
components. The major differences between hydrides of intermetallic com-
pounds and those of elemental metals are:
- larger variety of interstitial sites
- wide range of stability, which can be adjusted by substitutions
- tendency to decompose into a binary hydride and another intermetallic
compound (metastable)
- very reactive to hydrogen due to surface segregation, which prevents
passivation
- formation of hydrides of intermetallic compounds whose components do
not form binary hydrides and vice versa.
The first extended reports on the reaction of hydrogen with alloys and
intermetallic compounds were written by Beck [1.13] and by Pebler and
Gulbranson [1.14], who studied very stable hydrides and the detrimental effect
of hydrogen on the use of Zr alloys in nuclear reactions.
The discovery and description of hydride formation of Mg2Ni [1.15] and
FeTi [1.16] by Reilly and Wiswall and particularly of the easy hydride
formation of LaNi 5 by van Vucht et al. [-1.17] initiated a worldwide series of
studies of hydrogen storage in intermetallic compounds. The solution of
hydrogen and deuterium in LaNi 5 had already been studied a few years earlier
by Neumann [1.18]. Although he observed the formation of a new phase with
expanded lattice constants (those of LaNi 5H6)and the onset of disintegration of
the sample into small pieces which desorbed hydrogen gas, he did not realize the
consequences of this observation. Several years later hydrides of AB z type
Laves phase compounds were independently discovered by Shaltiel et al. [1.19]
and Ishido et al. [1.20]. Then hydrogen absorption was reported for oxygen-
stabilized compounds like Ho3Fe90:, [1.12] and Ti4Fe20 x [1.22] and already
much earlier by Beck [1.13]. Investigations of hydrogen absorption by
amorphous alloys were initiated by Maeland [1.23]. Finally the series of new
hydrides came to a - hopefully only temporary - stop with the successful
synthesis of Mg/FeH 6 by Didisheim et al. [1.24].
Volume I of Hydrogen in Intermetallic Compounds, "Electronic, Thermody-
namic and Crystallographic Properties", contains the chapters Preparation of
Intermetallics and their Hydrides, Thermodynamic Properties, Crystal and
Magnetic Structure, Electronic Properties, Heat of Formation Models, and
Magnetism and Superconductivity. Volume II, "Surfaces, Dynamics, and
Applications", contains the chapters Surface Properties, Dynamics, Kinetics,
Applications, and Experimental Techniques.
6 L. Schlapbach

Most chapters of these two volumes are structured as follows:

- Introduction
- Methods
- Results on binary hydrides
- Results on LaNi s type hydrides
- Results on FeTi type hydrides
- Results on Mg alloy hydrides
- Results on A B 2 Laves phase hydrides
- Results on other hydrides (amorphous).

1.3 Recent Highlights and Outlook

To round up the introductory chapter we would like to mention some
highlights in this field which should give the reader a taste of the most exciting
current research activities. Some possible new developments in the near future
are also indicated.

Phase Diagrams, Elastic Interactions, Hydrogen Ordering: Considerable efforts

are made to study hydrogen-metal phase diagrams and phase transitions, and
to understand thc underlying hydrogen-metal and hydrogen-hydrogen inter-
actions. The fundamental mechanism is the elastic expansion of the metal
lattice around the dissolved hydrogen atom and the resulting elastic interaction
between neighbouring hydrogen atoms. The theory concentrates on experi-
mentally well-known phase diagrams such as Nb-H, for which transitions from
disordered to disordered phases were successfully described by the lattice gas
model by calculating pairwise interactions. Now the models are extended to
include the effects of long-range electronic and many-body elastic interactions
on the occurrence of ordered phases on the one hand, and the effect of disorder
of the metal host lattice (disordered alloy, amorphous metal, disordered
impurities) on the phase diagram, on the other hand [1.25-27]. Detailed
experimental studies of the phase diagrams of rare earth (RE) hydrides have
been initiated particularly at low temperatures and in the concentration
ranges of solid solubility and from REH2 to REH 3 [-1.28, 29]. The occurrence
of miscibility gaps has been discussed [1.29]. In addition to the hydrogen
ordcring, which is known to occur in REHz+ x around 240 K, linear ordering
and hydrogen pairing were found in the solid solution range of LuHx and YH~
at low temperature, caused by an anisotropic hydrogen potential [1.30, 31].

Materials, Preparation Methods. Only very few new hydrides have been found
recently, but the successful synthesis of Mg2FeH 6 [1.24] opened further paths
for the discovery of ternary hydrides whose corresponding intermetallic
compounds do not exist. Implantation techniques [-1.32] allow the preparation
of hydrides with hydrogen contents higher than any known to date and with as
 Introduction 7

yet unknown properties. The development of a catalyzed reaction to precipitate

MgHz from an organometallic solution opened an extremely promising field
for producing highly reactive high surface area hydrides [1.33]. The develop-
ment of high pressure technology up to the 100 G P a range in diamond anvil
pressure cells is soon likely to lead to the synthesis of new hydrides [1.34].

Hydride Formation, Formation of Disordered, Amorphous of Quasi-Crystalline

Materials, Superlattices, Hydrogen as a Probe. The solution and rapid
diffusion of H in metals and the subsequent hydride formation together with
the rise of temperature due to the evolution of the heat of formation affect the
local structure and order of the host metal in many different ways:
The implantation of a few percent of H into A1 grains (in an A1-Ge mixture)
induces an Anderson type metal-insulator transition by increasing the local
disorder [1.35].
N b - T a superlattices (multilayered structures) not only show greatly
increased H solubility (as compared to bulk Ta and Nb), but also H-induced
strain modulation [-1.36].
The short-range order of intermetallic compounds can be destroyed
completely by hydride formation so that H absorption may well develop into
an important solid state reaction to produce amorphous metals [-1.37, 38]. As
other solid state reactions have already been successfully applied to form quasi-
crystalline materials [1.39], and since the rate of hydride formation is easily
controlled by parameters such as H z pressure, sample size and heat transfer,
hydrogen absorption might soon be used to form quasicrystals. Icosahedral
ordering was recently suggested in a hydrogen site occupancy model for
amorphous transition metal hydrides [-1.50].
Some hydrides of binary alloys are thermodynamically unstable and tend to
decompose into a stable binary hydride and a binary alloy of different
composition. Early stages of this decomposition can be viewed as local
disorder. The formation of"antistructure" Fe atoms (i.e. Fe atoms on Ti sites)
in the bulk of FeTi-like ordered alloys has been discussed [1.40]. Till 2
precipitates out of a hydrogenated amorphous CuTi matrix. A small angle
neutron scattering study of the aggregation of these Till 2 precipitates
demonstrated their fractal structure [1.41].
Disordered and amorphous systems have attracted much attention, and
considerable progress is being made particularly in the theory. Again, H-metal
systems provide challenging examples [1.5]. Amorphous metals absorb
hydrogen in amounts which are - apart from some interesting ex-
ceptions - comparable to those of their crystalline counterparts at compar-
able pressure and temperature. Many properties, e.g. pressure-composition
isotherms or diffusion, depend on the configurational disorder in the host
lattice, and particularly on the spread in energy of the hydrogen sites [1.42,
1.43]. Recently it was shown the other way round that the slope of pressure-
composition isotherms of hydrogen in amorphous metals can be used to probe
the hydrogen sites [1.44, 50].
8 L. Schlapbach

Electronic and Magnetic Properties. Theoretical approaches to the electronic

properties of stoichiometric hydrides of elemental metals by total energy
calculations look promising [1.45]. Band structure calculations are being
tentatively extended to nonstoichiometric and disordered hydrides and very
successfully to ordered hydrides of intermetallic compounds (Chap. 5). Purely
phenomenological correlations, for example, between electronic and thermo-
dynamic properties or between embrittlement and decohesion are being
replaced by semiempirical models based on the electronic band structure [1.46]
(Chap. 6). Hydrides of rare earth metals illustrate how advances in the under-
standing can sometimes lead to new contradictions: Band structure calcu-
lations for stoichiometric R E H 2 and R E H 3 support a metal-semiconductor
transition. However, explanations on the origin and nature of the gap and on
the opening of that gap at nonstoichiometric concentrations are very con-
troversial [1.473 (Chap. 5).
Hydrides of Ce are heavy-electron-like materials. A weak f - d interaction
seems, at low temperatures, to cause a narrow band at the Fermi level which
enhances the electronic specific heat by a factor of more than 2000 relative to
that of La hydrides [1.483. Core level spectroscopy revealed that YbH2.6 is a
mixed valence compound [1.49]. Several hydrides of rare earth metals and
rare earth intermetallic compounds show nal:row band phenomena.
Some analogies between hydrides and oxides allow for speculation about
the existence of high T~ hydrides: Hydrogen absorption can affect the Tc of
superconducting metals; some high T~ oxides absorb hydrogen and thereupon
change their T~; hydrides of alloys with inhomogeneous valence of the
constituents are known and their phonon frequencies as well as electron-
p h o n o n coupling can be modified over a wide range. Let's begin the search for
these high T~ hydrides!

References
1.1 Intl. Symp. on the Properties and Applications of Metal Hydrides V, Maubuisson,
France, 1986, ed. by A. Percheron-Guegan, M. Gupta, J. Less-Common Met. 129, 130,
131 (1987), with references to earlier conferences of this series
1.2 Hydrogen in Metals, Proceedings of the Intl. Symp., Belfast 1985, ed. by F.A. Lewis, E.
Wicke, Z. Phys. Chemic N.F. 143-147, (1986), with references to earlier conferences of
this series
1.3 Hydrogen Energy Progress VI, ed. by T.N. Veziroglu, N. Getoff, P. Weinzierl, Vol. 2 and
3, (Pergamon, New York 1986) with references to earlier conferences of this series
1.4 P. Jena, C.B. Satterthwaite (eds.): Electronic Structure and Properties of Hydrogen in
Metals (Plenum Press, New York 1983)
1.5 Hydrogen in Disordered and Amorphous Solids, ed. by G. Bambakidis, R.C. Bowman,
NATO ASI Series, Series B: Physics Vol. 136 (Plenum Press, New York 1986)
1.6 H. Buchner: Energiespeicherung in Metallhydriden (Springer,Wien 1982)
1.7 C.J. Winter, J. Nitsch: Wasserstoff als Energietrdger(Springer, Berlin 1986);
DECHEMA, Wasserstofftechnologie, ed. by D. Behrens, Frankfurt (1986)
1.8 D.G. Westlake, C.B. Satterthwaite, J.H. Weaver: Physics Today 31, 32 (Nov. 1978)
J.J. Reilly, G.D. Sandrock: Sci. Am. 242, No. 2, 118 (1980)
 Introduction 9

1.9 D,O. Northwood, D.G. lvey: Proc. Second. Int. Symp. on Hydrogen Produced .from
Renewable Energy, 1985, ed. by O.G. Hancock, K.G. Sheinkopf, p. 199, Florida Solar
Energy Center;
D.G. Ivey, D.O. Northwood: Z. Physik. Chemic N.F. 147, 191 (1986)
1.10 H. Oesterreicher: Appl. Phys. 24, 169 (1981);
K.N. Semenenko, V.V. Burnasheva: J. Less-Common Met. 105, 1 (1985)
1.11 Y. Fukai: Cryst. Latt. Def. and Amorph. Mater. 11, 85 (1985)
1.12 K.H.J. Buschow, P.C.P. Bouten, A.R. Miedema: Rep. Prog. Phys. 45, 937 (1982)
K.H.J. Buschow: "Hydrogen Absorption in Intermetallic Compounds" in Handbook on
the Phys. and Chem. of Rare Earths, ed. by K.A. Gschneidner, Jr., L. Eyring (North
Holland, Amsterdam •984) Vol. 6, Chap. 47, 1-111
1.13 R.L. Beck: Investigation of Hydriding Characteristics of Intermetallic Compounds, DRI
2059 (University of Denver, 1962)
1.14 A. Pebler, E.A. Gulbranson: Electrochemical Technology 4, 2011 (1966)
1.15 J.J. Reilly, R.H. Wiswall: Inorg. Chem. 7, 2254 (1968)
1.16 J.J. Reilly, R.H. Wiswall: Inorg. Chem. 13, 218 (1974), where reference to earlier work on
FeTi H is given
1.17 J.H.N. van Vucht, F.A. Kuijpers, H.C.A.M. Bruning: Philips Res. Rep. 25, 133 (1970)
l.l 8 H.-H. Neumann: "L6slichkeit von Wasserstoff und Deuterium in LaNis"; Ph.D. thesis,
Faculty of Chemistry, Technische Hochschule Darmstadt, 1969
1.19 D. Shaltiel, I. Jacob, D. Davidov: J. Less-Common Met. 53, 117 (1977)
1.20 Y. Ishido, N. Nishimiya, A. Suzuki: Denki Kagaku 45, 52 (1977) and Energy
Developments in Japan 1,207 (1978)
1.21 M.P. Dariel, M.H. Mintz, Z. Hadari: J. de Phys. 40, Suppl., C5-213 (1979)
1.22 K. Hiebl, E. Tuscher, H. Bittncr: Monatshefte f/ir Chemie 110, 9 (1979)
1.23 A.J. Maeland: "Comparison of Hydrogen Absorption in Glassy and Crystalline
Structures" in Hydrides for Energy Storage, ed. by A.F. Andresen, A.J. Maeland
(Pergamon, Oxford 1978)
1.24 J.-J. Didisheim, K. Yvon, P. Fischer, J. Schefer, M. Gubelmann, A.F. Williams: Z.
Kristallogr. 162, 61 (1983)
1.25 J. Peisl: Lattice Distortion, Elastic Interaction and Phase Transitions of Hydrogen in
Metals, Festk6rperprobleme XXIV, 45, ed. by P. Grosse (Vieweg, Braunschweig 1984)
1.26 C.K. Hall, A.I. Shirley, P.S. Sahni: Phys. Rev. Lett. 53, 1236 (1987);
C.K. Hall, C. Soteros, I. Macgillivray, A.I. Shirley: J. Less-Common Met. 130, 319 (1987)
1.27 W. Frenzl, J. Peisl: Phys. Rev. Lett. 54, 2064 (1985)
1.28 T. Ito, B.J. Beaudry, K.A. Gschneidner, T. Takeshita: Phys. Rev. B 27, 2830 (1983)
1.29 E. Kaldis, M. Tellefsen, R. Bischof: J. Less-Common Met. 129, 57 (1987)
1.30 O. Blaschko, G. Krexner, J.N. Daou, P. Vajda: Phys. Rev. Lett. 55, 2876 (1985)
1.31 I.S. Anderson, J.J. Rush, T. Udovic, J,M. Rowe: Phys. Rev. Lett. 57, 2822 (1986)
1.32 X.W. Lin, M.O. Ruault, A. Traverse, J. Chaumont, M. Salom6, H. Bernas: Phys. Rev.
Lett. 56, 1835 (1986)
1.33 B. Bogdanovie, K.-H. Claus, S. G/irtzgen, B. Spliethoff, U. Wilczok: J. Less-Common
Met. 131, 163 (1987)
1.34 H. Hemmes, A. Driessen, R. Griessen: J. Phys. C19, 3571 (1986)
1.35 P. N6dellec, A. Traverse, L. Dumoulin, H. Bernas, L. Amaral, G. Deutscher: Europhys.
Lett. 2, 465 (1986)
1.36 P.F. Miceli, H. Zabel, J.E. Cunningham: Phys. Rev. Lett. 54, 917 (1985)
1.37 K. Samwer, in [1.5] p. 173
1.38 K. Chattopadyhay, K. Aoki, T. Masumoto: Scripta Met. 21, 365 (1987)
1.39 W.A. Cassada, G.J, Shiflet, S.J. Poon: Phys. Rev. Lett. 56, 2276 (1986)
1.40 G. Hilscher, G. Wiesinger, R. Hempelmann: J. Phys. F 11, 2161 (1981)
1.41 Ph. Mangin, B. Rodmacq, A. Chamberod: Phys. Rev. Lett. 55, 2899 (1985)
1.42 R. Kirchheim, F. Sommer, G. Schluckebier: Acta Metall. 30, 1059 (1982)
R. Kirchbeim: Acta Metall. 30, 1069 (1982)
1.43 R. Griessen: Phys. Rev. B 27, 7575 (1983)
10 L. Schlapbach: Introduction

1.44 R. Griessen et al.: to be published

1.45 A.C. Switendick: J. Less-Commun Met. 130, 249 (1987)
1.46 M.S. Daw, M.I. Baskes: Phys. Rev. Lett. 50, 1285 (1983)
1.47 J. Shinar, B. Dehner, B.J. Beaudry, D.T. Peterson: Phys. Rev. B (1987) in press
1.48 L. Schlapbach, H.R. Ott, E. Felder, H. Rudigier, P. Thiry, J. Bonnet, Y. Petroff, J.P.
Burger: J. Less-Common Met. 130, 239 (1986)
1.49 St. Bfichler, L. Schlapbach, R. Monnier, L. Degiorgi: Proc. 14th Conf. on X-ray and
Inner Shell Physics, 1987, Paris (to be published)
1.50 J.H. Harris, W.A. Curtin, M.A. Tenhover: Phys. Rev. B36, 5784 (t987)
2. Preparation of Intermetallics and Hydrides

Annick Percheron-Gurgan and Jean-Marie Welter

With 30 Figures and 1 Table

This chapter describes the preparation and the characterization of the

intermetallic compounds and their hydrides. It is not an exhaustive review, but
by means of some examples, emphasis is put on the correlation between the
quality of base intermetallic compounds and the properties of their hydrides.
After a recapitulation on the thermodynamic properties of the most
representative families, the different preparation techniques on laboratory and
industrial scale are discussed.

2.1 Introduction
The aim of this chapter is first of all to draw the attention of scientists and
engineers who work in the field of hydrides to the importance of the synthesis
and characterization of the intermetallic compounds in determining the
properties of their hydrides. Indeed, the standard route to obtain the hydrides
consists in preparing first the intermetallic compound and then in hydrogenat-
ing it. This is usually acceptable because in most intermetallic phases the metal
atoms do not diffuse when hydrogen is dissolved in the lattice. Secondly, we will
give a short description of the different ways of preparing hydrides.
We do not give here an exhaustive review of the preparation of all the
different intermetallic compounds which are of interest since in most work
dealing with the hydrides of intermetallics little is said about the preparation of
the compounds and almost nothing about their characterization. So we want to
focus on general ideas needed to prepare well-defined intermetallic compounds
of high quality, and on the intrinsic properties of the hydrides. We will report
mostly on methods used in our own laboratories. General formation methods
for intermetallic synthesis have been well described by Brown and Westbrook
[2.1].
As various hydrides of intermetallic compounds are also used for industrial
applications, it is necessary to cover the preparation of the intermetallics both
for basic research and for large-scale applications. The prevailing points of view
in the two fields are different: in the first case it is essential to have as simple and
well-defined materials as possible (low impurity level, usually only two
elements...) whereas on an industrial level many elements have to be added to
the base material to tailor it for a given application. Furthermore, batch sizes
and economic considerations are drastically different.
12 A. Percheron-Gu~gan and J.-M. Welter

The chapter is therefore structured as follows: First, we will review the

thermodynamic properties of the most representative intermetallic compound
families which form hydrides. A knowledge of these properties is needed to
optimize the preparation of the compounds. For the other intermetallics a
review of phase diagrams will be given in the Sect. 2.2.
Second, we will discuss the preparation techniques of intermetallics which
are mainly solidification methods from a melt, the importance of several
parameters (purity of the starting materials, annealing treatments...) and the
characterization of the resulting materials, both on a laboratory and on
industrial scale.
The hydrogenation techniques will then be treated rather briefly since they
are almost identical to those used for binary hydrides [-2.2]. Finally, we will
illustrate with some examples the variation of the properties of the hydrides
induced by the impurity content, the homogeneity of the compounds and the
partial substitution of main components. We will concentrate on crystalline
materials and restrict ourselves to the following four major families of
intermetallic compounds and their hydrides:

Family Representativecompound

AB 5 LaNi 5
AB2 TiCr2
AB TiFe (FeTi)
ABo.5 MgNio.5 (Mg2Ni)

In the short hand formula B represents in general a late transition metal of

the iron group whereas A may be an alkaline earth, an early transition metal,
a rare earth or an actinide element.
The formulae of minor families can be found in Buschow et al. [2.3] and Ivey
et al. [-2.4].

2.2 Thermodynamic Aspects

of lntermetallic Phases and Their Synthesis
The question which arises first is: what is an intermetallic compound AB? It is
an intermediate phase between A and B with a different structure to the parent
elements and to their terminal solid solutions which is based either on
symmetry or order properties. In all cases the intermediate phase exhibits a
domain of homogeneity which may be of variable extent according to the
relative stability of the neighbouring phases. Having two constituents increases
the degrees of freedom of the material: besides the intrinsic variation in
composition, within the domain of homogeneity, various structural defects
such as vacancies and antiphase boundaries may be present. In comparison,
elemental materials may be considered as well-defined materials. As this is no
 Preparation of Intermetallics and Hydrides 13

longer true for intermetailic compounds, experimental reproducibility is more

difficult to achieve. Of course both kinds of systems have in common the
problem of purity, but here again intermetallics are in a less favorable situation
because impurities may either stabilize or destabilize the phase in a much more
dramatic way. Furthermore, whereas in pure elements the surface layers are
chemically identical to the bulk, a shift of composition exists between the bulk
and the surface of intermetallics [2.5]. Here too, impurities may play a more
important role in enhancing the segregation of one of the constituents.
As a consequence the preparation of the intermetallic compound, its
chemical and structural characterization and the study of its interaction with
hydrogen are very strongly correlated. For instance, a spatial fluctuation of the
composition can affect the solubility of hydrogen. Indeed, in a chemically
inhomogeneous binary system the chemical potential # of e.g. element A not
only varies with the local composition, but depends also on the variation of the
composition [2.6]

#(C) = #o(C) -I- K 2 V 2 c , (2.~)

where #o(C)denotes the chemical potential of a uniform system with composition
c, K is a constant and V2 the Laplace operator. Equation (2.1) shows that
fluctuations with short wavelengths affect /~(c) most. If the highly mobile
hydrogen is dissolved in such a material, it will try to have the same chemical
potential everywhere, as imposed e.g. by the gas phase. Because of the coupling
of the chemical potentials of the hydrogen and of the metallic components, this
situation can only be achieved through a local adjustment of the hydrogen
concentration: the total content of hydrogen may now differ considerably from
the content in a uniform system with a mean composition c.
To prepare these intermetallic compounds it is first necessary to study the
phase diagram: this gives us the basic information on the homogeneity range,
on the solidification reaction and on possible solid state transformations. These
data are of primary importance in determining the technique of preparation, the
heat treatment and the allowed impurity levels and composition shifts. For
binary systems the phase boundaries are satisfactorilywell known I-2.7-13]. Here
also, a careless preparation technique of the samples used in phase diagram
investigations is often the source of controversy over topographic details [-2.14].
The phase diagrams of the systems La-Ni, Mg-Ni, Ti-Cr, Ti-Fe are shown
for illustrative purposes in Figs. 2.1-4. These diagrams, and more specifically
the microregions around the compounds of interest, should only be considered
as a general guideline because many uncertainties remain. These involve first
the nature of the solidification reaction. Whereas the situation appears to be
quite clear for LaNis, which solidifies congruently and for MgzNi, which
crystallizes through a peritectic decomposition, it was long a matter of
uncertainty in the case of FeTi. For this compound both kinds of solidification
reactions were quoted in the literature and only recently was the peritectic
reaction clearly established [2.17]. For TiCr z one should first note the large
14 A. Percheron-GuOgan and J.-M. Welter

1
.~ ~r,,, i.~ 14541 Fig. 2.1. Phase

i
1400 ! diagram of the system
La-Ni [2.15, 16]
1200

1000
918

E 800 __

600
50

400,
0 20 40 60 80 100

La Atomic Percent Nickel Ni

Z
cq == 1454
o~
1300
1145

f
E
900 _ _ _ _

f
/ 760

500 507

Fig. 2.2. Phase

diagram of the system
100 Mg-Ni [2.7]
0 20 40 60 80 100

Mg Atomic Percent Nickel Ni

solidification interval (ATe-95 K) which leads to the solid solution. The

compound forms through solid state reactions which involve three polymor-
phic phases [,-2.18]. Until now no agreement exists on the exact nature of these
transformations.
The second aspect is the domain of homogeneity of the compounds: TiCr2
and FeTi have an extended homogeneity range [-2.17, 18], whereas LaNis and
Mg2Ni were thought to belong to the line-compound type. However, recent
studies have revealed that LaNi5 has a large domain of solubility at high
temperatures, but the exact phase lines are still unclear [2.15].
 Preparation of Intermetallics and Hydrides 15

1900 Fig. 2.3. Phase

I-
diagram of the system
Ti Cr [-2.19]
1700 t6-70
/
o 1500
1360
1300
2 / 12,~b
0. f i~" " C14 ~X~
E 1100 1190

9 0 0 882
C15_ ~--C 36
700
667 J
5OO
[
0 20 40 60 80 100

Ti Atomic Percent Chromium Cr

170o
~.1670
"~¢' 153(_,
1500
1427
/
1300
1~1
~1 1289 \
O. 1100
\
E ' 1085 I

I
i

21ZI
F i g . 2.4. Phase
diagram of the system
I Ti-Fe I-2.17]
0 20 40 6O 80 100

Ti Atomic Percent Iron Fe

A third point concerns the arrangement of the atoms in the compounds. The
data on the crystal structures and lattice parameters of hydrides and some
intermetallic compounds are reviewed in Chap. 4. A very complete collection of
crystallographic data of intermetallic compounds was published by Villars and
Calvert [-2.19]. But little information exists to indicate how off-stoichiometry
and thermal disordering affect the occupancy of the various sublattices of the
metallic components. The degree of ordering at the stoichiometric composition
has been analysed both experimentally and theoretically only in FeTi [-2.20,
2~]. It was found that the order parameter is one up to the melting point. In
LaNi5 and FeTi deviations from the stoichiometric composition seem not to
16 A. Percheron-Gu~gan and J.-M. Welter

create vacancies on one of the sublattices. In both cases the variation of the
lattice parameter across the range of homogeneity has been determined quite
accurately, but a good determination of the density is still missing. Thus the
conclusions have been drawn on a comparative basis. In analogy to YCo5
[2.22] and SmC% [2.23] an excess of nickelatoms, is accommodated in LaNis
by the replacement of lanthanum atoms by nickel atoms which associate with a
neighbouring nickel atom in the form ofa dumbeU. A 2 : 17 compound is formed
locally. In FeTi the excess atom is just substituted on the deficient sublattice
[2.21]. The question of vacancies and other structural defects is also a critical
aspect for mass transport in the compound. A knowledge of the transport
coefficient is needed to optimize the conditions for sample homogenization
beyond an empirical approach.
Phase diagrams of ternary or higher intermetallic compounds are almost
unknown. Therefore in such cases the adjustment of the composition to obtain
pure phases is mostly a question of trial and error. A guideline was in many
cases the search for pseudo-binary compounds [-2.36, 37, 47]. Another problem
introduced by the lack of knowledge on ternary compounds is the question of
how many impurity atoms a phase can support before being destabilized,
especially when the hydride is to be formed. This is especially true for the light

4 ''4.
~ ~ Z r O 2

1.a205
"~,.~"~ ./ MnO

~ ~ Cr203

Fig. 2.5. The standard free energies of

formation of the oxides (Ellingham
diagram) [2.27, 28]. Conversion fac-
tors: 1 eV/O atom = 24 kcal/mole of
O atoms--- 100.34 kJ/mole of O atoms
500 1000 1500 2000 2500
__T [K]
 Preparation of Intermetallics and Hydrides 17

elements like oxygen, nitrogen and carbon. Here most results exist in
connection with the system Ti-Fe [2.24-263; see Sect. 2.6.
The light elements C, N, O... are also the major source of contamination
during the synthesis of the compounds. The standard technique consists of
preparing a melt of the components which is then solidified. This is the only
technique used on a large industrial scale whereas on a laboratory scale other
techniques like evaporation/condensation or solid state reaction are also used.
One characteristic of the AB,, compounds is that the elements A form very stable
oxides, whereas the B elements have a much lower affinity for oxygen. This is
clearly revealed by the Ellingham diagram shown in Fig. 2.5 [2.27, 28]. The
analogy with the stability of the corresponding hydrides is striking. It is the
coupling of dissimilar elements which create the maj or metallurgical difficulties.
Whereas the metallurgy and shaping of the iron group elements are well
established techniques which are not too sensitive to crucible materials and
gaseous environment, the presence of the reactive element drastically compli-
cates the situation and new melting and shaping techniques have to be
developed. Eventually, one must also account for the differences in volatility of
the elements involved [2.29, 303.

2.3 Preparation of Intermetallic Phases:

Laboratory Scale
In this chapter we describe and discuss how to make well-defined samples
which will be used for investigating the intrinsic physical and chemical
properties of the intermetallic compound-hydrogen systems. Well-defined
means two things here: first, a material with a low level of impurities and
structural defects and with a high degree of homogeneity and stability;
secondly, an indication how well this goal has been achieved.

2.3.1 The Starting Materials

As the chemical purity of the final material depends to a large extent on the
impurity level of the constituents, it is worth selecting the starting materials
carefully, keeping in mind the costs of the substances, the requirements of the
experiment and possible alterations of the purity during further processing
[2.31]. As a guideline, Fig. 2.6 gives a survey of the extent to which the elements
can be purified nowadays. The aim of this chart is not to list world records in
purity, but to give the order of magnitude of foreign atoms present in high-
purity brands, which are commercially available in sufficient quantities at
reasonable prices [2.32]. Here all the impurities have been included, the main
group being the interstitial ones followed by the group of neighbouring
elements. Interstitials usually have a very harmful influence, whereas the
neighbouring elements affect the hydrogen related properties very little. One
should be aware that many commercial data sheets only quote the detected
18 A. Percheron-Gu~gan and J.-M. Welter

Fig. 2.6. Impurities contents in com-

mercially available high purity brands
of the elements. The scale is log-
1 arithmic: one c a s e = l atomic ppm,
two cases = 10 atomic ppm... [2.32]

IK 2o Scri V

RbJSr Y Zr NbMo lRulRhlPdiAglCdllniSnlSblTelllXel

I I I I I l I I
I I I I I I I I I I I
ISrnlEu[OdITbIDy [HoIEr [TmiYblLu [

metallic impurities. These are usually detected by optical emission spec-

troscopy and in a few cases by mass spectrometry. It is clear that an accurate
determination of e.g. the oxygen content in a very reactive material such as a
rare earth is a very arduous and expensive task because of the interference of the
surface oxide layer. Here nuclear activation methods can lead to more reliable
data. In the following we comment on the purity of a few selected materials.
They are listed according to increasing prices, which range from a few $/kg for
magnesium to more than thousand $/kg for lanthanum and chromium of
very high purity.
Magnesium: the purity of selected electrolytic magnesium batches can reach
99.9 %, for distilled magnesium it exceeds 99.99 %; here the main impurities
are Fe 30 w.ppm, Ni 5 w.ppm, Cr 5 w.ppm.
Nickel: electrolytic nickel, which is prepared in large quantities for superalloys,
can be obtained with a purity better than 99.995 %; in comparison to nickel
pellets it has a very low carbon content - typically less than 20 w.ppm.
Iron: electrolytic material has the highest purity at better than 99.995%;
whereas the level of metallic impurities is below 20 w.ppm, larger amounts of
light elements like O, C, C1 are present up to 200 w.ppm.
Titanium: electrorefined titanium crystals have a purity of 99.9%; the main
metallic impurity is usually AI at 100 w.ppm, but O and N can easily exceed
500 w.ppm.
 Preparation of Intermetallics and Hydrides 19

Manganese: electrolytic flakes can be obtained with purity 99.99%; the main
impurity is magnesium with 50-100 w.ppm; sometimes calcium and silicon
may be present up to 10 w.ppm; the typical oxygen level is around 50 w.ppm.
Zirconium: crystal bars obtained through the decomposition of the iodide
have a purity of 99.95 %; the main metallic impurities are Hf, 100 w.ppm and
Fe, 100 w.ppm and the light elements are usually present at a level below 100
w.ppm.
Lanthanum: calciothermic reduced lanthanum fluoride can be obtained with
less than 50 w.ppm of rare earth and common metal impurities; the oxygen
level usually exceeds a few 100 w.ppm.
MischmetaI: is an unrefined alloy of a number of rare earths elements,
typically: 50% Ce, 32-34% La, 13-14% Nd, 4-5% Pr and about 1.5% other
rare earths; this composition changes slightly according to the origin of the
ones [2.33].
Chromium: electrolytic chromium has less than 50 w.ppm of metallic impur-
ities, but an oxygen level as high as 3000 w.ppm; it can be reduced by distillation
or through an iodide decomposition process, but even here 100 w.ppm of light
elements are still present.

Batches of less pure materials (with the possible exception of lanthanum

and chromium) are only slightly cheaper. On the other hand, an increase of the
purity can only be achieved at very high costs. Therefore a post-purification
treatment to remove primarily the interstitial elements should only be under-
taken if one makes sure that all sources of contamination have been removed
from the following preparation steps. The most effective technique to remove is
a heat treatment under ultra-high vacuum [2.34] (which is usually not available
in industrial companies). Either oxygen leaves the metal as a volatile oxide
like in the IV B, V B elements and non-volatile rare earths [2.35], or the metal is
distilled and leaves the oxide as a residue as in Mg, Mn, and the volatile rare
earths [2.35]. The method of solid state electrotransport has produced rare
earth metals of higher purity than any other methods [2.35]. Carbon and
oxygen can be removed from the iron group elements by a fusion treatment in
wet and dry hydrogen.
Once the interstitials have been removed the reactive elements must
be handled carefully: this requires minimal exposure to air and no contact with
humidity and fingers. If extreme demands exist, the use of glove-boxes with high
purity inert gases is necessary.

2.3.2 Melting Techniques

Having obtained the pure components, the question arises of how to mix them
thoroughly to get a homogeneous material Due to the kinetic limitations of
solid state reactions, the components first have to be dissolved in a liquid phase.
In this section we will concentrate mainly on melts, because it is the usual
procedure and has the advantage of giving a high yield of material. The
20 A. Percheron-GuOgan and J.-M. Welter

subsequent solidification process is at least as important as appropriate

melting, so we will also make a few remarks about this problem.
The melt is prepared by heating the constituent elements in a suitable
container at least up to the melting temperature given by the phase diagram.
The following melting techniques are frequently used to prepare the alloys:

a) RF levitation melting in a cold crucible: best and very versatile method

to alloy reactive metals in high vacuum or inert gas atmosphere in quantities of
I 100 g. No temperature gradient, rapid cooling possible.
b) Arc melting in a water cooled copper hearth in an argon atmosphere in
quantities of 1-100 g. Strong temperature gradients across the sample which
may induce inhomogeneities; inert gas concentration of the solidified sample
might be high.
c) Resistance or RF induction heating in crucibles in vacuum, inert gas
atmosphere or in air under a protective layer to prevent strong oxidation
(quantities l0 g-100 kg). Refractory metals as well as oxide and non-oxide
ceramics are the most suitable materials for crucibles. Unfortunately the
presence of the reactive element in the liquid state usually leads to the corrosion
of crucibles made from ceramics and hence to a contamination of the melt. The
inertness of refractory metals like tantalum, molybdenum, and tungsten for
molten rare earth elements no longer holds when the iron group elements are
introduced in the melt.
Magnesium however is an exception. Because of the low melting tempera-
ture (923 K) and protective oxide layer, this element can be melted in almost
any crucible and the final choice will depend mainly on the nature of the second
element. In many cases a very convenient crucible material appears to be iron or
stainless steel. One advantage is that the lid can be welded on the crucible either
in vacuum or in an inert gas atmosphere. The reason for this precaution is that
magnesium has a high vapour pressure: it is approximately 0.4 kPa at the
melting point and the boiling point is reached at 1391 K. Many heating
techniques have been considered for this kind of crucible: they include the use of
standard tubular furnaces, RF-generators or by passing the current directly
through the crucible. In all cases the crucible should be rocked to ensure a good
homogeneity. If extreme metallic purity is not essential, iron crucibles can also
be used for MgzNi, Mg2Cu, and MgLao.lz.
In the next section we show a metallographic view of a Mgo.68Ni0.31Lao.ol
sample prepared by such a technique (Fig. 2.14). A very complete study of the
effect of crucible on the melt contamination during vacuum induction melting
of FeTi has been made by Sandrock et al. [2.24] showing the oxygen and metal
contamination for three common materials. The results are summarized in
Table 2.1.
Upon the melting of"Mischmetal-calcium-nickel" alloys of the LaNi5 type in
crucibles of Al20 3, Sandrock [2.37] noticed a rather low oxygen solubility of the
melt as long as the temperature did not exceed the melting temperature by more
 Preparation of lntermetallics and Hydrides 21

Table 2.1. Effect of crucible on melt contamination during

vacuum induction melting of FeTi [2.24]

Crucible O Contamination M Contamination

material wt. % wt. %

MgO 0.072 <1).001 Mg

A1203 1.14 1.23 A1
ZrOz 0.30 0.45 Zr

than about 100 °C. Best results were obtained by adding solid Mischmetal and
calcium to the Ni melt immediately prior to pouring. F o r the rare earths and
titanium-based c o m p o u n d s cold crucible techniques have proved to be most
convenient, also here the size of the batch is limited to a few cm 3. The cheapest
technique is arc melting. It has furthermore the advantage that the melt and
hence the solidified ingot can be somewhat shaped. Its disadvantage is that
inhomogeneitics due to segregation and foreign phases frequently occur. A bet-
t e r - but also much more expensive technique includes RF-heating because here
levitation effects can be exploited. Full levitation crucibles of the basket type are
ideally suited to prepare rapidly solidified samples e. g. in a piston and anvil type
system. Figure 2.7 shows a picture of a machined levitation crucible made from

Fig. 2.7. Water cooled cop-

per cold crucible for RF
induced heating and levi-
tation with titanium disc
getter [-2.38]
22 A. Percheron-GuOgan and J.-M. Welter

copper I-2.38]. The stems are watercooled. The copper contamination is

typically less than 50 ppm even for high melting temperature metals like
niobium. Furthermore the melt can be easily stirred either through mechanical
agitation of the crucible or through electromagnetic forces. Finally a very large
flexibility exists with regard to the gaseous atmosphere. This flexibility is lost
again when the melting is performed by bombardment with high energy
electrons. When inert gas is used - e.g. for arc melting or when volatile elements
are heated - high purities are imperative. Therefore it is useful to introduce into
the reaction chamber a getter material like titanium or zirconium which first
has to be activated. Figure 2.7 clearly shows the bright shining titanium disc
above the melt.
An important point is to ensure that the melt becomes homogeneous.
Difficulties arise when a solid crust remains in the cold part of the crucible or
when one component is trapped within its surface oxide layer. This occurs
frequently when rare earths are used in a poor atmosphere. Eventually the rate
of dissolution may be slow with small gains in mixing free energy [2.11]. An
excessive increase of the temperature is not necessarily the solution, because the
rate of parasitic reactions such as contamination and evaporation also
increases. Some help can be obtained from mechanically and electromagneti-
cally induced forced convection. On the other hand, an excessive gain in mixing
enthalpy can lead to an explosive reaction where some material is spilled out of
the crucible.
Once a homogeneous melt has been obtained, it must be solidified without
segregation. This problem has been discussed in various recent reviews I-2.39,
40]. Therefore we will limit ourselves hereto a few remarks. Segregation-free
freezing can only occur if the liquidus curve of the melt and solidus curve of the
intermetallic compound have a point of congruency. At this particular
concentration the equilibrium partition coefficient k, which expresses the ratio
of the solubilities cs and c~ of e.g. component B at a given temperature in the
solid and liquid phases, respectively, is unity. In reality, even for congruent
compounds, this ideal condition is never met. Minute shifts in composition or
the addition of dopants and impurities lead to a spatially varying redistribution
of the components during solidification. External parameters which influence
this redistribution are the heat flow on both sides of the liquid-solid interface
and the mass flow in the melt. The heat flow determines the rate at which the
latent heat of solidification is removed and hence the speed of freezing v.
Furthermore it governs to a large extent the stability of the interface. If the
latent heat is evacuated through the undercooled melt the growth front of a
solid nucleus becomes unstable and radially propagating dendrites develop.
The resulting solid morphology consists of strongly inhomogeneous equiaxed
grains. If the melt is overheated and the latent heat is extracted through the
solid, the interface remains stable and microsegregation is avoided. This holds
as long as the build up of the rejected component at the solidification front
remains small. Otherwise a thermodynamically unstable layer develops ahead
of the interface which decomposes tangentially into microregions of different
 Preparation of Intermetallics and Hydrides 23

T=[_ q-- Fig. 2.8. Partial phase diagram of an intermetallic

[ I compound solidifiedcongruently at (T~,Co)showing
| ~ the two-phase gap between the liquidus and the
I ~ tSe°limdU~a~drCtetcedn;egnit°a~i°Th;aj
het;yileisne~ereffr°efstehrt
] I/ solidification
front

"r
Tcm~

Ti~

Tum
_ i I I I I I
Cc kC= cSc= cli Coo/k

composition. A criterion which describes roughly the onset of this instability

can be derived with the help of Fig. 2.8 for a binary system. The evolution of the
temperature and of the solute concentration along the ordinate z ahead of the
solidification front can be described by the two functions T(z) and c(z),
respectively. Combined they give the trajectory T(e) in the phase diagram: it
starts at (To, coo)far ahead of the solidification front and ends at (Ti, el) on the
melt side of the interface. If the trajectory crosses the two-phase field an
instability occurs. To avoid this, the following condition must be fulfilled:

gradT d;
grad c I > mi' (2.2)

where mi is the slope of the liquidus at cl. The location of c I and the strength of
grade I depend on the importance of the buildup of the solute. Two extreme
cases can be considered. If the mass flow in the melt is very rapid because of
convective transport, almost no solute can accumulate in front of the moving
interface. Hence cl"c~ and grad el ---0 : The condition expressed by (2.2) is
always fulfilled. On the other hand, maximal pile-up occurs for a regime of pure
diffusive mass transport. Under stationary conditions cl~-c~/k and grad cl~-
-v(el-c~)/D=-vc~o(1-k)/kD. D is the diffusion coefficient and is usually
of the order of l0 -4 cmZ/s. As an example we can ask at which maximum speed
v, a TiCr 2 melt can crystallize in order to give a microsegregation-free solid
assuming a temperature gradient of 100K/cm. With mi=750 K/unit con-
24 A. Percheron-GuOgan and J.-M. Welter

centration and coo-c~i---0.1, one obtains v< 1.4 x 10 - 4 c m / s , ' ~ 0 . 5 cm/hour.

Although this condition would be strongly relaxed in reality because pure
diffusive mass transport is almost never achieved, the example shows the
advantage of low freezing speeds to avoid microsegregation. Of course v should
not be so low as to enhance macrosegregation. This takes place in the melt
layers which solidify first and last and have width of the order of D/vk. At very
high freezing speeds (v>10cm/s) the melt is quenched without spatial
concentration shifts of the components, but with these ultrarapid solidification
techniques only minute amounts of material can be prepared.

2.3.3 Annealing Treatments

Although the compounds which melt congruently can be obtained directly

from the liquid, an annealing treatment is useful to eliminate microsegregation
by solid state diffusion or to allow the system to establish a strictly
homogeneous composition and full structural order.
For a peritectic formation compound, annealing is in all cases required. The
annealing temperature should be below the peritectic temperature to dissolve
precipitates of neighbouring phases. This aspect was discussed by Busehow
[2.16] for the La-Ni system. Since the reaction rates are diffusion controlled,
high temperatures are preferred to shorten the annealing time.
The problem during annealing is to avoid excessive oxidation of the
material and evaporation of the more volatile compound (see Fig. 2.29).
Depending on the nature of the compound it is useful to anneal the sample
either in vacuum or in a high-purity inert gas atmosphere. In both cases a good
technique is to wrap the sample in an envelope of a getter material with a low
vapour pressure like zirconium or tantalum. The effect of the annealing on the
hydrogen absorption properties will be discussed in the Sect. 2.6. An example of
the influence of the conditions of annealing, time and temperature, is given in
Fig. 2.9 for LaNis_xMnx. The composition of the samples is analyzed by
electron microprobe. The 1150 °C annealing leads to a large dispersion of the
composition with some microprecipitation of secondary phase, whereas an
annealing at 900 °C for 36 to 72 hours allows a satisfactory homogeneity of the
strictly single phase to be reached. The effect of homogenization on the shape of
the isotherms can be seen on Fig. 2.10: the inhomogeneous compound annealed
for 3 hours at only 1150 °C exhibits very sloped plateaus with large hysteresis.
Another example is given by Sandrock E2.42] for the Mm Ni4.sA10. 5.
In the case of pseudobinary compounds such as the LaNi 5_xM~ series, the
limit of substitution varies according to the substitutional element: X(Ga, In,
Sn)=0.4 [2.43]; S(Si)=0.6 [-2.44]; S ( F e ) = l . 2 [-2.45, 46]; X(A1)= 1.3 [-2.47];
X(A1) = 1.5 [2.48]; X(Mn) = 2.2 [2.49]; S(Mn) = 0.99 [2.50]; X(Cu) = 5 [2.51,
52]; X(Co) = 5 [2.53]; X(Pt) = 5 [2.54]. These limits depend strongly on the heat
treatment (number of fusions temperature and duration of the annealing of the
samples) which has to be fitted to each family of compounds. For example this
 Preparation of Intermetallics and Hydrides 25

Fig. 2.9. Chemical homogeneity of

a LaNis_~Mnx alloys after casting
[~:z~] and annealing treatments at
1150 three temperatures for 3 ( ~ ) and 72
[m] hours. The bars represent the
compositional fluctuations as observed
P by electron microprobe. For samples of
900 ° == _,
very similar compositions the bars on

r -H- the temperature scale arc slightly

shifted for clarity [2.41]

800

0.25 0.5 1 2

- XMn

.J
2
Fig. 2.10. Isotherms of Lal.0
•Ni4.s3Mno. ~7 (A) annealed
1 2s°c ~T I for 3 hours at 1150°C
and of Lao.99Ni4.74Mno.z7
(I) annealed for 84 hours
0 at 900°C I-2.41]
0.25 0.5 0.75 1
~.Atom Ratio [H/M]

limit can be increased either by lowering the annealing temperature as for the
LaNi5 _ ~Mn~ system. In this case annealing at 800 °C leads to X(Mn) = 2 as the
limit of substitution, whereas for X ( M n ) = 0 . 9 9 with 1175 °C annealing a two
phase c o m p o u n d was observed [2.50].
F o r LaNi s - xFex only a high temperature annealing (1090 °C) for three days
enables a single phase to be obtained for X ( F e ) = 1. An example of the
disappearance ofmicroprecipitations is given in Fig. 2.11. This effect can explain
the different hydrogen contents given in the literature: 5.7 H / m o l e [2.56, 45],
4.5 H/mole [2.57] at 40°C under 10 atm; and 4 H/mole [2.42] at 25 °C under
10 atm for an as-cast compound.
 26 A. Percheron-GuOgan and .I.-M. Welter

Fig. 2.11a, h. Micrographs showing the disappearance of Fe~.sNi inclusions present in

as-cast LaNi4Fe compound (a) and after 3 days annealing at 1070°C (b) [2.55]

lOOg
I I
1001~
I I

a b
Fig. 2.12a, b. Comparison of multiphase annealed (2 months at 900 °C) LaCu 5 compounds (a)
and monophase LaCu 5 compound obtained by partial unidirectional solidification (b) [2.55]

In some cases a long annealing is not sufficient to get a single phase sample.
For example LaCu s, a compound which decomposes peritectically at 805 °C
[2.58] did not become a single phase even after 2 months' annealing at 900 °C as
we show in Fig. 2.12a. The electron microprobe analysis reveals three phases:
LaCus.9, LaCu4. 9 and some inclusion of almost pure copper (La0.osCuo.95).
Only a partial unidirectional solidification allows one to obtain a strictly single-
phase compound (Fig. 2.12b) [2.55].
Such annealing treatments are also useful to remove the structural defects
introduced during crushing and grinding of the materials. Indeed, it has been
shown that cold worked and annealed material react differently with hydrogen
[2.59].
 Preparation of lntermetallics and Hydrides 27

2.3.4 Single Crystal Growth Techniques

For fundamental research the best defined materials are single crystals. In the
case of many binary hydrides, like those of vanadium, niobium and tantalum,
single crystals have led to many interesting results, especially when the
dynamics of the base metal and of the hydrogen were studied. Unfortunately
such samples cannot be prepared in general for the ternary hydrides. The large
increase of the cell volume upon hydriding and the brittleness of the compound
create a decrepitation of the sample. Nevertheless single crystals are useful to
study the properties of the solid solution with hydrogen or just to make a
starting material of high quality.
Single crystals are usually prepared from the melt. Therefore many aspects
discussed in the previous section must also be considered here, e.g. contamin-
ation due to crucible corrosion or interaction with the reactive impurities of the
atmosphere will also play an important role. This is an even more serious
problem during crystal growth because of the long time that the material is kept
in the molten state or at high temperature. Indeed, to avoid segregation, the
solidification speeds are of the order of a few mm/h. The optimal situation is to
give up crucibles completely and to work in an atmosphere where the partial
pressures of the reactive gas molecules are less than 10 .6 Pa. This means that
floating zone techniques in ultra high vacuum should be used. A difficulty is the
preparation of the cylindrical feed rods. Because intermetallic compounds are
very brittle, it is not possible to swage them. A good method consists in casting
the rods into a sucking mold attached to an arc melting unit. These techniques
were successfully applied to prepare single crystals of LaNi s and FeTi.
Because of the peritectic solidification reaction of FeTi, a slight off-
stoichiometry with an excess of titanium had to be accepted. The components
were first melted by RF-induction heating in a water-cooled horizontal boat
crucible. Homogenization was achieved by levitation melting. The further
treatment of the homogenized knob depended on the specimen requirements. If
powder was needed, the knob was crushed in a mortar. This procedure im-
proved, if the material was, at least partially, hydrided. Because of the hard-
ness of iron-titanium, the best technique to obtain slabs and discs was spark
cutting. The minimal thickness which could be obtained was approximately
0.5 ram. The damaged and contaminated surfaces were usually cleaned with a
mixture of HF + HNO 3 -t- H20. Single crystals were grown in a vacuum which
had to be better than 10- ~ Pa. Otherwise a bluish layer immediately covered
the crystal.
LaNi 5 single crystals can be obtained by pulling (5 mm/hour) in ultra high
vacuum (10 7 Pa) by the Czochralsky technique (MgO-A1203 crucible)
from the melt of polycrystalline sample prepared by induction melting and
subsequent quenching in a water cooled boat at 2 x 10 .4 Pa. The starting
stoichiometry is LaNi4.97 to avoid Ni precipitations [-2.61].
Single crystals are usually shaped by acid sawing or by spark erosion. Here
also, the high reactivity of the compounds with hydrogen makes it mandatory
28 A. Percheron-GuOgan and J.-M. Welter

to minimize the presence of hydrogen. Therefore the best technique is spark

erosion cutting at low speed.

2.3.5 Noncrystalline Materials: Rapid Solidification and Solid-State Reactions

Maeland [2.62] has shown that under similar conditions amorphous or glassy
alloys can have larger hydrogen absorption capacities than their corresponding
crystalline alloys, e.g. in the case of TiCu and Ti2Cu. Furthermore often they do
not disintegrate into powder. Hydrides of amorphous alloys are also amor-
phous if upon hydride formation the recrystallization temperature is not exceed-
ed. Since these first results the formation and properties of amorphous metal
hydrides have become a quite active research area. Among the different alloy
systems studied we can quote Zr-Cu, Zr-Ni, Zr-Fe, Zr-Pd, and Zr-Rh. For
reviews we refer to [2.63-67].
Amorphous alloys are usually prepared by rapid quenching of the melt by
the melt spinning method or by the splat cooling technique [2.68]. These
ultrafast cooling methods allow the preparation of many amorphous alloys
which are normally crystalline, but restrict their form to thin ribbons or sheets.
On the other hand, the formation of amorphous alloys by solid-state reactions
[2.67], e.g. diffusion in layered structures, does not in principle limit the shape
or size of the samples. It is particularly interesting to note that hydrogen
absorption is also a way to form amorphous alloys and hydrides, e.g. in the
Zr-Rh system [2.67].

2.3.6 Synthesis from the Gas Phase and Thin Films

To avoid the problems due to the peritectic decomposition of the melt it may
sometimes be useful to use an evaporation-condensation technique in a closed
crucible. This is only possible if the components have a sufficiently high vapour
pressure at temperatures below the peritectic temperature and if no reaction
occurs with the crucible material.
This case can be illustrated by the preparation of Mg2Ni: in a recent
publication Post et al. [2.69] described the preparation of single phase MgzNi ,
by vapour exchange and diffusion reaction of elements in the absence of a liquid
phase, in a sealed stainless steel container under high purity argon. In previous
studies samples of MgzNi prepared by induction melting of the elements under
argon have been reported for multiphases including usually Mg, MgO, and Ni
or MgNi2, often in substantial amounts due to peritectic decomposition and
atmosphere contamination.
Thin films are extremely well suited for electron microscope and resistivity
studies. Good results have been obtained in the case of binary hydrides. They
would even be more suited in the case of ternary hydrides because on one hand it
is very difficult to prepare thin samples from a massive ingot and on the other
hand decrepitation problems could in principle be avoided. Little success has
been obtained until now, however, Adachi et al. [2.70] succeeded in preparing
 Preparation of Intermetallics and Hydrides 29

thin films of LaNi5 by conventional vacuum deposition: by heating LaNi5

powder at 2000 °C on a tungsten filament at 10-z Pa. Under these conditions
both the two components evaporated immediately and were deposited on quartz
plate. The thickness was about 3000A, the characterization confirmed the
composition LaNi 5 with presence of lanthanum oxide due to superficial oxygen
contamination. Livesay et al. I-2.71] also prepared AB5 alloys in thin films with
thickness less than about one micron, which remained coherent even with
repeated hydrogenation cycles.
The RF sputtering technique has been widely used to produce amorphous
materials; some examples are given for FeTi compounds [-2.72, 731.

2.4 Characterization of Intermetallic Phases

It is obvious that for a good chemical and structural characterization of the
material all the classical tools can be used. They include characterization by
analytical chemistry, diffraction techniques, the measurements of physical
properties and of the pressure-composition-temperature (pcT) curves in this
peculiar case.
Usually the nature of the main phase is obtained by x-ray powder dif-
fraction, which also enables one to check for the presence of minor phases, but
only at a level greater than 5 %. More detailed information is obtained by
metallographic examination either by optical or electron microscopy. Here, not
only minor phases at a level below 1% are seen, but also compositional shifts
due to segregation during solidification are revealed. Furthermore the electron
microscope can be equipped with an x-ray fluorescence analyser to give a
quantitative chemical analysis of the phases present. Electron microprobe
analysis can achieve the same results. Nevertheless one should be aware that the
level of detection depends on the element, but it is usually around 1% and that
the analytical information is averaged over a volume a few ~tm3.
For the metallographic preparation of the samples it is recommended to use
organic solvents. Otherwise, depending on the reactivity of the material,
artefacts may be induced by hydrogenation or oxidation due to the decompo-
sition of water. In some cases it is better to completely avoid chemical etching
and use only mechanical polishing (diamond paste under organic lubricant) to
prepare the surface for further examination.
If the material is a non-cubic structure, the use of polarized light for optical
microscopy examination is essential. An example is given in Fig. 2.13a which
shows the grain morphology of a LaNi5 sample prepared on an industrial scale
by Japan Metals and Chemicals. In contrast to this, the grain structure is not
revealed when the same sample is examined with unpolarized light (Fig. 2.13b).
A second example is a Mg2Ni type compound (Mg0.68Ni0.31Lao.00. The
sample shown in Fig. 2.14 is representative of a 0.5 kg batch melted in an
evacuated steel crucible and cooled down slowly (50 K/hour). The very different
nature of magnesium and lanthanum did not allow us to contrast the phases by
30 A. Percheron-Gu~gan and J.-M. Welter

Fig. 2.13a, b. Micrographs of LaNi 5 examined by using polarized light (100 × ) (a) and un-
polarized light (b) [2.74]

Fig. 2.14. Scanning electron micro-

graph of a slowly solidified partial-
ly substituted MgzNi (Mgo.6s
• Ni0.3aLao.ox) with energy disper-
sive x-ray spectra of the different
phases [2.75]

chemical attack. Therefore the polished surface was examined with secondary
electrons in a scanning electron microscope. This also had the advantage that
the composition of the different phases could be determined at least on a
relative scale by x-ray analysis. The micrograph of Fig. 2.14 reveals that besides
the major phase MgzNi and the minor phase M g l v L a 2 there is also some
magnesium forming an eutectic with Mg2Ni and various (Fe, Ni) inclusions
 Preparation of Intermetallics and Hydrides 31

resulting from a reaction with the crucible material. Neither Mg nor the
inclusions were revealed by x-ray powder diffraction. Special care was taken to
eliminate the fluorescence radiation of nickel and to improve the peak to
background ratio by using a graphite monochromator in front of the detector.
Therefore the limit of detection can be set as low as 5 %. It is good to point out in
the present context that many JCPDS-cards need revision. For example, the
card 1-1268 for Mg2Ni (up to 1982) still contains one magnesium line but many
important reflections of the compound are missing. This, of course, is again the
result of poor sample preparation. This point has also been noted by Post et al.
[2.69] (see Sect. 2.3.6) who have recently presented a refined set of x-ray data.
For the compounds which exhibit a fairly wide homogeneity range such as
LaNis ±o.3 metallographic examination and microprobe analysis are not able
to give accurate information about the stoichiometry. Chemical analysis and
measurements of the lattice parameters and of the magnetic susceptibility can
give good results [2.15, 76, 79].
Since the pcT curves are also very sensitive to the variation of the
composition [-2.15] and homogeneity of the intermetallic compounds, it is a
good way to assess the quality of the materials. This is particularly true to check
the stoichiometry deviation of LaNi 5 I-2.15] or the rate of substitution of one
component [2.55]. This was already mentioned in Sect. 2.3.3 and will be
emphasized again in Sect. 2.6 where we will report the correlations between the
properties of hydrides and the microstructure of intermetallic compounds.
The structural quality of single crystals can be derived from the widths of the
rocking curves measured with various types of monochromatic radiation like
neutrons, x- and 7-rays. The curves are obtained by rocking the crystal through

°.

(2:

t ,

: •

(2 s

\
1° 2.5' 5'

~ Rocking angle

Fig. 2.15a--c. Rocking curves of a FeTi single crystal obtained with neutrons (a), x-rays (b), and
y-rays (c) [2.603
32 A. Percheron-Gu~gan and J.-M. Welter

the beam in such a way that th e different subgrains successively fulfill the Bragg
condition. Figure 2.15 shows the rocking curves obtained with the three probes
(neutrons: 2=2.4 A; x-ray: 2 = 1.54 A; y-ray: 2=0.03 A) from the same FeTi
crystal. It is clear that each radiation gives a different information depending on
the beam dimension, angular spread and on the absorption and extinction
lengths.
For the analysis of thin films and their surfaces high and low energy electron
beams are very useful. The structure and the composition of thin films are best
checked in an analytical transmission electron microscope [2.77, 78] whereas
Auger spectroscopy is very sensitive to surface contamination and segregation.
For this last case magnetic measurement has emerged as a very sensitive
physical technique when a ferromagnetic component is involved in a weak
paramagnetic compound [2.79].

2.5 Preparation of Intermetallic Phases:

Industrial Scale
Although industrial applications of reversible metallic hydrides have not
blossomed as was expected in the seventies, there exists nevertheless a limited
market for which approximately 50 tons of material were prepared during the
last few years. Such amounts could no longer be produced on a laboratory
scale. Therefore various metallurgical companies started the fabrication of
intermetallic compounds with batch sizes of approximately 100 kg.
It is very difficult to get detailed information on the state of the art. It is
obvious that industry is less and less willing to describe how the fine tuning of
the melting, casting and shaping techniques is performed. Nevertheless some
general information could be obtained from various companies.
Historically the FeTi-based hydrides were the first ones to be prepared on a
large scale. Although they no longer represent the prototype system for
automotive applications, they are still of interest for large-scale, stationary
hydrogen storage units. Furthermore it is interesting for comparison with the
laboratory techniques described previously to show how industrial size batches
are made.
For example batches up to 50 kg of manganese-alloyed FeTi are prepared at
the Nfirnberg plant (FRG) of the Gesellschaft ffir Elektrometallurgie ]-2.80]. As
starting materials, vacuum remelted titanium sponge (99.5 %) and electrolytic
manganese (99%) are used. For iron a source with low carbon and oxygen
levels is selected: it is either sponge which has been treated by hydrogen (e.g. by
the H6gan~is process), or cathodes. The compound is prepared by a standard
vacuum (medium frequency) induction melting (VIM) process and is obtained
within one day. To minimize manganese losses during melting and casting the
vacuum unit is backfilled with argon up to approximately 0.5 bar. Ceramic
crucibles and steel molds are used. The ingot usually breaks apart and the
 Preparation of Intermetallics and Hydrides 33

pieces are crushed and ground to granules with a diameter less than 3 mm in an
inert gas atmosphere to avoid excessive oxidation.
The granules are packed in an argon atmosphere into plastic bags which are
hermetically sealed. All these precautions are necessary to obtain a material
with a low oxygen content. Because of the careful selection of the starting
materials and of the VIM procedure no deoxidation substances are used. A
tight control of the melting sequence and duration leads to a high yield of usable
batches of 80 %. Failures are mainly due to an oxygen contamination during
the process and in most cases its source remains unclear. The chemical
composition and purity are checked for each batch: metallic elements are
analyzed by x-ray fluorescence, gaseous impurities by inert gas extraction.
Occasionally a metallographie examination is made. The hydrogen absorption
properties are determined for each batch by measuring the pressure-
composition isotherm at room temperature. If titanium with a higher oxygen
content is used, it is necessary to deoxidize the melt by the addition of
mischmetal. This technique was developed at Ergenics (USA) [2.81 ] and has the
advantage of giving a compound with a good activation behaviour although no
manganese is added to the material. FeTi with niobium and zirconium
additives (Fe0.94Ti0.96Zro.04Nbo.o4) is also prepared by Japan Metals &
Chemicals on a medium size batch scale of approximately 10 kg [2.82]. Small
blocks of iron, titanium, zirconium, and niobium are cleaned with compressed
air and melted together in an arc or induction furnace. The working
atmosphere is argon. Crucibles are made from carbon. Trouble with the
crucible is the major reason for failures. After keeping the alloy for 20 minutes in
the molten state it is cast to obtain cylindrical blocks. Heat treatments,
crushing and grinding are performed as required by the customer and the final
material is sealed in plastic bags. The quality control of the starting materials is
based on the certificates of the producers, but for each batch the chemical purity
and composition are controlled and the pcT curve is measured.
For the more recently developed multicomponent TiMn2-based com-
pounds, low oxygen levels are difficult to achieve in a one stage VIM process,
unless excessive amounts of deoxidation substances are used. Therefore a two
stage process was developed by the Gesellschaft fiir Elektrometallurgie (FRG)
I-2.83]. To produce a compound e. g. of the type (TiZr) (MnVFeCr)z a fraction of
the manganese is first melted with ferrovanadium to make an oxygen poor
MnVFe alloy. The ingot is crushed and then remelted together with the rest of
the manganese and with titanium and zirconium. Small amounts of cerium-
mischmetal are also added for deoxidation purposes. With this technique a
yield of 90 % is obtained even for batches with sizes up to 250 kg. When a single
stage procedure is used, the yield does not exceed 50 %. Absorption isotherms
and a micrograph of a grain are shown in Figs. 2.16 and 2.17.
Rare earth (R) compounds of the type RNi5 -xMx with M being aluminium,
iron or manganese, are prepared e.g. by Santoku Metal Industries (Japan)
[2.85] with heat sizes of 100 kg. The purity of the rare earths is typically 99 %.
Melting is performed in an induction furnace using alumina crucibles. An
34 A. Percheron-Gu#gan and J.-M. Welter
100

lo f
f

/
J
g
/
0.1
0 0.5 1.5 2 2.5

H concentration [Wt % ]
Fig. 2.16. Desorption isotherms at 20 °C (o) and 60 °C (m) of the partially substituted TiMn2
type compound Tio.gaZro.o16Mnx.5oaVo.431Feo.oagCro.o49 prepared on industrial scale
[2.84]

Fig. 2.17. Micrograph for a crunched grain

of a batch of the compound described in
Fig. 2.16 [2.74]
 Preparation of Intermetallics and Hydrides 35

10 fT.

i'0,5
j
0.1
0 0.5 1
Fig. 2,18 Atom Ratio[H/M] Fig. 2.19

Fig. 2.18. Isotherms at 30°C (*¢), 50°C (m), and 80°C (A) of partially substituted LaNis
compound (LaNi4 vAlo3) prepared on industrial scale [2.86]
Fig. 2.19. Micrograph for a crunched grain of a batch of the compound described in Fig. 2.18
[2.841

important reason for failure is that the melt adheres to the crucible. The
melting atmosphere is argon. Deoxidation is made in vacuum obtained by
diffusion pumps. The temperature and the duration of the melt are tightly
controlled. The melt is cast into a copper mold. After shaping it is packed in
vacuum. The chemical purity and composition are checked by x-ray flu-
orescence, the phase purity by x-ray diffraction and pcT curves are usually
measured. The yield of usuable batches is approximately 97%. Similar
techniques are used by Ergenics (USA) [2.81] and by Japan Metals &
Chemistry (Japan) [2.82]. A characterization of LaNi4.vAlo. a material pro-
duced by this company has given the isotherms shown in Fig. 2.18 and the
micrograph given in Fig. 2.19.
It is interesting to note that with some care a material prepared on a large
industrial scale can achieve almost the same performance as a material
prepared on a laboratory scale. This is revealed by the sorption-rate curves
obtained with J M C material (Fig. 2.20a) and with a reference material prepared
in one of our laboratories (Fig. 2.20b).
A large variety of binary and pseudobinary compounds are made by Chuo
Denko Kogyo (Japan) [2.88] including Mg2Ni. The purity and shape of their
raw materials are: Ti (99.6%, sponge); Fe (99.98 %, flakes); Mn (99 %, rods);
Mn (99.9%, flakes); Ni (99.97%, granules); Mg (99.9%, ingots); Ca (99.8%,
granules) and A1 (99.7%, granules). These materials are - according to the
compound - loaded in alternate layers up to 100 kg in crucibles made from
36 A. Pereheron-Gukgan a n d J.-M. Welter

, t[s]
100 100 200 300

80

l/ / /V---

-Ii-

,oS°ili
, i!O/'./1
////5"
I-;"

k ,°V,
o
a o lO t[s] 20 30

100 t Is] 200 3o0

100

so
o

-r 40

2O

I 7o
b 0 10 20 30
t[s]

Fig. 2.20a, b. C o m p a r i s o n of a b s o r p t i o n rate of L a N i 5 s a m p l e s as a function of the n u m b e r

of cycles (1, 3,5..... 13). (a) industrially prepared sample (J.M.C.); (b)sample prepared in
laboratory by RF heating in a cold crucible [-2.87]

electrofused calcia. They are melted by high frequency induction heating and
kept in the liquid state for one hour. The atmosphere is either vacuum or argon
and no deoxidation materials are used. The melt is cast in a mold and a plate-
like block is obtained. The material is crushed to medium granules of
approximately 10 mm which are packed under argon and sealed in vinyl bags.
During the melting procedure the pumping time, the residual pressure, the
melting time, and the tapping temperature are controlled. Each batch is
analyzed and the peT curve is determined. The yield of usuable batches is
approximately 95 %. A batch is rejected when the hydrogen plateau pressure
deviates from the reference value by more than 0.5 bar.
An economically promising method of preparing FeTi compound was
studied by Shen et al. [2.89]. A chemical reduction was used to form FeTi from a
 Preparation of Intermetallics and Hydrides 37

F e - O - T i - O ore (illmenite), thus avoiding the process of chemical separation of

Fe and Ti. Other chemical methods to prepare FeTi, LaNis, and Mg2Ni from
chlorides or mixed oxides were also developed in China [2.90, 91].

2.6 Hydrogenation of lntermetallic Phases

The hydrogenation techniques will be treated rather briefly because they are
almost identical to those used for binary hydrides [2.92].
The usual way to hydrogenate the intermetallic compounds consists of
exposing the sample to hydrogen gas at a convenient pressure and temperature.
This is usually acceptable because in most intermetallic phases the
arrangement of the metallic components remains unchanged when hydrogen is
dissolved in the lattice. Nevertheless, a rapid decay of the metastable state
occurs sometimes even at quite low temperature like in Mg/CuH4, which
disproportionates to MgH2 and MgCu2 [2.93]. On the other hand, some
compound-like structures may only exist together with hydrogen. An example
is given by magnesium and iron. In the binary system they are almost
immiscible, but they form MgFe2H 6 in presence of hydrogen [2.94].
Intermetallic compounds are generally much more reactive towards
hydrogen than elemental metals and less sensitive to impurities in hydrogen
gas. Nevertheless the gaseous contaminents (mainly H 2 0 and 02) have an
inauspicious influence on the hydrogenation [2.95], thus a tight control of the
purity of the hydrogen is important. The apparatus should have a low
degassing rate and low leak rate [2.96]. In some cases it is necessary to use ultra
high vacuum equipment [2.34]. This is especially true when hydrogenation
takes place at high temperatures (> 200 °C) where oxidation is a very severe
problem. Secondly, high-purity hydrogen gas should be used, obtained e.g. by
the decomposition of a hydride. Nowadays containers filled with FeTi hydrides
are commercially available. They deliver suprapure hydrogen with an impurity
level well below 1 ppm [2.97, 98].
A guide-line for the choice of the pressure and temperature considerations
are the pcT diagrams. Isotherms with plateau pressures in the range 1-10 atm
are shown in Fig. 2.21 for the representative compounds mentioned in Sect. 2.2.
One should especially note the wide temperature range between - 9 0 and
300 °C according to the compound. Further isotherms are given in Chaps. 3 and
6 of this volume. Very thorough surveys of experimental results can be found in
[2.4, 16, 99-102].
The pressure-temperature-time (p-T-t) program is mainly dictated by the
passivity of the surface and by the resistance to decrepitation. The first property
governs the dissociative chemisorption of the hydrogen molecules and hence
the incorporation of the atoms in the lattice. The fundamental aspects for these
reactions are discussed in Chap. 2 of volume II.
Hydrogen transport in intermetallic compounds is usually a slow process.
This can be seen from the values of the diffusion coefficients given in Chap. 3 of
38 A. Percheron-Gu~gan a n d J.-M. Welter

/
100
Fig. 2.21. Comparison of desorption
FeTI isotherms of representative compounds
(4o~c at differenttemperatures

E 10
Y LaNi 5
2
14°°c)I ~
Mg2Ni
13oo~c)

~5 TiCrl. 8
(-90~)

0.1
0.5
Atom Ratio[H/M]

volume II. Nevertheless the sorption time is drastically reduced if decrepitation

occurs [-2.96]. This happens in compounds where the lattice expansion due to
hydride formation is so large that it cannot be accommodated by plastic
deformation. As a consequence the effective diffusion lengths are shortened.
Furthermore, new clean surfaces for hydrogen adsorption are created sponta-
neously. A drawback of this phenomena is the impossibility of preparing large
massive samples. To avoid an excessive dispersion of the freshly formed grains
an embedding technique with aluminium or copper can be very useful [2.103,
104]. This technique also has the advantage of improving the thermal
conductivity of the hydride bed.
Rare earth based compounds like LaNi5 and its substitutes and some
titanium and zirconium Laves phases have the advantage of an active surface
which is rather insensitive to the presence of oxygen and of a strong tendency to
decrepitation. Figure 2.22a shows a decrepitated grain of LaNis. With these
compounds one could basically start from the ingot and simply by hydriding
and dehydriding obtain a fine powder on the ~tm scale. Nevertheless, to
accelerate the kinetics the ingot should first be ground to grains with an
intermediate size of approximately 1 mm 3. Because grinding heats up the
particles, it is recommended to work in an inert gas atmosphere. With a
compound like TiFe a passive surface layer forms quite rapidly. Here it is
necessary first to activate the surface by running heating/cooling cycles up to a
few hundred degrees. Sometimes, judicious additions of different alloying
elements such as manganese lead to a material which is sufficiently active at
room temperature, and which does not necessitate any preliminary activation
treatment. Once hydrogenation has started, the material also decrepitates:
 Preparation of Intermetallics and Hydrides 39

Fig. 2.22. (a) Decrepitated grain of LaNis (3000 × ). (b) Decrepitated grain of FeTi

Fig. 2.22b reveals very nicely the peeling behaviour of decrepitation. The
specific surface areas of most intermetallic compounds decrepitated by some
hydrogen sorption cycles (about 10) amount to 0.1-1 m2/g[2.24, 79, 96, 105].
Once the hydrogenation has started, it is sufficient to keep the pressure
above the equilibrium pressure to have a high driving force. The amount of
hydrogen which is taken up by the sample can be followed by measuring the
weight increase of the sample or the pressure drop and the gas flow when
working at constant volume or pressure, respectively.
A modification of the gas-loading technique consists of dissolving the
hydrogen molecules in an organic solvent in which the sample is suspended
[-2.106].
Because hydrogenation may be quite inhomogeneous, especially when
different hydride phases can be formed, it is worthwhile checking the final
product by x-ray, and if possible by metallographic analysis. These two
techniques reveal mainly large inhomogeneities. At the present time no good
simple analytical method exists with a high spatial and/or compositional
resolution.
For these examinations, and in a more general way for many experiments, it
is necessary to expose the hydrided material to air. Precautions must be taken to
ensure that the fine powder does not start to burn or the hydrogen does not
escape. In many cases the hydride can be sealed quite effectively by the
adsorption of a layer of SO2 or CO [2.95, 103, 107]. Air should be introduced
into the reaction vessel progressively in small amounts using an inert gas as a
ballast. In all cases the hydrogen content should be controlled after the
experiments, e.g. by desorption at high temperature. Furthermore, it is also
40 A. Percheron-Gudgan and J.-M. Welter

Fig. 2.23. Scanningelectron micro-

graph of a LaNis-Teflon compo-
site electrode pressed on a nickel
grid (1000 x) [2.111]

necessary to check whether disproportionation processes have taken place as

a consequence of internal hydriding.
A very new method to prepare highly reactive magnesium and Mg2Ni with
large specific surface areas (,--100 m2/g) was described by Bogdanovie et al.
[2.108]. An organic solvent in the presence of soluble organo transition-metal
catalysts is used to hydrogenate magnesium powder. By doping the catalyti-
cally prepared magnesium hydride with small amounts of a nickel
organo metallic compound, a well defined MgzNi or hydride in amorphous
form is obtained.
Another way to charge the intermetallic compound is to use hydrogen ions
in an electrochemical medium. The electrolyte may be either an acid or a KOH
aqueous solution. The experimental set-up is quite inexpensive, but it is limited
to rather small samples [2.64, 78]. Furthermore, the technique is suited for
hydrides with a low equilibrium pressure [2.109]. Sometimes, to avoid
hydrogen losses from materials which form hydrides of low stability, it is
necessary to pressurize the electrochemical cell with hydrogen gas at the
equilibrium pressure of the hydride [2.110]. The limitations of the technique are
the difficulties in controlling the amount and spatial distribution of the
dissolved hydrogen. The problem associated with the decrepitation of the
cathode can be eliminated by mixing the active particles with metallic or
polymeric powder. Furthermore, the cathode must be pressed into an envelope
of porous metal to ensure a good electrical contact. An example of such a
mounting is shown in Fig. 2.23 where LaNi s was pressed into a nickel grid
[2.1113.

2.7 Concluding Remarks

To conclude this chapter we will give a few examples of how the preparation of
the intermetallic compounds affects their aptitude to form hydrides. This
 Preparation of Intermetallics and Hydrides 41

aptitude will be characterized by the plateau pressure, the hydrogen loading

capacity, the kinetics and the cyclic life-time.
An important factor which influences these properties is the level of
impurities contained in the compound. We do not know of any work where the
influence of metallic impurities at less than 100 at. ppm level has been studied.
To get a feeling for the importance of such impurities, it is necessary to use the
data obtained with metallic substitutions at a 1-10 at.% level. Using data
available in the literature [2.42, 112], the temperature dependence of the
plateau pressures at a HIM ratio of 0.5 is shown in Fig. 2.24 for various
substitutes in FeTi. Similar variations are found for non-metallic impurities like
C, O, and S as shown in Fig. 2.25. Typically 1 at. % of impurities induces a
variation of the order of I arm in the plateau pressure. A closer examination of
the data shows that at least for the metallic substitutions the plateau pressure
depends exponentially on the volume expansion of the compound due to the

TI°Cl
100 70 40 10
i i I i
25 \

T[*Cl
100 7O 40 25 10
i 'l I I I I
10
÷

~C 5

o. i

i
2.5
\
1 _r

0.5

2.75 3 3,25 3,50 2.75 3 3.25 3.50

lo001TI K1] 1000/'1" [ K -1]

Fig. 2.24. Stability of FeTiHo.s type hydrides Fig. 2.25. Influence of various oxygen ( / )
for various partial substitutions of iron or
titanium by other metallic elements. Data contents (wt.%=0.01--0.87) at 40°C
compiled from [2.112] and [2.113]. The solid [2.24], of various sulphur contents
line is representative for pure FeTi and has
been evaluated in [2.59]. (i,) FeTil_xAl~; (0.004 < S/Ti at. ratio < 0.08) at 25 °C [2,25]
x <0.1 ; (C3) FeTio.95Vo.os, (0) and of 2at.% carbon (i) (20°C<T<91 °CI
FeTio.gsCro.os; (o) FeTio.soZro.2o; (+) [2.26] dissolved in the base compound on
FeTio.ssSio.15; ( × ) FeTiMno.2 the stability of FeTiH0,5
 42 A. Percheron-Gu~gan and J.-M. Welter

Fig. 2.26. Stability of LaNi 5 type

hydrides for various partial substitutions
of nickel or lanthanum by other metallic
elements. Data compiled from [2.56]. (~)
La(Ni, Pt)5; (A) La(Ni, Cu)5; (~)
Yh(Ni, Al)s; (~) La(Ni, Mn)5; (~)
La(Ni, Fe)s; (o) La(Ni, Al)s; (n) RNis.
Plateau pressure [bar] is plotted versus
unit cell volume Vii of the intermetallic
compound
o.
¢:
•J 0

1 \ ~ ~ 'L "'-. •

-2
'\
\
\
-4

I \\
-6
80 84 88 92 96
, V= [~31

=E
~- e 8
.o
h-

E
~= 06

~ 0.4
Fig. 2.27. Influence of oxygen (11)
I [2.24] and sulphur (e) [2.25]
content dissolved in FeTi on the
0.2 maximum loading capacities
0 0.01 0.02 0.03
_ Alloy Impurities Content [ 0 or S/M]

foreign elements. This is clearly demonstrated in Fig.2.26 for LaNi5

[2.47, 113, 1141. Such an argument does not hold for the non-metallic elements:
here a second phase is precipitated. In the case of FeTi one has observed the
formation of TiC I-2.26, 42], FeTTiloOa [2.42], and Ti2S I-2.25]. This leads, of
course, to a shift of stoichiometry in the bulk material. All these impurities also
influence the shape of the pcT curves and, in particular, the loading capacity of
the compound. As an example we show in Fig. 2.27 how the capacity is reduced
by the presence of oxygen and sulphur in FeTi.
 Preparation of Intermetallics and Hydrides 43
10
Fig. 2.28. Comparison of
40°C isotherms of LaNis
8
prepared by RF melting in
cold crucible under vacuum
7 (o), are melting (n) and are

[,
5~_=
6
melting followed by an
annealing treatment at
1100 °C for i½ hours (zx).The
lattice parameters a and c in
/~ of the compounds are
3 respectively: (o) 5.014, 3.977;
(n) 5.005, 3.974; (/9 5.007,
3.973 [2.86]

0 L
0 3 4 5
, Atom Ratio [H/M]
100

50 _ _ _

¥
10 _ _

5
f .:
Z

1 _ : ¢

0.5
Fig. 2.29. Desorption isotherms at 25°C
of Mischmetal substituted calcium-nickel
compound (Cao.vMo.3Nis): As-cast (*) and
annealed for 24 hours at 1000 °C (I) and at
0.1 1100 °C (A) [2.42]
0 0.2 0.4 0.6 0,8 1.0 1.2
Atom Ratio[H/M]

Even if all impurities influence the plateau pressure and the loading
capacity in a similar way, the problem is more severe for the non-metallic
impurities. As we mentioned in Sect. 2.3.1 the level of foreign metallic elements
can be kept below 100 at. ppm. This is not true for non-metallic elements, either
because the starting materials are strongly doped with them or because they are
introduced during the process. Levels of 1 at. % frequently occur.
Another important factor which governs the hydrogenation behaviour of
the c o m p o u n d s is the solidification structure. This aspect is revealed in Fig. 2.28
44 A. Percheron-Gudgan and J,-M. Welter
6
Fig. 2.30. Hydrogen storage
capacities of LaNi5 samples
& from various suppliers
,L
during absorption cycling:
I -N (A) Research Chemicals
4 ____
X + \ sample, dry H2; (B)
Ergenics (C) Research
Chemicals; (D) Molycorp
samples are cycled in 12 arm.
hydrogen with about
I 10 Torr H20 [2.115]

2
0 30 60 90 120 150 180
• Cycles

which shows the different behaviour of LaNis obtained by melting together

lanthanum and nickel from the same batches in an arc melter and in a cold
crucible (see Sect. 2.3.2). Probably for homogeneity reasons the cold crucible
sample exhibits flatter plateaus at lower pressures, a smaller hysteresis loop and
an increased absorption capacity. By a heat treatment the properties of the arc
melt material are slightly improved. The same effect was previously mentioned
in Sect. 2.3.3. and illustrated in Fig. 2.10. Nevertheless, some precautions must
be taken when a material containing a highly volatile component is heated. This
is clearly seen in Fig. 2.29 where isotherms of three Cao.TMo.3Ni5 ( M =
mischmetal) samples, which have experienced different heat treatments
[2.42], are shown.
As a conclusion, we hope that we have illustrated the importance of the
metallurgical state of the base intermetallic compounds which will be used for
research and application purposes. According to our experience, a tight control
of the whole fabrication process needs a high degree of care. As a final example a
comparative study of the cyclic life-time of different commercially available
LaNi5 samples should be quoted (Fig. 2.30) [2.115]. Furthermore, we can only
repeat that a painstaking characterization of the material and of the process is
mandatory for the use of hydrides in basic research as well as in their
applications.

References
2.1 A. Brown, J.H. Westbrook: "Formations Techniques", in Intermetallic Compounds, ed.
by J.H. Westbrook (Wiley, New York 1967) Chap. 17, pp. 303-339
2.2 W.M. Mueller, J.P. Blackledge, G.G. Libowitz: Metal Hydrides (Academic, New York
1968)
2.3 K.H.J. Buschow, P.C.P. Bouten, A.R. Miedema: Rep. Prog. Phys. 45, 937-1039 (1982)
2,4 D.G. Ivey, D.O. Northwood: J. Mater. Sc[. 18, 321-347 (1983)
2.5 A.R. Miedema: Z. Metallkde. 69, 445-461 (1978)
2.6 J.W. Cahn, J.E. Hilliard: J. Chem. Phys. 28, 258-267 (1958)
 Preparation of Intermetallics and Hydrides 45

2.7 M. Hansen, K. Anderko: Constitution of binary alloys, 2nd ed. (McGraw-Hill, New
York 1958)
2.8 R.P. Elliott: Constitution of binary alloys. 1st Suppl. (McGraw-Hill, New York 1965)
2.9 F.A. Shunk: Constitution of binary alloys, 2nd Suppl. (McGraw-Hill, New York 1969)
2.10 N.V. Ageev: Diagrammy Sostoyaniya Metallicheskik Sistem (Metallurgiya, Nowsow
1969)
2.11 R. Hultgreen, R.L. Orr, P.D. Anderson, K.K. Kelley: Selected Values of Thermodynamic
Properties of Metals and Alloys (Wiley, New York 1963, ASM; Metals Park/04 1973)
2.12 W.G. Moffatt: The Handbook of Binary Phase Diagrams (General Electric Co.
Shenecdaty NY 1977)
2.13 Bulletin of alloy phase diagrams since (1980)
2.14 R.D. Shull: Bulletin of Alloy Phase Diagrams 4, 5 15 (1983)
2.15 K.H.J. Buschow, H.H. Van Mal: J. Less-Common Met. 29, 203-210 (1972)
2.16 K.H.J. Buschow: "Hydrogen Absorption in Intermetallic Compounds" in Handbook on
the Phys. and Chem. of Rare Earths, ed. by K.A. Gschneidner, Jr., L. Eyring (North
Holland, Amsterdam 1984) Vol. 6, Chap. 47, 1-111
2.17 J.L. Murray: Bulletin of Alloy Phase Diagrams 2, 320-334 (1981)
2.18 J.L. Murray: Bulletin of Alloy Phase Diagrams 2, 174~181 (1981)
2.19 P. Villars, L.D. Calvert: Pearsons Handbook of Crystallographic Data for lntermetallic
Phases, Vols. I-3 (American Soc. for Metals, Metals Park, Oh. 1985)
2.20 H. Huthman, G. Inden: Phys. Status Solidi A 28, K129-130 (1975)
2.21 G. Inden: Z. Metallk. 66, 648-653 (1975)
2.22 J. Schweizer, F. Tasset: J. Phys. F 10, 2799-2817 (1980)
2.23 D. Givord, J. Laforest, J. Schweizer, F. Tasset: J. Appl. Phys. 50, 2008-2010 (1979)
2.24 G.D. Sandrock, J.J. Reilly, J.R. Johnson: Proc. l l t h Intersociety Energy Conversion
Engineering Conference, Vol. I, 965-971 (1976)
2.25 R. Suzuki, J. Ohno, H. Gondoh: J. Less-Common Metals, 104, 199-206 (1984)
2.26 D. Khatamian, N.S. Kazama, F.D. Manchester, G.C. Weatherly, C.B. Alcock: J. Less-
Common Met., 91, 267-273 (1983)
2.27 Ellingham: Proceedings of the Electric Steel Conference Amer. Inst. M in. Met. Eng.
1949 in: "Aspects Physicochimiques de l'Elaboration des M~taux" ed. by D.W. Hopkins
(Dunod, Paris 1958)
2.28 K.A. Gschneidner, N. Kippenhan, O,D. Mc. Masters: Thermochemistry of the Rare
Earths, Rare Earth Information Center Report IS.RIC.6, 1973
2.29 A.N. Nesmeyanov, R. Gary: Vapor Pressure of the Chemical Elements (Elsevier,
Amsterdam 1963)
2.30 R.E. Honig, D.A. Kramer: in Techniques of Metals Research ed. by R.A. Rapp
(Interscience Publishers New York 1970) Vol. IV, Part 1
2.31 J.M. Welter: Metall. 28, 731-735 (1974)
2.32 J.M. Welter, H. Wenzl: Metall. 39, 116 125 (1985)
2.33 Y. Osumi, H. Suzuki, A. Kato, K. Oguro, M. Nakane: J. Less-Common Met. 74,
271 277 (1980)
2.34 H. Wenzl, J.M. Welter: "Properties and Synthesis of Nb-H Interstitial Alloys. Current
Topics" in Mat. Science, Vol. 1, ed. by E. Kaldis (North Holland, Amsterdam 1978)
2.35 B.J. Beaudry, K.A. Gschneidner, Jr.: in Handbook on the Physics and Chemistry of Rare
Earth, ed. by K.A. Gschneidner, Jr., L. Eyring (North Holland, Amsterdam 1978) Vol. 1,
Chap. 2, p. 173-232
2.36 Z. Zhou, H. Zhang, D. Tang, D. Zhou, X. Ping, J. Ni: Proc. 6th World Hydrogen Energy
Conf., Vienna 1986, ed. by T.N. Veziroglu, N. Getoff, P. Weinzierl. (Pergamon, Oxford
1986) Vol. 2, p. 898
2.37 G.D. Sandrock: Proceedings of the 12th Intersociety Energy Conversion Engineering
Conference, Am. Nuclear Soc., 1,951-958 (1977)
2.38 M. Beyss, J.-M. Welter, T. Kaiser: J. Crystal Growth 50, 419-428 (1980)
2.39 H. Biloni: "Solidification", in Physical Metallurgy, ed. by R.W. Cahn, P. Haasen (North
Holland, Amsterdam 1983) pp. 477 579
46 A. Percheron-Gu~gan and J.-M. Welter

2.40 W. Kurz, D.J. Fisher: Fundamentals of Solidification (Trans. Tech. Publications, 1984)
2.41 C. Lartigue: "Etude Structural et Thermodynamique du Syst6me LaNis_~Mnx-
Hydrog~ne" Th6se 36me cycle, Universit6 Paris VI, 1979
2.42 G.D. Sandrock: Proceedings of the 2nd World Hydrogen Energy Conference 3,
1625 1656, 1978, ed. by International Association for Hydrogen Energy
2.43 M.H. Mendelsohn, D.M. Gruen, A.E. Dwight: Mat. Res. Bull. 13, 1221 1224 (1978)
2,44 M.H. Mendelsohn, D.M. Gruen: in Rare Earths in Modern Science and Technology, ed.
by G.J. Carthy, H.B. Silber, J.J. Rhyne (Plenum, New York 1980) 2, 593-598
2.45 J. Lamloumi, C. Lartigue, A. Percheron-Guegan, J.C. Achard: in The Rare Earths in
Modern Science and Technology, cd. by G.J. Mc Carthy, H.B. Silber, J.J. Rhyne (Plenum,
New York 1982) 3, 487-488
2.46 J. Lamloumi, A. Percheron-Gu6gan, J.C. Achard, G. Jehanno, D. Givord: J. de Phys. 45,
1643 1652 (1984)
2.47 J.C. Achard, A. Percheron-Gu6gan, H. Diaz: Proceedings Second International
Congress on Hydrogen in Metals, Paper IE12, Paris, 1977
2.48 M.H. Mendelsohn, D.M. Gruen, A.E. Dwight: Nature 269, 45-47 (1977)
2.49 C. Lartigue, A. Percheron-Gu6gan, J.C. Achard, F. Tasset: J. Less-Common Met. 75,
23 29 (1980)
2.50 C.E. Lundin, F.E. Lynch: Proc. International Conf. on Alternative Energy Sources,
Miami Beach, FL, pp. 3803-3820, 1977
2.51 J.J. Reilly, R.H. Wiswall: Summary Rep. BNL 17136 Upton, NY: Brookhaven Natl.
Lab., p. 28, 1972
2.52 J. Shinar, I. Jacob, D. Davidod, D. Shaltiel: Hydrides Jbr Energy Storage, ed. by A.F.
Andresen, A.J. Maeland (Pergamon, Oxford 1978) pp. 337-352
2.53 H.H. Van Mal, K.H.J. Buschow, F.A. Kuijpers: J. Less-Common Met. 32, 289-296
(1973)
2.54 I.D. Weisman, L.H. Bennet, A.J. Mc Alister, R.E. Watson: Physical Review B 11, 82-91
(1975)
2.55 A. Percheron-Gu6gan, C. Lartigue, J.C. Achard: J. Less-Common Met. 109, 287-309
(1985)
2.56 J. Lamloumi: "Etude Structurale, Thermodynamique et Magn6tique du Syst&me
LaNis_~Fex-hydrog&ne", Th&se 3~mc cycle, Universit6 Paris VI, 1982
2.57 H.H. Van Mal, K.H.J. Buschow, A.R. Miedema: J. Less-Common Met. 35, 65-76 (1974)
2.58 S. Cirafici, A. Palenzona: J. Less-Common Met. 53, 199-203 (1977)
2.59 J.-M. Welter, G. Arnold, H. Wenzl: J. Phys. F13, 1773 (1983)
2.60 M. Beyss, T. Kaiser, A.J. Singh, J.-M. Welter: Proc. 2nd European Conference on
Crystal Growth, Lancaster (1979)
2.61 L. Schlapbach, A. Seiler, H.C. Siegman, T.V. Waldkirch, P. Z/ircher: Int. J. Hydrogen
Energy 4, 21-28 (1979)
2.62 A.J. Maeland: In Hydrides for Energy Storage, ed. by A.F. Andresen, A.J. Maeland
(Pergamon, Oxford 1978) p. 427-462
2.63 A.J. Maeland: In Hydrogen in Disordered and Amorphous Solids, ed. by G. Bamhakidis,
R.C. Bowman, NATO ASI Series B 136, p. 127 (Plenum, New York 1986);
G.G. Libowitz, A. Maeland: J. Less-Common Met. 101, 131 (1984);
A.J. Maeland, L.E. Tanner, G.G. Libowitz: J. Less-Common. Met. 89, 183 (1983)
2.64 K. Suzuki: J. Less-Common. Met. 89, ~83-195 (1983)
2.65 R.C. Bowman, J.S. Cantrell, A. Attala, D.E. Etter, B.D. Craft, J.E. Wagner, W.L.
Johnson: J. Non-Crystalline Solids 6162, 649-654 (1984);
R.C. Bowman, J.S. Cantrell, E.L. Venturini, R. Schulz, J.E. Wagner, A. Attala, B.D.
Craft: In Rapidly Quenched Metals, ed. by S. Steeb, H. Warlimont (Elsevier, Amsterdam
1985) p. 1541-1544
2.66 H. Bernas, A. Traverse, C. Janot: In Amorphous Metals and Semiconductors, ed. by P.
Haasen, R.I. Jaffee, (Pergamon, Oxford 1985) p. 435;
A. Traverse, H. Bernas: J. Less-Common. Met. 129, 1-12 (19871
 Preparation of Intermetallics and Hydrides 47

2.67 K. Samwer: J. Less-Common Met. 131, 210 (1987) and NATO ASI Series B136 on
Hydrogen in Disordered and Amorphous Solids, ed. by G. Bambakidis, R.C. Bowman
(Plenum, Oxford 1986);
H. Schr6der, K. Samwer, U. K6ster: Phys. Rev. Lett. 54, 197 (1985)
2.68 H.J. Gtintherodt, H. Beck (eds.): Glassy Metals, I, II, Topics Appl. Phys. Vols. 46, 53
(Springer, Berlin, Heidelberg 1981, 1983)
2.69 M.L. Post, J.J. Murray, G.J. Despault, J.B. Taylor: Mat. Res. Bull. 20, 337-342 (1985)
2.70 G.Y. Adachi, K.I. Niki, J. Shiokawa: J. Less-Common Met. 81, 345-348 (1981)
2.71 B.R. Livesay, J.W. Larsen: Proceedings of JIMIS-2, Hydrogen in Metals, November
26 29, 1979, Minakami, Japan, Suppl. Transact. Jap. Inst. of Metals 21, 345 (1980)
2.72 K. Nakamura: Scripta Met. 18, 793 (1984)
2.73 K. Sumiyama, Y. Hashimoto, Y. Nakamura: Transact. Jap. Inst. of Metals 24, 61 (1983)
2.74 E.M. Wiirtz: private communication (1985)
2.75 H. Baars: Diplomarbeit, Fachhochschule Aachen, Abt. Jtilich (1985)
2.76 J.F. Lynch, J.J. Reilly: J. Less-Common Met. 87, 225-236 (1982)
2.77 T. Schober, G. Arnold, A.W. Nellen: Z. Praktische Metallographie 16, H.7, 309 (1979)
2.78 T. Schober: Scri. Metall. 13, 107 (1979)
2.79 L. Schlapbach: J. Phys. F 10, 2477-2490 (1980)
2.80 Gesellschaft fiir Elektrometallurgie: private communication (1984)
2.81 Ergenics: Hy-Stor Marketing Leaflet (1984)
2.82 Japan Metals & Chemicals: private communication (1984)
2.83 Gesellschaft f/Jr Elektrometallurgie: Patent Application (1984)
2.84 Gesellschaft fiir Elektrometallurgie: private communication (1985)
2.85 Santocku Metal Industry: private communication (1984)
2.86 F. Briaucourt, F. Demany, A. Percheron-Gu6gan: private communication (1985)
2.87 L. Touron, A. Percheron-Gu6gan: private communication (1985)
2.88 Chuo Denki Kogyo: private communication (1984)
2.89 P. Shen, G. Wang, W. Yin, D. Song, D. Zhan: Rare Metals, The Chinese Society of
Metals, Vol. 2, 38 (1983)
2.90 P. Shen, G. Wang, D. Song, Z. Zhou, D. Zhang, H. Yuan: Proc. 5th World Hydrogen
Energy Conf., Toronto 1984, ed. by T.N. Veziroglu, J.B. Taylor, Vol. 3, p. 1303, Int.
Assoc. Hydrogen Energy;
P. Shen, Y. Zhang, S. Zheng, X. Feng, H. Yuan, S. Chen: Proc. 6th World Hydrogen
Energy Conf., Vienna 1986, ed. by T.N. Veziroglu, N. Getoff, P. Weinzierl (Pergamon,
Oxford 1986) Vol. 2, p. 831
2.91 D. Song, G. Wang, Z. Zhou, F. Wang, S. Li, S. Lu, J. Wu: Proc. 6th World Hydrogen
Energy Conf., Vienna 1986, ed. by T.N. Veziroglu, N. Getoff, P. Weinzierl (Pergamon,
Oxford 1986) Vol. 2, p. 845
2.92 Coy L. Huffine: In Metal Hydrides, ed. by W.M. Mueller, J.P. Blackledge, G.G.
Libowitz (Academic Press, New York 1968) Chap. 13, pp. 675 747
2.93 J.J. Reilly, R.H. Wiswall: Inorg. Chem. 6, 2220-2223 (1967)
2.94 J.J. Didisheim, P. Zolliker, K. Yvon, P. Fischer, J. Schefer, M. Gubelmann, A.F.
Williams: Inorg. Chem. 23, 1953 1957 (1984)
2.95 G.D. Sandrock, P.D. Goodell: J. Less-Common Met. 73, 161-168 (1980)
2.96 H.H. Van Mal: Philips Res. Repts. Suppl. 1, 1-85 (1976)
2.97 K.-H. Klatt, S. Pietz, H. Wenzl: Z. Metallkde. 69, 170-173 (1978)
2.98 HWT Gesellschaft f/Jr Hydrid- und Wasserstofftechnik mbH, D-4330 Miihlheim
2.99 R. Wiswall: "Hydrogen storage in Metals" in Hydrogen in Metals II, ed. by G. Alefeld, J.
V61kl, Topics Appl. Phys., Vol. 29 (Springer-Verlag, Berlin, Heidelberg 1978) Chap. 5,
pp. 201-242
2.100 W.E. Wallace, R.S. Craig, V.U.S. Rao: Adv. Chem. 186, 207-240 (1980)
2.101 H. Oesterreicher: Appl. Phys. 24, 169-186 (1981)
2.102 P.S. Rudman, G. Sandrock: Ann. Rev. Mat. Sci. 12, 271-294 (1982)
2.103 M. Ron, D. Gruen, M. Mendelsohn, I. Sheft: J. Less-Common Met. 74, 445-448 (1980)
2.104 J. Toepler, O. Bernauer, H. Buchner: J. Less-Common Met. 74, 385-399 (1980)
48 A. Percheron-Gu~gan and J.-M. Welter: Preparation of Intermetallics and Hydrides

2.105 L. Belkbir, E. Joly, N. Gerard, J.C. Achard, A. Percheron-Gu6gan: J. Less-Common

Met. 73, 69-77 (1980)
2.106 J.J. Reilly, J.R. Johnson: J. Less-Common Met. 104, 175-190 (1984)
2.107 D.M. Gualtieri, K.S. Narasimhan, T. Takeshita: J. Appl. Phys. 47, 3432 (1976)
2.108 B. Bogdanovic: Int. J. Hydrogen Energy 9, 937-941 (1984); 12, 863-873 (1987)
B. Bogdanovic, K.H. Claus, S. Gfirtzgen, B. Spliethoff, U. Wilczok: J. Less-Common
Met. 131, 163 172 (1987)
2.109 J.C. Achard, A. Percheron-Gu6gan, J. Sarradin, G. Bronoel: In: A.F. Andresen, A.J.
Maeland (eds.) Proc. Int. Symp. on Hydrides for Energy Storage, Geilo, 1977
(Pergamon, Oxford 1978) pp.485 490
2.110 M.H.J. Van Rijswick: In A.F. Andresen, A.J. Maeland (eds.) Proc. Int. Syrup. on
Hydrides for Energy Storage, Geilo 1977 (Pergamon, Oxford 1978) pp. 261-271
2.111 A. Percheron-Gu6gan: L'Actualit6 Chimique 2, 11-18 (1982)
2.112 H. Buchner: Energiespeicherung in Metallhydriden (Springer, Wien, New York 1982)
2.113 K.A. Gschneidner Jr., T. Takeshita, Y. Chung, O.D. McMasters: J. Phys. F 12, L1-L6
(1982)
2.114 M.H. Mendelsohn, D.M. Gruen, A.E. Dwight: J. Less-Common Met. 63, 193-207
(1979)
2.115 G.L. Holleck, J.R. Driscoll, B.E. Paul: J. Less-Common Met. 74, 379, 384 (1980)
3. Thermodynamics of
lntermetallic Compound-Hydrogen Systems
Ted B. Flanagan and W. Alan Oates

With 14 Figures and 3 Tables

This chapter concerns the thermodynamics of intermetallic com-

pound-hydrogen systems. The single phase and two condensed phase
situations are considered. Integral and partial thermodynamic quantities are
defined with reference to these systems. Experimental approaches are described
and compared. Special difficulties encountered with intermetallic compound
hydrides are discussed, e.g., effects of activation and annealing, sloping plateaux,
decomposition. Results for some representative systems are described. Phase
transitions in these systems are considered, as is the relationship between partial
and complete equilibrium.

3.1 Introduction
The suitability of intermetallic compounds for energy storage and other
technological purposes rests principally on their thermodynamic properties
and thus a knowledge and understanding of these properties is of paramount
importance.
The need for this review arises because the properties of intermetallic
compounds introduce features of behavior towards hydrogen not found in the
pure metal-hydrogen systems. For example, the range of stoichiometry over
which most intermetallic compounds can exist introduces trapping sites which
influence the dilute phase hydrogen solubility. The variety of interstices
available for hydrogen occupation in most intermetallic compounds introduces
complicating features in the statistical thermodynamic modeling of these
systems. The well-known instability of intermetallic compound hydrides
towards decomposition into a component metal hydride also leads to
complications in the thermodynamic description of these metastable systems.
The present review will not attempt to survey all aspects of the thermody-
namics of intermetallic compound-hydrogen systems but will concentrate on
the more fundamental aspects. Results for several prototype systems will be
described in detail. Other recent reviews, which overlap with the present one to
a certain extent, have been given by Buschow et al. [3.1], Rudman and Sandrock
[3.2], Ivey and Northwood [3.3] and Oesterreicher [3.4]. A comprehensive
tabulation of the plateau pressures and hydrogen capacities of intermetallic
hydrides is given in [3.1] and Chap. 6 of this volume should also be consulted.
50 T. B. Flanagan and W.. A. Oates

The words "intermetallic compound", "intermetallic hydride" and "inter-

metallic compound-hydrogen system" occur so frequently in this review that we
will use the abbreviations IMC (plural IMCs), I M H (plural IMHs) and I M C H
(plural IMCHs), respectively.

3.2 Macroscopic Thermodynamics

At low temperatures, IMCHs behave in a pseudo-binary, partial equilibrium or
para-equilibrium (PE) [3.5] manner. This type of behavior, which gives rise to a
maximum of two solid phases in equilibrium with hydrogen gas, is discussed in
Sect. 3.2.1. However, an I M C H may, under the right conditions, behave as a
ternary system in complete equilibrium (CE), when a maximum of three solid
phases may be in equilibrium with hydrogen gas. CE, which is most likely to
occur at higher temperatures, is discussed in Sect. 3.2.2. We shall stress later (in
Sect. 3.5.4) that the more commonly observed PE behavior is not independent
of CE, so that an understanding of the latter can provide insight into the more
commonly met PE behavior.
The following notation will be employed throughout:
The I M C is composed of the metals A and B and, in substituted compounds,
of metal C also. Metal A is assumed to have a stronger affinity for hydrogen
than B.
The composition of the host I M C is expressed as c = n~/(nA +nB) so that the
IMC may be written A 1-cB~• In substituted IMCs we use A1 -c(B1 _ xCx)~, where
x = nc/(nn + nc) and c = (nn + nc)/(nA + nn + nc). The H concentration is ex-
pressed as r = nH/(n A + riD) so that the I M H composition is A 1_cB~H,. Alterna-
tively, the H concentration is sometimes expressed in terms of 0, the fraction of
interstitial sites occupied. Sometimes the IMC will be denoted by M and the
I M H by M H r. Thus the compound LaNisH6. 4, for example, may also be
written as Lao.tvNio.s3Hl.07 or as MHI.07. The composition "a" refers to the
value ofr in the metal-rich phase which is in equilibrium with the hydride, and
"b" refers to the value o f t in the hydride phase which is in equilibrium with the
metal-rich phase, i.e., M H , is in equilibrium with MHb. Saturation occurs at the
composition r = y, i.e., MH r is a stoichiometric compound. In some systems
several such saturations occur at increasingly higher H concentrations as the
hydrogen chemical potential is increased.

3.2.1 Single and Two Condensed Phase Equilibria

a) Integral/Formation Properties
Figures 3.1-3 illustrate, schematically, the influence of H concentration on the
integral and partial thermodynamic properties G, S, and H. The integral
quantities are expressed per unit amount of M , e.g., A G/nM, which we denote by
AGm. Pure M and H2(g ) at 1 atm are chosen as the standard states, i.e., AG m
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 51

A
t O

~
~--

~
- -

AGm(b) l E
0
\\ A~JM(plat.) /
,.! , <
A~JHa(plat.) /,
,'~1>-,
j, i i

M ~ r_i> 6 MHy Ma F ~ I~ MHy

Fig. 3.1a, b

I b
I
I
I
I
I
i ~ Il/
I ASH2(a) iI
I / /
f
I
J
/
A~ff(MHy) ASM(b)J
j
i ASH2(plat.)
[ ///
.<] z~SM(plat.)
I / /
ASH2(b)~,

o!
/,ASM(a)

0 I I

F--~
b iMHy Ma r'-~
b MHy
Fig. 3.2a, b
Fig. 3.1. (a) Schematic plot of AG,~ against hydrogen content. The phase boundaries are shown
as the values of r at the points of common tangent. (b) Schematic plot of A#~ against hydrogen
content where A#t represents A#n2 or A#M. ( ), A#H2, and (. . . . ), A#u
Fig. 3.2. (a) Schematic plot of ASm against hydrogen content. ASH2(plat) is given by
2 ( b - a ) i [ASm(b)--ASm(a)] where AS,~(b)and dSm(a)are shown by the small filled circles. (b)
Schematic plot of d S~against hydrogen content where AS~represents ASn2 or AS,,. ( ), ASn2
and ( - - - ) , ASM.The values of ASn~and AS~ at the co-existing phase boundaries are indicated
by the small filled circles

refers to the free energy change for the mixing or formation reaction:
M(s)+ r/2 H2(g)= MHr(s), (3.1)
where s and g denote solid and gaseous states, respectively. Using these
standard states, the free energy of mixing at r = y is the same as the standard free
energy of formation of MHy, A GO(MHr), i.e.,
A G~MHv) = A Gin(y). (3.2)
52 T. B. Flanagan and W. A. Oates

i i
I I
I I /
~''HM(a)
_1_ j_ _
I I AHM(plat.)
AHM(b)%
\
AH~(MHy) \
AHH2(plat.) /
f
AHH2(b)¢"

/I
I
I
I I I
Ma
I'-..-.e,
b MHy I'4 a
r'-,-.>
b MHy

Fig. 3.3. (a) Schematic plot of AH,~ against hydrogen content. The values of AH,~ at the phase
boundaries are indicated by the small filled circles.(b) Schematic plot of AH~against hydrogen
content where AH~represents AHn2 or AH u. (-----), AHH2 and (---), AHu. The values of AHH~
and AHu at the co-existing phase boundaries are indicated by the small filled circles

b) Partial Molar Properties

When A G,, and r are used as variables, ihe partial quantities in a single-phase
region may be obtained from the tangent slope and tangent intercept to the
integral quantity [3.6]. The integral and partial quantities are related to one
another at constant temperature and total pressure by the following equations:

a G.(r) = (r/2) a ~.2(r) + a ~M(r) (3.3)

d[AGm(r)]
A#a2(r ) = 2 dr (3.4)

A#M(r) = A G=(r) -- r d[AG,,(r)]

dr where (3.5)

A#H~ = 2 # H ( r ) - # ~ = R T l n ( p . j a t m ) and, (3.6)

A#M = ~M(r) -- # ~ , (3.7)

where ~ refers to the structural form existing in the pure metal. A~H2 is the
property most accessible to experimental determination. The other partial and
the integral quantity are readily obtained from it as follows:

AGm(r) = ½ i 3#n~dr (3.8)

0

A #M(r) = A G,,(r) -- ½rA/~n2(r). (3.9)

 Thermodynamics of Intermetallic Compound-Hydrogen Systems 53

The chemical potentials of the individual metallic components may be obtained

from the equations:

A #A(r) = A Gm(r ) - i r A # . ~ -- c \(~ham~

---~c]~ , (3.10)

= Gin(r)- ½r#.2 +(1-c) (aA

\ 8cm'}
/r" (3.11)

Since A~u = (1 -- c)ApA + cA#B it can be seen that (3.10) and (3.11) are consistent
with (3.9).
When using models for the solid phases it is desirable to choose some
standard state for the hydrogen potential,/L~, in the particular phase of interest.
For pure metal-hydrogen systems the most convenient choice is the hypothet-
ical infinitely dilute solution standard state 1-3.7]. With this choice the hydrogen
potential standard state difference, A u~ = p~ -~#i~2,
1 ,, can be obtained from

A #H'*=!imfRTln(p~/~z~-~)l . (3.12,

where fl is the number of interstitial sites per metal atom. An excess H potential
may then be defined by

~A#H 2 - , (3.13)

where the term in square brackets is the H potential of the ideal solution.
Because of the multiplicity of the different types of sites for H occupation in
IMCs a suitable definition of an ideal solution and hence a reference state for H
in the solid is not as straightforward as for pure metals. The best solution is to
use (3.12) and (3.13) with fl being taken as the value of r equal to the
experimentally observed saturation composition, y.

e) Plateau Properties
When PE behavior by the I M C H is being considered the I M C behaves as one
component and the coexistence of the dilute and the pseudo-binary I M H phase
leads to one degree of freedom or a constant pressure of hydrogen at constant
temperature, the plateau pressure.
This co-existence of the two condensed phases may arise from either the
presence of H - H attractive interactions, which lead to a phase separation at low
temperatures, or from a structural difference between the I M C and the IMH,
which can occur in the presence or absence of H - H interactions. In practice, the
phase transformations are accompanied by hysteresis, which is discussed in
Sects. 3.3.3, 3.5.5 but in this section we will assume that the plateau pressures
represent equilibrium states.
54 T B. Flanagan and IV..A. Oates

The most important thermodynamic property from a technological view-

point is the relative hydrogen potential when MH, is in equilibrium with MHb,
since it is this which is related to the plateau pressure:

A/~H2(plat)= R TIn [PHz(plat)/atm]. (3.14)

A#H~(plat) is related to the integral quantities by:

2
A/x.~(plat) = ~ [ A G n ( b ) - AGm(a)] (3.~5)

and, for the related partial quantity for the metal:

A#~(plat) = ~ [bA Gin(a) - aA G,,(b)] (3.16)

whilst the integral plateau property at composition r is:

1
AGm(plat, r ) - [(r-a)AGm(b)+(b-r)AG,,(a)]. (3.17)
(b-a)

As may be seen from Fig. 3.1b, in the case of the chemical potential, the plateau
property is the same as the partial molar hydrogen quantity in the single phases
which are in equilibrium, i.e.,

Ap.2(plat) = A p.2( a) = A ix.~( b ) . (3.18)

However, this is not the case for the plateau entropy and enthalpy properties (cf.
[3.8, 9]), i.e.,

ASa2(plat ) ¢ ASH2(a) ~ AS.2(b ) (3.19)

AHH~(plat) 4=A H.~(a) 4=A Hu:(b ) . (3.20)

This distinction may be seen by comparing Fig. 3.1b with Figs. 3.2b and 3.3b.
Frequently, the experimentally determined values of AS.2(plat ) and
AHH~(plat) remain approximately constant over fairly wide ranges of tempera-
ture, even though substantial changes in a and b may occur over this
temperature range. Rudman and Sandrock [3.2] suggested that S.2(plat ) is small
compared with S~2 and, since the latter varies strongly with temperature,
ASH2(plat) should also vary with temperature. They conclude that any
experimental observation to the contrary reflects inaccuracies in the measure-
ments. This is not necessarily so, however, as can be illustrated for the case of a
regular interstitial solution [3.6] where a and b vary markedly with tempera-
ture. In this case it is readily shown that AS,~(plat) is given by:

AS~(plat) = 2AS~ = 2S~ - S,~ 2 . (3.21)

 Thermodynamics of Intermetallic Compound-Hydrogen Systems 55

I I I ! I Fig. 3.4. Plot of SH2/Ragainst temper-

20 ature for the gas phase (,_.~JR), the
GAS -.----'-
solid phase (2S~n/R)and the difference
15 between these two quantities
GAS-SOLID

5
SOLID
o e.=,OOOK
20( 400 600 800
T/K

If the principal contribution to S~ is assumed to be vibrational and this is

approximated by the entropy for an Einstein oscillator then, as shown in
Fig. 3.4, we can understand how ASR2(plat) can remain approximately constant
over a quite wide temperature range. Similar arguments apply to the
approximate constancy of AHn2(plat) with temperature.
An alternative way to approach the plateau equilibrium is to consider the
reaction

2 2
b - a M H a + H 2 - b - a MHb (3.22)

and use the equilibrium condition, ~ ni# i = 0 so that

2
b - a (~MH~-- #Mn°)- #.2 = O. (3.23)

For the same reaction under standard conditions we have

2
b -----a(l~mib -- I-t~r~,)-- Iz~2 = A G~(a~b) (3.24)

so that, by subtraction

A G"(a~b) = A#n2(plat ) = R T in [pn~(plat)/atm] (3.25)

since M H , and M H b are in their standard states.

It should be noted that in (3.25) we are dealing with the standard free energy
change for (3.22) and not free energy change, which is, of course, zero.
The relations between the plateau properties and the formation properties
are important and, for a system with a single plateau, are readily obtained by
56 7:. B. Flanagan and W. A. Oates

C ..-1,.
Fig. 3.5. Log[pHJatm.] against c
5(
I
012
I I
0.4
I
0,6 0.8 1.0 where c = nN',/(nL,+ nNi) for the
' ' I' LN~Hy La-Ni-H system under conditions
of para-equilibrium at 373 K from
0 L3NHy Oates and Flanagan I-3.10]. For
-I- clarity of presentation not all of the
LaH x
E L NHz
phase fields have been indicated.
-5 +
The small filled circles indicate the
L3NHy hydrogen pressures where the vari-
LN 5
-10 + ous hydride phases form under PE
o~ L3NHy
Ni conditions. L = La and N = Ni in the
0 + formulas for the various intermetal-
LN
-15 lic compounds
LaHx+L3N L3N
-20 +
LN
La + L3N
-25
La L3N LN LN2 LN3.5 Ni
LN~.4 LN3LN 5

considering the reaction

M(s) + y/2 H2(g) = MHr(s) (3.26)

for which it can be shown using (3.6) and (3.7) that

A Gy~MHy) = y/2 A #n2(plat) + A/~M(a) - A #Mn~(b). (3.27)

As can be seen from Fig. 3.1a both the second terms are fairly small when a ~ 0
and b ~ y , and they also offset one another because of their different signs in
(3.27). Consequently, a good approximation, especially at low temperatures, is:

A GT(MHr),~ y A#n~(plat), (3.28)

so that Aktn2(plat) m a y usually be identified with the stability of the hydride. In

some I M C H s multiple plateaux are found. This gives rise to no significant
differences in the thermodynamic treatment from that discussed above.
Another feature of interest is the hydriding behavior of different I M C s
belonging to the same binary alloy system. The type of isothermal represen-
tation shown in Fig. 3.5 for PE in the L a - N i system is useful in that it permits
the isothermal behavior of the whole system to be taken in at a glance [3.10].

3.2.2 Three Condensed Phase Equilibria

Three condensed phases m a y co-exist in an I M C H if the system is in CE. The

I M C A 1 _cBc dissolves H to the composition A 1 - cBcHa which then breaks down
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 57

into AH x and another IMC with H in solution, AI -c,Bc,Ha,.The standard state

reaction equation for CE is

2c
yAc2C~'A, _~B~(s)+ H2(g) = AHr(s) + ~ A, _ c,B~,(s), (3.29)

where Ac=c'-c and c'>c.

The standard free energy for this reaction, A G~, can be obtained from the
standard free energies of formation of the the three solids in (3.29). However, in
order to calculate A/~n: for the three condensed phase equilibrium it is necessary
to know A#AH,, APA,-cBc,and A#A~_o,no,as a function of H concentration, so
that they can be obtained at AHr, AI-cBcH,, and Ax_c,B~,H,,, respectively.
Since, in general, this information is not known, it is not possible to calculate
A#n2 accurately. Nevertheless, as is discussed in Sect. 3.5.4, calculations based
on taking A#H2=AG~ can be useful in predicting the PE behavior to be
expected in the same system.

3.3 Experimental Aspects

Some experimental difficulties are encountered in the determination of the
thermodynamics of IMCHs which are not found for pure metal-H systems. For
example, deviations in the stoichiometry of the starting material can cause
differences in the dilute phase solubilities and the observed plateau pressures.
The preparation of IMCs and their hydrides is described in Chap. 2 and should
be referred to. Only the special problems related to the determination of the
thermodynamic properties of IMCHs will be stressed in this section.

3.3.1 p-r-T Measurements

In single phase regions/I~H 2in the solid can be determined from the equilibrium
hydrogen pressures as a function of r through the use of (3.6). AHH~and ASh2
can be obtained from the temperature dependence of A#H2 and r can be
measured volumetrically, gravimetrically or by monitoring the change in a
physical property whose variation with r is known. Errors in r can arise if the
starting IMC contains residual hydrogen following the attempted removal of
trapped hydrogen at the temperatures where the isotherms are being
determined.
Results from p-r-T measurements are usually presented as a series of
isotherms. Phase diagrams can be deduced from these p-r-T results, but the
precise locations of the phase boundaries are often difficult to determine
because of the hysteresis problem (see Sect. 3.3.3a) and because of the tendency
for the isotherms to change very gradually in the regions of these phase
transitions.
58 T. B. Flanagan and W. A. Oates

3.3.2 Calorimetric Measurements

a) Reaction Calorimetry
The heat evolved or absorbed on the addition or removal of a given amount of
hydrogen, nH~, to an IMC can be measured calorimetrically. An advantage of
IMCHs for such measurements is that the kinetics are generally fast, thereby
increasing the accuracy of the calorimetry. If nH~is small enough, the measured
heats approach AHn2 in the single phase regions, so that either AHH: or AH m
can be measured. It should be pointed out that, with this technique, the largest
errors often arise from the determination of the Ann: which causes the heat
absorption or evolution. This is particularly the case for measurements of
AHn2 in those regions of an isotherm where the pressures are relatively large
and change rapidly with r.
Different types of calorimeters have been used for studies on IMCHs.
Lebsanft and Wenzl [3.11] employed a heat flow calorimeter of the Tian-Calvet
type in their investigations of the Fe-Ti-H system, whilst Murray et al. [3.12]
used a twin cell calorimeter of similar type. Bowerman et al. [3.13] used a single
cell, quasi-isothermal calorimeter manufactured by the LKB Instrument
Corporation for high pressure applications. Hubbard et al. [3.14] used a single
cell Rossini-type iso-peribolic flow calorimeter. Murray et al. [3.12], Bowerman
et al. [3.13] and Hubbard et al. [3.14] all investigated the LaNis-H system and
in addition Murray and his group have examined the CaNis-H and MgNi2-H
systems [3.15, 16].
The heat correction due to work, (pV), in these calorimeters has been
discussed by Nace and Aston [3.17] and more recently by Boureau and Kleppa
[3.18]. This correction, Vc,~Ap,where Ap = py-Pi, has been employed by most
workers in this field except for Murray et al. [3.12], who argue that no
correction is needed for the twin-cell calorimeter. The correction is negligible in
the plateau regions where there is zero pressure change but it is important in
single phase regions. The need for the correction should be re-examined in the
light of the work of Hill [3.19], who considered the analogous problem for
adsorption and concluded that it is not possible to make a correction for the
work term except in the two extreme cases where the gas is added reversibly or
completely irreversibly.

b) Thermal Analysis
Differential thermal analysis has been employed by Shilov et al. [3.20] to
determine AHn2(plat) from the temperature at which the decomposition of the
hydride occurs. Alternatively, AHH2(plat) values can be determined from the
area under the DTA curves [3.21]. These methods are possibly useful for rapid
surveys of systems but in view of the relative ease with which values of
AHn2(plat) can be determined from van't Hoff plots, these methods seem to be
of limited utility.
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 59

e) Heat Capacity
Only a few heat capacity measurements have been made on IMHs. Ohlendorf
and Flotow [3.22] determined the heat capacities of LaNi 5, LaNisHo.36, and
LaNisH6.39 from 5 to 300 K. Activated LaNi 5 was used in the preparation of
the hydrides and it is not known ff this activation affected the results due, for
example, to strain removal during heating. Standard entropies at 298 K were
evaluated from the results and the vibrational and electronic contributions
were also evaluated from the low-temperature heat capacities.
Wenzland Pietz I-3.23] measured the heat capacities of Feo. 5Ti0. s H, samples
with r =0.31, 0.47, and 0.66 at 280 K. From their measured heat capacities they
calculated the heat capacities due to H and, from these, the values of the
Einstein temperatures. It was observed that the heat capacities due to H
increased (Einstein temperatures decreased) rather markedly with r.

3.3.3 Problems in Obtaining Accurate Thermodynamic Data

a) Hysteresis
Hysteresis, which is almost always observed in metal-hydrogen systems, leads
to the results that (i) pf > Pd, and (ii) af > a d, bf > bd, where the subscripts f and d
refer to formation and decomposition, respectively.
There are thus two sets of "thermodynamic" parameters which correspond
to the reactions:

2 MHaf 2 MHb~ (3.30)

bf-af + H 2 - ~br-a
'

2 MH,d 2 MHbd
(3.31)
-ha-- ad + H2 - bo - ad

where af > a d and br > ba.

For example, two AHn2(plat) are obtainable from:

- R¢3lnpf
O(1/T) =AHH~(plat, f) and (3.32)

- Ra In Pd
O(I/T) AnH2(plat, d). (3.33)

Hysteresis can be expressed quantitatively by the free energy difference,

AGrh(hyst) = RTln(pf/pd). This is small relative to the enthalpy changes for the
above reactions so that only small differences are found between AHn2(plat)
determined from van't Hoffplots ofpf and Pd. It has been argued [-3.24, 25] that
calorimetrically measured enthalpies will not be influenced by (i) but will be
60 T. B. Flanagan and W. A. Oates

affected by (ii) and that entropies determined from the temperature dependence
of Pr or Pd will not be affected by (i) but will also be affected by (ii).
It has also been argued by some [3.26, 27] that pa closely represents the
equilibrium plateau pressure, whereas Flanagan and co-workers [3.28, 29]
have presented evidence that the equilibrium pressure lies somewhere between
pf and Pd (see Sect. 3.5.4 for more details). In the case of the P d - H system, where
the data are probably more accurate than for any other metal-hydrogen system,
the equilibrium plateau pressures, calculated from the single phase thermody-
namic data, seem to lie nearly half-way between pf and Pd [3.30], thus favoring
the Flanagan model.

b) Characterization of the Starting I M C - Activation and Annealing

The IMCs employed in thermodynamic investigations of the IMCHs are
generally not those resulting from the initial preparation and annealing
treatment. They are usually first activated by subjecting the IMC to a series of
hydriding-dehydriding cycles and, although it is still not entirely clear from a
fundamental viewpoint what occurs during this treatment, from a practical
viewpoint the material is mechanically disintegrated into a finely divided
material which reacts very rapidly with hydrogen. Different IMCs are affected
differently by activation. In the case of LaNis, x-ray examination of the
dehydrided, unannealed IMC [3.31] reveals that there is considerable line
broadening due to strain which does not appear in the unactivated powder.
This broadening is greatest in the (hhO) and (0k0) directions and is zero in the
(00/) direction. It has been shown [3.32] that energy is released upon heating
activated, dehydrided LaNis. This is presumably due to the release of the strain
energy. The strain energy introduced during activation may also be responsible
for the effect of activation on the plateau pressure pf (see below).
Hempelmann et al. [3.33] have recently shown from inelastic neutron
scattering spectra that activated, annealed LaNi5 exhibits well-defined discrete
localized modes in the dilute phase whereas the unannealed, activated material
does not. They supposed that the defects in the activated, unannealed material
lead to a nearly continuous distribution of frequencies. The tendency for LaNi5
hydride to decompose upon repeated activation cycles has recently been
discussed by Goodell [3.34], who suggests that repeated cycling leads to
disordering of the LaNi5 structure. This was also suggested earlier by Flanagan
et al. [3.35] on the basis of the non-ideal solubility behavior in the dilute phase
of activated LaNis. With reference to decomposition of LaNis during cycling,
Sect. 3.5.4 should also be consulted and Chap. 6 of volume II of this work and a
recent contribution of Busehow [3.36].
It is the behavior of the defect-free material which is important for a detailed
theoretical understanding of the system, whilst the behavior of the activated
material is important for an understanding of processes occurring under
technological conditions. Other than for the FeTi-H system [3.37], there
appear to be no unequivocal data available for the dilute metal-rich region for
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 61

systems which show large differences in thermodynamic properties between

annealed and activated material. Results for the LaNis-H system [-3.38, 39] are
conflicting.
It is also possible to study the thermodynamic properties of a material which
has been annealed after activation, i.e., when the IMC is in a finely divided state.
The annealing of activated, H-free LaNi 5 markedly reduces the dilute phase
solubility in LaNi5 and leads to a transition from non-ideal behavior,
characteristic of trapping and other factors introduced by activation, to nearly
ideal dilute-phase behavior [-3.38, 39]. Flanagan and Biehl [3.40] also found
that the plateau pressure p~ was markedly increased by this annealing treatment
which commences at relatively low temperatures (< 400 K) suggesting that the
effect on Pr may arise from micro-strain. There was only a small effect of
annealing on Pd which is not an unreasonable observation because the
formation of the hydride phase results in activation and decomposition and
therefore always reflects the behavior of the activated IMC.
It is clear that this annealing effect is important because it can introduce
variations in plateau pressures pf, even for IMCs which are otherwise identical.
Rapid hydriding and the accompanying self-heating might introduce
localized annealing which could lead to anomalous plateau behavior.
The generality of the annealing-after-activation effect has not been
established. It has been found to be small in L a C o s - H [3.41] and non-existent
in ErFe2-H [3.39]. The effect of annealing on activated TiFe appears rather
different from that of the ANi5 compounds (see below). It would seem that it will
be a factor in those IMCHs where the effects of activation appear to be
profound, e.g., when it leads to mechanical disintegration as in the ANis-H
IMCHs, and not so important in those IMCs where activation does not so
markedly disrupt the sample, e.g., E r F % - H [3.39].
Besides the influence of activation on the measured thermodynamic
properties, structural disorder, small variations from the desired IMC
stoichiometry or second phases can also be unwanted sources of complicating
behavior in studies of IMCH thermodynamics.

c) Sloping Plateaux
Sloping two-solid-phase tie lines or "plateaux" are frequently observed in
IMCHs. The phenomenon seems to be most pronounced in the case of ternary
IMCs and in non-stochiometric binary IMCs. Sloping plateaux can result from
inhomogeneities in the IMC [3.42] but may have a more fundamental origin
[3.43]. lvey and Northrup [3.44] have shown from x-ray diffraction that the
hydrogen content of a homogeneous hydride phase increases with hydrogen
content along the sloping plateau of the Zr(FexCrx_x)2-H system. This
observation seems to rule out inhomogeneities in the starting material as the
Source of the sloping plateaux because otherwise the x-ray reflections of the
hydride phase should increase in width with H content due to the distribution
of hydrogen concentrations in the hydride phase of the inhomogeneous sample.
62 T. B. Flanagan and W. A. Oates

Whatever its cause, it must be allowed for when AHH2(plat) values are obtained
from van't Hoffplots ofpf or Pa, i.e., pf and Pd should be evaluated at the same r
value at various temperatures.
It should be mentioned that when sloping plateaux are present, the slope for
pf appears greater than that for Pd. However, if A#H2, i.e. logp, instead of p or
p~Z2, is plotted against r, the apparent differences in slope for hydride
formation and decomposition become smaller or disappear entirely.

d) Decomposition
The thermodynamics of three condensed phase equilibria has been discussed in
Sect. 3.2.2. Such a three phase equilibrium will arise if the IMC breaks down in a
hydriding situation to a component metal hydride and a further IMC. This
complete equilibrium CE is possible only when there is sufficient atomic
mobility of the metallic components. This will always occur at high tempera-
tures but CE in IMCHs seems to be able to occur quite often at temperatures
well below those which might be expected from a consideration of the
properties of the pure IMC. It seems that mobility is induced by the hydriding-
dehydriding pseudo-binary phase transformation and this influences the ability
of the system to switch over from PE to CE at unexpectedly low temperatures as
compared to those where significant diffusion of the metal atoms is expected to
OCCUr.
As long as the temperatures are low and the hydrogen gas used is extremely
pure, decomposition of most IMCHs should not be a serious problem F3.34],
especially since thermodynamic investigations on IMCHs are generally carried
out with IMCs which have been cycled only a few times. The decomposition of
IMCHs is discussed in Sect. 3.5.4 from a theoretical viewpoint and from a
practical viewpoint in Chap. 5 of Volume II of this work.

e) Amorphization
It has been discovered recently that amorphization of some IMCs can occur
during hydriding [3.45, 46]. This represents intermediate behavior between CE
and PE and, unless its role is known, its presence can lead to erroneous
interpretation of thermodynamic data for hydriding. It appears that the
amorphous hydride which is formed must be quite stable because upon removal
of the hydrogen the sample crystallizes, i.e. the hydrogen stabilizes the
amorphous structure. A o k i et al. I-3.47, 48] have found that hydrogen-induced
amorphization occurs in most of the RNi2 IMCs where R is a rare earth. The
detailed mechanism of the amorphization process is not understood as yet.
This seems the appropriate place to mention that the hydrogen absorption
characteristics of amorphous (glassy) and crystalline IMCs have been com-
pared, e.g. for the case of TiCu 1-3.49]. No evidence has been found for a plateau
pressure region in the isotherms of the amorphous materials. It is not known if
there is a plateau region when hydrogen is gradually added to an IMC which
 Thermodynamicsof Intermetallic Compound-HydrogenSystems 63

undergoes the amorphization reaction. Models for the absorption of hydrogen

by amorphous materials have been proposed by Kirchheim [3.50] and Griessen
[3.51].

3.4 Results for Representative Systems

It is not possible to review all of the available thermodynamic data for IMCHs.
Most of these data have been derived from plateau pressures and their
temperature dependence. Griessen and Riesterer, in Chap. 6 of this volume, and
Buschow et al. [3.1] have tabulated many of these results. For a few IMCHs
more detailed thermodynamic data are available for the single phase regions
and from reaction calorimetry in both the single and two phase regions. It is
these results upon which we will concentrate here. In addition we have selected
some IMCHs as representatives of a given type of behavior, e.g., ErFe2-H has
been chosen to illustrate multi-plateau behavior and ZrV2-H to illustrate
continuous solid solution behavior over a wide range of hydrogen contents. (In
this review the plateau thermodynamic properties for hydride formation and
decomposition will be given with reference to the formation reaction, i.e.,
AH(plat) will be negative for both processes - see (3.32) and (3.33).)

3.4.1 L a N i s a n d S o m e L a N i s - B a s e d Systems

Three calorimetric investigations on activated LaNi~-H have given results for

AHH2(plat) which are in substantial agreement (Table 3.1). The available
calorimetric data for activated LaNi 5 in the dilute phase are in qualititive
agreement only, i.e., there is agreement that IAHn2r,>rAHH2(plat)l. The
differences in the results of the different investigators I-3.13, 14, 52] reflect the
different annealing and evacuation conditions employed on the samples. The
large values observed for IAHH2I~are consistent with the non-ideal (not obeying
Sicvert's Law) pressure-composition behavior observed in this region. This
non-ideality has been attributed to trapping effects originating from the
activation treatment and, possibly, from some non-stoichiometry in the
starting material I-3.35, 38]. A discussion of the isotope effects for LaNi 5-H (D)
for the plateau region is given in Sect. 3.5.2.

Table 3.1. Results for AHn~(plat)for activated LaNis-H

dHn~(plat,I) AHii2(plat,d) Ref.
[kJ mol-1 H2] [kJ mol-1 H2]
- 29.4+ 0.8 - 29.8+ 1.2 [3.13]
- 32.3_+0.06 -- 31.8+ O.10 [3.52]
- 31.6 + 0.2 -- 31.4 + 0.2 [3.14]
64 T. B. Flanagan and W. A. Oates

No calorimetric studies have been carried out in the dilute phase using well-
annealed LaNi5 samples. One p-r-T study in this region [3.39] gave
AH~n2= - 10.8 kJ per mole H2, but the results on which this value was based did
not intersect the origin when plotted as pl/2 versus r, i.e., there appears to be
some trapping effect present, although the authors [3.39] argued that this did
not affect their thermodynamic results because the traps were saturated. In
another p-r-T study [3.38] at higher temperatures and smaller H con-
centrations, a value of AH~2 = - 32.0 kJ per mole H2 was obtained. Although
these results apparently obeyed Sievert's Law, it is possible that the measure-
ments still contained a contribution due to traps since the agreement with
Sieverts' law was only over very small hydrogen contents.
Ohlendorf and Flotow [3.22, 52] evaluated the entropies of LaNi5Ho.36 and
LaNisH6.39 from measurements of the heat capacities between 5 and 300 K.
ASH~(plat) calculated from the equation
2
ASH2(plat) = ~ - a [S(b)- S(a)] - ~ (3.34)

was found to be in good agreement with ASH~(plat) obtained from plateau

pressure measurements.
The thermodynamic properties of LaNi5 can be altered by the partial substi-
tutional replacement of either the La or Ni. Investigations have concentrated
upon the variation of the H capacity and the plateau properties as a function of
the type and amount of the substituted atom. Reviews of these results have been
given elsewhere [3.53 55]. The only calorimetric measurements are those of
Hubbard et al. [3.14], who investigated La(Nig.sAIo.5)-H and La(Ni4A1)-H.
They found that AHH~(plat, f) increased in exothermicity upon substitution of
A1 for Ni: from - 38 to - 4 6 kJ per mole H2 for La(Ni4.sAlo.5) and La(Ni4A1),
respectively. The former value agrees well with the result of Mendelsohn et al.
[3.55], obtained from decomposition plateau pressure measurements. Gener-
ally, it seems that substitution affects the value of AHH2(plat) more than that of
ASH2(plat).

3.4.2 TiFe

Reilly and I4qswall [3.56] found that two hydride phases form from cubic TiFe:
the B-phase (TiFeH2) and the ~-phase (TiFeH1.9). Table 3.2 summarizes the
results of Reilly and Wiswall [3.56], and Wenzl and Lebsanft [3.11]. The latter
authors obtained their plateau enthalpies calorimetrically for activated TiFe.
The values found by Wenzl and Lebsanft [3.11] for the e/fl plateau are
somewhat smaller in magnitude than those found by Reilly and I4qswall [3.56]
but they confirm that the f l - ~ transformation is more exothermic than the
e - f l transition, which is somewhat unusual for a multi-plateau system.
The results of Reilly and V~swall [3.56] suggest that the fl-phase co-existing
with the e-phase is TiFeH1.4 during hydride formation, and TiFel.0 during
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 65

Table 3.2. Thermodynamic data for the TiFe-H system

AHH2(plat,d) AHn2(plat, 0 ASH2(plat,d) Ref.
[kJ mol - 1 H2] [kJ mol- 1 H2 ] [J K- 1mol- a H2]
~/3 -28.12 - - 106.2 [3.563
]~/?, -33.30 - 103.6 [3.563
~//~ -24.0+0.8 -23.8+1.0 - [3.11]
fl/7 -33.8-1-5.0 -32.0+3.6 - [3.11]

decomposition. Schefer et al. [3.57], whose isotherms for the TiFe-D2 system
resembled those found by Reilly and 144swallfor hydrogen, introduced the
terminology fll for TiFeDa. 0 and f12 for TiFeDI. 4. Schefer et al. reported
different distributions of the D atoms in the fll and the f12 phases within the
same orthorhombic structure and Reidingeret al. [3.58] have characterised the
two fl structures by x-ray diffraction. However, it is possible that hysteresis may
be responsible for the apparent existence of the two fl-phases. It is known for
systems with only one hydride phase that bf > b d. For a system exhibiting two
plateaux, this hysteresis effect could give rise to the appearance of different
phases during hydride formation and decomposition. It is possible to estimate,
from the desorption isotherms of Reilly and 144swallin the pure fl-phase region,
that (O#n2/ar)r ~ 10000 J. If this value is combined with a hysteresis free energy
of 1720 J per mole H 2 [3.11], then a value of Ar=0.17 is obtained, which is the
order of magnitude of the observed difference in compositions of the fl "phases".
In Chap. 4 of this text more details of the structural aspects of TiFe and its
hydrides are given.
Reilly et al. [3.59] found that the 7-phase can be formed directly from the
e-phase if the activated TiFe is annealed at 800°C prior to the absorption
isotherm determination. The plateau pressure Pf(e-7) was found to be
approximately the same as that for pr(e- ~) for unannealed samples. Goodellet
al. [3.60] have found that isothermal cycling of TiFe 148 times leads to greatly
increased plateau pressures for t h e / / - 7 transformation.
Because Wenzl and Lebsanft [3.11] employed an activated sample, their
results for AHH2in the metal-rich phase were found to be large in magnitude,
reflecting trapping at defect sites introduced during activation. In contrast,
Welter et al. [3.37] found that AHH2was endothermic ( + 21.6 kJ per mole H2)
for the annealed material.
Zhirnova and Mogutnov [3.61] investigated the metal-rich phase by p-r-T
methods in TiFe, presumably unactivated, at elevated temperatures (673 to
1273 K). Their results, described by the equation

- AHn2 = 4452 + 2 . 1 0 T - 6.84 x 10- 3T 2 (3.35)

R
give a value of AHn2= -- 37.2 kJ per mole H2 at 298 K but yield positive values
above 975 K. Their results are therefore intermediate between the values with
66 T. B. Flanagan and W. A. Oates

no trapping [3.37] and those with trapping [3.11], although the extrapolated
value at 298 K seems unlikely, in that it is more exothermic than AHn~(plat).

3.4.3 ZrVz and Some ZrMnz-Based Systems

ZrV2, a Laves phase with the C15 structure, does not form a separate hydride
phase at temperatures at or above 298 K at the H2 pressures normally
encountered. It does, however, dissolve large amounts of H in solid solution.
The most complete thermodynamic data for the system are those of Pebler and
Gulbransen [3.62], although it should be noted that these results may have been
affected by the presence of a second phase in the starting material. Their results
for AHH~ and ASn~ are shown in Fig. 3.6, where a minimum in ASH2 at r -~ 0.45 is
seen to be present. The stoichiometry of ZrMn2, a Laves phase with the C14
structure, can be altered from ZrMnl.s to ZrMn3.s [3.63] with concomitant
changes in plateau pressures and H capacity. Sinha and Wallace [3.64] have
shown that Mn replaces Zr in the hyper-stoiehiometric ZrMn2 + x. Both Zr and
Mn can also be partially replaced by other elements [3.65-70].
The ZrMn 2 +:,-H system exhibits only one plateau. The interesting feature,
however, is that the "plateau" is sloping, with the slope increasing with
increasing x [3.71]. Thermodynamicdata, obtained by both p-r-T and
calorimetric methods, exist only for this plateau region [3.63, 72, 73]. The
values reported for ZrMn2.a at r = | are given in Table 3.3. The discrepancy
between these values was attributed [3.72, 73] to irreversibility effects, which
give rise to uncertainties in the values obtained from van't Holt plots in the two-
phase region. However, provided attention is paid to aliquot size and the effect
of hysteresis, irreversibility should not give rise to such discrepancies. A re-
determination [3.74] of the plateau properties for the ZrMn2 +x-H systems for
x=0.2 to 1.5, using both p-r-T and calorimetric methods has shown no

-100 -40
e,i £
I 'T
"T__ 0
0
E E
-%
-..)
<--- 0
'~
-r
-120 o o -60 £
U) I
<

Fig. 3.6. Plots ofdHn2 and ASh2 against

-140 -8O hydrogen content for the ZrV2-H sy-
I [ t I stem [3.62]
(3 (3.2 0.4 0.6 0.8 1.0
H/M
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 67

Table 3.3. Thermodynamic data for ZrMn2.8 at r = 1

Method and Ref. AHn2(plat, d) ASr~2(plat, d)

rkJ mo] - 1 He ] [J K - 1 m o l - 1 H2 ]

p-r-T [3.63] - 17.8 --50.6

Cal. [3.72] + p-r-T - 29.9 (cal) - 91.3 (cal + p-r-T)

significant difference in the AH~2(plat ) values obtained by the two methods. It

seems, therefore, that the values obtained by p-r-T, which are given in Table 3.3
may be questioned for reasons other than lack of reversibility.
As pointed out by Wallace et al. [3.63], discrepancies also exist between
results obtained by different investigators for the plateau pressures themselves
on ostensibly the same ZrMn2+x-H system. For example, van Essen and
Buschow [3.71] report Pd =0.1 atm at 323 K for ZrMn2, whereas Shaltiel et al.
[3.75] report 0.007 atm. Majorowski and Flanagan [3.74] find a value closer to
that of the former investigation. There is a corresponding discrepancy between
the results of Wallace et al. [3.63] and van Essen and Buschow [3.71] for the
ZrMnz.s-H system. These discrepancies do not appear to be caused by the
effects of annealing. Whilst van Essen and Buschow did find a marked effect of
annealing the IMC at 1050°C on the subsequent isotherms, the effects on the
plateau pressure are not as big as the differences between different
investigations.

3.4.4 ErFe2

Although ErFe2, like ZrVz, possesses a C15 Laves phase structure, its behavior
towards H differs markedly. ErFe2-H forms five different hydride phases, as
illustrated by the results of Kierstead [3.76] in Fig. 3.7. No structural

10 4

1o3
~~ 10 2
E
E 101

a
/ Fig. 3.7. Plot of the equilibrium hydro-
gen pressures against hydrogen content
for the ErFe2-H system at 293 K [3.76].
The five plateau regions are indicated
by the roman numerals
1%- I 1
4.0
I

H / ErFe2
68 T. B. Flanagan and W. A. Oates

transformations are involved for the first four hydrides which form but, when
the hydride corresponding to ErFe2H4 forms, there is a rhombohedral
distortion of the unit cell. Kierstead has determined the thermodynamic
properties from p-r-T measurements on this system in the temperature range
273 to 353 K [3.76]. The values of AHH:(plat) for the first four plateaux are all of
a similar magnitude, - 47.9 +_3.0 kJ per mole H2, so that the different A#n2(plat)
values for these four plateaux are determined principally by the different
ASH:(plat) values.

3.4.5 Mg-Based IMCs

These IMCs constitute a rather special group from the viewpoint of their
potential for hydrogen storage due to their favorable ratios of weight of
hydrogen to weight of metal in the IMCHs. Another unusual aspect of some of
these Mg-based IMCHs is that the IMCH can be more stable than its
decomposition products. Both MgH2 itself and the Mg-based IMCHs are
rather stable and whether decomposition of the IMCH to MgH 2 and another
IMC or to a pure metal component occurs or not depends upon small
differences in the stabilities. The similar stabilities of the hydride phases can be
appreciated from the values of AHH2(plat) for MgH 2 and MgzNiH4 which are
- 70.0 kJ per mole H 2 [3.77] and -- 64.4 kJ per mole H 2 [3.781, respectively.
ReilIy [3.79] has shown that at 562 K

Mg2Cu(s)+2H2(g)~Mg2CuH4; AG~(plat)> 1.7 kJ per mole H 2

2Mg2Cu (s) + 3H2(g) ~MgCu2(s) + 3MgH2(s),
AG~(plat) = 1.7 kJ per mole H2

whereas

Mg2Ni(s) + 2H2(g)--*MgzNiH4(s); AG#(plat) = 1.0 kJ per mole H2

2Mg2Ni(s) + 3H2(g)--* MgNi2 + 3MgHe(s);
AG'~(plat)=2.1 kJ per mole Hz.

It can be seen that the IMCH is the thermodynamicallyfavored state for Mg2Ni
but not for Mg2Cu. Mg2Ala has also been shown to decompose upon hydriding
to MgH2 and A1 [-3.80].
Didisheim et al. [,-3.81] have recently prepared Mg2FeH 6, which is a very
hydrogen-to-metal rich IMCH, from the reaction of Mg/Fe mixtures with
hydrogen at 773 K and 60 bar. Unfortunately this IMCH does not seem well
suited to hydrogen storage because of its high stability which may result from
the presence of [FeH6]-4 ions in the structure as shown by the authors from
physical measurements and neutron diffraction.
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 69

3.5 Microscopic Thermodynamics

3.5.1 Configurational Models

Configurational models attempt to give an interpretation of the composition-

dependent thermodynamic behavior of solutions in terms of simple interactions
between the component atoms. The other contributions to the partition
function are considered independently. Since exact calculations of the configur-
ational partition function are probably impossible in a complex systems such as
IMCHs, gross simplifying assumptions must be made in the calculations. As
will be discussed below, these assumptions may result in physically unrealistic
models of IMCHs.

a) Site and Interaction Energies

In strictly pairwise model calculations on ternary metal-hydrogen solutions it is
necessary to know the metal-hydrogen and H - H interaction potentials. A
modified approach, which has been used by most workers in treating alloy-H
systems, is to use the local environment model [3.82, 83], in which the various
possible sites for H absorption are allocated "site energies", Fo(j), where j
denotes a particular type of interstitial site. In its simplest form, the local
environment model assumes that the number of different kinds of metal atoms
which constitute the first co-ordination shell surrounding the interstitial site
determine the value of Fo(j). Then ifj represents a tetrahedral site with two A
atoms and two B atoms in the first co-ordination shell, the site energy is
I/0([A2B2] ). The reference state for the various Vo(j) is most usefully taken as
Hz(g, 1 atm).
In applying the model to finite H concentrations the H - H interaction can be
considered pairwise with interaction energies V~, where i is the ith nearest
neighbor. V~is taken as zero at infinite separation. In principle, it should be
possible to predict by a priori methods the preferred sites for H occupation and
the values of Vo(j)and V~required in the model calculations. A few attempts have
been made along these lines and some useful results have emerged. Switendick
[3.84], for example, has shown that V~--*+ oe when the H - H separation is
< 2.1 A, i.e., there is a hard core repulsion inside this radius. This calculation is
in agreement with experimental observations.
For the most part, however, information on Vo(j) and V/has come from a
knowledge of the preferred sites for occupation by H, the H capacity and the
thermodynamic properties, although, with this more empirical approach,
difficulties can arise in distinguishing between the relative importance of Vo(j)
and V~in contributing to the observations. Multi-plateaux, for example, can be
explained by assuming either equal or different Vo(j)for the different sub-lattices
(see below).
Westlake [3.85] has used a completely geometrical method to interpret
IMH structures. He postulates that there is a minimum hole size for H
70 T B. Flanagan and W. A. Oates

occupation (r h > 0.33/~) in pure metals or IMCs, although hydride formation

does not occur until rh > 0.40 A. Because of the lattice expansion accompanying
H absorption, sites initially too small may become available for H occupation
at higher H concentrations. Although Westlake does not make the claim, it
might be expected, on the basis of this postulate, that the larger the hole size the
lower V0(j), so that, in compounds where the A atom is larger than the B atom,
Vo([A2B2])<Vo([AB3])<Vo([B4] ). A further postulate of Westlake is that
V ~ ~ at separations of less than 2.1 h and is zero outside this hard core radius.
In the case of the [AzB2] sites in some IMCs possessing the C15 structure this
hard core radius may extend out as far as the 6th nearest neighbor sites. Using
these assumptions concerning Vo(]) and Vi, Westlake has been able to interpret
many observed structural features of IMCHs [3.86, 87]. Yartys et al. [3.88]
have given a rather similar interpretation of IMH structures to that of Westlake.
A slightly different view, however, has been presented by Somenkov and Irodova
[3.89] for the C15 Laves phases who interpret the observed structures as
implying that, when the lattice size is small, the [A2B2] sites are preferred but, in
large lattices, Vo(]) for the [AB3] becomes lower than that for the [A2B2] sites.
The crossover occurs at a lattice parameter of --7.9 A in these compounds.
Shinar et al. [3.90] have attempted to estimate the actual magnitudes of the
Vo(j) by using an "imaginary binary compound" concept. It is assumed that the
fraction ofa H atom which is bonded to an A atom is determined by the fraction
of A atoms in the first shell of metal atoms surrounding the H atom. Thus, for
example, the site energy per H atom for [A2B2] sites containing n H atoms in
the compound AB 2 is given by the sum of the enthalpies of formation of AH,/2
and B2H,/2 divided by n. Using this procedure the Vo(]) values vary with H
concentration. The enthalpies of formation of the imaginary binary compounds
are obtained using Miedema's method (see Sect. 3.5.3a), although Westlake
[3.85] has noted that this does not appear to have been applied correctly by
Jacob and Shaltiel [3.91]. Buschow et al. [3.1] have also pointed out that any
estimation of the V0(]) should somehow take into account the stability of the
IMC as well as the binary hydrides. In the case ofa Nb-Pd IMC, for example,
large negative values of the Vo(j)would be predicted from the imaginary binary
hydrides and thereby significant IMH stability, whereas the reverse stability
rule would predict, correctly, high Vo(j) values and hence an unstable hydride.
Whilst the approximate range of the hard core part of V~is known, there is
little information available on the longer range behavior of the H - H
interaction. In IMCHs where phase separation occurs, thermodynamics
requires that there must be a lowering of AHH2 with increasing r. In the pairwise
interpretation this means that there must be an effective H - H attractive
interaction outside the hard core radius. However, it is possible that, as in the
case of pure metals [3.92], an indirect interaction is primarily responsible, i.e.,
the lowering of AHn2 comes from the lattice expansion on H absorption, which
could be interpreted as implying that the V0(j) values decrease with r.
In principle, once Vo(j)and Viiare known they can be used in the calculation
of the configurational thermodynamic properties. From the foregoing it seems
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 71

that, at present, the information on Vo(j) and V~ is more qualitative than

quantitative so that they are used as fitting parameters in the thermodynamic
models. The models which have been used in attempting to interpret the
experimental results on I M C H s and which are discussed in the next few sections
are of varying degrees of sophistication.

b) Hard Core, Constant Vo(j) Models

The simplest model is to assume that all the Voq) are equal and that V~is zero
outside a cut-offradius but infinite inside that radius. The hard core assumption
means that the H - H attractive interaction, required for phase separation, is
assumed to be indirect. Only the configurational entropy has to be calculated.
The calculation of the hard core configurational entropy associated with H
occupation on a set of sites of the same energy is straightforward only if there is
no blocking, i.e., only when the hard core radius does not extend beyond the
nearest neighbor interstitial site. In the case of no site blocking:

--Sn/R=ln(0) (3.36)

-- SM/R = fl ln(l - 0), (3.37)

where fl is the number of interstitial sites per metal atom.

It is worth emphasising that there is an effect on AI% as well as on A/~n~as a
result of this configurational entropy, even though the metal lattice itself has
not undergone any configurational change (cf. Pasturel et al. [3.93]).
When blocking of near neighbor sites occurs, the calculation of the hard
core entropy is more complex. In the case of blocking without overlap the
entropy can be calculated exactly by simply substituting zr/fl for 0 in the above
equations, where z is the number of near neighbor sites blocked. However, a
better calculation of the entropy allows for the overlap of the blocked sites. This
calculation is not straightforward so that Monte Carlo methods must be used
to calculate how z is reduced with increasing H concentration [3.94], although
Boureau [3.95] has given an approximate method for calculating Sn.
In the case of IMCHs, an improved model allows for the presence of the sites
of different environment and V0(]) which form different sub-lattices in the
ordered IMC. The calculation of the configurational entropy is now straight-
forward only if the sub-lattices are assumed to be independent, if the differences
in V0(]) are neglected and if it is assumed that no blocking takes place. With
these assumptions:
-- S m
R - ~. gl [01 In 01+ (1 - 0i)In(1 - 0i)], (3.38)

where 0i is the fraction of type i sites occupied, gi is the degeneracy of the type i
sites and S,, is the integral configurational entropy.
72 T. B. Flanagan and W. A. Oates

Wallace et al. [3.96] used (3.38) in an attempt to give support to one

particular structural model for LaNisH 6 by comparing calculated and
experimental entropies. However, as Achard et al. [3.97] correctly pointed out,
the calculation of the configurational entropy in an I M C H is more complex
than this since it is necessary to allow for the blocking effect within and between
the different sub-lattices. They attempted to do this by using the following
equation for the thermodynamic probability, 12:

¢ ~,Tavail'~ T
(3.39)
i r~i'L z Y i -- t~i]"

where n i is the number of H atoms on type i sites and N~wn is the number of
available sites of this type.
Pasturel et al. [3.86] calculated N~v"il from:

N~v"il= f ~ N i - ~ f q n s, (3.40)
J
where f~ and fo are blocking factors for the type i sites due to the presence of H
atoms on type i a n d j sites, respectively, N~ is the total number of type i sites and
nj is the number of H atoms on t y p e j sites. Although (3.39) and (3.40) are an
improvement over (3.38) they are still inadequate, in that they are correct only
for the case of blocking without overlap and, more importantly, are restricted to
mixing on sub-lattices of the same energy.
As indicated previously, phase separation at low temperatures can be
accounted for in terms of the preceding hard core entropy calculations if the
attractive H - H interaction is assumed indirect. For example, in the simplest
case of small hard core mixing on sites of equal Vo(])the hydrogen potential is
given by:

(3.41)

where W is the interaction coefficient. When W is constant and negative, phase

separation will occur at temperatures < - W/4R. However, when W is constant
the isotherms and phase separation envelope are symmetric about 0 = 1/2. A
linear composition dependence of W permits asymmetry about 0 = 1/2.
In this same approximation the plateau properties are given by [3.6]:

AHn2(plat) = A H ~ + W/2 (3.42)

ASrh(plat) = AS~2. (3.43)

The solvus compositions, a and b, can also be obtained from (3.41) (see [3.6]).
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 73

e) Multiple Plateaux Models

An interesting feature of some IMCHs is the occurrence of multiple plateaux as
the H concentration changes. Even more interesting is that these multiple phase
changes occur sometimes when only the one type of site is being occupied. A
good example of the latter is the ErFe 2 system where, at least at the lower H
levels, only the [AzB21 sites are occupied [3.761.
The simplest way to explain the multi-step isotherms is to assume that the
interstitial sites in the I M C can be considered as independent sub-lattices
possessing different Vo(]) values. The distribution of H atoms between the
different sub-lattices must then be calculated. Jacob et al. [3.981 assumed a
Boltzmann distribution for calculating the distribution of H atoms on the
different sub-lattices at room temperature from a knowledge of Vo(/) values
calculated by the method of Shinar et al. [3.90]. However, the H atoms are
partitioned within and between the different types of sites according to a Fermi
and not a Boltzmann distribution. At the large H concentrations of interest in
IMCHs large discrepancies will exist between distributions calculated from
these two distribution functions.
Since the hydrogen potential must be equal on each of the sub-lattices,
equations similar to (3.41) can be written for each sub-lattice. These equations
incorporate the Fermi distribution of the H atoms.
In the absence of H - H interactions, the configurational AHH~ is given by
[3.99]:

AHn~= 2 ~ Vo(j)p*, where (3.44)

1

f1 {exp [(Vo(j)- ½A#th)/kTl}/{ 1 + exp [(Vo(j ) - ½A#H~)/kT]} 2

(3.45)
P* = Y, {exp [(Vo(j)-½A/~.)/kT]}/{1 + exp E(Vo(j)-}A#.~)/kT]} 2'

where fj is the fraction of type j sites.

In the presence of indirect H - H interactions the term 2WO can be added to
AHH2 obtained from (3.44).
Kierstead [3.1001 has used this model for calculating the multi-step
isotherms in some I M C H s although he did not use (3.44) and (3.45) for the
partial energy. Figure 3.8 illustrates, for the two-sub-lattice case, how A#Hj,
AHHj, and ASn2 vary with H concentration for selected V0(])and W values. The
experimental results of Kierstead for the DyCo3-H system [3.1011 exhibit
features similar to those shown in Fig. 3.8.
Equation (3.41) provides four adjustable parameters for each sub-lattice for
the fitting of experimental results, namely, AH ~, AS ~, W, and fl, the number of
interstitial sites per metal atom (which is needed to relate r to 0). If allowance is
made for a linear temperature dependence of AH", AS~, and IV,,the number of
parameters is increased to seven, and if a linear dependence of W on 0 is also
introduced, the parameter count is increased to eight. Thus if there are m such
74 T. B. Flanagan and IV.. A. Oates

L I I I I I I I I

2oL / I0

I
I

k
t~

:E \ \
<~ \ ~'e~
k--" I~
rr ,q

<q

-5

/ -10
0 0.4 0,8 1,2 . 1,6 2,0

Fig. 3.8. A/ta2,AHH2,and ASh2plotted as a function of hydrogen content for a two sub-lattice
model where Vo(1)=0, Vo(2)/RT=10 and W=0, The latter has been chosen for simplicity;
suitable negative values of W will lead to loops in the A#u2r relationship, i.e., to horizontal
plateaux

sub-lattices then between 4m and 8m parameters become available for fitting to

experimental results. With such a plethora, it is not surprising that the model
can be made to fit experimental results quite well. The fitting may not be unique,
however, as shown by the fact that Kierstead first [3.101] interpreted his
experimental results for DyCo3 using a 4 sub-lattice/6 parameter model but
subsequently [3.102] interpreted these same results using a 3 sub-lattice/8
parameter model. Both involve a total of 24 parameters. As was indicated by
Kierstead himself, the value of this type of approach is more to give a
representation of the data rather than to suggest that such models bear any
close relation with physical reality.
According to the multiple site model discussed above, multi-step isotherms
are explained in terms of the different Vo(])for the different types of interstitial
site. However, in several IMCs the H atoms occupy only one type of site but
multiple plateaux are still observed. ErFe a is a case in point. A model due to
Rees [3.103] attempted to explain how multi-step isotherms could arise from H
absorption on only one type of site. In this model the emphasis is placed on V~
rather than V0(j),i.e., the occupation of one site is envisaged as creating adjacent
sites of higher energy. Subsequent occupation of these new types of site creates a
further set of still higher energy and so on. It is worth noting that the calculation
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 75

of the partition function for this model is difficult and was not done correctly by
Rees (see [3.94]). Kierstead [3.104] has used the Rees model for interpreting the
multi-plateau behavior of some IMCHs as an alternative to the multiple sub-
lattice model. Again, however, the number of adjustable parameters is large.
From the above discussion it seems that multi-step isotherms can be
interpreted as arising principally from either the effect of different Vo(j)or from
the effect of VvIn actual IMCHs it is likely that both factors play a role with the
relative importance varying from system to system. It is possible that the
relative magnitudes of ASH2(plat) for the different plateaux in one I M C H may
be of value in distinguishing between the relative importance of Vo(j) and Vii.
This can be illustrated by considering a two-plateaux system. For the two-sub-
lattice model then ASn2(plat ) ~ AS~I2for each plateau. However, where the steps
in the isotherm are due to a repulsive V~then, taking the simplest assumption of
ideal mixing

2
ASn~(plat) = ~ [ASm(b)- AS~(a)] (3.46)

2R
= A S ~ -- b - ~ [b In b + ( y - b) In (y - b) - a In a

- ( y - a) l n ( y - a)]. (3.47)

The effect of (3.47) is to make ASn2(plat) more negative as the average H

concentration, f=(a+b)/2, increases. These different predictions regarding
ASn2(plat ) from the two models are shown schematically in Fig. 3.9.
As discussed in Sect. 3.4.4, AHn2(plat) is approximately constant for the
different plateaux in the ErFez-H system. However, as can be seen in Fig. 3.10,
ASH2(plat) steadily becomes more negative with the average H concentration. If
y is taken as 5 for the occupation of equivalent [AzBz] sites in the C 15 structure
and AS~2 is taken as - 1 0 4 JK-1 per mole H 2 [3.39] the calculated values of
ASn~(plat) shown in Fig. 3.10 are obtained. Qualitatively, the calculated and
experimental results agree, confirming the expectancy that, where the same
interstices are being occupied, ASn2(plat ) should become more negative in
multi-plateau systems with increasing ?. On the other hand, if different kinds of
interstices are being occupied as ~ increases then, to first order, ASr~(plat) will
remain approximately constant. For the LaCos-H system AStir(plat) increases
slightly in passing from the first plateau to the second. This seems to be
consistent with the isotope effect (see Sect. 3.5.2), which indicates a looser
binding in the second hydride phase and, therefore, a less negative value of
ASn~(plat ). For the RCo3-H systems ( R = D y , Tb, Lu, etc.) with the PuNi 3
structure Kierstead [3.105] finds that dSn2(plat) does not vary from system to
system or for different plateaux in the same system. This again indicates that
different types of sites may predominate for each plateau region, a conclusion
which is consistent with structural information [3.106].
 76 T. B. Flanagan and W. A. Oates

(a) (b)
ii i i
I
I
I
I
I

t-. i
--.., ,, j
E "- I E
cO I
<~ -. I I
~,JoJ I
I
I

I I
I
I r I, 1I
1 I I
0 a bl ae b2 Y O a I bl Y~ a2 b2y2
p ---,..
Fig. 3.9a, b r--~

70 ,,
-

Fig. 3.9a, b. A schematic comparison of

A S m against hydrogen content for two
:£ multi-plateaux models: (a) single sub-
"7
5 lattice (Rees model [3.103]) and (b) two
"T
E -90 o
sub-lattice model. The values of
"e"
ASH2(plat)can be seen to decrease mark-
edly for model (a) but not for model (b),
%-~ see (3.46)
-110
<]
Fig. 3.10. dSn2(plat) plotted against the
average hydrogen contents of the
plateaux for the ErFe2-H system. (o),
-130 experimental data from Kierstead [3.76]
1 I I I and (A), calculated results from (3.47)
0 2 4
F

3.5.2 Non-Configurationai Contributions

Strictly speaking configurational models incorporate any non-configurational
contributions to the partition function in the adjustable site energy values and
assume that they are independent of composition. It is of interest, however, to
have some knowledge of the electronic and, particularly, the vibrational
contributions.
Little is known about the lattice dynamics of IMCHs. Some information is
available from heat capacity measurements [3.22, 23], inelastic neutron
scattering measurements [3.33, 1071 and from measurements of the isotope
solubility ratio. The neutron scattering measurements of Shapiro et al. on the
optic modes in FeTiH~ [-3.107] reveal rather complicated spectra which are
strongly dependent on both temperature and H concentration. Because of the
difficulties in preparing single crystal hydrides, no measurements of the
coherent scattering from IMCs containing deuterium have been made and,
consequently, no theoretical interpretations of the lattice dynamics of I M C H s
have been attempted.
In spite of the obvious limitations, it is possible to calculate the expected
results for the isotope effect at constant H(D) concentration expressed as the
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 77

:f
6.C
"D
q~

e~
4.C
.....
e,i 2.C

C c

--2,1

- 4 .~

-6,1
16oo 1 5'0 0 260o
elk
Fig. 3,11. Plot of R T in [PD2/PH2]for conditions of infinite dilution against On where On = hv/k
for 300 and 500 K from Oates and Flanagan [3.108]

isotope ratio, IR = R T In (PD2/PH2),by using the Einstein oscillator model and by

assuming 0n = l/r200 • Some calculated results are shown in Fig. 3.11 [3.108].
The values of the optical frequencies obtained in this way do not agree exactly
with those obtained from neutron scattering in the case of pure metal-H
systems and reasons for the discrepancy have been suggested [3.108]. Thus the
main value of measurements of the IR is in the observation of trends. It can be
seen from Fig. 3.11 that in the region of room temperature, the IR changes from
positive to negative values as the optic mode frequency increases. The IR can
also provide information on the occupation of octahedral versus tetrahedral
sites. Thus it is known from inelastic neutron scattering measurements on pure
metal-H systems that On_~1200-1800 K for H in tetrahedral sites, whereas
0n~-650-1000 K for H in octahedral sites. As can be seen from Fig. 3.11, these
differences are sufficient to have a significant effect on the IR near to room
temperature.
From the viewpoint of a theoretical understanding, the IRs are best
determined at infinite dilution. Unfortunately, this is not feasible for IMCHs,
because most dilute phase data are affected by trapping of H at defects.
Consequently, the only reliable IR information for IMCHs refers to the plateau
region. Although a full analysis of the isotope effect of the plateau region is quite
complex, generally it can be shown that the IR is somewhat greater for the
plateau region than for the infinitely dilute case. The differences are, however,
not expected to be great and thus similar conclusions regarding the type of site
occupation, etc. can be drawn.
Figure 3.12 shows some results for the IR in different IMCHs. Biris et al.
[3.109] found that the IR for LaNi 5 changed from positive to negative values
78 T. B. Flanagan and W.. A. Oates

4.8
Fig. 3.12. Plot of RTln(pD2/Pn2)against
f t i I temperature for infinitedilution at sel-
ected values of 0n. zx, Lao.4Ceo.6Nis
"T
1,-3.111]; (o), LaNi5 [3.109] and (e),
4. LaNis [-3.111].The experimentalvalues
,/
"5 all refer to the plateau pressures
~ / o /

~2
c ~ ~: /o /

I-"
01 -0. ~

l I , I I I i l
180 220 260 300 340
T/K

near room temperature as shown in Fig. 3.12. Andreyev et al. [3.110] extended
this work to 195 K and, as also shown in Fig. 3.12, their results are consistent
with those of Biris et al. [-3.109].
Dayan and Dariel [3.111] reported what they considered to be an
anomalous isotope effect for the (Lao.4Ce0.6)Ni 5 system. Their data are also
shown in Fig. 3.12, where it can be seen that these results are those to be
expected for tetrahedral site occupation.
There are also a few results for the IR in multi-plateaux systems. Kuijpers
[3.112] found, for LaCo5, that IR = 410 J per mole H 2 and IR = 2000 J per mole
H2 for the first and second plateaux, respectively. These values are both within
the range expected for occupation of tetrahedral interstices and suggest a
looser binding of H to the sites which are being occupied in the second plateau
region, as is also expected. However, it should be noted that, in the CaNi5
system, the opposite behavior has been observed [3.113] in that the IR
decreases as the system progresses to a higher plateau.

3.5.3 Phase Transitions

Until fairly recent times, the phase relations in pure metal-H systems appeared
relatively simple but it is now realised that this is not the case. Even the phase
diagram of the P d - H system, previously thought to be the simplest of all and
well established, is now known to be rather more complex and remains in some
doubt [3.114]. Partial phase diagrams for IMCHs are only just becoming
available [3.89] and much remains to be done in this area. Nevertheless, it is
already clear that they exhibit many of the features observed in the diagrams for
pure metal-H systems. Figure 3.13, for example, shows the phase diagram for
the ZrCr2-D system. It can be seen that, similar to many pure metal-H phase
diagrams, there is a "gas/liquid" c~-ct' phase transition, with order/disorder
reactions occurring at lower temperatures.
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 79

Fig. 3.13.The phase diagram for the

ZrCr2-D system. The solid lines
represent regions established ex-
300 perimentally by diffraction techni-
ques and the dashed lines represent
extrapolated behavior as given by
~+0~;
Somenkov and Irodova [3.891
~d
I--- 2 0 0
I
I

t m ~
, 0~+~ 'P'"
1O0
I ~+a"
i
/I I I I I I
0 0.4 0.8 1.2 1.6

D/(Zr+Cr)

a) The ~ - ~ ' Transition

Yartys et al. [3.88] consider that this "gas/liquid" condensation transition, as
opposed to a "gas/solid" structural transformation, occurs in most IMCHs at
temperatures where they are likely to be used technologically. Thus the
observed plateau pressures and solvus compositions will usually pertain to this
type of transformation. It also seems probable that, just as in the case of pure
metals [3.92], the condensation arises, principally, from the indirect H - H
attractive interaction resulting from the lattice expansion associated with the
introduction of H into the host lattice. The direct repulsive interactions, which
operate at short range, will influence the position of the critical composition
through their influence on the configurational entropy. Another effect of these
short range repulsions is to give rise to short range order, which has been
observed in neutron diffuse scattering in the ZrV2-D system [3.88].
A#H2(plat) and the solvus compositions for the c~-~' transition should
follow from a full knowledge of the configurational and non-configurational
partition functions in the single phase regions. However, in view of the fact that
our understanding of these, at present, is unsatisfactory for predictive purposes,
more direct methods have been suggested for estimating the technologically
important plateau properties. All rely on the approximate relation (3.28)
between A#n2(plat ) and AG~f(MHy). The reader is referred to Chap. 6 of this
volume for details of the various methods which have been suggested for the
estimation of AHH2(plat ). However, a method proposed by the present authors
[3.10] for the direct estimation of A#n2(plat) is discussed briefly in Sect. 3.5.4.

b) Order/Disorder Transitions
Various interstitial superstructures, formed by H atom rearrangement and
characterised by a wide range of wave vectors, k, have already been found in
80 T. B. Flanaganand W. A. Oates

IMCHs [3.81]. In the ZrV2-D system, for example, ordering with k =(1/2 1/2
1/2), (1 0 0), 0, and (1 0 0) has been found at MD 2, MD3, MD,~, and M D 6,
respectively [3.89]. It appears that superstructures with k = 0 or (1/2 1/2 1/2)
are found at higher temperatures and over wider ranges of stoichiometry than
those with the other wave vectors [3.88].
The simplest type of ordering, orientation ordering, occurs when k =0.
However, even in this case, because of the different types and complex
arrangements of the sites available in IMCs, calculations of the superstructures
are quite involved. The theory of ordering on simple lattices has been
developed, by amongst others, Khachaturyan [3.115] and, recently, Irodova
[3.116] has used Khaehaturyan's theory to predict the structures which may
appear due to orientational ordering on the [-A2B2] sites in C 15 Laves phases.
The 96 sites per unit cell form 24 interpenetrating fcc Bravais lattices. Irodova
showed that the matrix symmetry restricts ordered structures to nine types.
When the further restriction of the blocking of first and second nearest
neighbors is imposed, only two of the nine remain and only one of these permits
the composition AB2H 4 to be achieved. Gufan and Shirokov [3.117] have
adopted a slightly different approach and showed that ordering at the
composition AB2H 4 is possible in 13 ways which differ in symmetry but which
can be reduced to 8 if it is assumed that the H - H interaction decreases fairly
slowly with increasing separation. In all cases the superstructure has a
tetragonal symmetry and belongs to the space group C6h. It is clear that such
comparisons of the theoretical predictions of all possible superstructures in a
system with those observed experimentally will undoubtedly lead to a better
understanding of H - H interactions in IMCHs.

3.5.4 Complete and Partial Equilibrium

As indicated in Sect. 3.2.2, A#H2(plat) for the decomposition reactions under CE

can be calculated approximately from AG~ for the reaction (3.29). The accuracy
of this approximation will be similar to that in using (3.28), i.e., it will be better
at low temperatures where the mutual solubilities of the metal and hydride
phases are lower. Fig. 3.14 shows the CE phase diagram for the L a - N i - H
system calculated on the basis of this approximation and should be compared
with Fig. 3.5, which shows the PE phase diagram for the same system. The
various A#H~(plat) for CE are lower than for the equivalent Apn2(plat) for PE. It
has been suggested [3.10, 118] that the differences between these A#H2(plat)
values for CE and PE will be fairly small so that those for PE can be estimated
with a fair degree of confidence from the calculated ones for CE. This means of
calculating the plateau pressures in pseudo-binary IMCHs provides an
alternative to the methods discussed in Chap. 6.
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 81

C ..-i,,.
Fig. 3.14. The complete equilib-
O 0.2 0.4 0.6 0.8 1.0 rium phase diagram for the
5 , i i i i i i i i
La-Ni-H system at 373 K from
L a H x + Ni Oates and Flanagan I-3.10-1.The
0 filled small circles represent the
hydrogen pressures where the
LaHx+ LaNi 5 various phases are expected to
-5 decompose
L H x ÷ LN3, 5
L H x t- L N 3 LN 5
-10 LH x + L N 2
+
I-H x -I- LN1.4
LNz
Ni
E~
o -15 LHx+ LN Ll~i +
LN 4- LN3

L H x + LaN L3N + LN 2
,LNt,4
-20
+
La + L3N LN
-25
La L3N LN LNt.4LN2 LN&5 Ni
LN3 LN5

3.5.5 Hysteresis
General accounts of hysteresis have been given recently by Flanagan and
Clewley [3.24] and McKinnon [3.25]. Sinha and Wallace [3.119] have also
recently developed a model of hysteresis based on elastic misfit between the
hydride and the dilute phase. However, this model seems to be at odds with the
fundamental point that hysteresis represents the conversion of work to heat,
i.e., hysteresis must involve irreversible effects and therefore cannot be due to
elastic deformation.
Hysteresis is believed to arise mainly from the irreversible plastic work
which is needed for the accommodation of the dilute phase/hydride phase misfit
during both hydride formation and decomposition [3.120]. In the initial
activation process the expansion cannot be accommodated by dislocations and
mechanical disruption occurs (brittle fracture) in most IMCs whereas in pure
metal-H systems the expansion can generally be accommodated by punching
out of misfit dislocations [3.121]. Following the first hydriding/dehydriding
cycle in IMCs, further expansions and contractions during cycles of hydriding
and dehydriding are presumably accommodated largely by dislocation cre-
ation. After the first hydriding/dehydriding cycle the dislocation density does
not increase markedly and this implies that the dislocations which are created
are subsequently annihilated leading to a constant, large dislocation density.
Park and Flanagan [3.122] recently observed an important effect of the
interface velocity on the subsequent, time-independent plateau pressures of the
LaNis-H system. Rapid, initial interface velocities lead to smaller values of pr
and larger values Ofpd as compared, in both cases, to the effect of small interface
velocities. This effect provides an explanation for the previously unexplained
results of Goodell et al. [3.60], who found that dynamic isotherms differed from
82 T. B. Flanagan and W. A. Oates

static isotherms for several IMCHs. It turns out that the former isotherms were
determined under conditions which corresponded to slow interface velocities
and the latter to rapid, initial interface velocities. The knowledge of the
existence of this effect is important because otherwise serious errors can be
introduced into thermodynamic parameters for the plateau regions. For
example, if values of Pr are determined under fast interface conditions at one
temperature and slow interface conditions at another temperature, erroneous
data will be obtained for the plateau thermodynamic parameters.
Park and Flanagan I-3.1233 offered an explanation for this effect based on
local equilibrium considerations at the interface, i.e., at the interface #a = #~,
#n = #~, and #H = #~ whereas in the bulk only the last of these equalities holds.
The nature of the local equilibrium established depends upon the initial
interface velocity and this, in turn, affects the final value of A#rh(plat). If their
explanation is correct, this will lead to another form of hysteresis in multi-
component solid-H systems. The fact that decomposition of IMCHs can occur
at relatively low temperatures as a result of repeated cycling supports the model
because it indicates enhanced mobility of the metal atoms at the moving
interface. Hillert [3.124] has pointed out that such a hysteresis will also occur in
the transformation of austenite containing alloying elements but his concern
was with the transformation kinetics rather than the stable, time-independent
plateau pressures of interest here.

3.6 Conclusions
Until recently most "thermodynamic" investigations of IMCHs have been
restricted to plateau pressure measurements in the vicinity of room tempera-
ture, i.e., the minimum information which is required for the technological
applications of these materials. When models have been used for the
interpretation of the results they have usually been too facile to be applicable
to the quite complex systems which are involved.
More thorough investigations of both the thermodynamic properties in the
single phase regions and of the phase diagrams are, however, now being
conducted and it is hoped that a better understanding of the thermodynamic
properties will ensue from these experimental studies. On the theoretical side
the complex structures of IMCHs, with their multitude of possible environ-
ments for H atoms means that models for the configurational and vibrational
contributions to the partition function will also have to become more
sophisticated.
Considerable improvements in both experimental and theoretical investi-
gations can be anticipated in the next few years. The same rapid progress which
has been made in the case of pure metal-H systems in the last decade can be
predicted with confidence.
Acknowledgements. TBF is grateful to the National Science Foundation and WAO to the
Australian ResearchGrants Schemefor financialsupport of their researchon metal hydrides.
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 83

References
3.1 K.H.J. Buschow, P.C.P. Bouten, A.R. Miedema: Rep. Prog. Phys. 45, 937 (1982)
3.2 P.S. Rudman, G. Sandrock: Annu. Rev. Mater. Sci. 12, 271 (1982)
3.3 D.G. Ivey, D.O. Northwood: J. Mater. Sci. 18, 321 (1981)
3.4 H. Oesterreicher: Appl. Phys. 24, 169 (1981)
3.5 A. Hultgren: Trans. ASM 39, 915 (1947)
3.6 W.A. Oates, T.B. Flanagan: J. Mater. Sci. 16, 3235 (1981)
3.7 M.L. McGlashan: Chemical Thermodynamics, Vol. 1. Special Periodical Report,
Chemical Society (London) 1973, Chap. 1
3.8 D. Ohlendorf, H.E. Flotow: J. Less-Common Met. 73, 2937 (1980)
3.9 W.E. Wallace, D. Ohlendorf, H.E. Flotow: J. Less-Common Met. 79, 157 (1981)
3.10 W.A. Oates, T.B. Flanagan: Mater. Res. Bull. 19, 1397 (1984)
3.11 H. Wenzl, E. Lebsanft: J. Phys. F 10, 2147 (1980)
3.12 J.J. Murray, M.L. Post, J.B. Taylor: J. Less-Common Met. 73, 33 (1980)
3.13 B.S. Bowerman, C.A. Wulff, T.B. Flanagan: Z. Phys. Chem. 116, 197 (1979)
3.14 W.N. Hubbard, P.L. Rawlins, P.A. Connick, R.E. Stedwell, P.A.G. O'Hare: J. Chem.
Thermo. 13, 785 (1983)
3.15 J.J. Murray, M.L. Post, J.B. Taylor: J. Less-Common Met. 90, 65 (1983)
3.16 M.L. Post, J.J. Murray, J.B. Taylor: Int. J. Hydrogen Energy 9, 137 (1984)
3.17 D.M. Nace, J.G. Aston: J. Am. Chem. Soc. 79, 3619 (1957)
3.18 G. Boureau, O. Kleppa: J. Chem. Thermo. 9, 543 (1977)
3.19 T.L. Hill: J. Chem. Phys. 17, 520 (1949)
3.20 A.L. Shilov, L.N. Padurets, M.E. Kost: Russ. J. Phys. Chem. 57, 338 (1983)
3.21 M.H. Mintz, D. Hiershler, Z. Hadary: J. Less-Common Met. 48, 241 (1976)
3.22 D. Ohlendorf, H.E. Flotow: J. Chem. Phys. 73, 2937 (1980)
3.23 H. Wenzl, S. Pietz: Solid State Comm. 33, 1163 (1980)
3.24 T.B. Flanagan, J.D. Clewley: J. Less-Common Met. 83, 127 (1982)
3.25 W.R. McKinnon: J. Less-Common Met. 91, 293 (1983)
3.26 N.A. Scholtus, W.K. Hall: J. Chem. Phys. 39, 868 (1963)
3.27 C.E. Lundin, F.E. Lynch: In Hydrides .for Energy Storage, ed. by A.F. Andresen, A.J.
Maeland (Pergamon, Oxford 1978) p. 395
3.28 T.B. Flanagan, B.S. Bowerman, G.E. Biehl: Scripta Metall. 14, 443 (1980)
3.29 T.B. Flanagan, W.A. Oates: J. Less-Common Met. 92, 131 (1983)
3.30 T. Kuji, W.A. Oates, B.S. Bowerman, T.B. Flanagan: J. Phys. F 13, 1785 (1983)
3.31 A. Percheron-Guegan, C. Lartigue, J.C. Achard, P. Germi, F. Tasset: J. Less-Common
Met. 74, 1 (1980)
3.32 T.B. Flanagan, T. Schober, H. Wenzl: to be published
3.33 R. Hempelmann, D. Richter, G. Eckold, J.J. Rush, J.M. Rowe, M. Montoya: J. Less-
Common Met. 104, 1 (1984)
3.34 P.D. Goodell: J. Less-Common Met. 99, 1 (1984)
3.35 T.B. Flanagan, C.A. Wulff, B.S. Bowerman: J. Solid State Chem. 32, 321 (1980)
3.36 K.H.J. Buschow: Mater. Res. Bull. 19, 935 (1984)
3.37 J.-M. Welter, G. Arnold, H. Wenzl: J. Phys. F 13, 1773 (1983)
3.38 J.F. Lynch, J.J. Reilly: J. Less-Common Met. 87, 225 (1982)
3.39 T.B. Flanagan, N.B. Mason, G.E. Biehl: J. Less-Common Met. 91, 107 (1983)
3.40 T.B. Flanagan, G.E. Biehl: J. Less-Common Met. 82, 385 (1981)
3.41 T.B. Flanagan, S. Majorowski, J.D. Clewley, C.N. Park: J. Less-Common Met. 103, 93
(1984)
3.42 G.D. Sandrock: In Hydrides for Energy Storage, ed. by A.F. Andresen, A.J. Maeland
(Pergamon, Oxford 1978) p. 353
3.43 W.A. Oates, T.B. Flanagan: Scripta Metall. 17, 983 (1983)
3.44 D.G. Ivey, D.O. Northrup: Scripta Metall. 19, 1319 (1985)
3.45 Y.L. Yeh, K. Samwer, W.L. Johnson: Appl. Phys. Lett. 42, 242 (1983)
3.46 K. Samwer, Y.L. Yeh, W.L. Johnson: J. Non-Cryst. Solids 61--62, 631 (1984)
84 T. B. Flanagan and W. A. Oates

3.47 K. Aoki, A. Horata, T. Matsumoto: Proc. 4th Int. Conf. on Rapidly Quenched Metals,
Sendai, August 24-28, 1981, Japan Inst. of Metals, 1982, p. 1649
3.48 K. Aoki, T. Yamamoto, T. Matsumoto: Scripta Metall. 21, 27 (1987)
3.49 A.J. Maeland, L.E. Tanner, G.G. Libowitz: J. Less-Common Met. 74, 279 (1980)
3.50 R. Kirchheim: Acta Metall. 30, 1069 (1982)
3.51 R. Griessen: Phys. Rev. B 127, 7575 (1983)
3.52 J.J. Murray, M.L. Post, J.B. Taylor: J. Less-Common Met. 80, 210 (1981)
3.53 C.E. Lundin, F.E. Lynch, C.B. Magee: J. Less-Common Met. 56, 19 (1977)
3.54 G. Busch, L. Schlapbach, A. Seiler: In Hydrides for Energy Storage, ed. by A.F.
Andresen, A.J. Maeland (Pergamon, Oxford 1978) p. 263
3.55 M.H. Mendelsohn, D.M. Gruen, A.E. Dwight: J. Less-Common Met. 63, 193 (1979)
3.56 J.J. Reilly, R. Wiswall: Inorg. Chem. 13, 218 (1974)
3.57 J. Schefer, P. Fischer, W. Halg, F. Stucki, L. Schlapbach, A.F. Andreson: Mater. Res.
Bull. 14, 1281 (1979)
3.58 F. Reidinger, J.F. Lynch, J.J. Reilly: J. Phys. F 12, 149 (1982)
3.59 J.J. Reilly, J.R. Johnson, J.F. Lynch, F. Reidinger: J. Less-Common Met. 89, 55 (1983)
3.60 P.D. Goodell, G.D. Sandrock, E.L. Huston: J. Less-Common Met. 73, 135 (1980)
3.61 V.V. Zhirnova, B.M. Mogutnov: Russ. J. Phys. Chem. 56, 1313 (1982)
3.62 A. Pebler, E.A. Gulbransen: Trans. Met. Soc. AIME 239, 1593 (1967)
3.63 F. Pourarian, H. Fujii, W.E. Wallace, V.K. Sinha, H.K. Smith: J. Phys. Chem. 85, 3105
(1981)
3.64 V.K. Sinha, W.E. Wallace: unpublished results
3.65 H. Ocstcrreicher, H. Bittner: Mater. Res. Bull. 13, 83 (1978)
3.66 H. Fujii, F. Pourarian, V.K. Sinha, W.E. Wallace: J. Phys. Chem. 85, 3112 (1981)
3.67 V.K. Sinha, F. Pourarian, W.E. Wallace: J. Phys. Chem. 86, 4952 (1982)
3.68 F. Pourarian, W.E. Wallace: J. Less-Common Met. 91, 223 (1983)
3.69 V.K. Sinha, W.E. Wallace: J. Less-Common Met. 91,229 (1983)
3.70 A. Suzuki, N. Nishimiga, S. Ono: J. Less-Common Met. 89, 283 (1983)
3.71 R.M. van Essen, K.H.J. Buschow: Mater. Res. Bull. 15, 1149 (1980)
3.72 A.T. Pedziwiatr, R.S. Craig, W.E. Wallace, F. Pourarian: J. Solid State Chem. 46, 336
(1983)
3.73 F. Pourarian, V.K. Sinha, W.E. Wallace, A.T. Pedziwiztr, R.S. Craig: Proc. lnt. Symp.
on Electronic Structure and Properties of Hydrogen in Metals, Richmond, VA, USA
(Plenum, New York 1983) p. 385
3.74 S. Majorowski, T.B. Flanagan: to be published
3.75 D. Shaltiel, I. Jacob, D. Davidov: J. Less-Common Met. 53, 117 (1977)
3.76 H. Kierstead: J. Less-Common Met. 70, 199 (1980)
3.77 B. Vigeholm: J. Less-Common Met. 89, 136 (1983)
3.78 J.J. Reilly, R.H. Wiswall: Inorg. Chem. 7, 2254 (1968)
3.79 J.J. Reilly: In Hydrides for Energy Storage, ed. by A.F. Andresen, A.J. Maeland
(Pergamon, Oxford 1978) p. 301
3.80 M.H. Mintz, Z. Gavra, G. Kimmel: J. Less-Common Met. 74, 263 (1980)
3.81 J.-J. Didisheim, P. Zolliker, K. Yvon, P. Fischer, J. Schefer, M. Gubelmann, A.F.
Williams: Inorg. Chem. 23, 1953 (1984)
3.82 C. Wagner: Acta Metall. 21, 1297 (1973)
3.83 R.B. McLellan: Scripta Metall. 16, 745 (1982)
3.84 A. Switendick: Z. Phys. Chem. N.F. 117, 89 (1979)
3.85 D.G. Westlake: J. Less-Common Met. 91, 1 (1983)
3.86 D.G. Westlake: J. Less-Common Met. 91, 275 (1983)
3.87 D.G. Westlake: J. Mater. Sci. 19, 316 (1984)
3.88 V.A. Yartys, V.V. Burnasheva, K.N. Semenenko: Russ. Chem. Revs. (Engl. trans.) 52,
299 (1983)
3.89 V.A. Somenkov, A.V. Irodova: J. Less-Common Met. 101, 481 (1984)
3.90 J. Shinar, I. Jacob, D. Davidov, D. Shaltiel: In Hydrides for Energy Storage, ed. by A.F.
Andresen, A.J. Maeland (Pergamon, Oxford 1978) p. 337
 Thermodynamics of Intermetallic Compound-Hydrogen Systems 85

3.91 I. Jacob, D. Shaltiel: J. Less-Common Met. 65, 117 (1979)

3.92 G. Alefeld: Ber. Bunsenges. Physik. Chem. 76, 746 (1972)
3.93 A. Pasturel, B. Brion, P. Hicter, A. Percheron-Gu6gan, J.C. Achard: J. Less-Common
Met. 86, 19 (1982)
3.94 W.A. Oates, P.T. Gallagher, J.A. Lambert: Trans. Met. Soc. AIME 245, 47 (1969)
3.95 G. Boureau: J. Phys. Chem. Solids 42, 743 (1981)
3.96 W.E. Wallace, H.E. Flotow, D. Ohlendorf: J. Less-Common Met. 79, 157 (1981)
3.97 J.C. Achard, C. Lartigue, A. Percheron-Guegan, J.C. Mathieu, A. Pasturel, F. Tasset: J.
Less-Common Met. 79, 161 (1981)
3.98 1. Jacob, J.M. Bloeh, D. Shaltiel, D. Davidov: Solid State Commun. 35, 155 (1980)
3.99 W.A. Oates, R. Ramanathan: Proc. 2nd Internat. Congr. on Hydrogen in Metals, Paris
1977 (Pergamon, Oxford 1978) p. 2All
3.100 H.A. Kierstead: J. Less-Common Met. 71, 303 (1980)
3.101 H.A. Kierstead: J. Less-Common Met. 73, 61 (1980)
3.102 H.A. Kierstead: J. Less-Common Met. 84, 253 (1982)
3.103 A.L.G. Rees: Trans. Faraday Soc. 50, 335 (1954)
3.104 H.A. Kierstead: J. Less-Common Met. 75, 267 (1980)
3.•05 H.A. Kierstead: J. Less-Common Met. 96, 133 (1984)
3.•06 B.D. Dunlap, P.J. Viccaro, G.K. Shenoy: J. Less-Common Met. 74, 75 (1980)
3.107 S.M. Shapiro, F. Reidinger, J.F. Lynch: J. Phys. F 12, 1869 (1982)
3.108 W.A. Oates, T.B. Flanagan: J. Chem. Soc., Farad. Trans., Part I, 73, 407 (1977)
3.•09 A. Biris, R.V. Bucur, P. Ghete, E. Indreea, D. Lupu: J. Less-Common Met. 49, 477 (1976)
3.110 B. Andreyev, V. Shitikov, E. Magomedbekov, A. Shafiev: J. Less-Common Met. 911,161
(1983)
3.111 D. Dayan, M.P. Dariel: Mater. Res. Bull. 16, 137 (1981)
3.112 F. Kuijpers: Philips Res. Rep. Suppl. No. 2 (1973)
3,113 G.D. Sandrock, J.J. Murray, M.L. Post, J.B. Taylor: Mater. Res. Bull. 17, 887 (1982)
3.114 R.A. Bond, D.K. Ross: J. Physics F 12, 597 (1982)
3.115 A.G. Khachaturayan: Theory of the Structural Transformations in Solids (Wiley, New
York 1983)
3.116 A.V. Irodova: Soviet Phys.-Solid State 22, 1493 (1980)
3.117 Y.M. Gufan, V.B. Shirokov: Soviet Phys.-Solid State 23, 1992 (1981)
3.118 W.A. Oates, T.B. Flanagan: Metall. Trans. 16A, 139 (1985)
3.119 V.K. Sinha, W.E. Wallace: J. Less-Common Met. 91, 239 (1982)
3.120 T.B. Flanagan, B.S. Bowerman, G.E. Biehl: Scripta Metall. 14, 443 (1980)
3.121 H.K. Birnbaum, M.L. Grossbeck, M. Amano: J. Less-Common Met. 49, 357 (1976)
3.•22 C.N. Park, T.B. Flanagan: J. Less-Common Met. 94, L1 (1982)
3.•23 C.N. Park, T.B. Flanagan: Scripta Metall. 18, 683 (1984)
3.124 M. Hillert: The Mechanism of Phase Transformations in Crystalline Solids (Inst. Metals,
London 1969) p. 231
4. Crystal and Magnetic Structures of Ternary
Metal Hydrides: A Comprehensive Review

Klaus Yvon and Peter Fischer

With 4 Figures and 3 Tables

The crystal and magnetic structures of hydrides and deuterides of binary metal
compounds as determined by neutron diffraction analysis are reviewed.
Emphasis is placed on the distribution of hydrogen (deuterium) atoms in the
metal atom host structure. At room temperature these distributions are
characterized by thermal disorder and preferential occupation of large
interstices which are surrounded by atoms belonging to hydride forming
elements. The metal atom arrangements are usually similar to those of the binary
metal compounds, except that their lattices are generally expanded and their
symmetries reduced. At low temperatures the hydrogen (deuterium) atoms tend
to order, thus inducing structural phase transitions and further symmetry
lowering. Efforts to rationalize observed hydrogen (deuterium) site occupancies
and maximum hydrogen (deuterium) contents in terms of empirical models are
reviewed. The influence of hydrogenation (deuteration) on the magnetic
structures is discussed.

4.1 Introduction
Most ternary metal hydrides 1 derive from binary metal compounds which
absorb hydrogen by filling interstices of their metal atom network. The
resulting hydrogen concentrations per unit volume are often higher than that of
liquid hydrogen, which makes ternary metal hydrides attractive candidates for
hydrogen storage (see Chap. 6, Vol. II). One of the major factors limiting
hydrogen storage capacity is the reluctance of hydrogen to occupy all available
interstices in the metal atom network. In order to investigate this behaviour
detailed structure data are necessary. Precise atomic coordinates are also a
prerequisite for energy-band calculations (see Chap. 5).
In this chapter the presently known (January 1986) crystal and magnetic
structures of ternary metal hydrides are reviewed. Only those compounds are
discussed for which hydrogen atom positions have been determined by neutron
diffraction experiments. Compared to other methods which provide structural

1 Throughout this article the terms hydrogen and hydride(s) are also used for the terms
deuterium and deuteride(s), respectively. This choice should not lead to major confusion
because significant structural differences between ternary metal hydrides and deuterides
have not so far been reported.
88 K. Yvon and P. Fischer

information, such as x-ray and electron diffraction, diffuse and inelastic neutron
scattering, nuclear magnetic resonance, M6ssbauer, Raman and infra-red
spectroscopy, neutron diffraction is unique as it gives direct and precise
information on the atomic coordinates of hydrogen atoms. Moreover it yields
direct and detailed information on the magnetic structures and order
phenomena.
In contrast to a previous critical review [4.11, emphasis in this review is
placed on completeness. The only major restriction concerns the type of
hydrides covered; apart from a few exceptions only metallic hydrides which
derive from binary metal compounds are included, i.e. compounds in which
different metal atom constituents occupy different crystallographic equipoints.
Therefore, the following classes of hydrides are not included (most recently
studied representatives with known hydrogen atom distributions are given in
parentheses): elemental hydrides (/~I-VzH and/?-V2D [4.2], La(Ce)D3 [4.3],
Ce(Pr)D, .gs [4.4], YbD 2 [4.5], jS-UH(D)3 [4.6], PuD x [4.7]), hydrides which
derive from pseudo-binary alloys (Ti~_yVyD~ [-4.8]) or amorphous alloys
(ZrNiD1. s and Zr2NiD4. 3 [-4.9], CuxTil_x(H,D)r [-4.10]), saline hydrides
(BaLiH(D)3 [-4.11]), non-metallic complex hydrides (K2ReH 9 [4.12]), ternary
hydrides containing only one metal constituent (D(H)Nb6111 [4.13], ZrBrD
[,,4.143, TbBrD 2 [4.15]), ternary metal hydroxides (NazPt(OD)6 [,4.16]),
hydride oxides (DAI1101v [4.17], Ni(OD)2 [,4.18]), hydrogen bronzes
(H (D)TaO 3 [4.19]), hydride carbides (nitrides, oxides) with close-packed metal
atom arrangements (TiCxHy [--4.20], ZrO0.4Do.1 [4.21 ]), and ternary hydrides
which generally do not derive from stable binary metal compounds (SrzRuD 6
[,4.22], K2PtD4 [-4.23]).
Previous structure reviews of metal hydrides have mainly focussed on
elemental hydrides [,4.24-27], in particular those of group V transition elements
and Pd [,4.28], and actinides [4.29], transition metal hydride complexes [4.30],
hydrides of compounds between transition metals and p-elements [4.31], and
hydrides of selected binary metal compounds [4.1, 32-35]. Structural aspects of
ternary metal hydrides have also been discussed in general reviews [4.3642]
and contributions to international symposia on the properties and applications
of metal hydrides [-4.43].
This review is organized as follows. In Sect. 4.2 experimental aspects such as
sample preparation, neutron diffraction setup and structure refinement
methods are discussed. In Sect. 4.3 a comprehensive list of ternary metal
hydride structures [-4.44-113] is given and their characteristic features are
discussed. Emphasis is placed on those features which are of relevance for the
understanding of hydrogen sorption properties, such as the distribution, the
number and the type of metal atom interstices occupied by hydrogen. In
Sect. 4.4 attempts to rationalize the observed hydrogen site occupancies and
maximum hydrogen contents are reviewed. Most attempts have been based on
two empirical models of which one emphasized the importance of geometrical
restrictions [4.114] and the other nearest neighbour interactions [-4.115, 116].
Work based on the former model was reviewed recently [4.114]. Section 4.5
 Crystal and Magnetic Structures of Ternary Metal Hydrides 89

concerns magnetic ordering of ternary metal hydrides as determined by

neutron scattering. Section 4.6 presents a conclusion.
Atomic coordinates (including error estimates) and magnetic order para-
meters (magnetic moments and ordering temperatures as determined by
neutron diffraction) for selected ternary metal hydride structures are sum-
marized in Appendix 4.A and 4.B, respectively. Most of these data cannot be
retrieved from existing crystallographic data files. Interatomic distances and
structural drawings can be derived from these data by using standard
crystallographic computer programs and thus are not given unless they are
essential for the understanding of the text.

4.2 Experimental Method

4.2.1 Sample Preparation

Most ternary metal hydrides discussed in this review were prepared by a two-
step procedure which consisted of the synthesis of the binary metal compound
(mainly by arc melting or inductive heating) and subsequent hydrogenation (see
Chap. 2). Exceptions are Mg2FeD6 [4.66] and Mg2CoD~ [4.65] which were
directly synthesized from the elements by a sintering technique. Most hydrides
were stable under atmospheric pressure, while others were kept under high
deuterium pressure (FeTiDx [4.76, 79]), or were "poisoned" by air and gases
such as CO, and S O 2 to prevent desorption (FeTiD x [4.77, 117], LaNi4A1D4.1
[4.53], LaNisD 6 [4.88]). Hydride formation sometimes required activation of
the binary metal compound (FeTiDx [4.73-79, 117]). For some systems, several
hydrogenation-dehydrogenation cycles were necessary to obtain homoge-
neous, fully hydrided samples (MgzNiH4 [4.59, 61, 63, 118]). For others
hydrogenation led to a segregation into various binary metal compounds
and/or elemental hydrides (for example: 5LavNi 3 +48D2--*3LaNi s + 32LAD 3
[4.82]), or to the formation of amorphous alloys (for example: LazNivDx [4.32
and references therein, 4.37]). The quality of the non-hydrogenated samples
was often checked by x-ray powder diffraction methods, but rarely by more
sensitive methods such as microprobe analysis or optical metallography (LaNi 5
[4.35]). Thus the presence of impurity phases was usually only detected after
completion of the neutron diffraction experiment (ZrVzDx [4.48-50]). For
some systems single-phase samples were not obtained because of the difficulty
in synthesizing homogenous binary metal compounds, such as Mg2Ni which
usually contains several percent of Mg and MgNi2 phases unless it is prepared
by gas-phase reaction [4.119]. In view of the relatively large sample size
required for neutron diffraction experiments (typically 1 ema), sample inhomo-
geneity is a major factor limiting the accuracy of the structural results. Another
factor is the loss of crystallinity during hydrogenation.
The hydrogen content of the samples was usually measured prior to (or
after) the diffraction experiment. Due to sample problems and the limitations of
90 K. Yvon and P. Fischer

the models used for structure refinement (Sect. 4.2.3), it often differed signifi-
cantly (by up to 10%, for example in ZrVzD4. 5 [4.50] and 7-TiFeD1.9 [4.76],
see remarks in Appendix 4.A) from that determined from the diffraction
experiment (Sect. 4.3.2).

4.2.2 Neutron Diffraction

Except for fl-Mg2NiD 4 [4.120-123], refinements of the metal atom substruc-

tures by x-ray diffraction have not been performed, although such experiments
would have been desirable in order to detect lattice distortions and reduce the
number of atomic parameters for structure refinements by neutron diffraction.
Thus most neutron diffraction data were analyzed with a view to determining
positional co-ordinates of both hydrogen (deuterium) and metal atoms.
Structure investigations by electron diffraction were only rarely performed as
for example on the system TiFe-H(D) [4.124], and on MgzNiH 4 [D. Nor6us,
1985, private communication]. These studies did not yield atomic coordinates.
Most neutron diffraction experiments have been performed with nuclear
steady-state reactors by using medium resolution two-axis diffractometers,
crystal monochromators (usually graphite, or Ge) yielding wavelengths
between typically 0.9 and 2.6 A, and single detectors. Only a few studies have
been performed with high-flux reactors such as that at the Institut Laue-
Langevin (ILL) in Grenoble (diffractometer D1B, and high-resolution angle-
dispersive diffractometer D1 A [4.125]),using a multidetector system [4.126], or
with pulsed neutron sources such as the Intense Pulsed Neutron SpaUation
Source (IPNS) at the Argonne National Laboratory, using an energy dispersive
time-of-flight (TOF) method [4.127].
Due to the lack of suitable single crystals, all studies on ternary metal
hydrides reported so far have been performed on powders and thus at a certain
expense of accuracy. Moreover the resolution and peak to background ratios of
most diffraction patterns (if stated) are relatively low and suffer from both
sample dependent properties (lack of crystallinity, microtwinning, micro-
strains, etc., leading to anisotropic line broadening) and structural properties
(disorder, short-range order, thermal diffuse scattering, etc. leading to non-
uniform background, see Sect. 4.2.3). Typical values for the angular resolution
are Ad/d~-4x 10 - 3 - 1 0 -2 (x-ray Guinier method: Ad/d~-10 -3) whereas the
cut-offvalue s = sin(0/2) .... varied typically between s = 0.4 A - 1 and s = 0.5 A - 1
(see values stated in Appendix 4.A, x-rays CuKe: s = 0.6 A - 1). High-resolution
studies such as those performed on fl-LaNisDx [4.87] with the constant-
wavelength spectrometer D1A (Ad/d=2 x l0 -a, s=0.51 A -1) at ILL, and on
Zr3 V aODx [4.47] and fl-Mg 2NiD 4 [4.64] (A did = 3 x 10-3, s = 1.19 A-1) with
the TOF spectrometer at IPNS are rare.
The samples are usually measured in cylindrical vanadium containers of 0.8
to 1.5 cm diameter at ambient pressure and temperature, or within quartz tubes
at higher pressures and temperatures. The in situ variation of these parameters
 Crystal and Magnetic Structures of Ternary Metal Hydrides 91

is easier to perform for neutron diffraction (pressures up to 120 bar,

temperatures between 1.3 and 600 K) than for x-ray diffraction, which explains
the scarcity of the latter studies. On the other hand the precision of the lattice
parameters as determined by constant-wavelength neutron diffraction is
usually lower than that determined by x-rays because of the uncertainty in
wavelength calibration (A2/2=3x 10-4). Absorption is generally not an
important problem. It can be determined by transmission measurements, and
the corrections applied are those for cylindrical samples for which the product
of linear absorption coefficient and sample radius is usually smaller than
/~R = 0.2 (except for compounds containing certain nuclei such as B, Sm, Gd
etc.). Polarized neutron diffraction studies which would enable separation of
magnetic from nuclear, and coherent from incoherent scattering, thus yielding
the most detailed information on magnetic density distributions, have not yet
been performed because of a lack of single crystals.
Most neutron diffraction studies have been performed on deuterides
(because of the much more favorable ratio between coherent and incoherent
cross-section of deuterium compared to that of hydrogen), at ambient
temperature and pressure, and at fixed deuterium concentrations. Only a few
studies have been performed on hydrides (TizNiH x [4.45], ZrVzH3. 7 [4.51],
ZrTiI.s6H3. 8 [4.53], ErFezH3. 5 [4.57], MgzFeH 6 [4.66], Nb3SnH [4.81], ~-
Mg2NiHo. 3 [4.100], ZrNiH2.98 [4.108], Pd3Po.sHo.17 [4.112], Pd6PHo.39
[4.112]) and mixed hydride-deuterides (TiFeHo.vsDo.22 [4.75],
Zro.35Tio.65Ho.v6Do.24 [4.55]), or as a function of temperature (ZrVaD3. 6
[4.49], Y6Mnz3D23 [4.69], 7-TiFeDa.8 [4.78], Ti~.2Mn~.sD3.1 [4.98], pres-
sure (TiFeD x [4.76], /?-LaNisD x [4.87], Pd6PDx [4.113]), and deuterium
content (Zr3V3OD x [4.47], ZrVzDx [4.48], ZrCr2D~ [4.51], YhZraD~ [4.58],
Y6Mn23D~ [4.67], Ho6Fea3Dx [4.72]), and/~-LaNisD ~ [4.82, 87]. Except for
one compound (/?-La6°NisD7./[4.88]), all hydrides studied contained natural
metal isotope mixtures.

4.2.3 Structure Refinement

Lattice parameters, atomic coordinates, amplitudes of thermal vibrations, and

hydrogen site occupancy factors have usually, but not always (see below), been
refined with the Rietveld powder diffraction profile fitting method [4.128] by
using computer programs of which some allowed the simultaneous treatment of
several phases [4.129, 130] (for reviews of the Rietveld profile fitting method
see [4.131], and for its application to neutron powder diffraction see [4.132]
for angle dispersive, and [4.133] for energy dispersive setup). The main
problem with this method as applied to ternary metal hydrides is the
complexity of the crystal structures (Y6Mn23D23 at 4 K has 56 positional and
isotropic thermal parameters, according to [4.69]; LaNiD6.7 has 36 atomic
parameters, according to [4.87, 88]; for the number of refined parameters see
values of n listed in Appendix 4.A), and the low resolution of most experi-
92 K. Yvon and P. Fischer

mental diffraction patterns (Sect. 4.2.2). Other limiting factors are the bad
crystallinity of the samples (CeRuzD s [4.56]), the presence of impurity phases
(Fe and MgD z in MgEFeD 6 [4.66], MgD 2 and MgNi 2 in Mg2NiD 4
[4.59-64]), anisotropic line broadening in the diffraction patterns due to
microstrains (fl-LaNisD5_ 7 [4.53, 86-88], see also Ni(OD)z [4.18]), and
microtwinning (fl-Mg2NiD 4 [4.64, 123]), uneven background due to short-
range order of the hydrogen atoms (ZrVED3. 6 [-4.49]), anharmonicity (leading
to multisite models such as for fl-LaNisDs_ 7 [4.87, 88]), and structural
disorder between the metal atoms (TiI.zMn~.sD3. l [4.98]). The agreement
indices after profile refinement, Rwv and R~ (for definitions see [4.134]) vary
typically between 0.08 and 0.12, and 0.03 and 0.08, respectively, whereas the
errors of the atomic positional parameters lead to uncertainties in the
interatomic distances of the order of +_0.02 •. Due to insufficient data sets
and inadequacies of the refinement models some structural parameters such as
the temperature factors and hydrogen site occupancy factors have been only
poorly defined. The former are often constrained by the assumption of an
overall temperature factor. The latter are usually not constrained although
they are not always consistent with the total hydrogen content as determined
by desorption measurements (see remarks in Appendix 4.A).
A controversy exists with respect to the reliability of the estimated standard
deviations (e.s.d.'s) as derived by the profile fitting method (for recent work see
[4.135] and references therein). Experimental evidence from cubic metal
hydride structures suggests [4.48, 59] that some of the e.s.d.'s have been
underestimated by a factor of up to three with respect to those derived from the
same data from refinements based on integrated intensities. This discrepancy
could be due to model errors [4.135] affecting the calculated integrated
intensities, but no satisfactory method of correcting the e.s.d.'s for these errors
has been proposed as yet. A reliable determination of e.s.d.'s is of particular
importance for metal hydride structures because ambiguous, or conflicting
results have often been reported as to their lattice symmetries (y-TiFeD 2
[4.76-79]), space group assignments (fl-LaNisDs_ v I-4.32, 53, 83-90]), lattice
dimensions (Pd6PD~ [4.113], ThNi2D2 6 [4.102]), and hydrogen atom distri-
butions (fl-Mg2NiD 4 [4.62, 64] and fl'-Mg2NiD 4 [4.59-61], see Sect. 4.3.2).
On the other hand the profile refinement method yields very accurate atomic
parameters provided it is based on high-resolution data, such as those of
Zr3V3OD ~ [4.47] (yielding anisotropic temperature factors) and fl-Mg2NiD 4
[4.64]. The high accuracy obtained on the latter compounds is presumably a
consequence of the more favorable ratio between the number of observations
and number of refined atomic parameters.
Refinements based on integrated intensities have been performed e.g. for
cubic fl'-Mg2NiD 4 [4.59, 61] and ZrV2D ~ [4.48], hexagonal LaNisDs_ 7
[-4.53, 83, 86, 87, 89, 90] (showing anisotropic line broadening), and Qrtho-
rhombic ZrNiH3 [4.108]. The main problem with this method as applied to
non-cubic structures is the treatment of overlapping peaks. Hydrogen atom
positions and approximate occupancy factors (without error estimates) have
been derived by inspection of nuclear density maps on ~-TiFeDo.0s 7 [4.73]
 Crystal and Magnetic Structures of Ternary Metal Hydrides 93

and Nb3SnH [4.81 ], and from a comparison between observed and simulated
powder diffraction patterns for Ti2NiH~ [4.45], HfVaD 4 [4.52], Th2A1D ~
[4.105], and ZrNiD [4.109].

4.3 Structural Results

During the past 25 years over 70 neutron diffraction studies on more than 50
ternary metal hydride systems have been reported. A comprehensive list of
these studies [4.45-113] is given in Table 4.1. The list is complete up to, and
including, 1985 to the best of the authors' knowledge. Reports which were not
easily accessible to the authors may have been omitted. Most of these studies
were performed on compounds having the following binary metal atom ratios
(typical representatives are given in parentheses): AB(TiFeDx, ZraVaOD x,
TiCuD x, ZrNiHx), AB2(ZrV2D ~, ZrMn2D x, ZrBe2D~, ThNizD ~, Mg2NiD4),
AzB(Ti2NiHx, Th2A1D x, Zr2PdDx), ABa(HoNiaDx,Pd3PDx), AaB(NbaSnH~),
A6Bz3(H%Fez3Dx),ABs(LaNisD~), and AB7 (Y2Fe14BD~), where A denotes a
transition or rare-earth element, and B a transition element or a p-element. A
study of the various types of constituents reveals that all hydrides investigated
so far contain at least one element which is capable of forming a stable binary
hydride, such as Mg, Ti, Zr, Hi', V, Pd, Th, Y, and the rare-earths. This aspect
presumably constitutes a general trend for metallic hydrides since all known
binary metal compounds which do not contain one of these elements are poor
hydrogen absorbers [4.136, 137]. As shown in Sect. 4.3.2 this characteristic
behavior is also reflected by structural features and can be explained in terms of
preferential hydrogen atom occupancies (Sect. 4.4.4).
The compounds listed in Table 4.1 are grouped according to the structure
types of their (hydrogen free) metal atom matrices which are listed in order of
decreasing symmetry. Each hydride structure is characterized with respect to
space group, lattice parameters (including those of the hydrogen-free metal host
structure), and metal atom environment of the occupied H(D) atom sites
(number and type of surrounding metal atoms, point group symmetry). For
convenience the H(D) sites are generally listed in order of decreasing
occupancy. For characteristic compounds (those printed with bold letters)
explicit atomic coordinates are given in Appendix 4.A. Lattice transformation
temperatures, T~, are indicated in both Table 4.1 and Appendix 4.A. Magneti-
cally ordered hydride structures are listed in Appendix 4.B. The latter contains
information on magnetic moments and Curie (or N6el) temperatures, Tc (or
TN), if known from neutron diffraction.
The structural features of ternary metal hydrides are summarized in the next
section.

4.3.1 Metal Atom Substructure

Compared to elemental metal hydride structures which are characterized by a

few, relatively simple and usually highly symmetrical, metal atom arrangements
94 K. Yvon and P. Fischer

:n
..o

o _< - - , - - ~ - - , - - ~ - ~ - ~ , - - ~

'm
"'~"~~'~'~'~ ~ ~ ~"~ ~ ~ ~"Z ~ ~ ~

8
LI II ~ II II ]1 N II II II II II II II II

"0

"0
Z

[- -u

o
N~q N ~ N ~q
 Crystal and Magnetic Structures of Ternary Metal Hydrides 95

~.~ ~ ~
.~ ~ II o

o<

o -~ = < -~
0
o o ~.~ ~
-~= ~ ~~ ~ ~ ~:

~.~ ,-~ . ~ ~ ~
II II II

.-~ ~ ~

0 .~ 0

.~-

"0
e.l e,l ¢-i ~ ~-i
o'~
~ - - . ~
,~.o~

~.~ ~=
•~ ~ ~
~ ~.=
 96 K. Yvon and P. Fischer

o
e'-,i ¢'-q e - q

'~'"o' %" ~'-~ ~,'~"~,~,-~,-~,'-'~,z,-'~"~

,q .q e~ ',:t 'q

oo
r.i • c,i
II II II N ~:" II ~" II ~" II II II [I II II [I
..1

~a

± ±

E E E ~ ~ ~ z~EE
~a

,.q

Z f.
..--,.

8
tn on
,¢
Z~
~o
 Crystal and Magnetic Structures of Ternary Metal Hydrides 97

~"~'-'-'~ ~'~ '~'~"~ "'"- ~',..~'~ ~e.q ., ~4~' .,.:.. ,,o _~.',o .-.,., ~r-,~
,,.-q e-~

,...,..~,..= ,..~ o o o o o o o o o o ~ ' ~

,..q. m
V II ~
h- ~- II II

II II ~- II II ~ II II

~4 o o ~ ,.,,-, o r.4 ~i ,,o ~ ~ o', o',, o',

II Jl II It II II II N II N "~ II II II

~. t-.-- ~ ~ ~ r.,- r--- r.-- ~ "e. ~ r-- r,.-

,-.1 ,,I ~ ,,1 .-q ~ .I ~ ,,1 ,,I

,,o

ox ,.o
 98 K. Yvon and P. Fischer

II II II II II II II II II II
<_

II II II ~ II ~ 11 ~" II ~ II ~ II II II ~ II II II II ~ II II II

8
v
%
0

0
 Crystal and Magnetic Structures of Ternary Metal Hydrides 99

~, ~,~ " . . _ -- ~a~ ~ . ~ _ ~ ~ , . . _ ._. ~ ,..~ ,-~ ~ , ~ ~ .~. _.~ ~ - ~

~.~ ~ ~ ~ ,--.~ ~ ~-.~ ~ ~C~a,~ ¢~D .-, -~-~-.~_

ooeq

%%
~ ~ .~'~
~ Z . ~ . - ~~ z' ,~ . ' ~ ~ . " ~ ~ ' z-~ . ~~, - ~ ~ d . . . . ~ ~ "~ ~ ~
0

o ~ m

II II II

II

0% ~ O0 P "~

II ~ II II N II II II II II c~ II II II II ~-" II II

• o-, ~ ~. ~. ~, o~ ~.

ZZZ Z
.a ~ ~ ~, ~, .~ Z ,
100 K. Yvon a n d P. Fischer

v
:2

II

It II II II ~ II II II II N II II
e~
e~

H I1 H 11 II H I[ [I II H II II

c) (.) ~ (.) r) r..) ~ ~ ~ r.)

o
"4
~ z ~ ~ ~ z
[-
 Crystal and Magnetic Structures of Ternary Metal Hydrides 101

L
¢$ ca ¢) ¢~

¢-q "---- ("q " - - - r q '~.~ r"q

.~ ~ ~~ ~
E EE EEtaEtqEEE :~ :~:= s ~

~'~ t~ ~.~ t~l

,~1- ~ " t-~

I[ II II II i II N ~J " ~ II II ('q II II N II II II

r-- m~

H H fl II II II II II II II II II

.~ . .'-:.

z :< ~ '~ S ~ S

¢)
~q
CCCC . . . . ,~.~

<

.-M

v
e~ m
® ~. g~

ea
102 K. Yvon and P. Fischer

>.,
..m

¢.<

~ ~.-a~ ~ ~ ,.,~ ~ ~'~,

..a
,,= hh

Z"
=, ?-...,

~= .2:, ~
,,r '4 ~ ~...4 .4 II II If I] II il
"-T
',£) ~,,6 ~ ,.g u4 .4

It iL II ~ tt ~ it ~ il g-,

r--r--.$ 7 ~ 7 ~ ,~"="7
]i It il '~" II ~ " If ~ II ,m.
,-a

..--,

~ "< n..., m., n.,

~"w

,m
I=
".m
@ 'q.

[- %.) >., b~ ~ ~ t,.,a ,..a

 Crystal and Magnetic Structures of Ternary Metal Hydrides 103

(mostly face- or body-centered cubic), ternary metal hydride structures show a

great variety of complex metal atom arrangements of various (usually low)
symmetries. Apart from the usual expansion and distortion (see below) most
arrangements are similar to those of the corresponding hydrogen free binary
metal compounds. Thus ternary metal hydride structures can be conveniently
classified according to the structure types of their corresponding (undistorted)
binary metal compounds as shown in the following list (structure types in
italics; representative hydrides in parentheses; same sequence as in Table 4.1):
Cubic structure types: 7i2Ni and filled Ti2Ni (also called t/-carbide) type
(TizNiHx, Ti4Fe2ODx, Zr3V3ODx); MgCu 2 type (often called cubic Laves
phase) (ZrV2Dx, HfVzD4, ZrCr2D x, ZrTil.86D(H)3.s, CeRu2D 5, ErFe2H3.5,
ThZr2D~); CaF 2 type (fl, fl'-Mg2NiD4, Mg2CoDs, Mg2FeD6); Zh6Mn23 type
(Y6Mn2aD~, Th6Mn23D~, Ho6Fe23D~); CsCl type (e-TiFeDo.os 7,
fl(fl')-TiFeDl(1.4) , y-TiFeD_2, TiFeo.gMn0.1Dx) and CraSi-type (Nb3SnH).
Hexagonal structure types: CaCu 5 type (LaNi5Ds_ 7, and substitutional
derivatives La(Nil_xB~)sD~, with B = M n , A1, Si, Cu, Co; RCosD_~ a, with
R = Pr, La, Nd, Ce; fl-CaNisD0.77), MgZn2 type (also called hexagonal Laves
phase) (ZrMn2D3, ZrMoFeD2.6, Ti(Vo.4Mno.6)~.s7D2.36, Til.2Mnl.sD3.x);
MgzNi type (c~-MgzNiDo.3); AlB 2 type (ZrBe2D1.49, ThNi2D~); PuNi3 type
(HoNi3DI.s, LaNi3D2); and PdIsP2 type (PdasP2Do.46).
Tetragonal structure types: 7-iCu(7-i)(= B1 I) type (TiCuDo.96); CuAlz type
(Th/A1Dx); MoSi z type (ZrzPdD~), and NdzFe~4 B type (YzFe14BD3.5).
Orthorhombic structure types: CrB type (ZrNiD(H)~, ZrCoD3, LaNiD3.7) ,
and Fe3C type (Pd3Po.8Dx).
Monoclinic structure types: Pd6P type (Pd6PDx).
Upon hydrogenation most binary metal compounds undergo a lattice ex-
pansion (between 2 and 3A3 per H atom [4.138]); for example LaNi s whose lat-
tice expands by up to 25 vol % at the composition LaNisD 7. The exceptionally
strong expansion of CeRu z (37% for CeRu2Ds) is due to a hydrogen induced
electronic transition as shown by XPS measurements [4.139]. A lattice
contraction upon hydrogenation has so far only been observed for ThNiz
(--2.2% for ThNizD2). For some hydride structures the space group symme-
tries at room temperature are identical to those of the binary metal compounds,
such as for cubic Zr3V3OD~, x = 1.86, 2.85, 4.93; ZrVzD~, x = 1.5, 2.8, 4.9; and
Ho6Fe23Dx, x = 1.5, 8.2, 15.7; for hexagonal ZrMnzD~.o; tetragonal ThzA1D4;
orthorhombic LaNiD3.7, and monoclinic Pd6PD~. For others these symme-
tries differ - see hydrides marked "def." (deformed) in second column of
Table4.1 - e.g. /~-TiFeD1, 7-TiFeD2, flI-PrCosD3.6, pn-NdC%Dz.s,
/3m-CeCosD2.ss, /%LaNisD 7, /%CaNisDo.77, ThNiaD ~, and ZrNiD, where
hydrogenation leads to a lowering of lattice symmetry but does not induce a
major reconstruction of the metal atom substructure. Only a few hydrides with
reconstructed metal atom arrangements, or hydrides which do not derive from
stable binary metal compounds have been reported to date (see below). For
some hydride structures the lattice symmetries depend upon hydrogen content
104 K. Yvon and P. Fischer

a-TiFeD~,o.o6 ,6'-Ti FeD~I

• "Y"~""

t11

Fig. 4.1. Projections of the crystal structure of the

a, fl, and ?-hydrides of TiFe. Large open circles Ti
at hight 1/2, small open circles Fe at height 0,
small filled circles D at 1/2, large filled circles D
at 0. Unit cell marked by dotted lines
7"-TiFeD 2

(Ho6Fez3D x, ZrV2D~, TiFeD~, LaNisD~) and upon temperature (HW2D4,

ZrV2D3.6, Y6Mn23D23, Th6Mn23D16 ). In general the symmetry decreases as a
function of hydrogen content (TiFeD x, see below) and increases as a function of
temperature ( Y 6 M n 2 3 D 2 3 , Z r V 2 D a . 6 , M g 2 N i H 4 ) .
Until now only a few systems that contain two (or more) well-defined
hydride phases have been structurally characterized. An example is the TiFe-D
system which contains a cubic (solid solution) phase of approximate compo-
sition TiFeD0.06 (called ~), one (or possibly two) orthorhombic phases of
approximate composition TiFeD ~ 1 (one called fl [4.74, 75] or fll [4.75], and
the other f12 I-4.76]), and a hydrogen-richer monoclinic [4.76-79] or prob-
ably orthorhombic phase [4.79]) of approximate composition TiFeD~2
(called 7). As shown in Fig. 4.1 the distortion of the overall metal atom
substructures increases as a function of the overall hydrogen content. Another
example is the LaNis-D system which contains a hexagonal (solid solution)
 Crystal and Magnetic Structures of Ternary Metal Hydrides 105

e-phase of approximate composition LaNisDo. 4 [4.82], a trigonal [4.32,

33, 83-85, 89] or hexagonal [4.86-881 /3-phase of approximate composition
LaNisDs. 5 and a hydrogen-rich hexagonal [4.87, 88] or orthorhombic [4.90]
phase of approximate composition LaNisD v.
For some systems (/?-LaNisD__6, 1% and v-FeTiDI_z, /~-MgzNiD4) the
space group symmetries are not yet fully established. For/?-LaNisD 6 five differ-
ent space groups have been proposed (P6/mmm [4.86], P63rnc [4.87, 88], P31m
[4.83--85,891, P321 [4.53], and Cmmm [4.1,90]). Of these only one (Cmmm)was
capable of explaining the observed anisotropic line broadening, but was judged
unlikely from refinements on integrated intensities [4.87]. The other space
groups required the assumption of microstrains [4.86, 87] or particle size effects
[4.881. A weak diffraction line previously not included in structure refinements
was interpreted [4.87, 88] by doubling the hexagonal cell along e, and by
assuming the (non-centrosymmetric) space group P63mc. For /%FeTiD x
various orthorhombic space groups have been proposed [4.74-76]. For
y-FeTiDz the published monoclinic structure [4.76-78] has been described in
terms of an orthorhombic space group with fewer parameters [4.79]. The
structure of the hydrogen storage material ~q-Mg2NiD4 at room temperature
was originally described as tetragonal [4.118]. Later it was described [4.122,
140, 141] as orthorhombic (space groups P21212 [4.122] or P2221 [4.140]) and
monoclinic (space groups C2/m [4.621, la (or Cc) [4.63, 121, 1221, or Cm
[4.120]. Recently an x-ray structure refinement which took into account
anisotropic line broadening due to microtwinning has suggested [4.123] that
its structure is monoclinic with space group C2/c. Based on this model,
hydrogen atom positions were later successfully derived from high-resolution
neutron diffraction data [4.64]. The presence of microtwinning in this
compound has been confirmed by electron diffraction [D. Nor6us, private
communication].
An example of a well-characterized hydride structure in which hydroge-
nation induces significant shifts of the metal atoms while the lattice expands
isotropically is cubic Zr3V3OD x [4.47]. Reconstructed metal atom substruc-
tures were found for the following hydrides: cubic ZrTil.86D3.s which derives
from a hexagonal close-packed alloy; orthorhombic ZrCoD 3 which derives
from a cubic compound with CsC1 structure, and monoclinic Mg2NiH 4 which
derives from a hexagonal compound with complex structure. On the other
hand tetragonal Mg2CoH5 and cubic Mg2FeH6 do not derive from stable
intermetallic compounds. They were included in this review because of their
close structural and bonding relationships to Mg2NiH 4, and because they have
provided insight into the factors which limit maximum hydrogen content in this
type of compound (Sect. 4.4.3). Diffraction evidence for reconstructed metal
atom arrangements also exists for other metal hydrides (for a recent list see
[4.42]). An interesting example is TiCrl.sHs.3 for which a disordered
CaF2-type metal host structure was proposed [4.142] while the hydrogen-free
alloy has a MgCu2-type structure. However this compound has not been fully
characterized.
106 K. Yvon and P. Fischer

4.3.2 Distribution of Hydrogen Atoms

In contrast to elemental hydride structures which are often ordered, ternary

metal hydride structures are usually disordered at room temperature. Conse-
quently their local hydrogen atom configurations around the metal atoms are
unknown. Examples of compounds with ordered hydrogen atom distributions
at room temperature and thus known hydrogen configurations are MgzFeD6,
Mg2CoD 5, and Mg2NiD4, as shown in Fig. 4.2. In the first two of these the
transition metal atoms are surrounded respectively by octahedral (as in
S r 2 R u D 6 [,4.22]), and square-pyramidal (as in Sr2IrD 5 [,4.143]), hydrogen atom
configurations. In Mg2NiD 4 they are surrounded by a tetrahedral configu-
ration, as shown from analysis of high-resolution neutron diffraction data
[4.64], in contrast to the square-planar [-4.62] and disordered [-4.63] configu-
rations reported previously from analysis of medium- and low-resolution data.
Examples of compounds with partially or nearly ordered hydrogen atom
distributions at room temperature are TiCuDo.96, LaNiD3.7, ZrNiD,
ZrNiH2.98 , ThzA1D4, y-TiFeD2, and ZrVzD3. 6.
One of the most characteristic and intriguing structural features of ternary
metal hydrides are their preferential hydrogen site occupancies. As can be seen
from Table 4.1, the hydrogen atoms exclusively occupy those interstices of the
metal atom network which allow them to be in contact with at least one hydride
forming element such as Mg, Ti, Zr, Hf, V, Ni, Pd, Th, Y, or rare-earths.
Moreover, in structures which contain several types of interstices those
surrounded by a larger fraction of hydride-forming elements, or those sur-
rounded by an element forming a more stable binary hydride, appear to be
preferentially occupied. For example in the ZrVzD~ system the [Zr2V2] holes
are populated at lower overall deuterium concentrations than the [-ZrV3] holes,
and in the hydrides of TiFe the [,Ti4Fe2] holes are filled at lower overall
hydrogen concentrations than the [,TizFe4] holes. In the ZraVaOD~ system
most of the hydrogen atoms first fill the [,Zr3V ] and [,ZrzV2] holes, while the

0,f,

Mg2FeD6 Mg2CoD5 Mg2NiD4

Fig. 4.2. Transition metal environments in Mg2FeD6, Mg2CoDs, and Mg2NiD 4 (from [4.64]).
Large circles Mg, small circles D
 Crystal and Magnetic Structures of Ternary Metal Hydrides 107

[ZrV3] and [V4] interstices mostly remain empty even at high hydrogen contents.
In LaNisD x most of the hydrogen atoms occupy the [La2Ni2] (or [La2Ni4])
and [LaNi3] holes, while the [Ni4] holes are only filled at high hydrogen
concentrations. Preferential hydrogen site occupancy was also observed with
amorphous alloys such as ZrNiDI. 8 [4.9] for which neutron diffraction analysis
has shown occupation of tetrahedral [Zr4] holes. The above trends correlate
with macroscopic sorption properties and constitute the main basis for a model
[4.115] which was used to rationalize hydrogen site occupancies and relative
stabilities in terms of attractive nearest-neighbour interactions (Sect. 4.4.4).
According to another model [4.114] the hydrogen site occupancies can also be
interpreted in geometrical terms by assuming that hydrogen prefers to occupy
large interstices and avoids small interstices, and by correlation effects between
hydrogen atoms (Sect. 4.4.1).
The metal atoms surrounding the hydrogen atom sites form polyhedra of
various shapes and symmetries (for number and type of metal atoms, and point
group symmetries see last column in Table 4.1). The polyhedra which occur
most frequently are deformed tetrahedra ("tetrahedral sites") and octahedra
("octahedral sites"). The hydrogen sites within these polyhedra usually have low
symmetry (mostly 2, m or 1).
Tetrahedral sites are occupied in hydrides belonging to the following
structure types (a representative compound and the nature of metal atom
environment are given in parentheses): Ti2Ni (Ti2NiH, [Ni4] ) and filled
Ti2Ni(Zr3V3ODx, [Zr3V], [Zr2V2], [ZrV3]); MgCu2(ZrV2Dx, [Zr2V2],
[ZrV3]); Th6Mn23(Y6Mnz3Dx, [Y3Mn], [Y2Mn2], [YMn3]);
Cr3Si(NbaSnH~, [Nb4]); CaCus(fl-LaNisD~, [La2Ni2], [LaNi3], [Ni4]);
MgZnz(ZrMnzD3, [ZrzMn2]); PuNia(HoNi3Dx.8, [HoNi3], [HozNiz]);
CuAI2(Th2A1D ~ [Th4]); MoSi2(Zr2PdD ~, [Zr4]); TiCu(Ti) (TiCuH~ [Ti4]);
Nd2Fe14B (Y2Fel4BD~, [Y3Fe], [Y2Fe2]) and CrB(ZrNiH 3, [Zr3Ni]).
Octahedral sites are occupied in: TiENi(TizNiH , [Ti6]);
Th6Mnz3(Y6Mn23Dx, [Y6]); CsCl(fl(fl')-FeTiD~, [Ti4Fe2]; y-FeTiD2,
[TigFe2], [Ti2Fe4]); CaCu5 (LaNi5D~, [La2Ni4]); and Pd6P(Pd6PDx, [Pd6]).
Square-pyramidal sites are occupied in: CaF2(Mg2NiD4, [Mg4Ni];
Mg2CoD 5, [Mg4Co]; MgEFeD 6, [MggFe]); F%C(Pd3Po.sDx, [Pds]); and
Pd6P(Pd6PD~ [Pds]).
Triangular, or triangular bi-pyramidal sites are occupied in the
hydrides Th6Mn/3D16([ThEMn3]), LaNisDa([LaNi/], according to [4.89]),
ZrBezD1.49 ([Zr3Be2]), and ZrNiH2.98([Zr3Ni2]), whereas
linear hydrogen coordination was found in LaNiD3.7([Ni2]) and
°~-Mg2NiHo.3([Mg2]) •
In the discussed ternary metal hydrides no example of double occupancy
of a hole such as that found [4.14-] in the halide ZrBrD exists as yet. Due to a
lack of resolution, the exact locations of the hydrogen atoms within the co-
ordination polyhedra are not always known. A typical example is LaNisD 6 in
which the deuterium atoms occupying the octahedral [La2Ni4] interstices
appear to be displaced from the centre in various directions, depending on the
108 K. Yvon and P. Fischer

2-sile mode[ .

t_a --I
P31m P521
D1 (3c) (6g I)
Dz(6d) (6g =)

5 - s i l e model /~
!
P61mmm
• D1 ( 3 f )
x Ds(12n)
o D3(6m)
u D4 (4h)
• D 5 (12o)

7 - s i t e model_

P63m c
o O1 (2b~)
¢ • D2(2b 2)
• D316c.I)
o D,~(6%)
A D~ (6%)
A D6(6c4 ]
x D7(12d }

(110) pLane (010) ptane

Fig. 4.3. Deuterium atom sites and space group symmetries of various multi-site models
proposed for the structureof fl-LaNisDx (adapted from [-4.87]; Wyckoff positions [4.44] in
parentheses)

model used for structure refinement (Fig. 4.3). Conflicting results with respect
to the deuterium atom displacements and site occupancies have also been
reported for other systems such as Y6M%3D23 [4.67-69], /~-TiFeD~
[4.74-76], and ZrV/D(H)3.6 [4.49 51]. Some of these ambiguities could be due
to anharmonic thermal motion of the H atoms (Y6M%3D23,/%TiFeDI) or
a reduction of cell symmetry (ZrVzH3.6) due to hydrogen atom ordering. An
interesting example of an ambiguous hydrogen atom distribution is the
disordered fl'-MgzNiD 4 modification for which two models with respectively
square-pyramidal [MggNi ] [4.59, 60] and square-planar [Mg2Ni2] I-4.61]
hydrogen site coordination have been reported which both allow an equally
good fit to the diffraction data. The latter model was considered less likely
I-4.144] in view of its unusually short magnesium-hydrogen distances (Sect.
4.3.3), and electronic properties [4.145].
In disordered metal hydrides the hydrogen atoms undergo large amplitude
thermal vibrations (rms. values of up to 0.1 A at room temperature, see B values
 Crystal and Magnetic Structures of Ternary Metal Hydrides 109

in Appendix4.A). Residual nuclear density in Fourier difference maps

(LaNisDx [4.35]) and temperature dependent site occupancy factors (ZrV2D3.6
[4.49]) suggest that these vibrations are strongly anharmonic. Anharmonicity
is presumably one of the factors responsible for the systematic difference
between the hydrogen content found by diffraction methods and that measured
by desorption (see remarks in Appendix 4.A).
In some compounds structural phase transitions occur below room
temperature (see TI values given in Table 4.1). These transitions are generally
due to ordering of the hydrogen atoms which leads to a reduction of lattice
symmetry. Examples are ZrV2D3.6, HfVzD 4, Y6Mnz3D23, and Th6MnzaD ~6.
Compounds with ordered hydrogen atom distributions at room-temperature
and disordered distributions at high temperatures are fl-Mg2NiD 4 and
Mg2CoD 5. Their high temperature structures resembles those of
Mg2FeDa(K2PtC16 type), and K2PtD4 [4.23].

4.3.3 Interatomic Distances

In view of relatively low precision (typically _0.02A, see Sect. 4.2.3) the
interatomic distances of the ternary metal hydride structures characterized to
date reveal only qualitative trends.

Metal-Metal Distances. At the hydrogen-rich phase limit these distances are

longer by about 0.1 A., on the average, than those in the corresponding binary
metal compound. This suggests that the metal-metal bonds are weakened
during hydride formation. Examples of the reverse behavior are found in
Zr3V3OD1.38, ZraPdDt.98, and Y6Mnz3Dx, in which some metal-metal
distances (V-V, Zr-Zr, and Mn-Mn, respectively) are shortened by about 0.2 A
with respect to those in the hydrogen-free compounds.

Metal-Hydrogen Distances. These correspond closely to those in the corre-

sponding elemental hydrides, except that they may be anomalously short-
ened due to partial occupancy of the hydrogen atom sites. For example some
Zr-D distances in Zr3V30Dl.s6([Zr-D]=l.94A ) are shorter than
those in binary ZrDz([Zr-D 1=2.09A), while some metal-hydrogen dis-
tances in fl'-Mg2NiD,([Ni D] = I .49(3) A) according to [4.59] and
~-FeTiDo.o57([Fe-D]=l.49/~) are close to the sum of covalent radii
([Ni-H] = 1.47 A, [Fe-H] = 1.49 A). Such short distances may not be charac-
teristic for all hydrogen atom sites belonging to a particular crystallographic
location because of possible local distortions in their metal atom environments,
as was suggested [4.146] for ~-TiFeDo.os7, and found [4.147] in binary
transition metal hydrides (for example in NbHx). On the other hand, the
unusually short metal-hydrogen distances reported in an early structure
model of LaNisDT([Ni-D]=I.17A [4.83]), and those recently re-
ported for disordered fl'-Mg2NiH4([Mg-D ] = 1.66 A_ [4.61]) are presumably
a consequence of limited data sets (LaNisDT) and deficiencies of the structure
11 o K. Yvon and P. Fischer

models (fl'-MgzNiD4) used for structure refinement. The Ni-D dis-

tance in the former compound is smaller than the atomic radius of
the metal atom (r[Ni] = 1.24/~) whereas the Mg-D distance reported for
the latter compound is significantly shorter than that in binary
MgDz([Mg-D ]=1.95A), and that reported [4.59] previously for
fl'-Mg2NiH4([Mg-D]=2.31 A) by using a different structure model. The
conflicting results obtained for fl'-MgzNiD4 from diffraction data of similar
resolution were analyzed in [4.144] and found to be mainly due to the
particular ratio between the neutron scattering lengths of the metal nuclei
(bNi-2bMg). Short metal-hydrogen distances in ordered (or nearly ordered)
hydrides occur in ZrNiHz.98([Zr-H ] = 1.95 A), MgzCoDs([Co-D ] = 1.52/~)
and Mg2FeD6([Fe-D] = 1.56 •). The distances in the two latter hydrides, and
also those in FeTiDo.057, are of particular interest because they involve metal
atoms which do not belong to elements forming stable hydrides (Sect. 4.4.4).
Most metal-hydrogen distances in ternary metal hydride structures can be
rationalized by 12-fold coordinated metal atom radii [4.148] and a fixed H
atom radius of 0.40A [4.114]. However, descriptions in terms of normal-
valence cation radii (or crystal radii [4.149]) and fixed H atom radii of 1.1 A
[4.150] (or 1.27/~ [4.24]) such as those used [4.24, 150] to rationalize metal-
hydrogen distances in binary metal hydrides presumably fit these distances
equally well.

Hydrogen-Hydrogen Distances. A characteristic feature of ternary metal

hydride structures is their minimum hydrogen atom separation. In ordered
hydrides this separation was never found to be smaller than 2.1 A, a limit
already noticed [4.151] in binary metal hydrides. Apparent exceptions are
Th2A1D4([D-D]=1.79A according to [4.1053, see also [4.31]), and
fl-Mg2NiD4([D-D] = 1.75 A, according to [4.62]) which could be due to the
low resolution of the reported diffraction data (Th2A1D4) and the inadequacy of
the model used for structure refinement (fl-Mg2NiD4). Well-documented
examples of hydrogen-hydrogen distances shorter than 2.1 A have so far only
been reported for (non-metallic) complex hydrides such as KzReH9([D-D]
= 1.87(2)A [4.12]).
In disordered hydrides partially occupied hydrogen atom sites are almost
always separated by less than 2.1 ]~. Typical examples are cubic ZrVzD 5 and
hexagonal ZrMnzD3 in which neighboring D atom sites are separated by about
1.3 ,~. As shown in Fig. 4.4 these sites are connected to form quasi-infinite three-
dimensional networks, thus providing possible diffusion paths through the
structures. Since the occupancies of the hydrogen atom sites are usually lower
than 50% (see Appendix 4.A), a minimum hydrogen atom separation of about
2.1 ]k was assumed [4.52, 114] to be effective also in these and other disordered
hydrides. A compound showing hydrogen-hydrogen contacts which are
possibly shorter than 2.1 ,h is Y6Mnz3D23for which full occupancy (without
e.s.d) of a site (32f3) which is only 1.3/~ away from another partially occupied
site (96kl) was reported in one study [4.67], and 62% occupancy in another
 Crystal and Magnetic Structures of Ternary Metal Hydrides 111

ZrMn2D 3 ZrV2 D5

?
Fig. 4.4. Distribution of hydrogen atom sites in hexagonal ZrMn2D3 and cubic ZrV2D5 (from
[4.1]). Large circles Zr, small filled circles and squares D atom sites. The V and Mn atoms are
omitted. Solid lines indicate possible diffusion paths in the structure

study [4.69]. Experimental support for the assumption of a minimum

separation between hydrogen atoms in disordered hydrides came mainly from
the observation of diffuse intensity peaks close to 2/(2 sin0) = 2 A in the neutron
diffraction patterns of hydrides with MgCu2 type metal atom substructure such
as HfVzD 4 [4.52], ZrV2Dx [4.48, 50], ZrCr2D3. 8 [4.51, 54], CeRu~D 5 [4.56],
the disordered modifications of Mg2NiH4 [4.59, 61] and Mg2CoD5 [4.65],
and of Tit.2Mna.sDa. 1 [4.98]. No such peaks were found in the patterns of the
ordered low-temperature phases of HfV2D, [4.52], ZrV2D3. 6 [4.49], and
ZrVzH3. 7 [4.51]. In these last hydrides, the shortest separations between
occupied hydrogen atom sites are about 2.1 A, compared to 1.3/k found in the
corresponding disordered modifications. Diffuse neutron diffraction intensity
peaks were also reported for ThNi2Dx [4.102].

Hydrogen-Non-Metal Distances. Except for Ti4Fe2OD2.25([O-D]=l.68Jk

[4.463) and ZrOo.4DoA ([O-D] not specified in [4.213 but presumably smaller
than 2.0A) these distances are larger than 2.0A (examples:
Zr3VaODx: [ O - D ] =2.01 A, Pd6PDx: [ P - D ] =2.94A, PdaPo.sDo.15 : [ P - D ]
= 3.51 A). For the last compound, an unusually small hydrogen-non-metal
distance is found 0-P-D] = 0.58 A_)which was attributed [4.111] to the presence
of defects on both deuterium and phosphorus sites.
112 K. Yvon and P. Fischer

4.4 Stability Aspects

Many attempts have been made to rationalize the stabilities and maximum
hydrogen contents of ternary metal hydrides (see Chap. 6). Electronic band-
structure calculations have so far only been performed [4.145, 152-155] for
hydrides having simple and highly symmetric crystal structures such as
cubic TiFeDx, ff-Mg2NiD4, MgzFeD6, ZrVEDx, and tetragonal MgzCoD 5
[see Chap. 5]. While these calculations were most valuable for the
understanding of electronic properties, their level of sophistication did
not allow heats of formation, relative site stabilities, or maximum hydrogen
contents to be derived. The results for TiFeDx suggest [4.152] that its
stability is mainly controlled by a lowering of empty metal states by the
hydrogen potential, similar to the behavior suggested [4.151] for elemental
metal hydrides. The results on ZrV2D x were claimed [4.155] to be consistent
with the experimentally observed hydrogen site occupation (mainly [ZrzV2]
interstices). The calculations on fl'-Mg2NiD 4 were performed [4.145] on two
disordered structure models of which one [4.59, 60] assumed octahedral and
the other [-4.61] linear surroundings of nickel atoms by hydrogen. The results
and comparison with XPS spectra [4.145] favor the former model However in
view of recent structure work [-4.64] on the ordered modification of
fl-Mg2NiD 4, new calculations based on the actual hydrogen atom distribution
(tetrahedral configuration around nickel) and a reinterpretation of the XPS
spectra are desirable. The calculations performed [4.154] on MgzFeD 6 predict
semiconducting behavior and confirm that the previously postulated band-
filling effect [4.59, 66] is an important factor limiting hydrogen capacity.
Various semi-empirical attempts to rationalize enthalpies of hydride
formation and maximum hydrogen contents have been reported. Some are
based on the inverse relationship found [4.136] between the thermal stability of
metal hydrides and the heats of formation of the corresponding binary metal
compounds ("rule of reversed stability"), while others are based on correlations
between the heats of formation of ternary metal hydrides and the stabilities of
the corresponding elemental hydrides [4.156], or some suitable electronic
parameters of the hydrogen-free metal host structure [4.157]. The influence of
entropy effects have been discussed [4.158] for LaNisDx and isostructural
hydrides [4.159], while the relative importance of geometric and electronic
contributions to the thermodynamic properties have been evaluated [4.160] for
body-centered-cubic metal hydrides. Correlations were pointed out between
atomic volume contraction during binary metal compound formation and the
stability [4.161] and maximum hydrogen content [4.162] of the corresponding
ternary metal hydride. A model for calculating enthalpies of hydride formation
has been proposed [4.163] which takes into account both electronic and
geometrical factors. For some systems, the maximum hydrogen content can be
rationalized by decomposing the ternary metal hydride into elemental
hydrides, such as [4.37]:
 Crystal and Magnetic Structures of Ternary Metal Hydrides 113

LaNisHs. 5 = LaH 3 + 5NiHo.5,

or by decomposing the ternary metal hydride into hydrides of structurally

related binary metal compounds. For example [4.164]:

A B 3H x + y(PuNi3type) = ½lABs] H~(CaCu5 type) + ~ [AB2]Hy(MgZn2type).

A multiplateau theory with consideration of hydrogen interactions has been

proposed [4.165] which allows experimental isotherms to be fitted.
Apart from a few exceptions the above methods are all structure indepen-
dent. Their success is presumably due to the relatively small contribution of
structure-dependent terms to the total energy of hydride crystals, as confirmed
by theoretical [4.166] and semi-empirical [4.163, 167] considerations, and by
the crystal chemistry of binary metal compounds [4.168]. Consequently these
methods fail to rationalize (or predict) structure-dependent features such as
relative stabilities of potentially occupied hydrogen atom sites and maximum
hydrogen content. Attempts to rationalize these two features have mainly been
based on empirical considerations, such as the availability of interstitial holes of
sufficient size, electrostatic repulsion acting between hydrogen atoms or
between hydrogen and non-metal atoms, electronic factors such as band filling,
and attractive nearest-neighbor interactions between hydrogen and metal
atoms. These attempts are summarized in the next section.

4.4.1 Interstitial Hole Size

Correlations between hydrogen equilibrium pressures and intersticial hole size

in AB, AB2, and AB5 compounds [4.169] and cell volume in AB5 compounds
[4.170] suggest that hydrogen prefers to occupy large interstices. Hole sizes
were defined [4.114, 171] by the largest sphere which can be inserted into the
holes. The radii of the metal atom spheres were either derived [4.171] self-
consistently by assuming that these spheres touch each other, or taken [4.114]
from a tabulated set of 12-fold coordinated metal atom radii [4.148 and 4.168,
p. 151]. Only holes for which the radius of the inscribed sphere exceeded a
threshold value of about 0.40 A were found to be occupied [4.114]. This value of
the radius constituted one of the two major parameters of the so-called
"geometrical model" [4.114] which was used to rationalize hydrogen site
occupancies and maximum hydrogen contents (Sect. 4.4.2). Its application
showed that the requirement of minimum hole size was fulfilled for nearly all
metal hydride structures studied so far by neutron diffraction analysis (for
example: Zr3V3OD x [4.172], TiFeD x [4.146], ZrBezD x [4.1731 Y6Mn23Dx
[4.174], Sr/RuD 6 [4.175], LaNisD x [4.176], Pd3P1 _,,Hr [4.177], ZrVzDx and
ZrMn2D~ [4.138], TizNiDx [4.178]). A compound for which this requirement
was not fulfilled was fl'-Mg2NiH4, suggesting [4.175] a possible deficiency in its
114 K. Yvon and P. Fischer

structure model. For hydrides of Th6Mn23-type alloys, the occupancy of a

particular site (site 96jl, see Appendix 4.A) has been investigated [4.179] along
the lines of this model and found to be critical for the tetragonal lattice
distortion. However, there also exist hydride structures containing holes of
sufficient size which are nonetheless not occupied by hydrogen, such as some of
the [Mg4Co ] holes in Mg2CoDs, some of the [Mg4Ni ] holes in//-Mg2NiD ,,
the [Zr3Pd ] and [Zr2Pd2] holes in Zr2PdDl.7o, the [M6] holes ( M = Th, Ho)
in M6Mnz3D16 and Ho6FezsD~ (x = 12.1, 15.7), the triangular-prismatic [Pd6]
holes in Pd3P l_x and Pd6P, and some of the tetrahedral [ Z r z M 2 ] holes
(M = V, Mn, Ti) in disordered ZrM2D3, ordered ZrVzD3. 6 [P. Zolliker, private
communication], and ZrTil.stDs. 8. For some compounds, repulsive interac-
tions between the D atoms are presumably responsible for the non-occupancy
of these holes (Sect. 4.4.2). For others (such as Pd3P l_~Hy), repulsive interac-
tions between hydrogen and non-metal atoms have been invoked [4.177].
There also exist compounds which absorb little hydrogen under normal
conditions, such as LaPt 5 and ThNi 5 (both CaCu 5 type [4.180]) whose metal
atom substructures presumably contain holes which are sufficiently large to be
filled by hydrogen atoms. In these compounds other factors apart from
interstitial volume appear to govern hydrogen site occupancy. Thus the
requirement of minimum hole size may be a neccessary but not a sufficient
criterion for hydrogen occupancy, and the maximum hydrogen content
calculated by purely geometrical considerations may be higher than that
observed (see below).

4.4.2 Electrostatic Repulsion

The observation of a minimum hydrogen atom separation of about 2.1 A has

been interpreted in terms of electronic effects [4.151], electrostatic repulsion
acting between the hydrogen atoms [4.50, 114], blocking effects [4.181,182],
or antibonding effects [4.150]. This separation corresponds to about twice the
pseudo-fluorine-like hydrogen radius of 1.1 A used [4.150] to rationalize
metal-hydrogen distances in elemental hydrides, and is consistent with anion-
like behavior of hydrogen. Anion-like behavior of hydrogen in ternary metal
hydrides was also suggested from a study of their magnetic properties [4.1833.
The existence of a minimum hydrogen atom separation in metal hydride
structures correlates with empirical considerations which suggest [4.137] that
the maximum hydrogen content corresponds to similar volume concen-
trations of hydrogen. The 2.1 A separation constituted the second major
parameter of the geometrical model which was used [4.114] to rationalize
hydrogen site occupancies and hydrogen capacities. According to this model,
the maximum hydrogen content of a particular metal structure corresponds to
the maximum number of spheres which can be packed into interstitial space
without violating the conditions of minimum sphere size (0.4 A) and sphere
separation (2.1 A), regardless of the nature of the metal atoms. Satisfactory
 Crystal and Magnetic Structures of Ternary Metal Hydrides 115

agreement with observed hydrogen site occupancies and maximum hydrogen

contents was observed for most systems investigated by this model, e.g.
ZrVzD 6 and ZrMnzD2.75 [4.138], LaNisD6 [,,4.31, 176], Zr3V3ODx [4.172],
YrMn2aD2a [4.174], PdaPl_xHy [4.177], ZrBe2D1. 5 [-4.173], Ti2NiD x, and
Ti4Fe2ODx [4.178]. In contrast, the agreement is less satisfactory for com-
pounds for which such an analysis has not yet appeared in the literature, such
as Mg/CoD 5 and /3-Mg2NiD 4. For both structures the model predicts a
maximum hydrogen content of six atoms per formula unit [P. Zolliker, 1985;
private communication]. Furthermore, predictions arising from this model
depend on the availability and accuracy of experimentally determined lattice
parameters and metal atom positions [4.178]. Thus a priori estimates of
maximum hydrogen contents based on this model are not always possible, as
is the case for LaCosD~(x=9 at 1500atm. [4.184]), for which structural
information at high deuterium pressure is not yet available. The model also
cannot predict the onset of structural phase transitions due to ordering of the
hydrogen atoms (ZrV2D~), and the occurrence of lattice distortions, as was
discussed [4.179] for Y6Mn2aD~ and related deuterides. Successful attempts
to rationalize maximum hydrogen content by the related exclusion principle
[4.181] were reported [4.67] for hydrides containing mainly tetrahedral
interstices such as those having the MgCu2, MgZn2, and Th6Mn23 type metal
host structures. The previously proposed "blocking model" [4.182] has been
mainly applied to binary hydride structures for which it allows the calculation
of configurational entropies of interstitial atoms. Evidence for repulsive
interactions between hydrogen and other non-metallic p-elements such as
oxygen and phosphorus have been reported [4.31,172, 177] for Zr3V3OD~,
Pd3Po.sDo.15, Pd6PD ~, and other compounds not covered by this review such
as hydride-oxides (Ti2OH~ [4.186]) or hydride-carbides (ZrCl _xDy [4.187]),
which were discussed in a previous review [,4.31]. Short oxygen-deuterium
distances have to date only been reported for Ti4FeaOD2.25([O-D ] = 1.68/~,
[4.46]) and ZrOo.aDo.x with hcp host structure ([O-D] not stated, D atoms
filling tetrahedral holes next to octahedral holes occupied by O atoms, [4.21]).
Repulsive interactions between hydrogen and metallic p-elements such as A1
and Sn have been found to influence the hydrogen content in the substi-
tutional series La(Ni, A1)sD ~ [4.31,170, 185], and in Nb3SnH ~ [4.31, 81]. On
the other hand the adverse influence of A1 on the maximum hydrogen content
in the former series was interpreted in geometrical terms by invoking [4.176]
the larger atomic size of A1 compared to that of Ni, and in terms of electronic
factors by invoking a non-availability of d-band bonding [4.180] (see below).

4.4.3 Electronic Factors

The particular bonding situation in non-metallic hydrides such as MgzNiH4,

Mg2CoHs, and MgzFeH 6 allows their maximum hydrogen contents to be
rationalized in terms of purely electronic factors. Their crystal structures
116 K. Yoon and P. Fischer

(Fig. 4.2) are considered [4.59, 65, 66] to be formally built by divalent Mg 2÷
cations and tetravalent complex anions [NiH4] 4-, [Coils] 4-, and IF'ell6] 4-,
respectively. Since the latter anions all satisfy the 18-electron rule [4.188] the
hydrogen content of the above compounds was assumed to be determined by
band filling effects [4.59]. Subsequent bandstructure calculations [145, 154]
confirm this view, but also stress the partially covalent nature of the magnesium-
hydrogen interaction I-4.145]. The usefulness of this approach in predicting
new ternary metal hydride systems was initially demonstrated by the discov-
ery of MgzFeH 6 and MgCoH 5 which were synthesized [4.65, 66] several years
after their possible existence was suggested [4.59]. Mg2FeH 6 is of particular
interest because of its unusually high volume efficiency for hydrogen storage
(more than twice that of liquid hydrogen) and its low material cost. For
Mg2NiH 4 predictions from the 18-electron rule prompted a re-investigation
of its low-temperature structure for which an unusual square-planar configu-
ration of Ni by hydrogen atoms had been proposed [4.62]. A high-resolution
neutron diffraction study recently confirmed [4.64] that the Ni atoms had the
expected [4.59, 188] tetrahedral configuration (Fig. 4.2).
Band-filling arguments were used to rationalize [4.180, 189] maximum
hydrogen contents in Ni-based CaCus type hydrides and electronic factors
have been invoked [4.190] to explain the variation of hydride stability in the
substitutional series Zr(V1 _xCrx)2Hr.

4.4.4 Attractive Nearest-Neighbor Interactions

Preferential hydrogen occupancy of metal atom interstices that are surrounded

by hydride forming elements have been interpreted in terms of a model [4.115]
that assumes the existence of attractive interactions between hydrogen and its
nearest-metal neighbors ("imaginary binary hydride model" [4.191,192]).
Relative stabilities of potentially occupied hydrogen atom sites are estimated by
calculating local formation enthalpies of suitably partitioned imaginary binary
hydrides, and by assuming a Fermi or Boltzmann type energy distribution
[4.116, 194]. The calculations are based on available structure data and
empirical constants given by M i e d e m a [4.167] which have previously been
used [4.137] to rationalize the stability of binary metal hydrides. Justification
for this approach comes mainly from empirical considerations such as the
crystal chemistry of intermetallic compounds, and less from theoretical
calculations [4.155]. Good qualitative agreement between observed and
calculated hydrogen site occupancies has been observed for MgCu2-, and
MgZn2-type hydrides [4.48, 191, 192] (for a review of the results on these
compounds see [4.193]), and also for Th6Mnz3-type hydrides [4.192] (except
for Y6Mn23D23 [4.69]), for LaNi5Dx [4.159] and for ZrNiDx [4.192, 194].
Quantitative agreement was observed [4.48] for ZrV2, which could be
fortuitous [4.114, 195]. Generally speaking, the imaginary binary hydride
model succeeds in predicting relative site occupancies for all those compounds
and interstices which are also successfully rationalized by the geometrical
 Crystal and Magnetic Structures of Ternary Metal Hydrides 117

model. On the other hand, it failed for those interstices which could not be
rationalized by the geometrical model such as the empty [M6] holes (M = Th,
Ho] in Th6Mn23D~6 and HorFe23D x (x=12.1, 15.7) and some empty
tetrahedral [Zr2M2] holes (M = V, Mn) in disordered ZrMn2D3 and ordered
ZrV2D3. 6. It is also unable to explain why some compounds such as ZrPd2,
LaPt5, and ThNi 5 are bad hydrogen absorbers [I. Jacob, 1986; private
communication]. This parallel behaviour was suggested [4.1] to be a conse-
quence of a correlation which exists between atomic size and the formation
enthalpies of binary metal hydrides, i.e. elements with large atomic size tend to
have larger enthalpies of hydride formation than those with small atomic size.
In contrast to the geometrical model, the imaginary binary hydride model did
not allow rationalization of the maximum hydrogen content. For ZrV2Dx this
failure has been attributed [4.48] to the neglect of geometrical restrictions for
hydrogen site occupancies, i.e. minimum hydrogen separation due to repulsive
interactions, entropy effects [4.1963 similar to those investigated in
/~-LaNisD6. 5 [4.158], and the weakening of metal-metal bonds due to lattice
expansions. The last of these effects appears to be the structural basis for "the
rule of reversed stability" [4.1363 which suggests that the weakening of the
metal-metal bonds occurs at the expense of metal-hydrogen bonds.

4.5 Magnetic Ordering

The influence of hydrogen absorption on the magnetic properties of interme-
tallic compounds is reviewed in Chap. 7 and previously in [4.42]. Typical
hydrogen storage systems such as TiFe and LaNi 5 as well as their hydrides do
not show long-range magnetic order at low temperatures (e. g. TiFeHx [4.76]).
Those ternary metal hydrides for which long-range magnetic order was found
from neutron diffraction experiments are summarized in Appendix 4.B. The
list gives information on the magnetic structures including ordered magnetic
sublattice moments and the corresponding ferro- (or ferri-) and antiferro-
magnetic ordering temperatures, Tc and TN, respectively. Usually these
temperatures agree well with those determined from bulk magnetic measure-
ments. The most extensive neutron diffraction work concerns cubic Laves
phase compounds such as RFe2H(D)x ( R = H o , Er, Tm) [4.57], and
A6B23H(D)x systems (A = Y, Th, Ho, Er, B = Mn, Fe; for a review see [4.197]).
In RCosH(D)x systems the magnetic moment of cobalt decreases as the hydro-
gen content is increased [4.42, 92]. In ferrimagnetic ErFe2(H , D) with Curie
temperature Tc=574K the magnetic moments of the Er atoms at low
temperatures show nearly their full free-ion value of 9#B and they are coupled
antiparallel to the moments of the Fe atoms of magnitude 1.6#B. Upon
deuteration to ErFe2D3. 5 the Curie temperature decreases to 440 K. While the
Fe moment remains unchanged upon deuteration the Er moment at satu-
ration is reduced to 4.3/~8. It vanishes at 295 K as the temperature is increased,
presumably due to the weaker exchange interactions between Er ions
compared to those between the Fe ions [4.57]. This system also shows a small
118 K. Yvon and P. Fischer

isotope effect (ErFe2H3. 5: Tc = 4 5 0 K). More dramatic changes of magnetic

properties due to hydrogenation were found for A6M23 systems. For example
the Pauli paramagnet Th6Mn23 changes on deuteration to a ferrimagnet with
a Curie temperature of Tc=329 K at the composition Th6Mnz3D30 [-4.71].
The ferrimagnetic compound Y6Mn23 (Tc = 486 K) changes on deuteration to
an antiferromagnetic compound with TN=175K at the composition
Y6Mn23D23 [4.69]. Of particular interest is also the superconducting com-
pound CeRu2 exhibiting short-range ferromagnetic fluctuations. It shows an
electronic transition (often considered as a valence change between Ce 4 + and
Ce 3+) upon hydriding. Cubic CeRu2D 4 is non-superconducting and orders
antiferromagnetically below TN,~ 1.4 K [4.56].

4.6 Conclusions
The structures of ternary metal hydrides differ greatly from those of elemental
metal hydrides in terms of their diversity and complexity. Their lattices are
considerably expanded and often distorted with respect to those of the
corresponding binary metal compounds but their overall metal atom arrange-
ments do not change much as a function of hydrogenation. The lattice
expansion usually (but not always) leads to a weakening of the metal-metal
bonds at the expense of metal-hydrogen bonds. So far, no evidence exists for the
formation of hydrogen-hydrogen bonds. The hydrogen atoms occupy predomi-
nantly tetrahedral and octahedral interstices in the metal atom network, are
mobile at room temperature, and are usually disordered over several different
sites which are connected to form three-dimensional diffusion paths through
the structures. The number of available interstices in the structures generally
greatly exceeds the number of hydrogen atoms absorbed. At low temperatures
the hydrogen atoms tend to order thus inducing structural phase transitions.
The interstices which are preferentially occupied are large, and are surrounded
by at least one atom of a binary hydride forming element. Their occupancies
can be rationalized in terms of geometrical factors (atomic size), attractive
nearest-neighbor interactions (metal-hydrogen bond formation), repulsive
second-nearest-neighbor interactions (mainly between the hydrogen atoms),
and electronic factors (band-filling effects), thus providing a qualitative
understanding at the atomic level of those factors which limit hydrogen
sorption capacities. Clearly, more neutron and x-ray investigations, possibly
on single crystals, more sophisticated structure models (taking into account
effects due to anharmonicity and microtwinning etc.), and diffraction data of
higher resolution are desirable in order to gain a quantitative understanding.
This is also true for the study of magnetically ordered hydrides, in particular
with respect to investigating correlations between chemical and magnetic
ordering. Finally the detection of significant isotope effects (hydrogen versus
deuterium) on both static and dynamic properties of ternary hydrides would
be of interest.
 Crystal and Magnetic Structures of Ternary Metal Hydrides 119

~ ,~ o~=~

.~. r-~ w'~

ox eq~'N .~0

II , , 0~

~ ,o~
. o ~ ~,,o o

%
~s

~ ~ o~
0 0

0,~'~ ~,-~ ~ ;~., ~ ~ ¢~.

R

"0

e~

om 0 ~ ~.~ ~-.~ ~1 ~ , . 0 , - ~ . . ~ ,~Q t.-,~.~

~'-~ ~ ~-~ 0 ~ ' . ~ , :~,~'~ ~1 ~ ~

>
120 K. Yvon and P. Fischer

b£~ cI
o~

$
¢-q , .
o.
II ~..~ :-. II

~"0
tNog

II II

II ii li
~'~ II

l~ ~ 0 0 ~ 0
0 ~ ~ "" "~
~o ~ C ~ '° o-o ~

II
0

.., o, ~ °.
o .~.~
~ ~
II
0

0 '.,~ 00

. ~ ~;:..c~ _ • ,,6 II

~.~ , o

~'~ ~ ~.=~
"t~ ..~

.~

0
 Crystal and Magnetic Structures of Ternary Metal Hydrides 121

b~
• ~ A A A A ~

~ O O '~ ~ V V V V

~ ~;~ ~ ~ , ~ o.~
e o ~o o ~- ~ - ~
II I I ~ o ~ O

"a~ II II II II oc~ ~.~ ~ ~o /~ o, ~ ~ II II II II o ~

~ - ~ • ..~ ~ ~ "~" II ~ II " . - - - ' ~ - ' ~ - - ~ - - ~

'~D ~ ' J "~-~ t ~'~
~ o. o. ~ .~-
cSoooc5
II II II II II ~ ~ ~ II II II II II II II II ~ I "~ II II tl II II
~ ~.~ ~ ~ ~ -~ • 0 ~

"Qo

oo ~o ~ ~ ~ ~ ~ o ~ ~ oo ~ ~ ~ ~ ~" ~ o

e-,i cxl , ~,~ c~

~N a
.... •~ o ~ . ~

m
o='~
~ ~.~
~.~ ~oo
,..~ = ~

~.=o.
~ ~ ~'~

X
8
122 K. Yvon and P. Fischer

~. ~ ' ~
e-q ~.-, .~ ~ ~ ~ ~
~ .--.~ ~ . o ~

- ~:.~ ~

•~ "~" "~ ~r'~ u'~ ~.~"~--" ~ ~C~, ~ "~" "~ ~ , ~ . ~

~o

•~ c',4

~.~
~~
g
~ . ~
o ~ o ~ ~

'.~ 0

II "~ ~ ~ ~..~-~ .~

o ~ ~.~'~ ~

,~
!
.~ ~-~ 0 ~ I~ . . .

0-~ il
II ~
~ '
o
-
~ ~-,-
ii
~.%
li
,i

~ o o ~
..~ . ~ ~.~

~'~ n
.-~.~

i;

.,e
 Crystal and Magnetic Structures of Ternary Metal Hydrides 123

~ c5 ~-i ~ c4 rq ~
II II II II II II II ~

~II. ~I~. ~N II II ~k,~ ~'~'~-~- ~ ~ ~ ~

t'--- t",l t'-- t-.- t'--- oo ~'~ u'~ ""~ ¢"1 " ~ ~r~_" ~.~ t'-- "-" "~ ~-" ~

~c5~5~5c5o~;c5c5 c5 ~.= ~ 1 ~II II II

bl bl Ixl bl

,-'~ e--. ~ /-~..~ ~-x ~--. ~'~, ~ ~, ~--x t-,q -- *-~ ~ ~-,.~. o ~ ~

~1 II II II II II II II ~ II II II II II II II II "~ II II ~. II II II II ~'~

c5

¢,q

o. ~.'a- O

e-

8
 124 K. Yoon a n d P. Fischer

oo
,.q o _o

~r-6
:~" ..-

~:~0 ''~

~'~ L.-q .-.~

II If

II ~ II II 2 II ff lr rl ~ II II tl
,, ~ r-,.o o . m: ~ N N

ft il tl II Ir ~ ~.E II II ~ II ~ II II II

• ~ ,,.:..~ .~

I:~ ,e

ii G ,,d 7 ,A

~',= ~

.~

,m
 Crystal and M a g n e t i c Structures of T e r n a r y Metal H y d r i d e s 125

c~ ~ . .-, ,.o . . ,;:: ~ .~ ~-- ~ ~- ~--o ,-'-

~o,,~11 II 11"7. ' ' 7 , I~ ~ :::'~
, II~ H . II
,£~t:,., II c::,c:,
II rl ..9.o .~ ,::~
~ . ~ ~ ~'~ c~ . . . . . . . ~ .~ ~ o ~'~

• "~ .,.~ r~ ~'~ ~

¢'q t"q ~ "~ o ~ ~ ..-~ ~'~" ~ ml

~ ~1=~ i~ ~ o "~'~
o o~-~o~ ~ -'o "° o -~_--. .

I[ ;~ @
e- ,.o

0o0 0 o
o. ~ ' ~ I~ o o o ~

~ ~ ~=~
•"~ ~,"~ ' ~ ,.~,

I~1.~ ~ ' ~ , o ~;.-.~

6 ~ ° ~.-~
o, .- o 1~ ,~ ~ o ~
~",~ ~ 0

~,.~ ~ m ~ ~ II ~ ~

oo

e,i
126 K. Yvon and P. Fischer

$$
~.~?
II II II tl c~
~J

II II o II II
II

(~oo ~'~

r~-

II II ~

~ d d d e e d d d ~.
II II II II o II II II II II II II II II •
N N N N ,.~, ~

N ~ t-I INI m
2 ¢,q

II

3zNES ~ -~

• • . M ~ . •

~~ ~
t~
".~

8
e~

Z
 Crystal and M a g n e t i c Structures of Ternary Metal Hydrides 127

~5

II II ~ ~ ~ iL
:~

$
en

~ --~o il II o ~- II
0
II

.T.~$
:s :s
II II II ~ ~ ~ II 11 ]I II 11 II II

ooo [I

:5 ~ o. o
o o~
il ~ ~ il

t-q ~...,

~ o4~-"
"~" ~ '~o. . ~ I[
N 11 [I II

O @ 0

0 --

o
 128 K. Y~on a n d P. Fischer

II
0

b~

II H

~.~ o. ~.
0

~., If

',~- t'N t"N

","-~ C , r' " t'l

il II II II e II II II II II II II ~ = II II

II II

t~
'7-,

~ s 2~ g "

g
..~
r-.
d~
t~
 Crystal and Magnetic Structures of Ternary Metal Hydrides 129

o~b ~
O

~ ~ O~

,,.<

~ ~5 ~ ~5 ~5 e5 ~ c5
O

r~

O'~ tg'~ t"~ I'~ ~ ~ OO 0"3

eq

Ixl bl t'-I ~ bl t,,1 tq ~-~ bl .o=

,d

+.a

II II II II E

"6
,..a
r.t3

t~
130 K. Yvon and P. Fischer

4.B Appendix
Magnetic structures of ternary metal hydrides and deuterides studied by neutron diffraction

Compound" Ref. Magnetic ordering b

CeRuzD5 [4.56] Antiferromagnetic; ferromagnetic (111) planes stacked in a

+ +-- sequence along [111]
#ce=0.6(2) #n at 120mK
HoFe2D3. 5 [4.57] Ferrimagnetic ("fanned"), Tc= T~°=295 K
vc -_ _ 1.9 (3) #~
#n~ = 5.3 (4)/h~, #~,t
Remarks." Temperature dependence of magnetic moments
determined
ErFe2D z [4.57] Ferrimagnetic ("fanned"); T~r = 369 K
Er --
p.,~, - 7.5 #B, # ~Fte ---- - - 1.8 #~
Remark: Temperature dependence of magnetic moment of Er
determined
ErFezD(H)3.5 [4.57] Ferrimagnetic ("fanned"),
Tc = 440 K (deuteride), 450 K (hydride)
T~r = 295 K (deuteride), 300 K (hydride)
#~t = 4.3 (4) #a (deuteride), 4.9 (4)/z n (hydride)
#Y~ = - 1.6(3)#B (deuteride), - 1.5(3)/~ B (hydride)
Remark: Temperature dependence o f magnetic moments
determined
TmFezHa. s [4.57] Ferrimagnetic ("fanned"), T T M = 250 K
Tm__ 3.8 (4) ItB
,//sat --

Remark: Temperature dependence of magnetic moment of Tm

determined

Y6Mn23Dz3 [4.69] Antiferromagnetic, TN= 172(7) K

4K 78K
~//Mn 1 = -- #Mn2 = 3.68 ,%, 3.60 ,un

#~,9 = - ~Zuns= 1.60 #n, 2.08 #n

Remark: Zero ordered magnetic moments at other Mn sites
Th6Mnz3D3o [4.71] Ferrimagnetic at 4 K
#M.x = -- 3.5(-) #B, #M,2 =0.8(1) gin, ,uM,3 = 1.8(1) ~B,
#M,4 = 1.5(1) #B
Ho6Fez3D1. 5 [4.72] Ferrimagnetic
4K 295 K
/z,° = - 10.2(2) It. - 5.7(2) ktn
#v¢1 = 2.2 #B 2.2 /~i~
/Iv, 2 = 0.9 (2) #B 1.2 (2) #B

" Listed in same order as in Table 4.1; see also Appendix 4.A
b/*=ordered magnetic moment in units of Bohr magneton, kt,; /A,t=ordered magnetic
moment at saturation, i.e. for temperature T close to zero; negative signs indicate opposite
directions o f magnetic moments; TN(C)=N6el (Curie) temperature determined by neutron
diffraction; TcR= ordering temperature of rare-earth sublattice; values within parentheses
indicate estimated standard deviations
 Crystal and Magnetic Structures of Ternary Metal Hydrides 131

4.B Appendix (continued)

Compound ~ Ref. Magnetic ordering b

/2F~a= 1.3(2)#. 1.4 (2)/2.

#F~4 = 2.0(2)/2, 1.9 (2) #,
Ho6Fe23Ds.2 [4.72] Ferrimagnetic,
4K 78 K 295 K
#no = -- 9.1 (2) #i~ - 8.3(2) #B -4.2(2)/2,
#F~i= 2.2 /2, 2.2 /2, 2.2 #~
/2Fe2 = 1.8 (2)/2, 1.8 (2) #. 1.6 (2) #a
#v~3= 0.9(2)/2. 0.8 (2)/2. 0.8 (2) #u
/2w4= 2.1 (2)/28 2.1 (2)/2. 1.9(2) #n
I--Io6Fe2 3D t 2A [4.72] Ferrimagnetic,
4K 295 K
/2no.lS~) =--10.2(2)/2~ --4.1(2)/2,
#Hobtl6*7 = -- 10.2(2) #B --4.1(2)#.
/2Fo~(4~,) = 2.0(2)/2, 2.2(-)/2,
#Fe2a (8c) ~ 2 . 5 (2) #B 2.0 (2)/2.
/2Fe2b{16f) = 2.5(2)#, 2.3 (2) # .
/2Fe3tl6nl) = 1.1 (2)/2~ 1.1 (2)/2.
/2v~,m6,~1 = 2.5(2)/2, 2.0 (2)/2.
Remark: Ho momcnt reorientation from [I00] to [101] between
295 and 4 K, Fel and Fe3 moments reorient similarly whereas
Fe2 and Fe4 moments remain within the basal plane.
Ho6Fe23D,s.7 [4.72] Ferrimagnetic,
4K 295 K
#no = - 10.2(2)/2n -4.3(2)/2,
/2Fel = 2.2(-)/2u 2.2(-)/2n
#ve2= 2.4(2) 2.3(2)/2n
#F~a = 2.0 (2) # . 1.8 (2)/2,
/2Fo4= 2.2(2)/2B 1.9(2)/2,
Er6Mn23D23 [4.68] Ferrimagnetic, #j[Ic; at I50 K:
/2Er(,te) =--2.5(3)/2,, /2M,(2b) =--2.7(3)/2,
PE,Sh) = 1.2(3)#B, #M,~,~.} = 3.4(2)#,
#Mn(8f) = 0
0.8(2)/2.
#~.{16.o =
#M,(,6,=} = 0.9(2)#,
fl'-PrCosD3. 6 [4.92] Ferromagnetic, &lie; at 300 K: #co=0.83(4),us
4.2 K 77 K
/2Pr =1.6(1) /2R 0.50(4)/2a
#co, = 0.12 (8) #. 0.98 (6) #n
#co=. a = 0.79 (6) #n 1.28 (5) #B
fln-PrCo sD2.9 [4.92] Ferromagnetic, #llc; at 300 K: #Co= ].05(5)#,
flLLaCosD3.35 [4.92] Ferromagnetic,/24te; at 300 K:/2co= 1.14(5)#n
flm-CeCosDz.s5 [4.92] Ferromagnetic,/2/rc; at 300 K: #c0=0.98(3)#.
ZrMn2D3 [4.200] Antiferro- or ferrimagnetic, TN{Cl= 175(5) K
Til.zMnt.sD3.1 [4.98] Ferromagnetic, ]2Mn(6h) perpendicular to c,
at 80 K: #M,=0.31 #~
132 K. Yvon and P. Fischer

Acknowledgements. The authors thank Professors S. Rundqvist and L. Schlapbach for

valuable suggestions, and Drs. Y. Andersson, D. Fruchart, I. Jacob, and P. Zolliker for a
careful reading of the manuscript. Errors which may remain in the text and tables are the sole
responsibility of the authors.
Support by the Swiss National Energy Research Foundation (NEFF) and the Swiss
National Science Foundation is gratefully acknowledged.

References
4.1 K. Yvon: J. Less-Common Met. 103, 53 (1984)
4.2 Y. Noda, T. Kajitani, M. Hirabayashi, S. Sato: Acta Crystallogr. C 41, 1566 (1985)
4.3 R.A.Alikhanov, V.I. Buzin, N.I. Kulikov, M.E. Kost, W. Sikora, L.S. Smirnov: J. Less-
Common Met. 101,291 (1984)
4.4 R.R. Arons, J.W. Cable: Sol. State Commun. 55, 835 (1985)
4.5 B. Lebech, N. Hessel-Andersen, S. Steenstrup, A. Schroeder Pedersen: Acta Cryst.
C39, 1475 (1983
P. Fischer, J. Schefer, K. Tichy, R. Bischof, E. Kaldis: J. Less-Common Met. 94, 151
(1983)
4.6 W. Bartscher, A. Boeuf, R. Caciuffo, J.M. Fournier, W.F. Kuhs, J. Rebizant, F.
Rustichelli: Solid State Commun. 53, 423 (1985)
4.7 W. Bartscher, A. Boeuf, R. Caciuffo, J.M. Fournier, J.M. Haschke, L. Manes, J.
Rebizant, F. Rustichelli, J.W. Ward: Physica 130B, 530 (1985)
4.8 H. Nagel, R.S. Perkins: Z. Metallkd. 66, 362 (1975)
4.9. H. Kaneko, T. Kajitani, M. Hirabayashi, M. Ueno, K. Suzuki: J. Less-Common Met.
89, 237 (1983)
4.10 B. Rodmacq, Ph. Mangin, A. Chamberod: J. Phys. F: Met. Phys. 15, 2259 (1985)
4.11 A.J. Maeland, A.F. Andresen: J. Chem. Phys. 48, 4660 (1968)
4.12 S.C. Abrahams, A.P. Ginsberg, K. Knox: Inorg. Chem. 3, 558 (1964)
S.C. Abrahams, K. Knox: J. Phys. 25, 461 (1964)
4.13 A.N. Fitch, S.A. Barrett, B.E.F. Fender, A. Simon: J. Chem. Soc. Dalton Trans. 4, 501
(1984)
4.14 S.D. Wijeyesekera, J.D. Corbett: Solid State Commun. 54, 657 (1985)
4.15 H.J. Mattausch, A. Simon, K. Ziebeck: J. Less-Common Met. 113, 149 (1985)
4.16 G. Bandel, M. Milliner, M. Tr6mel: Z. Anorg. Allg. Chem. 453, 5 (1979)
4.17 J.M. Newsam, A.K. Cheetham, B.C. Tofield: Solid State Chem. 60, 214 (1985)
4.18 C. Greaves: J. Appl. Cryst. 18, 48 (1985)
C. Greaves, M.A. Thomas: Acta Cryst. B42, 51 (1986)
4.19 M.T. Weller, P.G. Dickens: J. Solid State Chem. 58, 164 (1985)
4.20 I.S. Latergaus, V.T. Em, I. Karimov, D.Ya. Khvatinskaya, S.K. Dolukhanyan: Inorg.
Mat. 20, 1420 (1984)
4.21 S. Mukawa, T. Kajitani, M. Hirabayashi: J. Less-Common Met. 103, 19 (1984)
4.22 R.O. Moyer, Jr., C. Stanitski, J. Tanaka, M.I. Kay, R. Kleinberg: J. Solid State Chem. 3,
541 (1971)
4.23 W. Bronger, G. Auffermann, P. Mfiller: J. Less-Common Met. 116, 9 (1986)
4.24 G.G. Libowitz: The Solid-State Chemistry of Binary Metal Hydrides (W.A. Benjamin,
Inc., New York, Amsterdam 1965)
4.25 W.M. Mueller, J.P. Blackledge, G.G. Libowitz: Metal Hydrides (Academic, NewYork,
London 1968)
4.26 P. Eckerlin, H. Kandler: in Landolt-B6rnstein, Vol. 6 Structure Data of Elements and
lntermetallic Phases, ed. by K.-H. Hellwege, A.M. Hellwege (Springer, Berlin, Heidel-
berg 1971) pp. 138 174
4.27 R.R.Arons et al.: In Landolt-Biirnstein I11, Vol. 12c (Springer, Berlin, Heidelberg 1982)
pp. 372-405
 Crystal and Magnetic Structures of Ternary Metal Hydrides 133

4.28 H. Asano, M. Hirabayashi: Z. Phys. Chem. NF 114, 1 (1979)

H. Asano, M. Hirabayashi: In Metal Hydrides, ed. by G. Bambakidis (Plenum, New
York 1981) pp. 81-103
M. Hirabayashi, H. Asano: ibid., pp. 53-80
4.29 J.W. Ward: Physica 130B, 510 (1985)
4.30 R.G. Teller, R. Bau: In Structure and Bonding, Vol. 44, pp. 1-82 (Springer, Berlin,
Heidelberg, New York 1981); see also
R. Bau, M.Y. Chiang, D.M. Ho, S.G. Gibbins, T.J. Emge, T.F. Koetzle: Inorg. Chem. 23,
2823 (1984)
4.31 S. Rundqvist, R. Tellgren, Y. Andersson: J. Less-Common Met. 101, 145 (1984)
4.32 A.F. Andresen: In Hydrides for Energy Storage, ed. by A.F. Andresen, A.J. Maeland
(Pergamon, Oxford 1978) pp. 61-72
4.33 A.F. Andresen: J. Less-Common Met. 88, 1 (1982)
4.34 K.N. Semenenko, V.V. Burnasheva: J. Less-Common Met. 105, 1 (1985)
4.35 A. Percheron-Gu6gan, C. Lartigue, J.C. Aehard: J. Less-Common Met. 109, 287 (1985)
4.36 G.D. Sandrock: In Hydrides for Energy Storage, ed. by A.F. Andresen, A.J. Maeland
(Pergamon, Oxford 1978) pp. 353-393
4.37 H. Ocsterreicher: Appl. Phys. 24, 169 (1981)
4.38 G.K. Shenoy, B.D. Dunlap, P.J. Viccaro, D. Niarchos: Adv. Chem. Series No. 194, ed.
by J.G. Stevens, G.K. Shenoy, Chap. 23, pp. 501 521 (1981)
4.39 H. Buchner: Energiespeicherung in Metallhydriden (Springer, Wien, New York 1982)
4.40 K.H.J. Buschow, P,C.P. Bouten, A.R. Miedema: Rep. Prog. Phys. 45, 937 (1982)
4.41 D.G. Ivey, D.O. Northwood: J. Mater. Sci. 18, 321 (1983)
4.42 K.H.J. Buschow: In Handbook on the Physics and Chemistry of Rare Earths, Vol. 6,
Chap. 47, ed. by K.A. Gschneider, Jr., L. Eyring (Elsevier, Amsterdam 1984)
4.43 Proceedings of the International Symposium on the Properties and Applications of Metal
Hydrides: J. Less-Common Met. Vols. 73 and 74 (1980), 88 (1982); 89 (1983); 101, 103,
and 104 (1984); 129, 130, and 131 (1987) 1
4.44 International Tables for Crystallography, Vol. A (D. Reidel, Dordrecht 1983)
4.45 H. Buchner, M.A. Gutjahr, K.-D. Beccu, H. S/iufferer: Z. Metallkde. 63, 497 (1972)
4.46 C. Stioui, D. Fruchart, A. Rouault, R. Fruchart, E. Roudaut, J. Rebi6re: Mat. Res. Bull.
16, 869 (1981)
4.47 F.J. Rotella, H.E. Flotow, D.M. Gruen, J.D. Jorgensen: J. Chem. Phys. 79, 4522 (1983)
4.48 J.-J. Didisheim, K. Yvon, P. Fischer, D. Shaltiel: J. Less-Common Met. 73, 355 (1980)
4.49 J.-J. Didisheim, K. Yvon, P. Fischer, P. Tissot: Solid State Commun. 38, 637 (1981)
4.50 J.-J. Didisheim, K. Yvon, D. Shaltiel, P. Fischer, P. Bujard, E. Walker: Solid State
Commun. 32, 1087 (1979)
4.51 D. Fruchart, A. Rouault, C.B. Shoemaker, D.P. Shoemaker: J. Less-Common Met. 73,
363 (1980); Phys. Status Solidi a 57, Kl19 (1980)
4.52 A.V. Irodova, V.P. Glazkov, V.A. Somenkov, S.Sh. Shilstein: J. Less-Common Met. 77,
89 (1981); Sov. Phys. Solid State 22, 45 (I980)
4.53 V.A. Yartys', V.V. Burnasheva, K.N. Semcnenko, N.V. Fadeeva, S.P. Solov'ev: Int. J.
Hydrogen Energy 7, 957 (1982)
V.A. Yartys', V.V. Burnasheva, S.E. Tsirkunova, E.N. Kozlov, K.N. Semenenko: Sov.
Phys. Crystallogr. 27, 148 (1982)
V.V. Burnasheva, N.V. Fadeeva, K.N. Semenenko, S.P. Solov'ev, V.A. Yartys': Acta
Crystallogr. A 37, C-182 (1981)
4.54 A.V. Irodova, O.A. Lavrova, G.V. Laskova, L.N. Padurets: Soy. Phys.-Sol. State 24, 22
(1982)
4.55 N.F. Miron, V.I. Shcherbak, V.N. Bykov, V.A. Levdik: Soy. Phys.-Crystallogr. 16, 266
(1971)
4.56 D. Fruchart, F. Vaillant, A. Rouault, A. Benoit, J. Flouquet: J. Less-Common Met, 101,
285 (1984)

1 These references are only partly considered in this review.

134 K. Yvon and P. Fischer

D. Fruehart, F. Vaillant, E. Roudaut, A. Nemoz, X.G. Tessema: Phys. Status Solidi

a 65, KI9 (1981)
4.57 J.J. Rhyne, S.G. Sankar, W.E. Wallace: In The Rare-Earths in Modern Science and
Technology, Vol. 1, ed. by G.J. McCarthy, J.J. Rhyne (Plenum, New York 1978)
pp. 63-68;
J.J. Rhyne, G.E. Fish, S.G. Sankar, W.E. Wallace: J. de Phys. 40, C5-209 (1979)
G.E. Fish, J.J. Ryne, S.G. Sankar, W.E. Wallace: J. Appl. Phys. 50, 2003 (1979)
G.E. Fish, J.J. Rhyne, T. Brun, P.J. Viecaro, D. Niarchos, B.D. Dunlap, G.K. Shenoy,
S.G. Sankar, W.E. Wallace: In The Rare-Earths in Modern Science and Technology,
Vol. 2, ed. by G.J. McCarthy, J.J. Rhyne, H.B. Silber (Plenum, New York 1980)
pp. 569-570
4.58 W. Bartscher, J. Rebizant, A. Boeuf, R. Caciuffo, F. Rustichelli, J.M. Fournier, W.F.
Kuhs: Proc. Aetinides 85, Aix-en-Provence, Sept. 1985. J. Less-Common Met. 121,
455 (1986)
4.59 K. Yvon, J. Schefer, F. Stucki: Inorg. Chem. 20, 2776 (1981)
J. Schefer, P. Fischer, W. H/ilg, F. Stucki, L. Schlapbach, J.J. Dididsheim, K. Yvon, A.F.
Andresen: J. Less-Common Met. 74, 65 (1980)
4.60 D. Nor~us, L.G. Olsson: J. Chem. Phys. 78, 2419 (1983)
4.61 J.L. Soubeyroux, D. Fruchart, A. Mikou, M. Pezat, B. Darriet, P. Hagenmuller: Mat.
Res. Bull. 19, 969 (1984)
B. Darriet, J.L. Soubeyroux, M. Pezat, D. Fruchart: J. Less-Common Met. 103, 153
(1984); see also [4.63, 99]
4.62 D. Nor6us, P.-E. Werner: J. Less-Common Met. 97, 215 (1984)
4.63 J.L. Soubeyroux, D. Fruchart, A. Mikou, M. Pezat, B. Darriet: Mat. Res. Bull. 19, 1119
(1984)
4.64 P. Zolliker, K. Yvon, J.D. Jorgensen, F. Rotella: Inorg. Chem. 25, 3590 (1986)
4.65 P. Zolliker, K. Yvon, P. Fischer, J. Schefer: Inorg. Chem. 24, 4177 (1985)
4.66 J.-J. Didisheim, P. Zolliker, K. Yvon, P. Fischer, J. Schefer, M. Gubelmann, A.F.
Williams: Inorg. Chem. 23, 1953 (1984)
J.-J. Didisheim, K. Yvon, P. Fischer, J. Schefer, M. Gubelmann, A.F. Williams: Z.
Kristallogr. 162, 61 (1983)
4.67 M. Commandre, D. Fruchart, A. Rouault, D. Sauvage, C.B. Shoemaker, D.P.
Shoemaker: J. de Phys. 40, L639 (1979)
4.68 C. Crowder, B. Kebe, W.J. James, W. Yelon: In The Rare-Earths in Modern Science and
Technology, Vol. 3, ed. by G.J. McCarthy, H.B. Silber, J.J. Rhyne (Plenum, New York
1982) pp. 473~476; for work on Er6Mn23D23 see: Amer. Inst. Phys. Conf. Proc. 89, 318
(1982)
4.69 K. Hardman-Rhyne, J.J. Rhyne, E. Prince, C. Crowder, W.J. James: Phys. Rev. B 29, 416
(1984)
4.70 K. Hardman, J.J. Rhyne, E. Prince, H.K. Smith, S.K. Malik, W.E. Wallace: In The Rare-
Earths in Modern Science and Technology, Vol. 3, ed. by G.J. McCarthy, H.B. Silber, J.J.
Rhyne (Plenum, New York 1982) pp. 477-478
K. Hardman, J.J. Rhyne, H.K. Smith, W.E. Wallace, S.K. Malik: J. Appl. Phys. 52, 2070
(1981)
4.71 K. Hardman-Rhyne, H.K. Smith, W.E. Wallace: J. Less-Common Met. 96, 201 (1984)
K. Hardman, J.J. Rhyne, K. Smith, W.E. Wallace: J. Less-Common Met. 74, 97 (1980)
4.72 J.J. Rhyne, K. Hardman-Rhyne, H.K. Smith, W.E. Wallace: J. Less-Common Met. 94,
95 (1983)
4.73 P. Thompson, F. Reidinger, J.J. Reilly, LM. Corliss, J.M. Hastings: J. Phys. F: Met.
Phys. 10, L57 (1980)
4.74 P. Thompson, M.A. Pick, F. Reidinger, L.M. Corliss, J.M. Hastings, J.J. Reilly: J. Phys.
F: Met. Phys. 8, L75 (1978)
4.75 P. Fischer, W. Hfilg, L. Schlapbach, F. Stucki, A.F. Andresen: Mat. Res. Bull. 13, 931
(1978)
 Crystal and Magnetic Structures of Ternary Metal Hydrides 135

G. Busch, L. Schlapbach, F. Stucki, P. Fischer, A.F. Andresen: In Hydrogen Energy

System, ed. by T.N. Veziroglu, W. Seifritz (Pergamon, Oxford 1979), Vol. 5, pp.
2615-2636; Int. J. Hydrogen Energy 4, 29 (1979)
4.76 J. Schefer, P. Fischer, W. Hfilg, F. Stucki, L. Schlapbach, A.F. Andresen: Mat. Res. Bull.
14, 1281 (1979)
L. Schlapbach, A. Seiler, F. Stucki, P. Ziircher, P. Fischer, J. Schefer: Z. Phys. Chem.
N F 117, 205 (1979)
4.77 P. Thompson, J.J. Reilly, F. Reidinger, J.M, Hastings, L.M. Corliss: J. Phys. F: Met.
Phys. 9, L61 (1979)
4.78 W. Schfifer, G. Will, T. Schober: Mat. Res. Bull. 15, 627 (1980)
W. Sch/ifer, E. Lebsanft, A. Bl/isius: Z. Phys. Chem. NF 115, 201 (1979)
see also [4.124]
4.79 P. Fischer, J. Schefer, K. Yvon, L. Schlapbach, T. Riesterer: J. Less-Common Met 129,
39 (1986)
4.80 D. Fruchart, M. Commandr6, D. Sauvage, A. Rouault, R. Tellgren: J. Less-Common
Met. 74, 55 (1980)
4.81 L.J. Vieland, A.W. Wicklund, J.G. White: Phys. Rev. B l l , 3311 (1975)
4.82 A. Furrer, P. Fischer, W. H/ilg, L. Schlapbach: In Hydrides for Energy Storage, ed. by
A.F. Andresen, A.J. Maeland (Pergamon, Oxford 1978) pp. 73-82; see also I-4.84]
4.83 A.L. Bowman, J.L. Anderson, N.G. Nereson: Proceedings of the 10th Rare-Earth
Research Conference, Carefree (1973) pp. 485-489
4.84 P. Fischer, A. Furrer, G. Busch, L. Schlapbach: Heir. Phys. Acta 511,421 (1977)
G. Busch, L. Schlapbach, W. Thoeni, Th. v. Waldkirch, P. Fischer, A. Furrer, W. Hfilg:
Proc. 2nd Int. Congress on Hydrogen in Metals (Paris 1977) 1D7, pp. 1-8
P. Fischer, W. H/ilg, L. Schlapbach, Th. von Waldkirch: Helv. Phys. Acta 51, 4 (1978);
see also I-4.82]
4.85 D. Nor6us, L.G. Olsson, P.-E. Werner: J. Phys. F: Met. Phys. 13, 715 (1983)
4.86 A. Percheron-Gu6gan, C. Lartigue, J.C. Achard, P. Germi, F. Tasset: J. Less-Common
Met. 74, 1 (1980)
C. Lartigue, A. Percheron-Gu6gan, J.C. Achard, F. Tasset: In The Rare-Earths in
Modern Science and Technology, Vol. 2, ed. by G.J. McCarthy, J.J. Rhyne, H.B. Silber
(Plenum, New York 1980)
C. Lartigue, A. Percheron-Gu6gan, J.C. Achard, F. Tasset: J. Less-Common Met. 75, 23
(1980)
J.C. Achard, F. Givord, A. Percheron-Gu6gan, J.L. Soubeyroux, F. Tasset: J. de Phys.
40, C5-218 (1979)
4.87 C. Lartigue, A. Percheron-Gu6gan, J.C. Achard, J.L. Soubeyroux: J. Less-Common
Met. 113, 127 (1985)
4.88 P. Thompson, J.J. Reilly, L.M. Corliss, J.M. Hastings, R. Hempelmann: J. Phys. F:
Met Phys. 16, 679 (1986)
4.89 V.V. Burnasheva, V.A. Yartys', N.V. Fadeeva, S.P. Solov'ev, K.N. Semenenko: Soy.
Phys. Dokl. 23, 97 (1978)
4.90 A.V. Irodova, M.E. Kost, L.N. Padurets, V.A. Somenkov, E.I. Sokolova, S.Sh.
Shil'shtein: Russ. J. Inorg. Chem. 26, 166 (1981)
4.91 L.D. Calvert, B.M. Powell, J.J. Murray, Y. le Page: J. Sol. State Chem. 60, 62 (1985)
4.92 F.A. Kuijpers, B.O. Loopstra: J. Phys. Chem. Solids 35, 301 (1974); J. de Phys. 32, CI
657 (1971)
F.A. Kuijpers: Philips Res. Repts. Suppl. No. 2 (1973)
4.93 C. Crowder, W.J. James, W. Yelon: J. Appl. Phys. 53, 2637 (1982)
4.94 J.C. Achard, A.J. Dianoux, C. Lartigue, A. Percheron-Gu6gan, F. Tasset: In The Rare-
Earths in Modern Science and Technology, Vol. 3, ed. by G.J. McCarthy, H.B. Silber, J.J.
Rhyne (Plenum, New York 1982) pp. 481-486
4.95 E. Gurewitz, H. Pinto, M.P. Dariel, H. Shaked: J. Phys. F I 3 , 545 (1983)
4.96 J.-J. Didisheim, K. Yvon, D. Shaltiel, P. Fischer: Solid State Commun. 31, 47 (1979)
136 K. Yvon and P. Fischer

4.97 H.W. Mayer, K.M. Alasafi, O. Bernauer: J. Less-Common Met. 88, L7 (1982)
4.98 D. Fruchart, J.L. Soubeyroux, R. Hempelmann: J. Less-Common Met. 99, 307 (1984)
4.99 J.L. Soubeyroux, D. Fruchart, A. Mikou, M. Pezat, B. Darriet: Mat. Res. Bull. 19, 895
(1984); see also [4.61]
4.100 D. Nor6us, P.-E. Werner: Acta Chem. Scand. A36, 847 (1982)
4.101 A.F. Andresen, K. Otnes, A.J. Maeland: J. Less-Common Met. 89, 201 (1983)
4.102 A.F. Andresen, H. Fjellvfig, A.J. Maeland: J. Less-Common Met. 103, 27 (1984)
4.103 Y. Andersson, S. Rundqvist, R. Tellgren, T.B. Flanagan: Z. Phys. Chem. N F 145, 43
(1985)
4.104 A. Santoro, A. Maeland, J.J. Rush: Acta Cryst. B34, 3059 (1978)
4.105 ,i. Bergsma, J.A. Goedkoop, J.H.N. van Vucht: Acta Crystallogr. 14, 223 (1961)
J.H.N. van Vucht: Philips Res. Repts. 18, 35 (1963)
4.106 A.J. Maeland, E. Lukacevic, J.J. Rush, A. Santoro: J. Less-Common Met. 129, 77
(1987)
4.107 D. Fruchart, P. Wolfers, P. Vulliet, A. Yaouanc, R. Fruchart, P.L'Heritier: In Proc. of
the E. C. Workshop Meeting on Nd-Fe Permanent Magnets, Oct. 1984, Brussels, ed. by
I.V. Mitchell (1985)
4.108 S.W. Peterson, V.N. Sadana, W.L. Korst: J. de Phys. 25, 451 (1964)
4.109 D.G. Westlake, H. Shaked, P.R. Mason, B.R. McCart, M.H. Mueller, T. Matsumoto,
M. Amano: J. Less-Common Met. 88, 17 (1982)
4.110 A.V. lrodova, V.A. Somenkov, S.Sh. Shil'shtein, L.N. Padurets, A.A. Chertkov: Soy.
Phys. Crystallogr. 23, 591 (1978)
4.111 Y. Andersson, S. Rundqvist, R. Tellgren, J.O. Thomas, T.B. Flanagan: J. Sol. State
Chem. 32, 321 (1980)
4.112 Y. Andersson, S. Rundqvist, R. Tellgren: J. Sol. State Chem. 52, 327 (1984)
4.113 Y. Andersson, S. Rundqvist, R. Tellgren, J.O. Thomas, T.B. Flanagan: Acta Crystallogr.
B37, 1965 (1981)
4.114 D.G. Westlake: J. Less-Common Met. 91, 1 (1983)
4.115 I. Jacob, D. Shaltiel, D, Davidov, I. Miloslavski: Solid State Commun. 23, 669 (1977)
I. Jacob, D. Shaltiel: J. Less-Common Met. 65, 117 (1979)
4.116 1. Jacob, J.M. Bloch, D. Shaltiel, D. Davidov: Solid State Commun. 35, 155 (1980)
4.117 J.J. Reilly, R.H. Wiswall, Jr.: Inorg. Chem. 13, 218 (1974)
4.118 J.J. Reilly, R.H. Wiswall, Jr.: lnorg. Chem. 7, 2254 (1968)
4.119 M.L. Post, J.J. Murray, G.J. Despault, J.B. Taylor: Mat. Res. Bull. 20, 337 (1985)
4.120 D. Nor6us, P.-E. Werner: Mat. Res. Bull. 16, 199 (1981)
4.121 S. Ono, H. Hayakawa, A. Suzuki, K. Nomura, N. Nishimiya, T. Tabata: ,i. Less-
Common Met. 88, 63 (1982)
4.122 H. Hayakawa, Y. Ishido, K. Nomura, H. Uruno, S. Ono: J. Less-Common Met. 103, 277
(1984)
4.123 P. Zolliker, K. Yvon, C. Bfirlocher: J. Less-Common Met. 115, 65 (1986)
4.124 T. Schober, W. Sch/ifer: J. Less-Common Met. 74, 23 (1980), see also [4.78]
4.125 A.W. Hewat, I. Bailey: Nucl. Instrum. Methods 137, 463 (1976)
4.126 R. Allemand, J. Bourdel, E. Roudaut, P. Convert, K. lbel, ,i. Jacobe, J.P. Cotton, B.
Farnoux: Nucl. Instrum. Methods 126, 29 (1975)
W. H/ilg, H. Heer, J. Schefer, P. Fischer, B. Bron, A. Isacson, M. Koch: Helv. Phys. Acta
57, 741 (1984);
P. Fischer: Neutron Diffraction Newsletter, Int. Union of Cryst. (1985) p. 15
4.127 .I.D. Jorgensen, F.J. Rotella: J. Appl. Cryst. 15, 27 (1982)
4.128 H.M. Rietveld: J. Appl. Crystallogr. 2, 65 (1969)
4.129 P.-E. Werner, S. Salom6, G. Malmros, J.O. Thomas: ,i. Appl. Crystallogr. 12, 107 (1979)
4.130 T.G. Worlton, ,I.D. Jorgensen, R.A. Beyerlein, D.L. Decker: Nucl. Instrum. Methods
137, 331 (1976)
4.131 A. Albinati, B.T.M. Willis: The Rietveld method, publication MPD-NBS-136,
Materials Physics Division, AERE Harwell (1980); J. Appl. Crystallogr. 15, 361 (1982)
 Crystal and Magnetic Structures of Ternary Metal Hydrides 137

4.132 A.W. Hewat: Proc. of Symposium on Accuracy in Powder Diffraction, Gaithersburg,

MD (1979), National Bureau of Standards Special Publication 567, 111 (1980); see
also: C.R.A. Carlow (ed.): High Resolution Powder Diffraction, Mat. Science Forum
(Trans Tech. Publ., Switzerland) Vol. 9, pp. 69-80 (1986)
4.133 R.B. Von Dreele, J.D. Jorgensen, C.G. Windsor: J. Appl. Crystallogr. 15, 581 (1982)
4.134 R.A. Young, E. Prince, R.A. Sparks: J. Appl. Crystallogr. 15, 357 (1982)
4.135 H.G. Scott: J. Appl. Crystallogr. 16, 159 (1983)
4.136 H.H. van Mal, K.H.J. Buschow, A.R. Miedema: J. Less-Common Met. 35, 65 (1974)
A.R. Miedema, K.H.J. Buschow, H.H. van Mal: J. Less-Common Met. 49, 463 (1976)
4.137 P.C.P. Bouten, A.R. Miedema: J. Less-Common Met. 71, 147 (1980)
4.138 D.G. Westlake: J. Less-Common Met. 90, 251 (1983)
4.139 J. Osterwalder, T. Riesterer, L. Schlapbach, F. Vaillant, D. Fruchart: Phys. Rev. B 31,
8311 (1985)
4.140 Z. Gavra, M.H. Mintz, G. Kimmel, Z. Hadari: Inorg. Chem. 18, 3595 (1979)
4.141 J.P. Darnaudery, M. Pezat, B. Darriet, P. Hagenmuller: Mat. Res. Bull. 16, 1237 (1981)
4.142 J.R. Johnson, J.J. Reilly, F. Reidinger, L.M. Corliss, J.M. Hastings: J. Less-Common
Met. 88, 107 (1982)
4.143 J. Zhuang, J.M. Hastings, L.M. Corliss, R. Bau, Chiau-Yu Wei, R.O. Moyer, Jr.: J. Sol.
State Chem. 40, 352 (1981)
4.144 P. Zolliker, K. Yvon: Mat. Res. Bull. 21, 415 (1986)
4.145 M. Gupta, E. Belin, L. Schlapbach: J. Less-Common Met. 103, 389 (1984)
4.146 D.G. Westlake: J. Mater. Sci. 19, 316 (1984)
4.147 H. Behr, H. Metzger, J. Peisl: J. Less-Common Met. 88, 159 (1982)
4.148 E.T. Teatum, K.A. Gschneidner, J.T. Waber: Los Alamos Natl. Lab., Los Alamos, New
Mexico, Rept. No. LA-4003 (1968); see also [Ref. 4.168, p. 151-1
4.149 R.D. Shannon: Acta Crystallogr. A32, 751 (1976)
4.150 J.D. Corbett, H.S. Marek: Inorg. Chem. 22, 3194 (1983)
4.151 A.C. Switendick: Z. Phys. Chem. NF 117, 89 (1979);
B.K. Rao, P. Jena: Phys. Rev. B 31, 6726 (1985)
4.152 M. Gupta: J. Less-Common Met. 88, 221 (1982)
4.153 D.A. Papaconstantopoulos, A.C. Switendick: J. Less-Common Met. 88, 273 (1982)
4.154 M. Gupta: J. Less-Common Met. 103, 325 (1984)
4.155 B.M. Klein, W.E. Pickett: J. Less-Common Met. 88, 231 (1982)
4.156 W.A. Oates, T.B. Flanagan: Mat. Res. Bull. 19, 1397 (1984)
4.157 R. Griessen, A. Driessen, D.G. de Groot: J. Less-Common Met. 103, 235 (1984)
4.158 J.C. Achard, C. Lartiguc, A. Percheron-Gu~gan, J.C. Mathieu, A. Pasturel, F. Tasset: J.
Less-Common Met. 79, 161 (1981)
4.159 J. Shinar, D. Shaltiel, D. Davidov, A. Grayevsky: J. Less-Common Met. 60, 209 (1978)
4.160 R. Burch, N.B. Mason: J. Less-Common Met. 63, 57 (1979)
4.161 F.L. Carter: J. Less-Common Met. 74, 245 (1980)
4.162 A. Yoshikawa, T. Matsumoto, K. Yagisawa: J. Less-Common Met. 88, 73 (1982)
4.163 A.L. Shilov, M.E. Kost, N.T. Kuznetsov: J. Less-Common Met. 105, 221 (1985)
4.164 B.D. Dunlap, P.J. Viccaro, G.K. Shenoy: J. Less-Common Met. 74, 75 (1980)
4.165 H.A. Kierstead: J. Less-Common Met. 96, 141 (1984)
4.166 A.R. Williams, J. Ktibler, C.D. Gelatt, Jr.: Phys. Rev. B 19, 6094 (1979)
4.167 A. Miedema, P.F. de Chatel, F.R. DeBoer: Physica I00 B, 1 (1980)
4.168 W.B. Pearson: The Crystal Chemistry and Physics of Metals and Alloys (Wiley, New
York 1980)
4.169 C.E. Lundin, F.E. Lynch, C.B. Magee: J. Less-Common Met. 56, 1.9 (1977)
4.170 M.H. Mendelsohn, D.M. Gruen, A.E. Dwight: J. Less-Common Met. 63, 193 (1979);
Nature (London) 269, 45 (1977)
4,171 C.B. Magee, J. Liu, C.E, Lundin: J. Less-Common Met. 78, 119 (1981)
4.172 D.G. Westlake: J. Chem. Phys. 79, 4532 (1983)
4.173 D.G. Westlake: Mat. Res. Bull. 18, 1409 (1983)
138 Yvon and Fischer: Crystal and Magnetic Structures of Ternary Metal Hydrides

4.174 D.G. Westlake: J. Mat. Science 18, 605 (1983); and Scripta Metall. 16, 1049 (1982)
4.175 D.G. Westlake: Sol. State Chem. 53, 130 (1984)
4.176 D.G. Westlake: J. Less-Common. Met. 91, 275 (1983)
4.177 D.G. Westlake: J. Less-Common Met. 103, 203 (1984); J. Less-Common Met. 107, 189
(1985)
4.178 D.G. Westlake: J. Less-Common Met. 105, 69 (1985)
4.179 G.A. Stewart, DC. Creagh: J. Phys. F 15, 1639 (1985)
4.180 Y. Chung, T. Takeshita, O.D. McMasters, K.A. Gschneidner, Jr.: J. Less-Common Met.
74, 217 (1980)
4.181 D.P. Shoemaker, C.B. Shoemaker: J. Less-Common Met. 68, 43 (1979)
4.182 W.A. Oates, J.A. Lambert, P.T. Gallagher: Trans. Met. Soc. AIME 245, 47 (1969)
4.183 W.E. Wallace: J. Less-Common Met. 88, 141 (1982)
4.184 J.F. Lakner, F.S. Uribe, S.A. Steward: J, Less-Common Met. 72, 87 (1980)
4.185 W.E. Wallace, E.B. Boltich: J. Sol. State Chem. 33, 435 (1980)
4.186 H. Goretzki: Naturwiss. 54, 163 (1967)
4.187 K. Yvon, H. Nowotny, R. Kieffer: Monatsh. Chem. 98, 2164 (1967)
4.188 D.M.P. Mingos, J.C. Hawes: In Structure and Bonding, Vol. 63 (Springer, Berlin,
Heidelberg 1986) pp. 1 63
4.189 K.A. Gschneidner, Jr., T. Takeshita, Y. Chung, O.D. McMasters: J. Phys. F. Met. Phys.
12, L1 (1982)
4.190 M.H. Mendelsohn, D.M. Gruen: J. Less-Common Met. 78, 275 (1981)
4.191 I. Jacob, A. Stern, A. Moran, D. Shaltiel, D. Davidov: J. Less-Common Met. 73, 369
(1980)
4.192 I. Jacob: Sol. State Commun. 40, 1015 (1981)
1. Jacob, J.M. Bloch, E. Gurevich: J. Less-Common Met. 88, 25 (1982)
4.193 D. Shaltiel: J. Less-Common Met. 73, 329 (1980)
4.194 I. Jacob, J.M. Bloch: Sol. State Commun: 42, 541 (1982)
4.195 T. Riesterer: J. Less-Common Met. 103, 219 (1984)
4.196 D.M. Gruen, M. Mendelsohn: J. Less-Common Met. 55, 149 (1977)
4.197 C.E. Crowder, W.J. James: J. Less-Common Met. 95, 1 (1983)
4.198 C. Geibel, W. Goldacker, H. Keiber, V. Oestreich, H. Rietschel, H. Wfihl: Phys. Rev.
B 30, 6363 (1984)
4.199 M.Yu. Belyaev, A.V. Skripov, V.N. Kozhanov, A.B. Stepanov, E.V. Galoshina: Solid
State Commun. 48, 1049 (1983)
4.200 J.-J. Didisheim, P. Fischer: J. Less-Common Met. 103, 267 (1984)
5. Electronic Properties
Mich61e Gupta and Louis Schlapbach

With 25 Figures and 2 Tables

5.1 Introduction
The dissolution of hydrogen in a metal lattice and the formation of a metal
hydride perturbs considerably the electrons and phonons of the host metal.
Accordingly the study of the electronic properties of a hydrogen-metal system is
not only of fundamental interest to understand the hydrogen-metal interaction
in the bulk and at the surface, but also sheds light on technologically important
parameters such as the heat of formation and heat conductivity and to some
extent also the phenomena of hydrogen embrittlement (decohesion) and
catalysis.
Until the late sixties, research on electronic properties of metal hydrides
involved mainly experimental work [5.1]. The first theoretical investigations
started in the early fifties with the pioneering work of Friedel in which he
studied the screening of a hydrogen impurity and the heats of solution of H in
noble metals [5.2]. However, up to the early seventies, oversimplified models
were often used to describe the bonding mechanisms of H in transition metals
[5.31. The first ab initio band structure calculations on binary hydrides
performed by Switendick [5.4] gave a new impetus and stimulated a lot of
experimental and theoretical work. From the theoretical results on concen-
trated binary hydrides several important features have emerged and the
hydrides are no longer considered, as they were in the early days, as interstitial
alloys but rather as real compounds. The role of the chemical bonding has been
emphasized and the crude rigid-band model in its protonic or anionic form is no
longer being safely applied to the electronic structure of the host metal to
interpret the experimental data on hydrides. For about ten years both
theoretical and experimental work was limited to hydrides of elemental metals,
e.g. Pd, Ti, rare earth metals and Th [5.4, 5]. Only a few years ago the first
results on the electronic structure of hydrides of intermetallic compounds
elaborated by band structure calculations (FeTi) [5.6] and experimentally by
photoelectron spectroscopy (ZrMn2) [5.7] and x-ray emission spectroscopy
(crystalline and amorphous alloys of Zr and V) [5.8-10] became available.
Recent reviews on theoretical [5.6] and experimental work on crystalline [5.11]
and amorphous [5.12] materials have been given.
After a brief description of the theoretical and experimental methods given
in Sect. 5.2, we shall essentially review in this chapter the progress made since
the appearance of the two volumes on Hydrogen in Metals [5.13] of this series.
140 M. Gupta and L. Schlapbach

Since then, the work on binary hydrides mainly of transition and rare earth
metals has been pursued. For stoichiometric compounds new results concern-
ing Fermi surfaces, superconductivity, and prediction of various observables
using realistic wavefunctions have been obtained. An effort towards a self-
consistent treatment of the crystal potentials has been made and some
conclusions have been obtained concerning the problem of sign of the charge
transfers. Also, theoretical studies of stabilities and of preferential occupation of
sites by hydrogen atoms by means of total energy calculations have started to
appear. The recent results on binary hydrides are described in Sect. 5.3 while
Sect. 5.4 is devoted to the very recent theoretical and experimental investi-
gations of the electronic properties of intermetallic hydrides. The essential
experimental and theoretical results are gathered in Tables 5.1 and 5.2. The
consequences of the electronic properties of hydrides for heats of formation,
superconductivity, and maximum hydrogen content will be discussed while
some features relevant to surface electronic structure, cohesion, and ernbrittle-
ment will be briefly mentioned.
Four different effects are in general relevant for understanding the changes
in the electronic structure of the host metal due to the H absorption: (i) The
generally observed expansion (or in some special cases, the contraction) of the
lattice often accompanied by a change in the crystal structure results in a
modification of the symmetry of the states and in the band widths. (ii) The
attractive potential of the proton affects those metal wavefunctions which have
a finite density at the H site and leads to the so-called metal-hydrogen bonding
band below the metal d-band. Furthermore it can pull below the Fermi energy
Ev some metal states which were located above E v in the pure metal. (iii) The
additional H - H interactions in the hydrides which have more than one H atom
per unit cell lead to new features in the lower portion of the density of states. (iv)
The presence in the unit cell of the additional electrons brought by the H atoms
and the inbalance between the additional number of electrons and the number
of new electron states shifted to below the Fermi energy of the host metal results
in a shift of the Fermi level.
We shall underline the similarities and differences observed between the
electronic structure of binary and ternary hydrides. One of the goals of the
analysis of electronic properties of intermetallic compounds and their hydrides
is to understand why some compounds of non hydride forming components
form a stable ternary hydride e.g. Fe-Ni alloys [5.14] or vice versa why a
compound of hydride forming components does not form a ternary hydride e.g.
ZrPd2 [5.15].

5.2 Theoretical and Experimental Methods: General Features

5.2.1 Theoretical Methods
In this section we shall briefly remind the reader of the most salient features of
some of the methods used in the study of the electronic structure of hydrogen in
 Electronic Properties 141

metals and indicate references for further details concerning the formalisms and
the numerical approaches.

The Local Density Functional Approximation. Since a wide variety of theoretical

studies ranging from the problem of hydrogen as an impurity in jellium to the
treatment of stoichiometric transition metal hydrides, rest on the local density
functional (LDF) formalism developed by Hohenberg, Kohn and Sham I-5.16l,
we shall first summarize the main ideas associated with this method. The basis
of the L D F is that the ground-state energy E of an interacting N particle system
in an external potential is a unique functional of the electron density p(r) and
that the ground-state electron density is the one which minimizes the energy
functional E[Q(r)]. The application of the variational principle leads to the self-
consistent solution of a set of N one-particle Schrrdinger equations

- v 2 + ~ r [ e ( r ) , d ~'~(r) = ~ (r) (5.1)

from which the wave function ~. and energy eigenvalues ei are generated. The
effective one-electron potential Verycan be expressed in terms of the external
potential, the average electrostatic potential and the exchange and correlation
potential Vx~[0(r)] which is the functional derivative of the energy functional
Ex~[0(r)]. In the LDF, the exchange correlation energy Ex~ for the interacting
inhomogeneous electron gas is approximated by:

Exc[o(r)3 ~- f o(r)ex~[O(r)] dr, (5.2)

where exJ-Q(r)] is the exchange and correlation energy per particle for an
interacting but homogeneous electron gas having the local density 0(r). Various
forms of the exchange and correlation potential can be used.
Within the LDF, a self-consistent calculation can be set up (i) by
constructing an initial guess of the charge density, for example for a solid by an
overlap of atomic charge densities corresponding to a given configuration or
the renormalized atom model; (ii) calculating the effective crystal potential
Veff[o(r), r]; (iii) solving the Schrrdinger equation (5. ~); (iv) calculating the new
charge density by summing the one-electron densities over the lowest occupied
states and returning to (i) with an admixture of the old and the new charge
densities until a satisfactory convergence criterion is reached. The total energy
of the system is then given by:

E = E en--½ i dar ~ d3r' -~(r)~(r')

- + I a~r0(r) {~xo[e(r)] - uxo[e(r)]}, (5.3)
noc~ Ir--r'l

where the second term corrects for the double counting of the electron
Coulomb self-interaction and the third term accounts for the exchange
correlation energy within the LDA approximation.
142 M. Gupta and L. Schlapbach

a) Stoichiometric Hydrides
Energy Bands. The methods which have been mostly used for studying the
electronic structure of stoichiometric hydrides are the augmented plane wave
(APW) and Korringa-Kohn-Rostoker (KKR) methods [5.17] in which the
wavefunctions are expanded in a set of energy-dependent partial waves. The
crystal potential is often assumed to have a muffin-tin (MT) shape: it has a
spherical symmetry inside touching spheres surrounding the atomic centres
and is constant in the remaining interstitial region. This approximation is
known to work rather well for closed- packed structure such as the fcc structure
often encountered in metal hydrides. For more open structures one can also
introduce the spherical approximation around vacant interstitial sites- as done
sometimes for the fluorite structure dihydrides - or use the so called warping
corrections to account for the departure of the potential in the interstitial region
from a constant value. The warping corrections can become rather important
for intermetallic hydrides such as e.g. FeTiHx and Mg2NiH 4 for which the non-
overlapping MT spheres occupy less than 50% of the unit cell volume.
Other band structure techniques such as the LCAO method [5.17] which is
based on an expansion of the wavefunctions on a set of energy-independent
fixed basis functions and on the use of general crystal potentials have not been
very much applied to the study of metal hydrides. The LCAO method has the
advantage - unlike the APW and K K R schemes - of leading to energy-
independent Hamiltonian matrix elements. However, it suffers from the
problem of necessarily limited basis sets and from the difficulty in the
evaluation of multicentre integrals.
Linear methods of band theory such as the linear augmented plane wave
(LAPW) or the linear muffin-tin orbital method (LMTO) [5.18] on the other
hand, combine the advantage of the use of energy-independent fixed basis
functions with those of the partial waves. They have been shown to be very
successful in the treatment of compounds having complex structure; their use
for the study of metal hydrides is just starting [5.19].

Density of States (DOS). Improvements on the accuracy of the DOS and

Fermi energy calculations have been made since earlier work on metal hydrides
[5.20]; the histogram method with coarse mesh is no longer used. The energy
eigenvalues calculated ab initio in a discrete mesh of k points in the Brillouin
Zone (BZ) [typically at 89 points in the (1/48)th wedge of the irreducible fcc BZ]
are then interpolated using symmetrized plane-wave expansions or sym-
metrized Fourier series expansions. The BZ integration can then be performed
by analytical methods. It is widely admitted that one of the most satisfactory
schemes is the so-called linear-energy tetrahedron method [5.21]. Accuracies
typically of the order of a few tenths of an meV arc obtained using these
techniques. Other analytical procedures such as the so-called Quad scheme
have also been used [5.22]; this usually differs by a few percent from the
tetrahedron method.
 Electronic Properties t43

The energy band interpolation can also be performed by the LCAO

method. The Hamiltonian matrix elements are expressed in terms of the tight-
binding energy integrals of Slater and Koster (SK) [5.23]. The numerical
accuracy achieved is often inferior to that of the above mentioned techniques; it
depends in particular upon the number of basis orbitals included in the
calculation, on the range of the interactions included (nearest neighbours or
next nearest neighbours) and on the number of ab initio k points used in the
fitting procedure. However, the interest of the SK scheme resides in its physical
meaning; the energy integrals obtained as parameters are extremely useful for
building tight-binding Hamiltonians which allow the treatment of a variety of
electronic structure problems [5.24].

Decomposed DOS and Charge Transfer Analysis. Decompositions of the DOS

into site and angular momentum components are now available for most of the
metal hydrides; they shed light on the bonding characteristics and are
extremely useful in interpretating many experimental data such as the x-ray
emission spectra for example. The K K R and APW wavefunctions are from
their definition perfectly suited for an angular momentum analysis inside the
MT spheres. This decomposition depends, however, on the a priori partition of
the real space into non-overlapping atomic spheres and further assumptions
have to be made for the interstitial charge. The weights of the angular
momentum decomposed DOS can for example be obtained by continuing the
radial integrations beyond the MT to a Wigner-Seitz sphere. DOS analysis in
terms of LCAOs on the other hand suffer from the ill-defined spatial extension
of the atomic orbitals; if these functions are quite extended a sizeable portion
of the charge is sometimes located in the neighbouring MT sphere. The
peculiarities of the different schemes should be kept in mind when numerical
values of sensitive qualities such as charge transfer are discussed as done in the
following sections. Charge transfer definitions are still a matter of debate; it is
however our opinion that in the discussion of these effects the most
meaningful quantity is the charge associated with spatially localized states
such as the transition metal d-electron charge.

Total Energy Calculations. Besides the large amount of theoretical work on the
cohesive energies of transition metals using simplified schemes [5.25-28], total
energy calculations obtained from ab initio band structure studies by means of
(5.3) or equivalent expressions, are now available for a variety of transition
metals (TM) [5.29] and transition metal compounds I-5.30]. However metal
hydrides have not been systematically investigated in this respect. Gelatt et al.
[5.31] first proposed a theoretical explanation of the trends in the heats of
formation of binary TM hydrides. More recently total energy calculations from
first principles have been performed for binary hydrides of the end of the series:
Co to Cu and Ru to Ag by Williams et al. [5.32], Methfessel and Kiibler [5.33]
using the augmented spherical wave (ASW) method, and for monohydrides of
Nb by Ho et al. [5.34] within the pseudopotential method. Detailed work on
144 M. Gupta and L. S c h l a p b a c h

binary hydrides has been recently reported by Switendick [5.35] who calculated
total energies as a function of lattice constant for PdH, NiH, Till2, and PdHz
using self-consistent APW band structure calculations. No such calculations
are however yet available for hydrides of intermetallic compounds.
The calculated heat of formation AH of say - a binary metal hydride is
defined from the reaction of the metal M with gaseous hydrogen

M +½xH 2--*MHx + A H M H ~• (5.4)

Thus A H M H x is obtained by substrating from the total energy of the metal
hydride EMHxthe total energy of the corresponding pure metal EM and the total
energy per hydrogen atom of an H2 molecule EH2
D
AHMnx- E~nx-- EM--~XEH 2 • 1
(5.5)
We wish to point out the difficulty of obtaining accurate theoretical predictions
since the measured values of AHMIax for binary TM hydrides are in absolute
value smaller than 1 eV/H atom while the first two terms in (5.5) are several
orders of magnitude larger.
In order to obtain accurate absolute values of AHMn x even the theoretical
method involved in the calculation of Ea2 becomes important. For reasons of
consistency, the same type of theoretical -treatment should be used to calculate
each of the three terms of (5.5). The problems encountered here are similar to
those found in the computation of cohesive energies of solids where one
subtracts the total energy of the atoms from that of the corresponding solid.
Large absolute errors are known to be made for example in the atomic energies
with the use of the LDF. However, due to cancellation effects, the accuracy of
the computed cohesive energies usually turns out to largely exceed the absolute
accuracies for the atomic or solid calculations. On the basis of the success
obtained in the calculation of cohesive energies for TM, one can expect that the
use of (5.5) in conjunction with band structure studies will correctly reproduce
trends and provide a microscopic understanding of the heats of formation of
hydrides.
Gelatt et al. [5.31] proposed that the main contributions to the heat of
formation of a metal monohydride are essentially: (i) The metal-hydrogen
bonding contribution which results in the lowering of metal states having an s
symmetry at the H sites. This term is approximated by the energy lowering of
the lowest band of the metal upon formation of the hydride; this band
accommodates two electrons. (ii) The increase in the binding of the metal
d-bands due to the attractive potential of the protons. This term is taken to be
the change in the average energy of the d-states located above the lowest metal
band; these states are assumed to accommodate n - 2 electrons where n is the
number of valence electrons of the pure metal. (iii) The shift in the Fermi energy
of the metal due to the presence of an addition electron in the hydride. It is
assumed here that the protonic rigid-band model is correct since this term is
 Electronic Properties 145

approximated by the Fermi energy of the metal. (iv) Coulomb repulsions due to
the increased charge density at the H sites lead to a decrease in the heat of
formation. These different contributions have been calculated by Gelatt
et al. I-5.31] using the renormalized atom theory. The trends in the variation of
the experimental curves are reproduced and ascribed essentially to the
variation of the first and third contribution. The analysis of the total energy
calculations of Williams et al. [5.32]have essentially confirmed that the general
upward trend in the enthalpy curve as Z increases in a TM series is due to a
decrease of the M - H bonding contribution while it is the variation of the metal
chemical potential which explains the dip in the AH curve observed for Ni and
Pd.
A very different simplified model has been proposed by Norskov et al. [5.36]
to evaluate the heat of solution of H in bulk metals as well as chemisorption of
H on metal surfaces, trapping of H at defects etc. by means of the effective
medium theory. One calculates the binding energy of an H atom immersed in an
homogeneous electron gas which has a density equal to the average value of the
host metal density around the H interstitial site. Interactions with the metal
core electrons, hybridization between H s-electrons and transition metal
d-electrons and a correction term for the inhomogeneity of a real metal are also
introduced. Using this model, the general trends in the heat of solution of H in
3d, 4d and 5d TM can be explained mostly by the variation of the first term.
Besides the nonnegligible role of the hybridization contribution, the large host
metal electron densities in the middle of the TM series - also associated with the
maximum in cohesive energies - appear to be responsible for the observed
maximum in the heat of solution of H. Another interesting feature of the model
of Norskov et al. [5.36] is the proposed explanation of the lattice expansion
observed in transition metals around the H sites and the lattice contraction in
alkali metals, in terms of the existence of an optimum host electron density
which leads to a minimum in the binding energy curve of a H atom in an
electron gas. The idea of the existence of an optimum host electron density
provides also a ground for the empirical criterion of Westlake [5.37] of a
hydrogen minimum hole size of 0.38 A.
Daw et al. [5.41] showed that the problem of decohesion and hydrogen
embrittlement can be treated by calculating the embedding energy ofa H atom
in a transition metal in an effective medium theory.
The need for a rational understanding of the observed trends and the
possibility of making predictions about metal hydride stabilities is particularly
acute for hydrides of intermetallic compounds of technological interest. In view
of the scarcity of theoretical work for these materials of complex crystal
structures, numerous empirical or semi-empirical models have been proposed.
Griessen and Driessen [5.38] have recently further simplified and proposed an
empirical expression for the heat of formation of binary hydrides AH which
depends linearly upon the difference AE = EF-- Es between the Fermi energy EF
and the centre of the lowest conduction band of the host metal Es (see Chap. 6).
This empirical relation has then been extended to ternary metal hydrides by
146 M. Gupla and L. Schlapbach

using results on TM alloys obtained from the tight-binding coherent potential

approximation (CPA) model of Cyrot and Cyrot-Lackmann 1-5.39]. Another
powerful scheme for predicting the properties of ternary hydrides has been
proposed by Miedema et al. [5.40] following his work on transition metal
alloys. However it has been shown [-5.30] that the microscopic picture in
which Miedema's work is based overemphasizes the properties of the
constituents rather than the bonding properties of the compounds.

b) Nonstoiehiometric Hydrides
Several approaches have been used for studying the electronic structure of
nonstoichiometric hydrides. The best one to date is based on the coherent
potential approximation (CPA) within the K K R method [-5.42]. We want to
mention however that in older work using the virtual crystal approximation
(VCA) the crystal potential of the disordered binary system is simply
approximated by the crystal potential of an effective ordered system by taking a
concentration weighted average of the atomic potentials for a two component
system. This approach is obviously too crude since the VCA potential is real
and periodic and thus leads to real eigenvalues and infinite lifetimes of the
associated eigenstates. In the average t-matrix approximation (ATA) of
Korringa, single-site potentials are complex leading to complex eigenvalues
and finite lifetimes. In the ATA it is the. scattering t-matrix at a given site which
is taken as the concentration weighted average of the single scattering
t-matrices [-5.43]. The CPA was first introduced by Soven [5.44] and has been
successfully applied to the study of various alloy systems. In this approximation
one assumes an effective ordered system by locating on every site some effective
potential which is determined by requiring that the replacement at a given site
of the coherent potential by the true site potential produces no further
scattering on the average. From the averaged Green's function of the system
G(r, r', E), quantities such as e.g. the densities of states and the charge densities
can be obtained. CPA calculations using the tight-binding method have been
performed for binary TM hydrides and alloys [5.45, 46]. Recently, the CPA in
its K K R form has been used to study alloys such as P d - A g - H for which the
disorder in both the metal and the hydrogen sublattices have been taken into
account [5.42]. In that case the CPA medium consists of two sublattices of
effective sites.

c) H Impurity in Metals
A substantial number of theoretical studies of a hydrogen impurity in simple
metals have appeared since the mid 70 s; most of them are based on the jellium
model. These self-consistent calculations of the screening of the proton in an
electron gas using the LDF approximation have shown the importance of
nonlinear effects as emphasized in the pioneering work of Friedel [5.2]. The
effect of the lattice ions is introduced using approximate treatments such as
local pseudopotential theory [5.17] or the spherical solid model [5.47]. Other
 Electronic Properties 147

theoretical treatments of H in simple metals using supercell calculations, K K R

scattering theory have also been used. Studies of heats of solution of H,
determinations of hydrogen site occupancies, activation energies for overcom-
ing potential barriers as involved in classical diffusion mechanisms, and
calculations of local mode vibration frequencies are available; attempts to
study lattice relaxation effects around the H impurity in simple metals have
also been made. The corresponding theoretical work on simple metals has
been recently reviewed [5.48]. The treatment of H as an impurity in TM is
much more complex than in simple metals due to the presence of the metal
d-electrons. The amount of theoretical work in this field has started to
increase: Ab initio calculations such as the K K R Green's function method
[5.49], the large supercell [5.20, 50] or cluster models [5.51, 52] have been
used as well as parametrized tight-binding Hamiltonians [5.53].

d) Calculation of Other Observables

From these theoretical studies on metal hydrides and H in metals, accurate

information can be obtained concerning properties related to the Fermi energy:
Fermi surface geometries, densities of states at Ev which can be related, in the
independent particle model, to the Pauli susceptibility )~p and the electronic
specific heat coefficient ?. Details concerning the nature of the bonding, the
charge transfer and to a lesser extent total energies and predictions of
vibrational frequencies are becoming increasingly available.
Using quantities obtained from ab initio band structure work such as the
single-site scattering phase shifts at the Fermi energy 61, the partial DOS of
angular character l, n~, evaluated at site ~:, and the total DOS at Er for one spin
direction N~(EF), one can obtain an estimate of the electron-phonon coupling
constant 2 and thus the superconducting transition temperature To. Following
the work of McMillan [5.54] for transition metals, an approximate expression
of 2 has been proposed for compounds with a large mass ratio between the
constituent atoms

"~ *~Metal "q- 2H -- t]Metal t]H (5.6)

MMetal(Oj2)ae + MH(O)e)optie'

where r/ is the "electronic contribution", M the atomic mass and (09 2) the
second moment of the renormalized phonon frequencies defined by McMillan.
Using the rigid ion approximation which ignores renormalization effects due to
lattice vibrations, Gaspari and Gyorffy [5.55] have shown that:

EF
t/~ ,-~ N ~(EF)Tr2 ~ 2(I+ 1) sin 2(6~+ 1 -- 6~) ,t~' (1)rrn'[(Ev)n'{+~vJ,
+~"~(1)t
~, 1rl~'
(Er)
-~rJ~" (5.7)

This expression will be used in the following sections to comment on the

superconducting properties of metal hydrides.
148 M. Gupta and L. Schlapbach

5.2.2 Experimental Methods

In order to understand the electronic structure of metal hydrides, properties

such as the total and partial (s, p, d, f ) densities of states at Ev, the position, shape
and width of the occupied and empty energy bands and their k-dependence, the
charge transfer and the transitions from metallic to nonmetallic behaviour and
related energy gaps should be measured and compared to theory. Essentially all
the experimental methods which are used to look at the electronic properties of
metals and alloys can also be applied to metal hydrides. However, severe
restrictions may be introduced by the shape of the samples. Hydrides of
intermetallic compounds are easily available in the form of powder, in some
cases as thin films or amorphous ribbons, but so far, never in the form of
polycrystalline or single crystal bulk samples. Furthermore the rather low
thermodynamic stability of metal hydrides clearly limits investigations at high
temperature and at low pressure (ultrahigh vacuum, UHV) according to the
pressure-composition isotherms.
The Pauli contribution to the magnetic susceptibility, and the electronic
specific heat coefficient 7 are proportional to the density of states at E v for a
free electron gas. In real metals electron-electron and electron-phonon
interactions cause large deviations and render quantitative evaluations un-
certain. Magnetic precipitates in the bulk or at the surface of intermetallic
compounds can cause large errors in the evaluation of susceptibility data
[5.56]. Whereas electrical resistivity measurements still yield very valuable
results for binary hydrides, e.g. rare earth hydrides, only very few measurements
have been performed on hydrides of intermetallic compounds because of the
difficulties in making contacts to the samples. Contactless measurements of Q
values or capacitances in resonance circuits allow the study of the variation of
the resistivity with temperature and thus to the identification of metallic or
semiconducting behaviour.
Nuclear magnetic resonance (NMR), a well-known method to study H
diffusion, also provides information on the electronic structure [5.57]. The
hyperfine fields at the site of a nucleus are affected, among others things, by
conduction electrons which cause a Knight shift; this can be measured by
NMR, M6ssbauer spectroscopy I-5.58] and muon spin rotation spectroscopy
(~tSR) [5.59, 60]. With NMR, the Knight shift at the nucleus of the host metal
and also that of the proton can be measured. The time dependence of the
hyperfine field is related to a relaxation mechanism and contributes a
conduction electron component (l/z1) to the spin-lattice relaxation rate.
(zl x T), where T is the absolute temperature, is determined by the s-p-d
conduction band density of states at Ev. However, relations between electronic
structure, relaxation rate, and Knight shift are not simple and often electronic
specific heat and magnetic susceptibility are also needed for a full interpretation
of the NMR data.
The "classical" methods mentioned so far have the advantage of requiring
only a reasonably simple experimental set up which nonetheless usually allows
 Electronic Properties ! 49

measurements over a wide range of temperature and hydrogen pressure.

However, these methods probe the electronic structure around E F only and do
not give direct information on hydrogen induced bonding and antibonding
bands etc.
The de Haas-van Alphen effect, i.e. the quantum oscillatory part of the
magnetization of a metal in an external magnetic field, probes the effective
electron mass and the curvature of the Fermi surface and is very well suited to
measure small changes in the electronic structure of dilute H-metal systems
[5.61, 62] Compton scattering of high energy x-rays, i.e. the inelastic scattering
of a photon with electrons, probes the electron momentum distribution of the
host metal and its variation due to hydride formation. H 2 partial pressures up
to the mbar level are possible [5.63]. Positron annihilation has also been used
to study the Fermi surface of TM hydrides [5.64] as well as their momentum
space properties [5.65].
Among the most powerful methods to test the electronic structure of
elemental metals, intermetallic compounds, glassy and crystalline alloys and
their hydrides are the spectroscopic methods of electron and x-ray photo-
emission [5.66, 67] inverse photo-emission [5.68] and also x-ray emission and
absorption [5.10, 69]. In photoelectron spectroscopy (PES) monoenergetic
radiation from an UV lamp (UPS, hv = 10-100 eV), from a tunable synchrotron
radiation source, or from an x-ray source (XPS), e.g. Mg K~ (hv = 1253 eV) or
AIK~ (by-= 1486 eV), is used to excite and emit electrons from a sample by the
photoelectric effect. An electron energy analyzer yields count-rate versus
binding energy curves which are called energy distribution curves or photo-
electron spectra. With UPS, the valence band and possibly high lying core levels
are probed with an energy resolution AE,~ 0.1 eV whereas with XPS, valence
band and core levels down to 1.5 keV are accessible (AE ~ 0.5-1.5 eV). The mean
free path of the photoelectrons in the solid determines the probing depth. It
varies between ~0.5nm for kinetic energies EK,~50eV to ~ 3 nm for EK
1.5 keV. The small probing depth requires ultrahigh vacuum instruments to
prevent contamination of the surface. Though hydrogen partial pressure up to
10- 5 mbar is possible, pressure is the most serious limitation of PES in studying
metal hydrides.
A valence band photoelectron spectrum resembles a one-electron density of
states curve, but they are not in fact identical, for many reasons: The spectrum
represents an excited state (one electron is missing) and many-body effects
provide screening of the photohole, the emission of electrons of different
quantum numbers and of different elements is enhanced or diminished by cross-
section effects and the limited instrumental resolution broadens the spectra.
Screening and correlation lead to particularly strong and interesting effects in
the PES spectra of the rare earths and actinides.
Within some approximations, photoelectron spectra yield directly the
position and width of the occupied bands, the variation of the emission at E F
accounts for changes in the DOS at E F and the core-level shifts indicate the
charge transfer and/or changes in hybridization. The use of photons of various
150 M. Gupta and L. Schlapbach

energies allows, to some extent, the evaluation of partial DOS owing to the
energy dependence of the cross-section. Soft x-ray emission spectroscopy
(SXES) is even better at distinguishing the partial DOS by selecting various
valence band-core level transitions. To look at the bulk electronic properties of
hydrides XPS has significant advantages over UPS: the larger probing depth
and the analysis of the core levels of all components make possible controls of
deviations of hydrogen concentration and of alloy composition at the surface.
The major methods for studying unoccupied states above EF are inverse
photoemission (also called bremsstrahlung-isochromat spectroscopy, BIS)
[5.68, 70] and x-ray absorption (XAS) particularly in the form of x-ray
absorption near-edge structure (XANES) [5.71]. In XANES monoenergetic
photons of variable energy are absorbed and excite bound electrons below EF
to empty states above E F. The structure near the absorption edge reflects the
empty, local, partial DOS, but unfortunately only with rather low energy
resolution.
If single crystal samples are available, the dispersion E(k) of the occupied
and empty energy bands can be measured by angle resolved photoemission and
inverse photoemission, respectively, (ARUPS, ARBIS).
In optical spectroscopies the complex dielectric function is probed by
measurements of absorption or refleetivity. The dielectric function is directly
related to the electronic band structure. Features in the dielectric function can
be related to electric dipole interband transitions [5.72].

5.3 Results for Binary Hydrides

5.3.1 Hydrides of Pd and Ni

a) Energy Bands and Densities of States

Stoichiometric palladium hydride has been the most thoroughly investigated of
all the binary hydrides. All the calculations are in qualitative agreement with
the results first obtained by Switendick [5.20] so we shall summarize only briefly
the main results. An improvement of the densities of states curves has been
achieved over the earlier calculations due to a better sampling of the BZ and
due to the use of more accurate integration techniques. However, a dispersion
of the theoretical results still exists concerning the position and width of the
metal-hydrogen bonding bands, the metal d-bands, the value of the DOS at EF,
and the values of charge transfer. Some examples are given in Table 5.1. An
effort towards a self-consistent treatment has been made but here too
differences are observed between different self-consistent calculations due to the
different treatment of the exchange term of the crystal potential and due to
numerical accuracy problems.
As shown in Fig. 5.1, the main differences between the calculated DOS of
pure Pd metal and that of PdH can be summarized as follows [5.20, 35, 73-75].
 Electronic Properties 151

Fig. 5.1. Slater-Koster density of states for Pd

(top, from [5.94]) and total and partial wave
2.0
analysis of the DOS of PdH [5.35]
..J
.d
LU
0
, 0
>= -I0 -5 E~:O 5 I0
~ . ~ :Htotal DOS
z 4.o
tl.

-r 2.0
i--
o
~n
u. o
o 4.0 P d d-like
W
~- 2.0
o I ~ .
~ 0.2 Pd p-like

m o
~ 0.2 Pd s-like ~ _
I--
N o
z 0.4
bJ
H s-like
a

0.2
,, ,zg , ,,
0
-I0 -5 EF:O 5 I0
ENERGY(eV)

(i) The appearance below the metal d-bands of metal-hydrogen bonding states
which, as discussed later, have been observed by several spectroscopic
techniques. These states are mostly already filled in pure Pd; in the hydride they
are lowered in energy by the H potential and hybridize strongly with H s-states.
However, metal states of p symmetry which are empty in pure Pd, also
participate in this metal-hydrogen bonding band. They bring about 0.24
electrons below Ee. (ii) The filling of the metal d-bands. The Fermi energy of
PdH is found to lie in the metal sp-band in a region of low DOS. This explains
the drastic reduction of the electronic specific heat and Pauli susceptibility
upon hydrogenation. (iii) The lowering of empty metal states below Er results in
a Fermi level position in the hydride which is necessarily closer to the d-bands
than predicted by the protonic rigid-band model. Assuming that the metal
d-bands are not affected around EF, only 1--0.24=0.76 electrons should be
accommodated above the Fermi level of the metal. It is interesting to notice
that experimentally the hydrogen desorbs easily for concentrations larger than
x=0.7; this seems to correspond to the complete filling of the 0.36 holes in the
d-bands, if one takes into account the 0.24 additional electrons at low energy.
For x > 0.7 the Fermi level of the hydride lies in the metal sp-band which has a
low DOS and it is energetically not favourable to shift the Fermi level towards
152 M. Gupta a n d L. Schlapbach

r',

w~ r~

g~
+1
¢..q

t<o6
t-~ ee~
e,i ¢-I e,i r-i

r..) t"q t'~ t'q

e,,~ ca

o~
"7
r--: ~ t'< t " ~ "~.- ',,~ o ~
m

t~

e-q. ~ .
¢-,q •e:l- ¢',q t"-q

3: r.~
r.~r~

I
ce

i
u5

[...,
 Electronic Properties 153

r ' ~ r'-.,-

e~
cq
e~ ~ eq

~ ~ ~ z z zzz
154 M. Gupta and L. Schlapbach

g.,,

oo o~
e,i ~ ¢q t-i e,i

¢,
t¢3

r/~ ~ r,~ r / ~
0

0 e.
& o

ell ~E
 Electronic Properties 155

i• "
~'
t~C#~:
~~

3s ~

~'~'~'....~" ~ ~..~-'~-" ~. ~-

~.~£ ~

~a

~4 ~ ,.~ o
•~ ~ .~ ~

o o 6 ~

a ~ .I 3S 8 d ~ sssx~
156 M. Gupta and L. Schlapbach

higher energies. (iv) The metal d-bands are narrowed by the 5% lattice
expansion effect; however, most of the d-states are not affected by the metal
hydrogen interaction, with the notable exception of the states located at the
bottom of the band which participate in the M - H bonding.
In recent calculations, [5.35, 73-75] analysis of the total DOS either into its
angular momentum components s, p, d at the metal and at the H sites, as shown
in Fig. 5.1, or analysis in terms of LCAO's are now available. Such details are
very important in interpreting the x-ray emission and absorption data and shed
some light on the hybridization of the states in the full energy range. The nature
of the low-lying bonding states for example is made very clear by means of such
analysis. It has been shown that when H occupies octahedral interstices of the
fcc metal lattice as in PdH, the H s-states interact with the metal d-states of e0
symmetry.
The results of calculations of the DOS of nonstoichiometrie Pd hydrides
[5.46] are given in Fig. 5.2, Ev is found closer to the top of the Pd d-bands.
Eastman et al. [5.76] observed in early room-temperature photoemission
studies of Pd films exposed to hydrogen a strong additional emission at 5.5 eV,
which was interpreted as the hydrogen induced band, but their spectra show no
decrease of the emission at E v. Other groups failed to see even that hydrogen
induced band due to the rapid desorption of hydrogen (see introduction of
[5.77]). These contradictions were eliminated by the first low-temperature
photoemission studies [5.77] of (a) properly hydrogenated Pd which showed
the hydrogen induced band at 8 eV, a strong decrease of the emission at Ev, a
decrease of the bandwidth of ~ 10% and a very weak core-level shift, in very
good agreement with theory and of (b) oxygen dosed Pd which showed the
5.5 eV emission observed in the earlier work.
Within the last couple of years the occupied and empty parts of the valence
band and the core levels of properly prepared PdHx (x = 0.6 to 0.8) were studied
using UPS [5.77, 78], XPS [5.77-79], EELS [5.79], SXES [5.80] and low
energy BIS [5.78]. As compared to pure Pd the following features were
observed (Fig. 5.3a, b). The UPS, XPS and SXES spectra reveal a weak and
rather broad hydrogen induced band centred at 7.5 eV to 8 eV. EF moves out of
the Pd 4d-band into the flat sp-band which is clearly seen from the
disappearance of the peak at E v in the BIS spectra (empty 4d-states) and from
the decrease of the emission at Ev by a factor of about 3 in the UPS and XPS
spectra. In agreement with the filling of the 4d-band and the reduction of the
DOS at E v, the shake up satellites of the Pd 3d and 2p core levels disappear and
the core levels become less asymmetric. The width of the 4d-band decreases by
almost 1 eV. The shift of the core levels is very small (0.15 + 0.10 eV) indicating a
very small, transfer of charge, if any, in agreement with theory.
A slightly larger core-level shift (0.3 eV) was found in a recent photoemission
study of Pd hydride which was prepared by ion implantation technique and
thus contained higher hydrogen concentration [5.81]. Compton profiles of
PdHo.72 confirm the existence of a hydrogen induced bonding band below E v
[5.63].
 Electronic Properties 157

30. C EF_ Fig. 5.2. Total and partial DOS

^20,[
~d. x:,.o ,~/Jl i of nonstoichiometric PdHx
[5.73-1 and number of electrons

~ 0.:

0"5!
cc
o.,'

Pd dl
~10.[
~2
10,(
Pd d2 ~

go.[

~20.[ P°" '=°'° ~ / / ~ i

~o.o

.,~
H's ENERGY (RY)
~ 0.; ,,

~o.s
~m 0.5 r_d p t

t Pd dl __

Lf
30.0
PdH X=0.6 ~ / ~!
^20.0
d
~lo.o

H S ENERGY (RY)

N 2~--~'" ~--~'~ '

~Q.S i

~ 0.5 Pd p "

~2o.ot Pcl (:11

~I0.0~
Pd (:12 ~
158 M. Gupta and L. Schlapbach

i i // i r ~
a ups r, uv- ISDCMR

itti hv • 9,7 eV
/ i
/

4 0 8 eV s/ /"

. ~-

ZlZ ev\\\\ , . . . , i ) l"' ....

408 ev

21 e
/i /
/ I /
b

/
x.e,
L I , i , , I , , , , I--}, I . . . . I . . . . I ~ i i h i i i ,, r i
-IO -5 o o 5 Io 34B 3a6 344 342 340 938 336 ~ " I0 5 O=E F
ENERGY REL,TO EF [eV] ~BINDING ENERGY {eV)

Fig. 5.3. (a) UPS and UV-isochromat (BIS) spectra of the occupied and empty part of the
valence band of Pd and PdHo.6s [5.78]. Upon hydride formation the Pd d-band becomes
narrower and shifts away from E~ (UPS emission decreases, UV-isochromat peak at E z
disappears). Hydrogen-induced bonding band appears at 8 eV in UPS and also SXES I-5.80]
spectra. (b) XPS spectra of the valence band and of the Pd 3d core-levels of Pd and PdHo. 65,
Upon hydride formation the width of the Pd d-band and the emission at EF decrease, the
hydrogen induced bonding band appears at 8 eV, the Pd 3d core-levels shift to 0.2 eV larger
binding energy and become less asymmetric I-5.78]

The only remaining controversy between theory and these spectroscopic

results concerns the empty antibonding states which yield a prominent peak
5 eV above Ee in the calculated DOS, but which are not seen in the BIS
spectra (see Fig. 5.3a).
H dissolves in Ni endothermally (e-phase) and nickel hydride (B-phase) is
rather unstable at room temperature. High pressure hydrogenation, cathodic
charging and ion implantation have been used successfully to prepare NiH x in
the composition range 0 < x < 2.0. Low temperature ion implantation leads to a
supersaturated e-phase at least up to x=0.85. The magnetization decreases
linearly with H concentration and reaches zero at x = 0.8, in agreement with the
fact that the electrolytically charged ]?-phase is nonmagnetic ([5.823 and refs.
therein.)
The main differences between the two isoelectronic compounds PdH and
NiH originate mostly from the differences already existing in the metal
matrices. The Fermi level of NiH [5.83] is found to be closer to the top of the
d-bands than in PdH; thus the DOS at EF is larger and the "d" character of the
states is more pronounced due to the fact that the number of holes in the
d-bands is larger in pure Ni than in Pd. This result has consequences for the
 Electronic Properties 159

hv (eV) 20 '30 40 50 60 80
i I i l J I
X,.~..... K £ X K F
E F = O ~ F

m
Z
w "

Z
~ _
z
N _
I

ff K I" X K F

K X K r
Fig. 5.4. Comparison of calculated energy bands of Ni and NiH, plotted along several high
symmetrydirectionsof the BZ [5.83], and energybands of ion implanted NiH~measured by
angle resolved photoemission [5.85]. The metal-hydrogen interaction leads to a strong
lowering of the Z"1 branch

Fermi Surface (FS) anisotropies and plays a role in determining the strength of
the electron-phonon interaction. In the hydride NiH, the low-lying metal-H
bonding states have a larger tail of metal s-character than in PdH; this is again
reminiscent of the differences in the host band structure. Photoelectron
spectroscopic investigations of electrolytically charged Ni hydride have failed
so far due to the fast desorption of H across a clean Ni surface even at ~ 100 K
and a hydrogen partial pressure of 10-5 [5.84]. However, low temperature ion
implantation in situ allowed the study of the valence band of a hydrogenated
Ni single crystal by angle resolved photoelectron spectroscopy along the F K X
direction of the Brillouin zone [5.85] (Fig. 5.4). In rather good agreement with
band structure calculations the band $1 is strongly lowered in energy, e.g.
from 3 eV in Ni to 8 eV in Ni hydride at the X~ point. Further information was
obtained from the 6 eV and 13 eV satellites of the valence band of Ni. Their
intensity depends strongly on the number of 3d-holes. In Ni hydride the 13 eV
satellite has disappeared completely and the intensity of the 6 eV satellite has
decreased by a factor of two, in agreement with a filling of 3d-states.
Quantitative conclusions cannot be drawn as the hydrogen concentration of
the sample was not known.
The empty part of the valence band of NiH0.85 was studied using XANES.
The structure found in the absorption edge reflects quite nicely the structure for
the p-projected calculated DOS [5.71] after an expansion of the theoretical
160 M. Gupta and L. Schlapbach

energy scale by 5 %, the origin of which is not yet clear. Similar energy rescaling
has been tentatively justified in terms of the influence of the core hole (see
[5.71]).
b) Fermi Surfaces (FS)
Since the Fermi level of stoichiometric PdH and NiH lie in the metal sp-band,
the Fermi surfaces of the hydrides are expected to be qualitatively similar
to those of the noble metals; they are multiply connected surfaces having the
shape of a warped electron sphere centred at the fcc BZ centre with necks along
the L(111) direction [5.74, 83]. Since the Fermi level of the hydrides is closer to
the top of the d-bands than it is for noble metals, the FS anisotropies and the d
character of the states at E r are expected to be larger. FS cross-sections have
been calculated for PdH and NiH. Fermi surface nesting features have been
invoked to explain the observed concentration-dependent short-range order of
PdDx for 0.7 < x < 0.78 in the vicinity of the (1, 1/2, 0) point [5.86]. Although the
exact dimensions of the nesting vector are questionable since they were
obtained by applying the rigid-band model to the bands of stoichiometric PdH,
this suggestion deserves further examination since it is based on more general
grounds and has been applied successfully to other materials.
Experimental FS studies by de Haas-van Alphen measurements are only
available for very dilute hydrides CuHx [5.87], PdHx and PdDx and PdTx
[5.88]. The hydrogen concentration dependence of the Dingle temperature
dTo/dx of some high-symmetry orbits of dilute CuH~ shows that the larger
values of dTD/dx are observed for the neck orbits in the L(Ill) directions.
Indeed, this result agrees with theoretical calculations since the corresponding
4p-branch of metal states at the L2 k point has an s-symmetry at the
interstitial octahedral site and is thus expected to be strongly affected by the H
potential.
The modifications of some extremal FS cross-sections of Pd in the presence
of dilute amounts of H, D, and T [5.88] reveal large isotope effects for the small
hole ellipsoids around the X and L points of the fcc BZ while the large FS sheet
around F is isotope independent. This experimental result raises the question of
the possible role of the zero-point motion of H (and D) on the band structure
[5.62]. However, from the available FS measurements, Bakker et al. [5.88]
estimate that in the dilute limit, the DOS at EF should be almost isotope
independent.

c) Study of Other Observables

Total Energies. Calculations of equilibrium lattice constants and heats of
hydride formation as defined in (5.5) are available for PdH and NiH in the work
of Williams et al. [5.32] and Switendick [5.35]. The calculated lattice constants
are in good agreement to within 1%-1.5% of the experimental values. The
calculated heats of formation of PdH are -50.16kJ/(moleH) [5.32] and
- 40.13 kJ/(mole H) [5.35] compared to the experimental result
- 20.06 kJ/(mole H). For NiH the calculated value is - 33.44 kJ/(mole H)
 Electronic Properties 161

compared to the positive experimental value + 5.02 kJ/(mole H). Accuracy in

the absolute values of AEMn is difficult to achieve. As seen in (5.5) a result of a
few tenths of an eV is expected as the difference between two large values EMa
and EM each being of the order of 100 000 eV 1,5.35]. This is thus at the limit of
the accuracy of the calculations. In addition, the absolute value of AEMn
depends sensitively on the theoretical value used for the energy of the hydrogen
molecule. Nevertheless, the calculations reproduce the observed trend of the
greater stability of PdH versus NiH. In this connection it has been noted by
Switendick I-5.35] that the endothermic value of NiH could be accounted for by
the difference between the value for magnetic and nonmagnetic configuration
(used in the calculation) of pure Ni metal. From their calculation V~lliams et al.
I-5.32] trace back the increased stability of PdH compared to NiH to differences
in the chemical potential of the host metal.

Hyperfine Interactions. Calculations of wavefunctions and their use in the

prediction of various observables have been rather scarce in the field of metal
hydrides. We shall however mention here a theoretical study of the contact term
of the hyperfine interaction at the proton site in PdH which gives interesting
information on the s-character of the Fermi surface states at the proton site.
The theoretical study was motivated by the interpretation of proton spin-
lattice relaxation times "q, measured in the temperature range between 3 and
77 K where the proton does not diffuse [5.89].
The nuclear spin-lattice relaxation can be viewed as a spin-flip scattering of
the conduction electrons due to the magnetic hypertine interaction with the
nuclear spins. The contact term of the relaxation rate is often written in the
convenient factorized form:

r 11 = 41rT~kBT(N~ Hhfs)2, (5.8)

where N, is the partial s-type DOS at E F for one spin direction and Hhf s the
hyperfine field at the H site.
The theoretical value obtained by generating 15 000 wavefunctions at the
FS leads to (N~Hhf~)theory=4.4X1015Gerg - l which is to be compared to
(N~Hh,)exp=(4.0+0.1)x 1015Gerg -1. The theoretical value is quite satis-
factory since it lies within 7 % of the experimental error bar; it provides valuable
information about the H site s-character of the states at Ev. It is to be noted, as
discussed below, that this feature plays a role in the value of the matrix elements
of the electron-optical phonon coupling and thus on the superconducting
properties of PdH. The NMR experiments I-5.89] show that for non-
stoichiometric compounds corresponding to x = H/Pd = 0.7, z 1T = 70 ___3 s K;
this increase over the value z l T = 4 8 + 2 s K of stoichiometric PdH could
correspond, if other factors are equal, to a 21% decrease in the value of the
partial H s-DOS at EF. This trend is corroborated by theoretical calculations
1,5.46] on non-stoichiometric compounds. It is also reproduced by applying in
the vicinity of E F the rigid-band model to stoichiometric PdH.
162 M. Gupta and L. Schlapbach

Electron-Phonon Interaction and Superconductivity. The study of the electron-

phonon coupling constant 2 of PdH is particularly interesting in view of the
superconducting properties of PdHx (for x >0.8) and of the existence of an
inverse isotope effect on the superconducting transition temperature T~
[T~(PdH)--,9 K while T~ (PdD),-~ 11 K] I-5.90]. It seems now well-established
from superconducting tunneling experiments as well as from theoretical work
[5.73, 91] on An that the electron-optical phonon coupling is essentially
responsible for the large value of T~in PdH. The inverse isotope effect in T~ has
been ascribed to anharmonicity effects [5.73] in agreement with inelastic
neutron scattering data. However, other experimental results suggest that the
value of the anharmonicity is too small to account for the observed differences
in T¢ I-5.90]. Other possible mechanisms such as the role of zero-type
amplitude effects on the band structure have not yet been quantitatively
investigated.
Another interesting question is why NiH, which is isoelectronic to PdH and
diamagnetic, is not superconducting down to 1 K [5.92]. Indeed, calculations of
the electron-optical phonon electronic term r/n defined in Sect. 2 show a 57%
decrease of this term from PdH to NiH [5.91]. This is due to the decrease of the
s-electron density at the H site for states at EF, the Fermi level of NiH being
closer to the top of the d-bands than in PdH. Moreover, the temperature
dependence of the electrical resistivity [5.92] indicates that the optical phonon
modes in NiH are much harder than in PdH, the Einstein temperature being
O~(NiH) ~ 1.80E(PdH), a factor which further contributes to the decrease of T~.
Inelastic neutron scattering data [5.93] of NiHo.75 confirmed the higher
frequencies of the optical phonons in NiH than in PdH. The spectrum presents
a sharp peak around 88 meV and a broad shoulder extending up to 110meV.
Recent calculations of ,~ and Tc for P d - A g - H alloys using the CPA [5.94]
have confirmed the role played by the variation of the H s-type DOS at EF on
the value of r/H.
We can conclude that due to the large amount of experimental and
theoretical work performed on PdHx and to a lesser extent on NiHx and alloys
with noble metals, these systems appear to be quite well understood and
provide an example of the electronic properties of transition metal
monohydrides.

5.3.2 Hydrides of Ti, Zr, and Hf

The group IVB metals Ti, Zr, and Hf form stable metallic dihydrides of the
CaFz-type structure (fcc) over a wide range of concentrations I-5.1]. In this
structure, the H atoms occupy tetrahedral interstices of the fcc metal lattice. For
hydrogen concentrations approaching x = 2 and below a critical temperature, a
tetragonal distortion of the lattice occurs. Most experimental and theoretical
studies of the electronic structure of hydrides of Ti, Zr, and Hf were performed
in relation to this fcc-fct lattice distortion.
 Electronic Properties 163

Early calculations [5.4] have shown, as proposed by Ducastelle et al. [5.95]

from their study of the temperature and H concentration dependence of the
electronic specific heat, thermoelectric power and magnetic susceptibility, that
the cubic to tetragonal distortion can be ascribed to a Jahn-Teller effect. Since
then, several non-self-consistent calculations for Till2 and ZrH2 by Gupta
[5.96], for Till x by Kulikov et al. [5.97] Fujimori and Tsuda [5.98] as well as self-
consistent results for Till2, ZrH2, HfH2 by Papaconstantopoulos and Switch-
dick [5.99] and studies of substoichiometric hydrides of group IVB metals
[5.46] have become available. Improvements in the numerical accuracy of the
DOS led to the conclusion that the Fermi level of group IVB metal cubic
dihydrides fails in a peak of the DOS which araises from a flat branch of
degenerate metal d-states in the FL direction of the fcc BZ. The quadratic
distortion lifts this degeneracy and leads to a lowering of the Fermi level and of
the DOS at Ev in agreement with a variety of experimental data mentioned
previously such as electronic specific heat [5.95, 100] and positron annihilation
studies [5.65]. N M R investigations of TiHx and ZrH~ for 1.5 < x < 1.9 have
been performed by Korn [5.101], and Bowman et al. [5.102]. The spin-lattice
relaxation rate and the Knight shift also indicate that the DOS at E F is lowered
by the fcc--+fct distortion. The strong modification of the states at E F associated
with the Jahn-Teller effect is also evidenced by the sign change of the Hall
coefficient from negative in the cubic phase to positive in the tetragonal phase
[5.103]; the sensitivity of the FS geometry to the distortion appears from the
analysis of the FS cross-sections of cubic Till2 [5.96]. Work function
measurements [5.104] reveal a discontinuity of -~ 0.2 eV accompanying the fcc
~fct transition of TiH~ and ZrH~.
Wavefunction calculations and partial-wave analysis of the DOS are now
available [5.46, 96, 99]. They have revealed clearly the details of the M - H and
H - H bonding and the coupling of the H s-states with the metal d-states of the
tzg symmetry when H occupies the tetrahedral interstices of the fcc metal lattice.
These calculations show that besides the low-lying metal-hydrogen bonding
band observed in monohydrides, a second band due to H - H and metal-H
interactions appears in dihydrides and leads to a double-peak structure in the
DOS below the metal d-bands. In group IVB metal dihydrides, this second low-
lying band overlaps the metal d-states.
Thephotoelectron spectra of MH~ x.6 (M = Ti, Zr, HI) obtained by Weaver
et al. [5.105] are plotted in Fig. 5.5. The d-band derived emission extends within
3 eV from EF while the hydrogen induced bands centred at -,~5.5 eV below EF
show, in agreement with theory, a double-peak structure, the second peak being
observed below 7 eV. Direct comparison with calculations is however hazar-
dous since the samples correspond to x ~ 1.6. The spectra taken for ZrH~,
x = 1.63 and x = 1.94 reveal the change associated with the fcc~fct distortion
especially around EF and also below 7eV. Self-consistent calculations on
dihydrides [5.99] locate the H induced states at 4.9, 5.4, and 5.7 eV respectively
for Till2, ZrH2, and Hfl-I 2 while non-self-consistent results [5.96] locate these
states about 1 eV lower. The effect of iterating to self-consistency is a lowering of
164 M . Gupta and L, Schlapbach

i 1 i
7,0
i i
6,3
i i i 1 i i i i
Fig. 5.5. Photoelectron spectra of
- - H F H I 5•6 Expl ai HfH 1.56, ZrH 1.63,g r i l l .94, and Till 1.5o
II I'l 55
after background subtraction com-
- - E r H 2 DOS It t, i' pared with the calculated DOS of ErH 2
II, IA and Till 2. The spectra for ZrH~ reveal
= I' Jx \.., ,I the change in the bandstructure as-
sociated with the fcc fct distortion
[5.1053
i h.2, ¢ , ,
I-- t
5.3
't.. \k S / \
--
o) ZrH x Exp! 78 ^
Z
U.I
--x-1.63 I' Z2 , ] ¢ ~ EF
--~-1.94 /,,l..j \ o
(~ III

I ,"/ \
0" i ,f." I i I I I \/ I ~ I , t
--Villi.50 Exp' l'i II
II 5.1 II
--Till z DOS I i I',
I~A At
"7'31

12 I0 8 6 4 2 O=E F
ENERGY BELOW E F (eV)

the d-bands and o f E v with respect to the metal-H bonding states. This lowering
is about 0.9 eV for ZrH2, but it amounts to 1.4 eV in TiH2. A peculiar ordering
of the H - H antibonding states at the BZ centre has been obtained in self-
consistent calculations on Till2 [5.99]. This is, however, not observed in self-
consistent results on other dihydrides [5.35].
An XPS analysis of the valence band of an oxygen contaminated sample of
ZrH1.7 reveals the hydrogen induced band at 7 eV [5.106]. All core levels, as
measured by different groups using UPS and XPS, shift 0.5-1.0eV to larger
binding energy, consistent with charge transfer to the hydrogen site (Table 5.1).
This result agrees with the trends observed in theoretical calculations. Apart
from this shift, the core level spectra do not show any hydride effect• The Ti 2p
core level was shown to shift approximately linearly with hydrogen con-
centration [5.107].
In soft x-ray L3 emission spectra ofZrHx (x = 0.5, 1.0, and 1.9) and ofTiHl.s,
hydrogen induced bands at 7 eV developed at the expense of the emission
just below E F [5.8, 108]. The implantation of 8 keV hydrogen into Zr [5.109]
results in the appearance of a distinct peak in XPS valence band spectra at
3.4eV, which was attributed to a (Zr-4d-H-ls)-band. As the position of this
peak does not agree either with band structure calculations of the dihydrides or
 Electronic Properties 165

Ti K edge
Zr K edge r ~
1

- Ti - Zr

0 0
0 ZO E- Eo (eV)/.,O -zo o 20 ~0
E- Eo (eV)
Fig. 5.6. XANES at the K edges of Ti, Till1:9, Zr, and ZrH 1.98showingfeatures of the empty
part of the DOS above EF [5.71]

with photoelectron spectra of ordinary bulk ZrH x, the authors suggest that it
could be due to different sites of the implanted hydrogen. However, upon
annealing of the sample to 600 °C the peak does not disappear nor shift to the
position of the hydrogen induced band at about 7 eV, but rather becomes more
intense. The exposure of clean Zr to 3000 Langmuir hydrogen
(1L=10-6sTorr) at room temperature results in rather weak effects in
photoelectron spectra which are qualitatively similar to those observed for
hydrides. It does not lead to a significant hydride formation [5.110]. XANES
K-edges of Ti and Zr hydrides (see Fig. 5.6) have less structure than do those of
the corresponding pure metals [5.71], a fact which has not yet been explained
by DOS calculations in this energy range.
Theoretical studies of the total energy of Till 2 versus lattice constant [5.35]
reproduce the existence of a minimum in the total energy curve. The lattice
constant is obtained with an accuracy better than 1.5% and the heat of
formation has been improved considerably over previous results. However, a
detailed analysis of the theoretical calculation reveals the sensitivity of total
energies to the approximation used for the treatment of core electrons. The
theoretical results [5.35] range from - 35.53 to - 52.25 kJ/(mole H) depending
on the core approximation used while experimental data range from -44.73 to
- 61.86 kJ/(mole H).
The electron-phonon coupling constant has been caclculated for dihydrides
of group IVB metals [5.91, 111]. Since in these dihydrides Ev falls at the bottom
of the TM d-bands, the value of the H s-DOS at Ev is very small, unlike in PdH
and to a lesser extent NiH. Using (5.7) one obtains negligible values of the
electronic term of the electron-optical phonon coupling ~/n.Moreover available
neutron scattering data [5.I 12] and a study of the temperature dependence of
the phonon resistivity [5.113] indicate values of the optical phonon frequencies
much larger than those in PdH. This leads to very small values of the electron-
optical phonon coupling '~H.Moreover, the electron-acoustic phonon coupling
is found to be smaller than in the corresponding pure metal. Thus, the
dihydrides are not found to be superconducting, in agreement with experi-
166 M. Gupta and L. Schlapbach

mental data [-5.90]. An increase of Tc of pure Zr (To = 0.7 K) has been reported
[5.114] after implantation of small doses of H and D with values of T~= 3.14 K
for H and 4.65 K for D. Nevertheless this observation does not contradict
previous data nor theoretical results since it concerns disordered dilute hcp
phase for which the electronic structure is expected to be drastically different
from that of a stoichiometric fluorite structure dihydride.
CPA calculations on substoichiometric hydrides (x < 2.0) indicate that the
H-induced states become less sharp and decrease in height as the H
concentration decreases; however, they do not shift in energy. The position of
EF on the other hand appears to shift to lower energies by about 0.27 eV from
x = 2.0 to x = 1.6 for TiH~ resulting in a large reduction of N(EF) [5.46].
Theoretical studies of the electronic properties of group IV hydrides of
symmetry lower than cubic remain up to date very scarce. However, recently a
band structure calculation of the tetragonal v-phase of ZrH has been performed
[5.115]; substantial differences from the DOS of ZrHz have been obtained. A
detailed analysis of the evolution of the states with the removal of the H atoms
rules out fully the validity of a rigid-band behaviour applied to the bands of
cubic ZrH2.
The calculated DOS at E r is found to be 0.88 electron states eV-1 (Zr
atom)- 1 for the v-phase compared to the value of 1.76 electron states eV- 1 (Zr
atom)-1 for cubic ZrHz [5.35]. A comparison with magnetic susceptibility
measurements [5.116] seems to indicate that the orbital contribution to the
magnetic susceptibility is large in this compound.
Hydrides of Ti, Zr, and Hf are a nice example to demonstrate the effect of
structural distortion on electronic properties. These effects are clearly visible in
calculated DOS and in a number of observables however, the limited
experimental resolution prevents detection of detailed effects in PES spectra.

5.3.3 Hydrides of Nb, V, and Ta

The group VB metals Nb, V, and Ta form monohydrides in a rather
complicated diagram of ordered and disordered phases, and dihydrides [5.117].
NbH x, e.g., exists as monohydride at 300 K in the orthorhombic/?-phase and
below 200 K in the pseudocubic v-phase and as dihydride in an fcc structure
(presumably CaF2). In the monohydrides, the H atoms occupy tetrahedral
interstices of the bcc Nb lattice. Band structure calculations of the monohydride
were first performed for v-NbH [5.118]. A single and rather narrow peak was
obtained in the DOS of the hydrogen-metal induced states, as usually observed
for other monohydrides. Photoelectron spectra ofNbHl, o, VH~ .o, and TaHo.8
however present a double-peak structure in the hydrogen induced states at
~ 5.5 eV and ~ 7.5 eV [5.118]. In fact very early SXES investigations of VD0.7
[5.119] already revealed a double peak at ~ 6.5 eV and ~ 8.5 eV below EF in the
V - L 3 emission spectra contrary to more recent SXES spectra of NbHo.8 and
VHo.4, which show only one hydrogen induced band [5.120]. Hydrogen
induced states were also observed in electron energy loss spectra of NbH~ and
 Electronic Properties 167

2.5 i i i i i I
Fig. 5.7. Calculated DOS in units of
o) ¥-NbH / ~ [states/eV unit cell] for (a) pNbH and
2.0 (b) /~-NbH compared with photo-
electron spectra. The experimental
1.5 /~ Expt. 11-I / / curve has been shifted in (b) to line up
with the DOS. The DOS curve cal-
I.O culated for /~-NbH shows the double
I--o.5 / peak structureof the hydrogen-induced
bonding band, in agreement with ex-
CO

~o.o :i-" \ y., E.,,,, periment I-5.34]

>,-
I-- b) ,B-NbH / ~
4.0

3.0 /t
1,00:~ , ~F' THEOI'~
-I0.0 -8.0 -6.0 -4.0 -2,0 I0,0 2.0
ENERGY(eV) EF, EXP.

VH x of unknown concentrations at 4 and 7 eV respectively [5.121] but with less

structure than in the photoelectron spectra. Several explanations were proposed
to account for the unusual double peak, among them the formation of a surface
dihydride. However, recent first principles pseudopotential calculations [5.34]
of both fl- and T-phases of NbH yield a more probable reason. As shown in
Fig. 5.7 the calculated DOS of the fl-phase is in good agreement with the
photoemission data. Although the orthorhombic distortion was ignored in this
calculation, the width and double-peak structure in the hydrogen induced
states is reproduced; it has been ascribed to a splitting of the bands due to
shorter distances between the H atoms in the fl-phase. These results demon-
strate the sensitivity of the hydrogen derived bands to the crystal structure and
lattice parameters.
The empty DOS was studied by x-ray absorption for monohydrides and
deuterides of V [5.122] and by XANES for VDo.72 and NbHo.93 [5.71].
Upward shifts of the K-edge and a suppression of fine structures j ust above the
edge were observed and explained by the filling of unoccupied d-states by excess
electrons of hydrogen and the lowering of empty p-states to a position below Ev,
respectively.
Surprisingly, the PES spectra of TaHo.8, Nb4D3, and VH 1.0 [5.118] do not
show a shift of the p core levels, in contradiction to the theoretical calculations
which indicate some charge transfer from the transition metal to the H sites.
The p- and d-spectra of NbHx of unknown composition produced by ion
implantation technique show a shift of 0.8eV [5.109].
Total energy calculations for both the/% and ?-phases of NbH have been
performed by Ho et al. [5.34]. A good agreement with available experimental
168 M. Gupta and L. Schlapbach

results is reported for lattice constants, bulk modulus (within 4.5 %) and heats of
formation. However, as indicated by the authors, the calculated heats of
formation are small, a few tenths, of an eV per H atom, thus numerical
accuracies and effects such as neglecting the orthorhombic distortion and
contributions from zero-point vibration energy of hydrogen in the solid and in
the hydrogen molecule can become critical. From total energy calculations as a
function of the H atom position in the unit cell, the optical phonon frequencies
have also been obtained for the first time for a transition metal hydride [-5.34].
The results are in rather good agreement with neutron scattering data [5.123];
the importance of anharmonic effects was also pointed out.
The fluorite structure dihydrides NbH 2 and VH 2 have been studied
theoretically [5.96, 99, 124]; an interesting behaviour is the disappearance of
superconductivity in the high To group V metals upon formation of hydride
phases [5.90] which was confined by recent heat capacity measurements
[5.125]. The essential features of the bands and DOS of NbH 2 and VH2 are
very similar to those of the isostructural group IV metal dihydrides discussed
previously; the Fermi levels of group V metal dihydrides fall however at higher
energies since one more electron needs to be accommodated in the metal
d-bands. The very small value of the H s-type DOS at E Fin conjunction with the
large values of the optical mode frequencies O~opt>120meV observed by
inelastic neutron scattering [5.123] lead to very small values of the electron-
optical phonon coupling constant. A drastic weakening of the electron-
acoustic phonon term ?~metalbetween pure Nb and NbH2 has also been obtained
[5.96]. From a detailed analysis of the structural changes in the metal lattice,
and from values of the DOS at EF [5.91] this reduction in ~me,,f has been
ascribed mostly to a weakening of the metal d-d interactions due to the lattice
expansion rather than to a density of states effect. The observed hardening of
the acoustic phonons upon hydrogen absorption [5.112] leads to a further
decrease of the electron-acoustic phonon coupling and explains why the group
V metal dihydrides are not superconducting down to ~ 1 K. The studies of
Knight shifts and nuclear relaxation rates in NbH2 [5.126] indicate clearly in
agreement with theory the small value of Ns(Ev) compared to Na(Er). Further
information concerning the variation of the DOS at E v as a function of
hydrogen content in VHx and NbH~ with x up to 1.93 is available from the
detailed low temperature heat capacity measurements [5.125]; in the absence of
theoretical results on hydrides in the full concentration range covered by the
experiment these results have been successfully compared with results on V
metal and with data on V/Cr-alloys.

5.3.4 Hydrides of Rare Earth Metals Including Sc, Y, and La

The trivalent RE metals as well as Sc, Y, and La are known to from metallic
dihydrides REH 2 and semiconducting (or poorly metallic) trihydrides REH 3
I-5.127, 128], with the exception of Sc, which probably does not form a
trihydride. The dihydrides are of almost ideal cubic CaF2-type structure where
 Electronic Properties 169

the tetrahedral sites are roughly filled and the lattice parameter is increased by
typically 8% as compared to the elemental metals. The subsequent occupation
of octahedral sites leads to cubic BiFa-type and hexagonal HoD3-type
structure for the trihydrides of the light (La to Nd) and heavy rare earth and Y,
respectively. Ce does not seem to be an exception. The divalent RE metals Er
and Yb form semiconducting dihydrides.
The phase diagram for the intermediate range REH 2 to REH a is not yet
known in detail. A generalized diagram [5.127] shows a solid solubility range of
the REH 2 phase from x ~ 1.9 to x ~ 2.3, then a range of coexistence of the REH2
and REH3 phases, and finally the REH3 phase. The light RE metals show a
rather extended solubility range of the REH2 phase which ends in a
tetragonally distorted phase at REH2.33 with slightly contracted lattice
parameters [5.129]. Very recently the tetragonal distortion was observed in
Cell x at room temperature from CeH2.2a to CEH2.64 and evidence for a
miscibility gap from CEH2.56 to CEH2.64 was given [5.130]. In fact tetragonal
distortions have been observed in many RE hydrides over the last ten years;
however, at low temperature only, i.e. in the range from 200K to 250K
[5.131-136]. They are thought to be caused by hydrogen ordering effects.
Down to 4K no further structural transitions have so far been observed.
Possibly the range of concentration for which the tetragonal distortion occurs
is more extended at low temperatures. Tentative phase diagrams for the range
Cell2 to Cell3 including low temperatures can be found in [5.130, 134].
The transition of the electronic structure from rare earth metals with a
(5d6s) 3 conduction band to metallic dihydrides and further on to nonmetallic or
at most poorly metallic trihydrides, obtained in the calculations (Fig. 5.8), as

i i i

48 LoH 2 LoH 3
Oc
, , , , '/t 16
f* t.o
// / Z
t 0
r~ rv-
-. 36
O3
W b3
_J
ta.l
03 h
ffl 24 8 o
[.U ¢r
l W
rn
I Zg

4.

Z ....
"' 0 I I r i I I I I

..3
0 .2 .3 E F .5 .6 .7 .8 R 0 .2 .3 .4 E .6 .7 .8 .9 Ry
I I l l l I I I I I [ i i I i L I i 1 i i 1

-4 -2 0 2 4 G eV -6 -4 -2 0 2 4 eV

F i g . 5.8. T h e total D O S of L a H 2 and L a H 3 ( , left hand scale) and the total number of
electrons ( - - , right hand scale). D O S units are [states of both formula unit] spins/Rydberg
E5.148]
! 70 M. Gupta and L. Schlapbach

' ' I ' ' ' I + ' "~1

~ a

// ~
J / ,,p
~\ y

~-- Yf
yH z
b
LoHI98
L ^',~
nu=~u
~5.7~ ~
/;~
h,~-.--J[I \
It \
4.
--Expt
- - DOS- Gupto
and Burger
YF
yH 3

~k,~a H2.2
~~4.7
H2.9
/ LoH248

~ Ce

,~CeH2j

.~CeH2.9
-4.9
I ,'I ~ Expt
LaH2.89 ,'i --GB t
hv= 1 9 B ~ -'-MH DOS
k Pr

~,~.Pr H 2.1

(Mg Ka) , ~ ..~ PsH~

..... I J.l..~;-'~ "~ I n e I N2~-~-..=--~ ~1

-8 -4 O=EF -I0 -8 -6 -4 -2 0
--BINDING ENERGY(eV) Initiol-Stole Energy (eV)

Fig. 5.9. (a) XPS spectra of the valence bands of Y, La, Ce, Pr and their dihydrides and
trihydrides. The dihydrides are metallic but show reduced emission at EF, whereas the
trihydrides are non-metallic. The hydrogen-induced bands appear at 5-6 eV (4f emission of
cerium hydrides at 2 eV; that of praseodymium overlap with the hydrogen-induced band)
I-5.11]. (b) Photoelectron spectra for LaH1.gs, LaH2.4~, and LaH2.s9 after background
subtraction compared with calculated DOS for LaH2 and LaH 3 (Gupta and Burger [5.148-1)
and LaH 3 (Misemer and Harmon [5.1491). The calculated bonding band centers agree with
experiments but their widths are significantly narrower [5.138]

well as the appearance of the hydrogen-induced bands centred around 5-6 eV

are easily seen in photoelectron spectra (XPS) of hydrides of Y, La, Ce, and Pr
(Fig. 5.9a). The double-peak structure of the hydrogen-induced band, typical
for dihydrides, is nicely resolved in the low photon energy spectra of LaH1.9s
(Fig. 5.9b) and Cell2. r The results of all valence band studies of rare earth
hydrides by PES are summarized in Table 5.1 [5.11,137-145]. F r o m optical
studies of interband transitions in ScH2_ x, YH2_x, and L u l l 2 _ x it was
concluded that the octahedral site occupation is not negligible in the dihydrides
[5.72].
The essential ideas put forward by Switendick from his early band structure
sutides of stoichiometric YH2, YH3 and rare earth di- and trihydrides [5.4]
have been confirmed by more recent calculations for Yell 9 and Y4H11 [5.146],
 Electronic Properties 171

ErHz [5.147], LaH2 and LaH 3 [5.148-150] Sell2 and YH2 [5.151], Cell 2,
Cell 3 [5.150, 152]. For all these band structure calculations the 4f electrons are
omitted.
Due to differences in the nuclear charge of the metal atom, the low energy H
induced states overlap the metal d-bands in the case of Sc and Y while a gap is
observed in the case of LaH2 and RE dihydrides. This explains why the divalent
RE dihydrides are semiconductors. For trivalent RE dihydrides only one
electron occupies the bottom of the metal d-bands; the DOS at Ev is much
reduced from its value in the pure metal in agreement with trends observed in
the electronic specific heat [5.127, 132, 136] and magnetic susceptibility data
[5.127]. In the trihydrides, a third band appears at low energy leading, as
observed in the photoemission spectra (Figs. 5.9a, b), to a further depletion of
the RE metal d-band. Wavefunction analysis shows that the additional third
band of RE trihydrides has essentially a metal des and octahedral H
s-character while the first two low-lying bands gain also, besides the metal d-tzg
tetrahedral H s-character dominant in the dihydride, an additional metal d-eg
octahedral H s-contribution.
Most theoretical calculations on RE trihydrides indicate the presence of a
gap between the three low-lying bands and the higher metal d-states, with the
exception of Kulikov et al. [5.150, 153, 154] who obtained an overlap;
their calculation, however is based on the use of a model Hamiltonian and the
authors themselves warn that the results are only qualitative. The value of the
energy gap in LaH3 appears to be reduced by the use of a less strong exchange
and correlation potential and by self-consistency effects. The theoretical value
of the gap - if any - remains an open question since it could be further affected
by spin-orbit coupling effects not fully included up to date. The use of a strong
exchange potential (ct = 1 in the X, scheme) leads to a narrowing of the bands
and results in a better agreement with the optical conductivity data a(og)
[5.138]. These data show that the onset of interband transitions occurs at
1.1 eV, in much better agreement with the value of 1.3 eV obtained from non-
self-consistent strong exchange potential calculations. The value of 1.75 eV
obtained from self-consistent Hedin and Lundqvist exchange correlation
potential calculation is too large. On the other hand, the strong exchange
potential which narrows the valence band width by 13% from the theoretical
value of 4.2eV [5.149] leads to a poorer agreement with the photoemission
experiments [5.138] which indicate a valence band width of about 6 eV. It has
been argued in this connection that the premature filling of octahedral sites in
LaH2 will lead to an increase in the theoretical band width [5.149]. The
narrowing of the bands lead to a DOS at EF for LaH 2 of 1.11 states/eV-cell
larger than the self-consistent value of 0.85states/eV-cell and in better
agreement with the experimental value.
The results of different band structure calculations and of measurements of
specific heat [5.132, 136] susceptibility [5.127], NMR Knight shift [5.155] and
photoemission [5.11, 137-145] show "in agreement" that the trihydrides are
nonmetallic or at most poorly metallic.
172 M. Gupta and L. Schlapbach
I I I I [ "' 1 I '" I t .v I [ i
Y3d3/2,5/2 Lo3ds/2 Ce3ds/2 Pr3ds/2

. . . . . I I I I I
162 160 158 [56 154 840 850 890 880 940 930 920
--BINDING ENERGY (eV)
Fig. 5.10. XPS 3d core-levels of Y, La, Cc, and Pr and of their dihydrides and trihydrides. The
evaluation of the chemical shifts from the measured core-level shifts is not straightforward as
satellites indicate screening effects [5.11]

Inverse photoemission spectra of light rare earth hydrides show weak

structures at 1-1.5 eV and 3-3.5 eV above Ev which can be attributed to empty
(spd) states [5.70].
All band structure calculations indicate that a charge transfer exists from
the metal to the hydrogen tetrahedral sites in the dihydrides of T M and RE
metals while the octahedral site in the trihydrides can be considered as neutral.
This general trend is corroborated by the experimental XPS data on the metal
core-level shifts obtained upon hydrogenation [5.11]; the core-level photo-
emission spectra (Fig. 5.10 and Table 5.1) are all shifted to larger binding energy
as compared to the elemental metals. The apparent shift from the metal to the
dihydride can be larger (e.g. Ce) or smaller (e.g. La) than the shift from the
dihydride to the trihydride. A quantitative determination of the chemical shift
from the observed spectra is not straightforward owing to final state effects in
the photoemission caused by screening of the core hole. A careful analysis
agrees with a charge transfer to the tetrahedral site and no further transfer to the
octahedral site [5.139, 140, 143, 144, 70, 156].
The theoretical values of charge transfers are still a matter of controversy. In
the K K R and APW methods, the charge transfer is best defined as the charge
difference inside a given M T sphere between the final self-consistent results and
the overlapping neutral atoms case. Such a definition which of course suffers
from the arbitrariness in the partition of the real space into several regions,
leads to charge transfers of the order of 0.2 electrons towards the hydrogen M T
 Electronic Properties ! 73

spheres at the tetrahedral sites. In tight binding and LCAO analysis, the
number of electrons per atomic orbital is used to define the ionicity; this
definition however does not take proper care of the spatial localization of the
charge; as example, such analysis leads to about 1.8 electrons in the Is
tetrahedral hydrogen in Cell2 which is certainly an overestimate of the ionicity
since if one looks at the interatomic separations in this compound, it appears
that the 1s wavefunction is quite extended spatially and part of the correspond-
ing charge would be located inside the Ce MT sphere.
The Fermi surfaces of the dihydrides are determined only by one band and
are therefore much simpler than those of the corresponding metals. Ignoring
relativistic effects, these surfaces are found to be multiply connected hole
surfaces formed of a warped cube centred around F with necks in the FL
directions [5.147, 148]. Strong nesting properties of this surface have been
noticed; the nesting vectors are along the 100 directions their exact dimensions
are expected to be somewhat modified by relativistic effects; nevertheless their
values remain close to the magnetic ordering vectors found experimentally by
polarized neutron scattering studies [5.157]. This result is already encouraging
and suggests that the magnetic ordering could be explained by the existence of
Fermi-surface induced divergences in the generalized susceptibility; further
theoretical investigations concerning the role of relativistic effects are needed to
confirm this point.
The drastic change in the characteristics of the FS states from the pure metal
to the dihydride together with the strong modification of the phonons has
several important consequences on the electron-phonon coupling and thus on
the superconducting properties as well as on the electrical resistivity. While fcc
La is a superconductor with Tc~ 6 K, LaHx with 1.8 < x < 2.36 does not show
any superconducting transition above 1 K [5.90]. The strong decrease of the
electron-acoustic phonon coupling from the pure metal to the dihydride
[5.158] is due (i) to a reduction of the electronic term ~/met,l which originates
from the decrease in the La d-d coupling due to the lattice expansion and to a
lesser extent to a reduction of the DOS at Ev (ii) to a strong increase in the
acoustic phonon frequencies [5.159, 160]. The Debye temperature essentially
doubles from La to LaH2.o3 [5.161]. This drastic change explains also the large
decrease observed in the room temperature resistivity of the dihydrides [5.162,
163]. Further modifications of the electronic structure with hydrogen con-
centration induce changes in the electron-phonon interaction; shifts in the
phonon frequencies have been reported from the study of infra-red and Raman
spectra of CeHx with 2 < x < 3 [5.164].
Whereas the electronic structure of stoichiometric dihydrides and trihy-
drides is rather well understood, there is considerable confusion in the
intermediate concentration range, particularly at concentrations approaching
trihydride composition and at low temperatures. A concentration dependent
metal to semiconductor transition was observed in Ce hydrides at x~2.75.
Furthermore the occurrence of a semiconductor to metal transition upon
increasing temperature was suggested. Results published up to 1979 are
174 M . Gupta and L. Schlapbach

i i I i i
Fig. 5.11. Low temperature photoelectron spectra of
i/ ~ Cell 2.-r Cell2. 7 for various photon energies h(o, showing the
/ L
hydrogen-induced band at 5 eV and the 4f-emission at
2eV. The intense peak at EF (--) grows below
,/i / T = 70 K, indicating a semiconductor to metal tran-
sition. Inset: Intensity of the peak at E v versus
temperature [5.168]
, , J
" /!
/ % I | i
/ A I I / T~w
i ~ It X,.2oe

..," ";AIA I I

,,\
/ ~ I llk..._.o,
/ ',../ ,,

, ~o 6'0 ~ov
8 6 4 2 O=EF
BINDING ENERGY(eV)

summarized in the review of Libowitz and Maeland I-5.127]. Never results on

specific heat and NMR Knight shift reveal poorly metallic or nonmetallic
behaviour at low temperature [-5.132, 155, 165]. But resistivity measurements
seem to indicate metallic behaviour at low temperature and less metallic or
semiconducting behaviour at T > 2 0 0 K [5.149]. The origin of this puzzling
behaviour is not known, and the fact that the metal-to-semiconductor
transition is observed already at x ~2.75 instead ofx = 3 is not fully understood
and cannot be explained by one-electron band structure calculations. Several
models were proposed: Fujimori and Tsuda [5.152] describe the temperature
dependent transition by a delocalized-loealized transition of electrons as-
sociated with the order-disorder transition of octahedral hydrogen, which is
related to the tetragonal distortion. Kulikov [5.153, 154] proposed that the
small band overlap and the resulting nesting of hole and electron pockets of the
Fermi surface lead to an excitonic insulator phase at low temperature (see also
discussion in [5.149]). Recent NMR results of Ce hydrides do not confirm a
temperature dependent transition. The results on the concentration dependent
transition are consistent with a Mott transition [5.166, 167]. Concerning these
theoretical models one should not forget that the metal-to-semiconductor
transition was detected in resistivity measurements of rare earth hydrides in
1959 and 1972 [5.127], i.e. at a time when the purity of rare earth metals and the
hydride preparation methods where rather limited as compared to the present
time. Accordingly it can be expected that a repetition of similar measurements
will eliminate most of the controversies.
 Electronic Properties 175

Recent photoemission studies on hydrides of Ce and La revealed a surface

semiconductor-to-metal transition at temperature below T ~ 70 K (Fig. 5.11)
I-5.168]. A depletion of oetahedral-like surface hydrogen sites is thought to
occur. This pushes d-states from the hydrogen-induced band back to Ev and
results in a metallic surface dihydride. The results indicate that at room
temperature the hydrogen concentration at the surface exceeds that of the bulk,
whereas at low temperature the opposite seems to be the case.
Resistivity anomalies were found in the solid solution ~ phase of the elemen-
tal metals [-5.169] and of the dihydride c~phase [,5.170] at temperatures around
150 K. They seem to be caused by a peculiar linear ordering tendency [,,5.169].
So far we have completely neglected the 4f-electrons, which are otherwise so
important for the behaviour of many RE compounds. The atomic-like 4f-
electron states are thought not to contribute significantly to the metal-
hydrogen bonding. Valence band photoemission spectra of Ce hydrides [-5.142,
168] show that the hydrogen bonding band does not contain any noticeable 4f
contribution. The 4f related emission does not seem to be affected by the
hydride formation. However, as a very recent analysis of 4f related features in
core-level photoemission and inverse photoemission showed, there is signifi-
cant 4f-conduction band hybridisation in the light rare earth hydrides [-5.70]
which could possibly improve the results of band structure calculations, if taken
into account.
The theoretical treatment of 4f-electrons as band states is very delicate and
controversial I-5.172]; in the case of lanthanum hydrides, the 4f-levels are found
to be 2.0 to 3.0 eV above EF [-5.149], whereas inverse photoemission shows 4f
related features 5.2 eV above Ev [5.70].
A nice and illustrative consequence of a weak 4f-conduction band
hybridisation is the large coefficient 7 of the electronic specific heat of Ce
hydrides ?(CeHa.6)=l10mJmole-I K - 2 (as compared to y(LaH2.6)
< 0.04 mJ mole- 1 K - z), which suggest that Ce hydrides are so-called heavy-
electron compounds [5.172, 168].
Whereas in Ce hydrides the Ce 4f-electrons and the valency of Ce are hardly
affected, hydrogen causes a valence change in Yb, which forms according to
optical re-emission measurements a semiconducting YbH1.94 (divalent, in
agreement with N M R data of Zogal [-5.173]) and metallic YbH2.65 (more than
divalent) [,,5.174, 175]. Very recent XPS and susceptibility studies proved that
YbHz.6 is a mixed valent metallic magnetic compound which does not order
down to 1.7 K [1.49, 5.84].
For magnetic rare earths, hydrogen does not directly affect the magnetic
moments. However, the decrease in the number of conduction electrons and
also the large lattice expansion lead to a decrease of the strength of the RKKY
indirect exchange interaction of the local moments mediated by the conduction
electrons. Consequently, the magnetic transition temperatures of the dihy-
drides are much lower than those of the pure metals. The type of magnetic order
changes with increasing hydrogen concentration. For example Cell x orders
antiferromagnetically for x < 2.0, ferromagnetically for 2.1 < x < 2.7 and again
176 M. Gupta and L. Schlapbach

antiferromagnetically for x > 2.8 [5.134, 176, 177]. A tentative magnetic phase
diagram for Ce hydrides was proposed recently by Arons et al. [5.178], who
also reviewed magnetic ordering properties of other RE hydrides EuH z is a
ferromagnetic semiconductor [5.179].

5.3.5 Hydrides of Ca and Mg

The alkaline earth metals Ca and Mg form transparent insulating saline
dihydrides [5.1]. Though these light weight hydrides are of potential interest for
hydrogen storage, very little is known about their electronic properties.
APW band structure calculations performed for Call 2 in a simplified CaF 2
structure together with photoemission results reveal C a - H bonding states
centred 5eV below EF and a large band gap [5.181]. The bonding band has
largely Ca d-, Ca p- and H s-character. The binding energy of the Ca 3p core
level is 1.3 eV larger than in Ca metal. These results indicate that H-metal
bonding in saline or covalent hydrides is not radically different from that of
metallic hydrides.
From Compton scattering experiments [5.182] and from different theoret-
ical approaches to the electronic properties [5.183] it is concluded that the
valence electrons of MgH 2 are not localized as in ordinary ionic crystals. The
estimated effective charges of the ions arc Mg 1'9°+ and H °'65-.

5.3.6 Hydrides of Actinide Metals

APW band structure calculations for U H 3 in the crystallographically slightly
different phases ~ and fl (both are based on the A 15 structure) [5.184] give
evidence of a fairly large bonding interaction between U 5f- and H s-electrons
[5.185]. As usual there is a hydrogen bonding band some eV below Ev, which
overlaps the U5f-levels, in good agreement with photoelectron spectra of
fl-UD3 [5.186]. This results in two types of 5f-electrons, itinerant and localized
(magnetic). No photon energy dependent spectra have been measured so far to
elucidate the 5f contribution in detail. The measured 4f, 6p, and 5f spectra of
UD 3 are all shifted by 2 eV to larger binding energy in comparison to pure
uranium. The electronic specific heat coefficient takes the values
y = 3 3 . 9 m J m o l e - X K -2 for fl-UD 3 [5.186] and ~ = 2 8 . 7 m J m o l e - l K -2 for
/~-UH3 [5.187] as compared to 7 = 9 . 8 8 m J m o l e - l K -z for U metal. This
increase of y by a factor of three upon the formation of the hydride probably
does not reflect a large increase of the DOS at EF but rather points to a weak
f - d correlation effect as in heavy electron systems.
Photoelectron spectra of ThH2 and Th4H15, performed at energies
20-35 eV with a small cross-section for f-electron emission, clearly show the
growth of the Th 6d-hydrogen bonding band at the expense of the conduction
band [5.188]. The conduction band emission for ThaH15 is weak, but non-
vanishing.
Specific heat measurements [5.189] indicate a 50% increase of ~ in going
from ThHz to Th4Hts. According to the valence band spectra this increase is
 Electronic Properties 177

not caused by an enhanced DOS at EF. The f-electron contribution to the

H-metal bonding band has not been studied in detail. The 6P3/2 core levels shift
from 16.6eV in Th [5.180] to 17.7eV in T h H / a n d 18.6eV in Th4H~5 [5A88].

5.4 Results on Ternary Hydrides

During the last decade much progress has been made in the understanding of
the electronic properties of binary transition and rare earth metal alloys. A
number of theoretical band structure investigations of ordered crystalline
alloys [5.41, 194, 195] as well as CPA studies or model calculations of dis-
ordered binary TM alloys [5.196-200] and numerous photoemission investi-
gations of crystalline [5.66, 201,202] and amorphous [5.194, 203-205] alloys
have fully ruled out the validity of crude models such as the rigid-band
approach and provided insight into the cohesion of these alloys. The im-
portance of hybridization between the metal d-states of the two components
in binary systems has been stressed. It leads to bandfilling effects which are
more important than those previously ascribed to charge transfers. Indeed,
XPS core-level spectroscopy [5.201] as well as theoretical results indicate that
charge transfer in these alloys is much smaller than previously thought. The
concept of charge transfer remains however controversial due to the ill-defined
spatial extension of the charge in most theoretical calculations and to the
difficulty of analyzing experimental core-level shifts.
The investigation of the electronic structure of ternary hydrides began
rather recently with photoemission [5.7] and x-ray emission [5.8-10] work,
often hindered by hydrogen desorption, and with theoretical calculations [5.6],
made difficult by the complexity of the crystal structures and in some cases by
the controversies concerning the hydrogen site occupancy in the lattice [5.206].
In what follows we shall review the results obtained for hydrides of the
intermetallic compounds of the type AB 5 (e.g. LaNis) A B (FeTi), and AB2 Laves
phase compounds (ZrV2), on Mg2Ni and related compounds and on some
amorphous alloys. Valence band and core-level data are summarized in
Table 5.2 together with results of some other ternary hydrides.

5.4.1 Hydridesof A B s Compounds(Haueke Compounds)

Intermetallic compounds of the general formula AB 5, where A is a rare earth, Y

or La and B a transition element have been investigated for the last fifteen years
mainly due to their technological importance as permanent magnets (SmCos)
and hydrogen storage material (LaNis).

LaNis. The intermetallic RB5 compounds including LaNi5 crystallize within

the hexagonal CaCu5 structure with space group P6/mmm.
178 M. Gupta a n d L. Schlapbach

• ¢-]. Z
,r;

0 ¢.q o~

~.~ 02
t.,q~ o
Y, z

oo r,--
e,l ¢q

¢'4 t"q t--q

,"-. t"q
;:m

o=
=. eq..~
,3
en
0
O~

¢~ t"q
©
k)

0
0
t"q 7,
06
e~
tr5
o eq

e.,
o tt-i ¢'q
6 L)

09

e4

z z
 Electronic Properties 179

0
o

09

r~
t~

¢',1 t'M ~ ~,1

u° ~ ~ Z'"
¢~1¢N

c'q

~2 ~2
~ : "l:J I ~ >

• :~o;
~,.~>

c-I

o9o~
Eo~o

o
o ~o

~-~ 0 , 0

~ b~M M MNIlM M M
180 M. Gupta and L. Schlapbach
-I0 -5 0 EF 5 -I0 -5 0 EF 5 I0
r T-7 ? ~( r - - - " - - - ~ - - -r~ -~r Fig. 5.12 (a) The total
20
L°Ni5 , LoNi5H 7 DOS of LaNi5 and its
character at the La, Nil
15
J, _~c
(basal plane), and Nin
sites (middle plane)
[-5.208]. (b) The total
DOS of LaNisH 7, its
IC
5 d-character at a La site
5 and at a Ni site in the
_o I basal plane and its
F-

Lo 1
c
s-character at a tetra-
1 6
hedral H site with
Ni2La 2 coordination
,9 1
i
[5.208]
1 4
~2 i
1
i 2
t

go
~4 o

F-
NiI ~l
$-
~2
vl

Do i

,4
4 Nill i l
.2
.O

&_
5
.B
2
,6
I ,4
,Z
0 0
-IO -5 0 EF 5 I0 -I0 -5 0 EF 5 IO
ENERGY (eV) ENERGY (eV)

The band structure of LaNi 5 has been studied by the spin polarized self-
consistent APW method [-5.207] and by the tight-binding recursion scheme
[-5.208]. Partial results using the L M T O method have also been reported
[-5.209]. The local DOS of d-type at the two different Ni sites and at the La site
and the total D O S obtained by Gupta [5.208] are plotted in Fig. 5.12a. The Ni
3d-bands are located at lower energies than the La 5d-states, as expected from
the relative position of the atomic d-levels. They give rise to a narrow and high
peak in the D O S of about 3.2 eV width, while band structure calculations for
paramagnetic Ni within the local exchange approximation lead to a d-band
width of the order of 3.8 eV. We thus observe a narrowing of the Ni d-bands
from pure fcc Ni to LaNi5 which we ascribe essentially to the decrease of the Ni
coordination number from 12 in pure Ni to 7.2 in LaNis, and to a lesser extent
 Electronic Properties 181

to a small increase in the Ni Ni distances ranging from 2.46• to 2.51 A in

LaNi 5 compared to 2.45 ,~ in fcc Ni. The detailed features of the Ni DOS at the
two different sites show of course some variation due to different atomic
environments. The La 5d-bands centred at higher energies are also narrower
than in fcc La due to the large increase in the La-La distances and to the
decrease in the coordination number from fcc La to LaNis. In LaNi5 each La
atom is surrounded by 6 La atoms in the basal plane at distances ~ 5.01 A and
by 2 La atoms in the c direction located at ,-~3.98 .~. This is to be compared to
dLa_La=3.75)~ in fcc La. The L ~ N i interaction is not very strong; it is
responsible for the presence of smaUer peaks observed in the local d-DOS
plotted in Fig. 5.12a. The Fermi level Ev of LaNi 5 falls in a rapidly decreasing
portion of the Ni d-bands which are not filled. The contribution of La d-states at
Ev is very small.
The calculated values of the DOS are rather large, they correspond in the
independent particle model and without enhancement to ~ = 2 8 m J (mole
LaNis) - 1 K -2 [5.208] and y=32.4mJ (mole L a N i s ) - I K -2 [5.207]. From
heat capacity measurements of LaNi 5 between 1.6 and 4.2 K Nasu et al. [5.210]
found ~=34.3mJmole -I K -z while Takeshita et al. [5.211] from their mea-
surements between 1 K to 5 K obtained y = 36.5 mJ mole- 1 K - 2. Data taken at
higher temperatures between 5 and 300K by Ohlendorf and Flotow [5.212]
lead to y =42.6 mJ mole-J K - 2. The experimental results, in agreement with
the calculations, point to a large value of the DOS at E F in LaNi s. This is also
consistent with the large value of the magnetic susceptibility Z=4.6
× 10-6emu/g [5.56]. It is to be noted that LaNi5 is reported to be Pauli
paramagnetic with strong Stoner enhancement [5.56]. Spin-polarized calcula-
tions [5.207], however, predict that LaNis should be a weak ferromagnetic with
a small spin moment of 0.69 IxB per formula unit almost entirely due to Ni. The
majority-spin 3d sub-band has been found to be filled unlike the minority-spin
3d sub-band which is well described by a rigid shift of about 0.35eV due to
exchange splitting. We should point out here the difficulty of the APW
calculation due to the complexity of the crystal structure and to the self-
consistency requirement imposed on the value of the magnetic moment.
Valence band photoemission spectra of LaNi 5 [5.66, 191, 213] show the
similarity of occupied bands with those of Ni, i.e. strong emission at E v and the
presence of the 6 eV satellite which both indicate that the Ni d-band is not
completely filled (Fig. 5113), in agreement with theory. A shoulder at 2 eV seems
to be caused by the La-Ni interaction. The valence band of LaNi 5 is slightly
narrower than that of Ni. The experimental width of ~ 3.2 eV (2.4 eV FWHM),
[5.66] agrees with theoretical values of 3.2 eV [5.208] and 3.4 eV [5.207].
Core-level photoemission spectra [-5.191,201 ] show that apart from a more
intense low binding energy satellite of the La 3d-peak, which is ascribed to a 4f t
final state, the position and shape of the La 3d and Ni 2p core-level peaks do not
change in LaNi 5 as compared to La and Ni, indicating that no significant
charge transfer takes place upon formation of the alloy. In agreement, soft x-ray
appearance-potential spectra do not reveal charge transfer [5.214]. Indeed, as
182 M. Gupta a n d L. Schlapbach

i i i i i .~ i
Fig. 5.13. XPS valence band spectra of La-Ni
/ compounds I-5.66]. The intensity of the 6eV
satellite and the high emission at E v in the
spectrum of LaNi 5 are comparable with those of
t" *'¢ : ;
Ni and indicate that the Ni d-band is not full

/'.'\

.,¢,'

B E (¢VI

mentioned previously [-5.66], the bonding in intermetallic compounds arises

more from hybridization than from charge transfer. We have no explanation for
the large charge transfer of 1.5 electrons from La to 5 Ni atoms calculated by
M a l i k [-5.207] for LaNis, while in contrast the same authors did not obtain a
charge transfer in the isoelectronic compounds GdNi 5.
LaNis Hydride. The hydrogen absorption is accompanied by a large lattice
expansion (Aa/a,~ Ac/c ~ 8% for/~-LaNisHx with 5 < x < 7). Some controversy
concerning the location of the H atoms in the lattice has developed amongst the
numerous crystallographic investigations which are described in detail in the
Chap. 4. However, recent neutron diffraction experiments on hydrogen-rich
phases of/~-LaNisHx with ( 5 < x < 7 ) [5.206, 215] have been interpreted in
terms of the P63 mc non-centrosymmetric space group, in which the 6-fold
c-axis of the CaCu5 structure becomes a 63 screw axis due to the ordering of the
H atoms. This ordering leads to a reduction of symmetry and a doubling of the
unit cell size along the c-axis. The H atoms occupy octahedral sites coordinated
with La2Ni 4 neighbours and tetrahedral sites with Ni 4 and La2Ni 2 neighbours.
The N i - H distances range from ~ 1.61 to ~ 1.68 ,~ and are shorter than in NiH
(dNi_ n = 1.86/k). These crystallographic data made possible the first theoretical
investigation of the electronic structure of LaNisH7 which is the only hydride of
LaNi 5 which has been investigated theoretically to date [-5.208].
The total DOS of LaNisH 7 plotted in Fig. 5.12b shows the formation of
metal-hydrogen derived states centred at ,-~5 eV below the centre of the Ni
d-bands. A DOS analysis at the different atomic sites indicates that the H atoms
interact mostly with Ni atoms despite the larger affinity of La for hydrogen. A
 Electronic Properties 183

similar situation occurs in FeTiH in which the F e - H bonding is dominant [5.6].

The Ni d-bands are found to narrow from 3.2eV in LaNi 5 to less than 2.5 eV in
LaNisH 7. This is due in part to the large lattice expansion and to the lowering
of the lowest portion of the Ni d-bands due to the N i - H interaction. Some of the
Ni d-states are also depleted from the main portion of the narrow d-bands and
participate in the antibonding Ni d-, H s-states which extend over the La
5d-bands and interact with them. This effect tends to oppose to the narrowing of
the La d-bands caused by the lattice expansion.
The Fermi level of LaNisH 7 is found to fall at higher energy compared to
that of LaNi 5. This is due in part to the presence of 7 supplementary electrons
brought by H atoms in the unit cell and also to a deformation of the Ni d-bands,
part of the Ni d-states being now found at higher energies in the N i - H
antibonding region. Although there are not too many of these states per Ni
atom, since there are five Ni atoms in the unit cell this corresponds to a non-
negligible number of electrons to be accommodated at higher energies. This
effect contributes to the increase in the value of Er. At the Ni sites the DOS at
E~ in the hydrides is much smaller than that in the intermetallic compound.
The Fermi level of LaNisH7 is found to fall in a rising portion of the DOS of
the La atom. Thus the total DOS at Er in the hydride remains large as in LaNi 5.
The theoretical result agrees with the trend observed in the measured value of
the electronic specific heat coefficient: Ohlendorf and Flotow [5.2/2] have
observed a small decrease from ; ~ = 4 2 . 6 m J m o l e - l K -2 in LaNi 5 to
7 = 40.4 mJ mole- i K - 2 in LaNisHr. 3 and Chunget al. [5.216] found a decrease
from 36.5 to ~ 3 0 m J m o l e - J K -2. It is to be noted that the theoretical
calculation [5.2081 points to a large difference in the nature of the states at EF,
which are largely Ni d-like in LaNi5 with no La contribution while in LaNisHT,
the La d-contribution becomes important. In LaNisH 7 the DOS at E r also has
some H s-character due to the antibonding N i - H states. Although the Ni
d-contribution per Ni atom at EF remains smaller than the La contribution, the
total Ni d-character at EF is substantial since there are five Ni atoms per unit
cell.
Magnetic susceptibility measurements reveal a strong decrease from
2=4.6 x 10 .6 emu/g for LaNi 5 to ~1.3 x 10 -6 emu/g for LaNisH 6 [5.56,
217] which is, however, caused by a sharp decrease of the Stoner enhancement
factor and only a rather small decrease of the DOS at E F.
No spectroscopic data are available for LaNi5 hydride due to the fast
desorption of hydrogen in vacuum. Thus an important part of the metal-
hydrogen bonding features described above have not been confirmed experi-
mentally. The theory suggests the importance of the Ni-d, H-s interaction and
a much weaker La-d, H-s interaction. This is not in contradiction with
M6ssbauer data [5.218] which samples only the metal s-states and indicates a
charge transfer from La to H and no transfer from Ni to H.
I~SRKnight shift measurements [5.219] reveal K/~ ~ 0 for LaNis, which is a
typical value for transition metals, and K# = + 80 ppm for LaNisH 6 which is
closer to the values for normal metals.
184 M. Gupta and L. Schlapbach

Fig.5.14. Valence band photoelectron

spectra of Ni, LaNi5, LaNis_xCu~,and Cu
h v = 40 ev A N~ [5.224]

g
[ A/"
g

13-
- w.'i?
I I I I I I I I
8 6 4 2 0
Iniliol Slole Energy Below E F {eV)

Resistivity values are rather controversial; the room temperature electrical

resistivity of evaporated LaNi5 films (300nm thick) was shown to be 7.6
x 10- 4 f~ cm - more than an order of magnitude larger than values measured
on bulk samples of CeNis, NdNis, and GdNi5 [5.220] and to decrease to 7.2
x 10 .4 f~ cm upon charging of the sample with hydrogen [5.221]. The opposite
trend, i.e. an increase of the resistivity upon hydride formation was observed by
Larsen et al. [5.222] also on thin films and by Walsh et al. [5.223] by EPR
experiments on powder samples.
Effects of Substitution. The partial or total substitution of La or Ni in LaNi 5 is
known to strongly affect the heat of hydride formation and the maximum
hydrogen content. In metallic hydrides such variations should be caused by
differences of the electronic structure of the host intermetallic compound. Two
simple models have been proposed to correlate the hydride formation
properties with the requirement of a partially empty Ni 3d-band either by
appropriate position of EF relative to this band (Fig. 5.14, [5.224]) or by the
number of unpaired d-electrons [5.225]. To date neither total energy calcu-
lations nor electron spectroscopic methods are sensitive enough to detect these
differences unambiguously: Weaver etal. [5.213] showed that the valence
band photoelectron spectra of CaNi 5, UNi 5, LaNi 5, and ThNi 5 are nearly
identical in agreement with electronic specific heat measurements [5.211], and
reveal no differences which could be related to the wide range of the stability
of the hydrides [5.226]. They suggest that crucial differences in the band
structure, which are too subtle to appear as substantial changes in angle
integrated spectra, may exist and propose angle resolved photoemission
studies.
 Electronic Properties 185

Photoemission spectra of various La-Ni intermetallics (Fig. 5.13 [5.66])

do show certain trends; however, no correlation with the trends of the heats of
hydride formation has been established. Spectra of hydride forming LaTNi 3
are rather similar to those of non hydride forming AINi.
The heat of hydride formation depends among others on the availability of
empty host metal states which can be pulled down or filled by a shift of Ev. The
empty part of the density of states is clearly important and should be studied e.g.
by inverse photoemission. The 6 eV satellite of the valence band spectra of Ni
compounds which is a measure of the empty part of the band, gives us a hint: it
is significantly more intense [5.66, 213] in the hydride forming LaNis than in
ThNi 5 and CeNi 5, which do not form stable hydrides.

5.4.2 Hydrides of FeTi

FeTi was considered for many years to be the most promising hydrogen storage
material and considerable effort was made to understand its structural,
electronic and activation behaviour. FeTi dissolves some hydrogen in the
c~-phase and forms the three distinct hydride phases/~I(FeTiH1),/?2(FeTiHI.4)
and 7(FeTiH~.9). Amongst the ternary hydrides, the hydrides of FeTi were the
first to be investigated theoretically. Band structure calculations are available
for FeTiH, FeTiH2 [5.6] and for the substoichiometric FeTiHx< 1 [5.227].

FeTi (e-Phase). This is a Pauli paramagnetic compound of cubic CsCl-type

ordered structure which is isoelectronic to Cr. Systems with CsC1 structure were
among the first intermetallic compounds studied using band theory. Band
structure calculations for ordered stoichiometric FeTi using SC-KKR formal-
ism [5.198a, 228], SC-APW [5.229] and SC-ASW with a local density
description of exchange and correlation [5.195] yield in rather good agree-
ment a total DOS curve with E v in the valley between occupied d-like bonding
states derived primarily from Fe and empty d-like antibonding states derived
primarily from Ti. The total DOS at E v is N(E)F = 0.46 states/eV and the width
of the occupied band is 6.5 eV [5.229]. A transfer of 0.18 electrons from Ti to
Fe is indicated [5.229]. Results of recent K K R - G F and K K R - C P A calcu-
lations [5.230] for the pseudobinary compound TiFexCol _x agree for x = 1
with the earlier calculations and emphasize the effect of antistructural atoms
(e.g. Fe on Ti sites) on the magnetic properties. For disordered FeTi the sharp
DOS features disappear, the valley around EF becomes filled and N(Ev)
increases considerably [5.198a, 228].
Most of the experimentally determined parameters agree qualitatively with
the results of band structure calculations: De Haas-van Alphen measurements
of the Fermi surface can be identified with a hole surface around the M point of
the BZ and an electron surface around the X point [5.231]. The measured hole
pocket volumes are 2.7 times smaller than the calculated values. The electronic
specific heat coefficient of annealed samples is small (7 = 0.18 mJ mole- 1K -2)
186 M. Gupta and L. Schlapbach

I-5.232]. The magnetic susceptibility has a value of 3.4 x 10-6emu/g [5.233]

almost independent of temperature. Photoelectron spectra of the valence band
show the main features of the band calculations 1,5.234]. Core-level spectra
show, as compared to those of elemental Fe and Ti, no shift of the Fe 3p- and Fe
2p-levels, but shifts of ,~0.5eV to larger binding energy for Ti 2p- and Ti
3p-levels were observed I-5.233, 234]. The asymmetry of the Fe 2p-line of FeTi is
by far less pronounced than in elemental Fe, in agreement with a low partial
density of Fe states at EF 1,5.84, 233]. From XANES studies, it is concluded that
the empty p-like DOS for Fe and Ti sites are very similar in shape to that of
elemental Ti [5.235].
Surface precipitates of Fe or Fe-Ti-O compounds formed after selective
surface oxidation greatly affect the magnetic properties 1,5.233]. Furthermore
antistructure Fe atoms on Ti sites may also influence the magnetic and
electronic properties, if indeed they exist in FeTi I-5.240]. However, the order
parameter as measured by neutron diffraction I-5.236] is S=I.00; the local
order around both Fe and Ti is good up to the sixth coordination shell (EXAFS)
1,5.235] and upon annealing, when antistructure atoms should disappear, the
electronic specific heat coefficient y increases 1,5.232] in contradiction to band
structure calculations, which reveal a much larger N(EF) for disordered FeTi
[5.228]. The electrical resistivity of FeTi samples with a residual resistivity ratio
0(300 K)/0(4.2 K) = 8 amounts to 16.7 + 0.5 ~tf~cm at room temperature. FeTi
shows the largest increase of the resistivity ever observed due to the dissolution
of hydrogen in the e-phase. The specific resistivity increases by AO = 8.2 ~tt)cm
per 1% hydrogen in FeTi. A linear increase up to the composition FeTiHo.o7
has been observed I-5.237].

FeTil. o (/~-Phase). Although FeTi crystallizes in the CsC1 structure, fl-FeTiH

has an orthorhombic structure 1,5.238, 239] which is a distorted version of the
tetragonal lattice formed from the doubling of the CsC1 cell of FeTi (see
Chap. 4). The H atoms occupy distorted octahedral intersitices with two iron
atoms as nearest neighbours and four titanium atoms as next-nearest
neighbours. The hydrogen atoms are located in chain-like positions in the
planes containing the Fe atoms. It is interesting to notice that the hydrogens
occupy octahedral sites in FeTiH instead of the more usual tetrahedral sites
characteristic of the bcc metals, and that they form chains. Further, the H atoms
are much closer to the two Fe atoms (Fe-H distance = 1.579 A) than to the four
Ti atoms (Ti-H distance = 2.163 A), despite the greater affinity of hydrogen for
Ti than for Fe.
Keeping in mind the remarks concerning the crystal structure, the band
structure results I-5.6] of FeTiH plotted in Fig. 5.15a can be understood (i) by a
folding of the bands of FeTi due to the doubling of the size of the CsC1 unit cell,
(ii) the expansion of the cell in the (110) direction of the CsC1 structure and
lifting of the degeneracies due to the orthorhombic distortion and (iii) by the
role of the hydrogen atoms in the lattice. In order to illustrate the first point we
have plotted in Fig. 5.15b the bands of pure FeTi (dashed lines indicate the
 Electronic Properties 187

E(Ry)
1.1
~s 1.0
0.9
O.9
~ 0.8
0.8 . . . . . . Y '7 EF~ ~ "'-" ..............
~13
~1~ 0.7

0,60"7
~ ~ - - - - ' ~ ~ ~-t-'--.,.--..-J
~J~:::~ .o r~'~ 0.6
"Is '0.5
-a~
m
_.J
~0.4

~ 0.3

,,=,o.= ~ i

a o~
r,~ ~ f0,1°~o.3
0.2

=: X a U A Z "~ , Y " T . z b Z~ M

Fig. 5.15.(a) The energybands E(k) of fl-FeTiHalong severalhigh symmetrydirectionsof the

orthorhombic BZ [5.6]. (b) The energybands of FeTi along the FM directionof the bcc BZ.
( - - ) . The dashed lines indicate the folding of the bands due to the BZ reduction in that
direction

folding of the bands) in the (110) direction of the CsCI Brillouin Zone (BZ) due
to the doubling of the unit cell size. Thus, at the BZ centre of fl-FeTiH we obtain
the states corresponding to the points F and M of the BZ of pure FeTi; in the
hydride, all degeneracies are lifted by the orthorhombic distortion. Since the
electronic structure of FeTi in the CsC1 simple cubic BZ is characterized by 10
metal d-bands overlapped and hybridized with a wider metal sp-band, these
will lead to 22 bands in fl-FeTiH. The comparison of Figs. 5.15a and b shows
that the presence of H in the lattice leads to a drastic lowering of the lowest lying
states; for example, at the BZ centre of FeTiH, the two low lying states which
correspond to F and M3 in pure FeTi are lowered by about 4 eV. The lowering
of metal states having an s-symmetry at the interstitial site by the H potential
which strongly scatters s-waves, is a feature common to all the metal hydrides,
as mentioned previously for binary hydrides. The second band in Fig. 5.15a
should be understood as resulting from the folding of the low lying metal-
hydrogen band, due to the doubling of the size of the unit cell. A detailed
analysis of the states modified by the metal-hydrogen interaction shows that
additional states which are empty in FeTi appear in the metal d-bands below
the Fermi energy in the hydride. This is the case for a branch of metal 4p-states
corresponding to the M5 point of the BZ of FeTi. These states lie at about 1.5 eV
above Ev in FeTi; they are lowered by about 3.4 eV by the H potential and lie
below EF in the hydride. The orthorhombic distortion leads to important
splittings in the d-bands; for example, at the BZ centre, the ['12 and F~5 states of
the cubic BZ are split by as much as 0.25 meV in the hydride. The low lying
metal-hydrogen bands are formed out of states already filled in the pure
intermetallic, which have been lowered by the H potential. The width of the
 188 M. Gupta and L. Schlapbach

ENERGY(eV)
96.0 -10.0 -6.o -6.0 -4.0 -2.0 0.0 2.0 4.0 30,0 63,o

70.0 49.0 , :':':~ Fesite ]

60.0 20.0 ~ 42.0

500 ~ ~ 350

30.0 10.0 ~ 21.0

20,0 J 14.0
10.0| .
L 50 7.0

0.o ~ o.o o.~ . . ~

-0.06 0,08 0.22 0.36 0.50 0.64 O.'lST 0~92 1.06 1.20 -6.06 0.08 0.22 6.36 0.SD 0.64 0.78
T OJ)2 1.06 1.20
ENERGY[Ry) EF=0"632RY ENERGY(Ry) EF
5.4Q 63,0
4.80
50o I _FeT,.t
4.20 460 [:i,i s.o
3.60 42.0

3.00 35.0

~ 2.40 28,0

~ 1.80
21.01 . . ~ . ~ ] , ,
1,20/ 14.0
0.60 7.0
0.0
-0.06 0.06 0,22 0.30 o.56 o,o4 o.v. T o:92 l O6 1,2o -0.06 0.08 0.22 0,36 050 054 078T 092 106 120
¢

ENERGY{Ry) EF ENERGY(Ry) EF

Fig. 5.16. The total DOS of/3-FeTiH and its partial wave analysis inside the Fe, Ti, and H
muffin-tin spheres. Units are [states of both spins/Rydberg formula unit] [5.6]

lowest metal d-states measured up to the energy of the valley is slightly smaller
in FeTiH than in FeTi due to the lowering of the lowest portion of the Fe
d-states by the H potential; the increase in the lattice parameter also plays a role
in the slight narrowing of the d-bands. Since additional states which are empty
in FeTi appear below Ev in the hydride, less than one electron brought by the H
atom has to be accommodated at the top of the d-bands and thus, although E v
is shifted towards higher energies in the hydride, the protonic rigid-band
model is not quantitatively correct. Furthermore, the metal d-bands are
deformed in going from the intermetallic to its hydride.
In Fig. 5.16, we plot the DOS of FeTiH obtained from an APW
calculation [-5.6]; it can be characterized (i) by a structure at low energy, re-
sulting from the metal-hydrogen interaction which is centred at 9 eV below the
 Electronic Properties 189

Fermi level, and (ii) the metal d-bands at higher energies in which we identify, as
in pure FeTi the two peaked structure characteristic of the bonding and
antibonding metal states in the bcc materials. The increase by 2.3 % of the Fe-Ti
bond length between FeTi and FeTiH results in a weakening of the bond and in
a larger overlap between the bonding and antibonding metal d-states in FeTiH,
the valley being narrower in the hydride. The number of states at Ev increases
substantially in going from FeTi to FeTiH; for the latter we obtained N(EF)
= 1.76 states of both spin/eV-FeTiH which corresponds to an unenhanced
electronic specific heat coefficient y = 2.02 mJ mole- 1 K - 2.
A partial DOS analysis into its angular momentum components inside the
H and metal MT spheres is plotted in Fig. 5.16. It reveals that the low lying
energy states have essentially a H s- and also a Fe d-character, corresponding to
states having mostly an eg symmetry; the Ti d-contribution and the metal s- and
p-components are substantially smaller. The lowest portion of the metal
d-bands is essentially dominated by Fe d-states while Ti d-states have their most
important contribution for energies larger than that of the valley in the DOS as
in pure FeTi. However, inspite of an increase in EF, the Fe d-character of the
DOS at Ev analyzed inside the MT spheres is almost three times larger than the
Ti d-contribution.
The lowering of metal states by the H potential has been recognized to be an
important factor for the stability of the hydride. As indicated above, these metal
states can be classified into two categories: (i) metal states already filled in pure
FeTi, like those forming the two low lying metal-hydrogen bands and (ii) metal
states previously empty in FeTi, like the 4p metal branch around the M5 point
of the FeTi, BZ, which is lowered by about 3.4 eV in the hydride. It is interesting
to notice that in bcc VH Switendick[-5.4] found that rectalp-states at the N
point of the bcc BZ, having an s-symmetry at the H sites, are lowered by the H
potential by as much as 5 eV. The H concentration appears to be an important
factor in the lowering of these metal p-states and could explain, in part, the
difference in the enthalpies of formation ofc~-and/%phases. For binary hydrides
such as PdH it has also been noticed that the H - H next nearest neighbour
interactions are crucial in the lowering below EF of the empty Pd 5p-states.
In FeTiH, the lowering of empty metal states brings, however, only about
0.15 electrons below Ev and thus the Fermi level of the hydride is shifted
towards higher energies, a factor which adversely affects the stability of the
compound.
With the aim of assessing the role of the H - H interactions along the chains
parallel to the (100) direction in FeTiH, we chose to break these chains by
locating the two H atoms at (0, 1/4, 1/4) and (0, 3/4, 3/4) instead of(0, 1/4, I/4)
and (0, 1/4, 3/4) as in/?-FeTiH. The breaking of the chains results in a shift of the
low lying metal-hydrogen bands towards higher energies by 0.2 eV on the
average. With this new location of the H atoms, in addition to the plane URTZ
which is doubly degenerate in B-FeTiH, we obtain in the BZ another plane
TYRS of doubly degenerate eigenvalues. The metal d-states and the position of
the Fermi level are essentially unchanged except for the states at the point Y
190 M. Gupta and L. Schlapbach

which are shifted up in the new structure and become doubly degenerate like the
point Z. Thus, from the study of the one-electron energy eigenvalues, we can
conclude that the H - H interactions along the chains appear to play a role in the
stability of the compound. Evidence for this role is given by the difference in the
heats of solution between the dilute e-phase (AH = 0.11 eV/H atom) and the
r-phase (AH = 0.16 eV/H).
We also chose to locate the H atoms in the octahedral interstices close to the
Ti atoms. We find, in this case, that the low lying bands are shifted up by 0.34 eV
on average. Since, as shown above, the low lying bands of fl-FeTiH are mostly
formed of H and Fe states, the raising of these states in the new structure can be
interpreted via the decrease in the strength of the H-Fe interaction due to the
new location of the H atoms close to Ti. In addition we observe sensible
modifications of the metal d-bands but the final conclusion concerning the
stability of the hydrides needs to await total energy calculations. Besides the
role of the electronic interactions, demonstrated here by the change in the
position of the low lying metal-hydrogen states, and also by the modification of
the d-bands, the size of the interstice available for H may also be an important
factor for the preferred site occupancy of H close to the Fe atoms.
The few available experimental results support the theoretical conclusions;
rapid desorption of hydrogen and difficulties related to the need of activation
has so far rendered photoelectron spectroscopic measurements of FeTi hydride
impossible. The exposure of FeTi to small doses of hydrogen induces enhanced
emission in the range 4-8 eV [5.234].
Both magnetic susceptibility and electronic specific heat [5.232] increase
upon hydride formation. Due to magnetic precipitates in the near surface
region formed in the activation process [5.233], and due to the effect of
antistructure Fe atoms possibly formed upon hydride formation [5.240, 232]
the estimation of N(Ev) from susceptibility and specific heat data is even more
difficult than usually. Values of Z = (7.5 _ l) x 10-6 emu/g for FeTiH,. o [5.233]
and ~ = 1.01 mJ mole- ~K - z for FeTi hydride of unknown composition [5.232]
were obtained. Both values indicate a significantly higher N(EF) in the hydride,
a trend which is in agreement with theory.
Information on the momentum distribution of the electrons has been
obtained by La'sser et al. [5.241] from their Compton profile analysis of FeTi
and the/3-phase FeTiHI.17. The main differences between the two profiles are
observed for small values of the reduced scattering vector (q < 2 a.u.) which
correspond to large values of the position vector in real space and can thus be
ascribed to bonding effects. Realistic wavefunction calculations are needed to
give a reliable interpretation of the data since crude estimates obtained from a
simple scaling of the profiles using the protonic model are in very poor
agreement with the experiment. Indications concerning the trends in the heat of
solution of H in the r-phase have also been obtained from the Compton profile
measurements [5.241].
M6ssbauer isomer shift (IS) measurements reveal a systematic decrease of
the s contact electron density 0~ at the metal site with hydrogen uptake for
 Electronic Properties 191

binary compounds such as TaHx, PdHx, as well as for hydrides ofintermetallic

compounds FeTiH x [-5.242]. For FeTiH~ unlike for PdHx, simple renormal-
ization of the 4s-wavefunctions due to the lattice expansion accompanying
hydrogen uptake did not appear to be sufficient to explain the observed
decrease in the contact s density [-5.242]. A microscopic calculation of 0~ at the
nucleus was performed [5.243] to interpret the data of Swartzendruber et al.
[5.242]. The calculated valence electron contribution Q~= 5.858 for fl-FeTiH is
almost identical to the theoretical value obtained for pure iron metal by
Callaway et al. [5.244] (Qs=5.823) and with the experimental result of
Shinohara and Fujioka (Os= 5.53 + 0.46) [5.245]. Although it should be pointed
out that the present calculation for fl-FeTiH [-5.243] and the calculation of
Callaway et al. [-5.244] for iron metal have been carried out using different
techniques, we obtain nearly the same value of0s (Fe). This result is in very good
agreement with the data of Swarzendruber et al. [-5.242] since for the FeTiHo.9
sample these workers did not observe any shift of the M6ssbauer line compared
with that for pure iron metal; indeed a negative IS ( - 0 . t 4 m m s -1) was
observed from pure iron metal to pure FeTi and this shift disappears with
hydrogenation when fl-FeTiH0. 9 is formed [-5.242].

FeTiH2. The v-phase has been shown to have a monoclinic structure [-5.238]
which can be regarded as a distortion of a tetragonal cell; the hydrogen atoms
are ordered at octahedral interstices. In the electronic band structure calcu-
lation [5.6], the monoclinic distortion with an angle of about 97 ° was ignored
and the lattice was treated as tetragonal.
A comparison of the DOS of FeTiH2 plotted in Fig. 5.17 with the DOS of
FeTiH reveals that on hydrogen uptake the metal-hydrogen related structure at
low energies grows both in width and in intensity. Indeed, this structure which
corresponds to two bands in FeTiH is formed from four bands in FeTiH2 and
thus accommodates twice as many electrons. The additional two bands found
for FeTiH/result from the presence of two additional H atoms in the unit cell;
the hydrogen states forming these low lying bands are also hybridized with the
metal d-bands. The Fermi energy of FeTiH2 falls above the valley of the DOS,
as for FeTiH. For FeTiHa we obtain N(EF)= 1.68 states of both spin per eV
per FeTiH2 molecule which corresponds to an unenhanced value of
V= 1.93 mJ mole- 1 K - 2 [-5.6].
The DOS analysis of FeTiH2 into its angular momentum components s, p,
and d, inside the MT hydrogen and metalspheres, in shown in Fig. 5.17. As for
FeTiH, the structure at low energy is dominated by the H s-states and also by
the metal d-states; a smaller contribution from the metal s- and p-states is also
observed. It is interesting to note that the contribution of the Ti d-states to the
low energy structure is much larger in FeTiH2 than in FeTiH. This is due to
the fact that in FeTiH 2 one of the four H atoms is located closer to two Ti
atoms rather than to Fe atoms.
Susceptibility measurements indicate a further increase for FeTiHL9 to
)C=(8.5 + 1)x 10 -6 emu/g at room temperature [-5.233]. The valence electron
 192 M. Gupta and L. Schlapbach
900

800 60,0 Fe Ti H 2

17.5
2'" Fe site
70 O 52.5
--d
60 0
.,"
.,," 15o (i
l
~ 45.0
~3
500 ; ," 12.5 i .~ 37.5
O I i
LU
GI
-.I 40.0 / lOQ 30.0

30.0 22.5

2OO :.' 5.0 15.fl

lO.O 2.5 7.5

o.o 0.0
0.0 0.15 0 . 3 0 0145 0.60 0.75 T0,g0
ENERGY ( R y ) =EF

20.0 Fe Ti H 2 16.0 F e Ti H 2
~ H site '"2 Ti s i t e
17.5 14.0
--d

15.0 o~ 12.0
0 O
CI
...a 12.5 . j 10.0
F-
10.0 rr 6.0
=<
7.5 6.0

5.0 4.0

2.5 2.0

0,0 0.0
0.0 0.15 0,30 0,45 0.60 0.75 0.90 0.0 0.15 0.30
0.3( 0.45 0.60 0.75 TO.90
ENERGY(Ry) IEF
ENERGY(Ry)

Fig. 5.17. The total DOS of FeTiH2 and its partial wave analysis inside the Fe, Ti, and H
muffin-tin spheres. Units are [-states of both spins/Rydberg formula unit] [5.6]

contribution to O~ decreases by 10.4% from fl-FeTiH to 7-FeTiH2. One can

calculate the corresponding IS difference

A/1 = ~(QFeTiH -- 0FeTIH2) , (5.9)

where the calibration constant e, which is related to the change in nuclear
radius during the transition, has the value given by Ingalls [-5.246]
~ = - 0 . 3 7 a g m m s -1. The theoretical result A e = 0 . 2 3 m m s -1, is in good
agreement with the experimental value [5.242].

A• = ~(QFeTiHo.9 -- QFeTiH1.7) 0.27 mm s - 1.

~--" (5.10)

The decrease in 0s(Fe) with hydrogen uptake from FeTiH to FeTiH2 cannot
be ascribed to the pure effect of a 4s-wavefunction renormalization due to
 Electronic Properties 193

volume expansion since, on the assumption that this effect is proportional to

- A volume/volume, it accounts for only half of the IS observed between the
dilute a-phase of FeTiH~ and y-FeTiHl. 7. The theoretical results show that
93.6% of the decrease in ~(Fe) from FeTiH to FeTiH 2 is due to a decrease in the
contribution of the low lying bands. Since, as discussed previously, the metal
states which contribute to these bands are very much affected by the metal-
hydrogen bonding we can conclude from the theoretical study [5.243] that the
depletion of the 4s metal states at the iron site on hydrogen uptake is mainly due
to the metal-hydrogen bonding.

Substoichiometric FeTiHx (x < 1). The tight-binding CPA method mentioned in

Sect. 5.2.1 has been applied to study substoichiometric hydrides of FeTiH x < 1.o
[5.227], using Slater-Koster (SK) fits to the band structures of FeTi [5.229] and
FeTiHl.o [5.6]. The DOS and its site and angular momentum components
studied for H concentrations x=0.1, x=0.8, and x=0.9 indicate a very
nonrigid band behaviour. The low lying hydrogen induced states develop from
previously occupied metal states while the additional electrons brought by the
H atoms fill up previously unoccupied states pulled down below EF. The
readjustment of the Fermi level, is small and appears to be nearly independent
of x, but the value of the DOS at EF increases with x [5.227] in agreement with
the heat capacity measurements [5.232]. Similar behaviour of the Fermi level
position has been found by Temmerman and Pindor [5.42] from their
K K R - C P A results on the P d - A g - H system.
One can say that most of the features of the bonding of H in the intermetallic
compound FeTi do not differ fundamentally from the results knowns for binary
TM hydrides. They are, however, more difficult to analyse due to the low
symmetry of the crystal structure. Further work on the stability and on
preferred site occupancy of H should be pursued.

5.4.3 Mg2Ni and Isoeleetronic Compounds and Their Hydrides

The hydrides MgaNiH 4, MgzCoH 5, and Mg2FeH 6 are characterized by

extremely high hydrogen content per unit volume and weight, and by very
strong H-metal bonding together with nonmetallic appearance. One of the
final goals of the investigations of these compounds therefore, is to learn how
this H-metal bond could be weakened.

MgzNi. MgzNi has a hexagonal unit cell consisting of chains of Ni atoms

separated by two layers of Mg atoms [5.247]. No theoretical investigation of
the intermetallic compound MgzNi is yet available, however electronic
structure investigations of several AB2-C 16 structure compounds by means of
the tight-binding method have been performed I-5.248, 249]. Photoelectron
spectra of the valence band show that the Ni d-band moves away from EF and
that it becomes narrower and more symmetric than in elemental Ni. The
194 M. Gupta and L. Schlapbach

intensity of the 6 eV satellite is smaller indicating a decrease of the number of

d-holes [5.66]. Core-level spectra exhibit almost no shift (__<0.2 eV) for the Ni
2p-level and a shift of ~ 0.4 eV to smaller binding energy for Mg 2s- and Mg 2p-
levels as compared to the elemental metals [5.201, 84]. The measured magnetic
susceptibility amounts to X= (0.9 _+0.2) x 10- 6 emu/g at room temperature and
increases below 80 K [5.250].

Crystal Structure of Mg2NiH 4. MgaNi absorbs hydrogen reversibly and forms

an a-phase up to H/M = 0.3 and a fl-phase which is based on the stoichiometric
Mg2NiH4 composition, for higher hydrogen concentrations [5.251]. The
formation of the fl-phase is accompanied by a structural transformation of the
metal lattice and by a substantial volume expansion A V of 4.80/~3 per H atom.
A number of crystallographic determinations of the different phases of
Mg2NiH x are available as discussed in Chap. 4. The interpretations of the data
have been in some cases controversial.
Up to date, theoretical calculations on Mg2NiH 4 have been undertaken
only for the cubic high-temperature phase (T>245°C). X-ray diffraction
[5.2523 and neutron diffraction [5.253] data have shown unambiguously that
in the cubic phase the metal atoms have an anti-fluorite-type arrangement in
which the Ni atoms form an fcc lattice while the Mg atoms occupy the 1/4
(1, 1,1) positions in the cubic cell. The location of the H atoms in this phase
has been the object of discussions [5.253, 254].
For the interpretation of the crystallographic data and for the understand-
ing of the electronic properties, it is interesting to keep in mind that Mg2NiH 4
belongs to a family ofisoelectronic ternary metal hydrides M'2MHx where M' is
a divalent simple or rare earth metal (M'=Mg, Ca, Sr, Eu, Yb) and M a
transition metal (M = Fe, Co, Ni, Rh, lr, Ru, etc.). In this family, the maximum
hydrogen content corresponds to x = 6, and one can notice that the maximum
value of x which can be achieved for given elements M' and M is such that the
total number of valence electrons is 18. Neutron diffraction analysis on
MgFeD6 samples [5.2553 have been interpreted in terms of the K2PtC16
structure in which the H atoms form octahedral cages around the TM atoms
with positions close to (1/4, 0, 0). The positions of the H atoms have been shown
to be consistent with M6ssbauer, Raman and IR measurements [5.2553. This
structure will hereafter be referred to as structure I. Another possible partial
occupancy of H sites close to (1/4, 0, 0), hereafter referred to as structure II, has
been obtained from neutron diffraction analysis on Mg2NiD 4 [-5.254] and
appears to be consistent with NMR data [5.2561. However, most of the neutron
diffraction data on the high temperature cubic phases of MgaNiD4 [5.253] and
Mg2CoD5 [5.257] are analyzed in terms of a disordered distribution of D
atoms at the summits of octahedral cages around the TM atoms. These H
locations are believed to result from the high temperature thermal average of
orientationally square-planar [5.258] or tetrahedral [5.2593 ions NiD2 4 and
square pyramidal complex ions CoD 4- observed respectively for Mg2NiD 4
and Mg2CoD 5 at room temperature.
 Electronic Properties 195

Electronic Structure of Mg2FeH6. For clarity we shall first describe the

theoretical APW band structure results of a compound corresponding to the
maximum hydrogen capacity Mg2FeH6 in which the hydrogen atoms form
octahedral cages around the transition metal atoms. The energy bands of
Mg2FeH 6 I-5.260] are plotted along some high symmetry directions in
Fig. 5.18a. In order to understand these results we should keep in mind the
following essential features of the structural data. The M - M distances are large
(dFe-Vo =4.56 A in Mg2FeH6 while dFe-F¢ = 2.48 A in bcc iron), therefore we
expect a weak (Md)-(Md) interaction and narrow Md-bands. The M - H
distances are short (dve-n = 1.54 A in Mg2FeH6) , and therefore strong (Md)--H
interactions and thus large ligand field splittings are expected. The M'-H
distances are much larger because the hydrogen atoms are located close to the
centres of the faces of the M' cubes (dug-H = 2.28 A in Mg2FeH6). We can thus
expect the electronic properties of the compounds under study to be strongly
dominated by the characteristic features of the octahedral transition metal
complex MH6 since much weaker M'-H and M - M interactions are expected.
In order to illustrate this point we have reproduced in Fig. 5.18b the typical
energy diagram expected for a free octahedral ML 6 complex where M is a
transition element and the ligand L has occupied s- and empty p-orbitals. In the
present case, since the octahedral complex is in a periodic array of metal atoms,
we expect of course that the Ms- and p-states are partially filled and thus are
located below the Md-orbitals. Nevertheless, we observe strong similarities
between the bands and the schematic diagram plotted in Figs. 5.18a and b. At
the Brillouin zone center F, we obtain in increasing order of energy (i) the F1
state which is a hybridized (M's)-(Ms) state lowered by the H s-interactions (this
state corresponds to the lalg molecular orbital level), (ii) a doubly degenerate
F12 state which is a bonding combination of H s-states and the Mdz2- and
dx2_r2-orbitals whose lobes point towards the hydrogen sites (this state
corresponds to the leg molecular levels) and (iii) a triply degenerate H s-state
which is non-bonding with respect to the transition metal atom (this state
corresponds to the i t au molecular degenerate levels). In the solid we thus obtain
six low lying bands which result essentially from strong (Md)-(Hs) and H - H
interactions, and the corresponding states are hydridized with the M sp-states
and the M' sp-states. A gap separates the first six bands from the next three-band
complex which is formed by the narrow M d(t2g)-states. At F, the F~5 state
corresponds to the molecular lt2g levels shown in Fig. 5.18b. The t2g orbitals
whose lobes are not pointing towards the hydrogen atoms are not affected by
the H s-interaction. Another energy gap separates the M d(t2g)-states from the
higher states in which we find the antibonding combination of the doubly
degenerate d~2 and dx~_y2 orbitals with the H s-states.
The total DOS for Mg2FeH6 is plotted in Fig. 5.18c. It is characterized by a
low lying structure due to strong M - H and H - H interactions with a weaker M'
sp-hybridization. These states accommodate 12 electrons. The remaining six
electrons of the compounds with 18 valence electrons fill the t2g manifold of the
split M d-states. The filled M d-bands are narrow and strongly peaked owing to
 196 M. Gupta andL. Schlapbach
M 6L

---R-

__p
_E
._.&d
A
~ o4

~ o,2

D.o

o=
a
X W r X

, ",? ./~,./

•k• 100

5O

0-0.2 .o.I o 0`;- . . .4 o5 oi~ 0'7 o e o'~

-0.~ -D.I O 0.1 0.l 0.3 0.4 O5 , I 0.7 0.8 0.9 ENERGY(R
ENERGY(Ry) EF

I 7
i,i'iL
jl

! 31i I~
2~.~ 6 ~
. : ., 'J~O.
3 o:,
: ' o ~ o:. o:, o:.
ENERGY(Ry)
3
o:,

-0.2 .O.1 0 0.1 0.2 0`3 0.4 0`5 0.6 O.T O.G 0,9
i
ENERGY(Ry)

Fig,5,18a-e,Captionseeoppositepage
 Electronic Properties 197

the large M - M distances. An energy gap of 1.8 eV separates the M d(tzg)-states

from the empty higher bands. The compound is thus found to be non-metallic.
The origin of the gap between the filled non-bonding t2g levels and the empty
antibonding [M d(e~)-H s]-states is due to the large ligand field splitting caused
by the strong interactions between the transition metal elements and their
octahedral environment of hydrogen atoms, the M - H distances being rather
short.
The partial wave analysis of the wavefunctions was performed inside the
M T spheres and the corresponding DOS for Mg2FeH 6 are plotted in
Fig. 5.18c. It is clear that the six low lying bands which form the structure at low
energies consist mostly of H s-states and M d-states (which correspond to the eg
symmetry). The contribution of M s- and p-states also appears on the low
energy side of this structure. The non-negligible contribution of the M' sp-states
should also be noted. The second narrow filled structure of the DOS consists
almost exclusively of M d-states of t2s symmetry. It is interesting to note the
contribution of the antibonding M d-states (of eg symmetry) to the H s-states in
the empty levels located above the energy gap. At higher energies we find mostly
Mg p-states.
It would be very instructive to compare the present results with soft x-ray
emission and absorption spectra. Some data of this type are available for
Mg2NiH 4 I-5.261] and Mg2CoH 5 I-5.262] which belong to the same family, but
such spectra have not been obtained for the compound under study here.

Electronic Structure of Mg2NiH4. As discussed previously the electronic

structure calculation for Mg2NiH4 in its cubic phase was performed using
structure I in which the H atoms partially occupy the summits of octahedral
cages surrounding the Ni atoms. F o r our calculation we took into account the
partial occupancy of the H sites by averaging the structure factor of the H
atoms, this preserves the cubic symmetry around the Ni sites. The essential
features are similar to those discussed for Mg2FeH 6. Apart from small
variations originating from the change in the lattice constants and atomic
species the main difference is found in the relative position of the nine low-
energy M - M , M - H and H - H bands. F o r example, at the BZ centre the triply
degenerate H s-state is found in Mg2NiH 4 to lie above the F12 and F~5 N i - d
derived bands while in Mg2FeH 6 the H - s derived bands lie between the F1: and
/'~5 states. Indeed, with increasing hydrogen content, the T M - H interactions
become stronger, the crystal field splitting of the T M d-bands increases and the
M - H states gain in stability. As a consequence, unlike in Mg2FeH6 the narrow
nonbonding Ni d-t2g states are not separated by a gap from the low-energy
metal-hydrogen bonding states.
q
Fig. 5.18.(a) The energy bands E(k) of MgzFeH6 along several high symmetry directions of the
fcc BZ [-5.260-1.(b) Schematic energy level diagram ofa octahedral ML 6 complex where M is a
transition element and the ligand L has occupied s and empty p orbitals. (c) The total DOS of
MgFeH 6 and its partial wave analysis inside the Fe, Mg, and H muffin-tin spheres. Units are
[-states of both spins/Rydberg formula unit] [5.260]
198 M. Gupta and L. Schlapbach

330 --1
250 J

300 240
200: 100 ~

lal (b)
o~
150
15 i ~ 75
(n
0e-,
e.,

i lOO

¢/)

0 .'J
-0.1 0 0,1 0.2 0.3 0,4 05
5 i 25

0
-0.1
....
....22...... i . ~
0 0,1
P.-

0.2 0.3
!i!1 0.4 0.5
o
8
.2 __.

ENERGY(Ry) ENERGY (Ry)

7.5
Id)
r'. 30 (c)
5.0

i i i

lO

o
-0.1 0 0.1 0.2 0.3 0.4 015
oi .

-0.1 0
f. j

0.1 , 0.2 0.3

,..j ,

0,4 0,5
E N E R G Y (Ry) ENERGY(Ry)

Fig. 5.19a-d. The total DOS of Mg2NiH 4 and its partial wave analysis inside the Ni, H, and
Mg sites. Units are [states of both spins/Rydberg formula unit] I-5.261]

The total density of occupied states of Mg2NiH 4 is plotted in Fig. 5.19. The
structure of width 3 eV on the low-energy side is due to the H s-, Mg s-, and
Ni s-interactions. This appears clearly from our partial DOS analysis
performed inside the MT spheres which is also plotted in Fig. 5.19. The
sharp structures in the DOS found at higher energies are due to the M d-states
[bonding d(eg)- and non-bonding d(t2g)-states ] and to the H - H interactions.
Since we have obtained nine bands of total width 8.1 eV separated by an energy
gap from the higher states, the compound MgzNiH 4 with 18 valence electrons is
found to be non-metallic. The width of the gap is 1.36eV. As discussed
previously for Mg2FeH6, the energy gap originates from a splitting of the
transition metal d-bands due to the strong ligand field of the H atoms.
A band structure calculation for MgzNiH 4 was also performed using
structure II in which the H atoms are located close to the (1/4, 0, 0) positions.
Using structure II, the N i - H distances are a factor of 21/2 larger than those in
 Electronic Properties 199

structure I while the Mg-H distances are a factor of 21/2 smaller. This leads to
pronounced differences in the electronic structure, which can be briefly
summarized as follows: (i) In structure II, the hydrogen atoms interact with
delocalized M and M' sp-states, principally with the magnesium atoms. The
interaction between the H s- and the Ni d-states is very weak, unlike the case of
structure I. The bonding has an important ionic component because we observe
a depletion of the interstitial charge in favour of the MT spheres. In contrast, in
the structure I we have emphasized the strength of the (Ni d)-(H s) interactions.
(ii) The Ni d-bands are not split by the ligand field because the Ni-H distances
are a factor of 21/2 larger than those in structure I. The filled Ni d-states are
separated by a gap of 0.8 eV from the higher bands. The hydride is thus found to
be non-metallic although the physical origin of the gap is different from that
obtained with structure I. Ni core-level spectra of Mg2NiH4 measured by SXES
[5.261] and XPS (Table 2, [5.84]), though suffering from oxygen contamina-
tion of the sample surface, both suggest a strong Ni-H interaction and favour
structure I.

Electronic Structure o f M g 2 C o H s. The APW energy bands and DOS of the high
temperature cubic phase Mg2CoH 5 have been obtained [5.2623. They com-
plete the study of the isoelectronic series Mg2FeH 6 [5.2603 and MgaNiH 4
[5.261,263]. The ordering of the bands of Mg2CoH 5 is found to be the same as
that of MgzFeH 6. Thus for compounds with respectively 5 and 6 H atoms,
unlike in MgzNiH 4, the triply degenerate H s-states at the BZ centre are located
between the split transition metal d(eg)- and d(tzg)-bands. The observed trend is
towards a larger splitting of the TM d-bands and a stabilization of M - H and
H - H induced states upon increasing hydrogen concentration. In MgaCoHs,
the narrow sharply peaked Co d(t2g)-states are separated by a small gap from
the low energy metal-H and H - H induced states. Like Mg2FeH6, MgaCoH 5 is
found to be non-metallic. A gap of 1.92 eV opens up between the filled Co d(t2g)-
states and antibonding (Co-d(eg), H-s)-bands.
Several experimental and theoretical results appear to shed some light on
the intriguing fact that the maximum hydrogen content in this family of
compounds corresponds to 18 valence electrons (v.e.). The theoretical results
indicate a filling of the low energy states by 18 v.e. and suggest that the filling of
the antibonding (TM-d(eg), H-s)-states is energetically unfavourable owing to
the existence of an energy gap. M6ssbauer, Raman and IR data [5.255] have
been obtained for MgzFeH 6 and have been shown to be consistent with the
presence of oetahedral low-spin iron (II) (Fell6) 4- ions. The calculation shows
that the TM-d(t2g)-states are filled leaving the antibonding (TM-d(eg), H-s)-
states empty. This picture confirms the experimental interpretation made in
terms of a low spin d 6 configuration. It is however to be noted that the ionic
picture proposed is certainly too schematic. We have emphasized in our
partial-DOS analysis that the low lying bands are not just H-derived states.
They contain also a strong TM (eg) bonding contribution together with a Mg
sp-component.
200 M. Gupta and L. Schlapbach

Despite the absence of precise experimental determinations of the energy

gaps which is due to the difficulty of having powder samples, the compounds
belonging to this family are known to be non-metallic from the preliminary
resistivity data of Moyer and coworkers [5.264] and Zolliker et al. [5.257].
The conductivities of the Ca-Ir-H, Sr-Ir-H, Ca-Ru-H, and Sr-Ru-H
systems range from 6.4x10 -8 to 2.6x10 -8 to 2 . 6 x 1 0 - 7 ~ - 1 c m - 1 ; the
conductivities of the C a - R h - H and Sr-Rh-H systems are also small but are
nevertheless much higher than those of the other systems; they are of the order
of 10 -1~)-1cm-1 [5.264]. Resistivity measurements using the four-point
method yielded values of the order of 109-109 ~ cm at room temperature for
compacted powders of Mg2CoH 5. The resistivity has been found to increase
by a factor of 7 or 8 at 4 K [5.257]. Moreover, the hydrides do not have a
metallic appearance, MgzNiH4 and Mg2FeH6 are respectively reddish and
olive-green powders. The low temperature phases of Mg2NiH 4 exhibit a
metal-like conductivity based on measurements with a conventional Volt-
Ohm meter at room temperature [5.265], or a semiconducting behaviour with
an activation energy of 0.05eV according to other authors [2.266]. The
theoretical results show also that the compounds are not metallic, the values
of the energy gaps are 1.8 eV, 1.92 eV, and 1.36 eV for Mg2FeH6, Mg2CoHs,
and Mg2NiH 4 (using structure I) respectively. These values should however be
modified by correlation effects which are expected to be important in these
narrow d-band materials; smaller values of the exchange potential than in the
strong X~ ~ = 1.0 exchange used in the calculation should reduce the predicted
energy gap.
Magnetic susceptibility measurements are difficult to analyse due to the
presence of significant amounts of free Fe, Co, and Ni precipitates [5.255,257,
250], However, the results appear to be consistent with a weakly paramagne-
tic or possibly diamagnetic behaviour. A value of Z = (0.5 + 0.2)x 10-6 emu/g
was obtained for Mg2NiH 4 [5.250]. M6ssbauer data on MgzFeH6 also
suggest a diamagnetic low spin iron (II) complex [5.255].
As mentioned in Sect. 5.1.2, x-ray emission spectroscopy allows an experi-
mental determination of the distributions around each type of atom of occupied
electronic states having a symmetry consistent with the dipole selection rules.
The densities of occupied Ni s-, d-; Co s-, d- and Mg s-, p-states in Mg2NiH 4
[5.261] and Mg2CoH5 I-5.262] have been investigated by a study of the Ni L~,
Co L~ and Mg Ko emission spectra at room temperature. After correction for a
strong oxide contribution in the Mg spectra, the experimental results have been
found to be in satisfactory agreement with the corresponding partial wave
analysis of the theoretical DOS discussed previously, although these corre-
spond to the high temperature hydride phases.
The experimentally determined occupied 3d-states have a FWHM of 2.35
_ 6.1 eV and 4.2 +__0.2 eV respectively for Ni and Co compared to the theoretical
values of 2.5eV for Ni (using structure I) and 4eV for Co. The two-peak
structure obtained in the theoretical DOS in Fig. 5.19 are not resolved
experimentally due to the large 2p3/2 core lifetime (1.3 eV for Co and 0.7 eV for
 Electronic Properties 201

Ni) and to the experimental resolution. The experimental Ni d-spectrum does

not agree with the theoretical results obtained with structure II which lead to
much narrower Ni d-bands. The experimentally determined occupied Mg
p-states have a F W H M of 3.8 _+0.2 eV and an additional peak of lower intensity
at 6.0_0.3eV and 6.5_+0.3eV below the valence band edge respectively for
MgzNiH 4 and Mg2CoH 5. This result is in perfect agreement with the
theoretical position of the Mg p-, H s-induced states.
An attempt to study the unoccupied states has been made for MgzCoH 5 by
an analysis of Mg K absorption and by reabsorption of the Co L, emission in
the case of cobalt [5.262]. The presence of empty antibonding TM d-states
above the energy gap predicted by theory [5.260, 263] has been confirmed.
These states are characteristic of the KzPtCI 6 type structure [5.260]. Moreover,
the observed empty Mg p-states are found to be separated by a gap of 2.0
+ 0.2 eV from the valence band edge, a result which agrees with the calculated
value 1.92 eV for Mg2CoH s.
Important features of the electronic properties of hydrides isoelectronic to
Mg2NiH4 are now understood. However, further studies of fundamental
properties such as electrical conductivity and also theoretical investigations of
the low temperature phases are still needed. Due to the peculiarities of the
crystal structures and to the presence of divalent metal atoms these ternary
hydrides have a rather unusual electronic structure.

5.4.4 The AB2Laves Phase, and Related Compounds and Their Hydrides
Many cubic (C-15) and hexagonal (C-14) Laves phase compounds (AB2) readily
absorb hydrogen and form extended solid solutions or hydrides over a very
wide range of stability and concentration. They not only show interesting
hydrogen storage properties but are also fascinating because of their electronic,
magnetic and superconducting properties (see Chap. 7).
The electronic structure of ZrV2, ZrFe2, and ZrCo2 was calculated self-
consistently using the APW method and the local-density form of exchange-
correlation potential [5.267], that ofZrV 2 also by L M T O calculations [5.268].
The energy bands and the DOS of ZrFe2 and ZrCo2 are very similar. Their
valence bands are dominated by occupied states of largely Fe d- and Co
d-character and by empty states of largely Zr d-character. In ZrV2 the Zr
d-states contribute more bonding states below EF than in ZrFe2 and ZrCo2.
The Fermi levels in ZrV2 (see Fig. 5.20) and ZrFe2 fall in a peak in the DOS and
N(EF) values are quite large.
Klein and Pickett [5.49] used the above results as the basis for a muffin-tin
Green's function study of dilute hydrogen impurities in ZrV2 and ZrCo2. For
the interstitial sites in ZrV2, larger charge density near the proton is found than
in the case ofZrCo2. UV photoelectron spectra of ZrV2 confirm that EFis close
to the top of the d-band, whereas in ZrCr2, Ev is shifted to 0.6 eV above the
d-band in a region of low DOS [5.269]. In a rigid-band-like filling of the
calculated DOS of ZrVz by the additional valence electrons ofZrCr2, Ev moves
 202 M. Gupta and L. Schlapbach

I i i (He I 2112 eV) Fig. 5.20. Valence band photo-

electron spectra of ZrV2 and
ZrCr2, showingthat the d-band of
ZrCr2 is shifted0.6eV away from
EF as compared with ZrV2. lnset:
Calculated total DOS of ZrVz
t with Fermi levels of ZrV2 and
"ZrCr2". EF of "ZrV2"was deter-
I---
mined by adding4 electrons to Ev
Z
LIA
of ZrV2, which corresponds to a
F- shift of 0.65eV, in agreementwith
Z
E~e,g~ (eVl the photoelectron spectra [5.267,
5 6 7 B 9 10 I~
5.269]
ZrVa ,~Er i
tolal DOS 1 1

ZrCr2

ZrVz "ZtCrz"
I I I I
8 6 4 2 EF=0
BINDING ENERGY (eV)

up 0.65eV, in good agreement with the UV spectra shown in Fig. 5.20.

Differences of the heat of formation of binary transition metal hydrides were
shown to be dominated by shifts of Ev. The measured shift of 0.6 eV accounts for
a change of A H of +60kJ/(moleH2) from ZrV2 to ZrCr 2 which compares
surprisingly well with measured values of -78kJ/(moleH2) for ZrV2 and
- 24 kJ/(mole H2) for ZrCr2, i.e. a change of 54 kJ/(mole H2).
In ZrT2 compounds (T= Mn, V, Cr) small core-level shifts of opposite sign
were observed for Zr and T, respectively, as compared to the values of the
elemental metals [5.7, 269, 270] (Table 5.2 and Fig. 5.21).
Hydrides of the hexagonal C-14 Laves phase ZrMn2 were the first ternary
hydrides studied by photoelectron spectroscopy [5.7]. A strong emission around
6.5 eV resulting from the hydrogen induced band and a 0.5 eV shift of the Zr 3d
core-levels to larger binding energy together with some broadening of the Mn
2p core-level were observed (Fig. 5.21), in agreement with strong Zr-H-
bonding. In disagreement with the XPS results, SXES spectra of ZrMn 2 showed
neither a profile change nor any hydrogen-induced states on hydride formation
[5.9]. As hydrogen induced features were observed in SXES spectra of more
stable hydrides, we suppose that a rapid desorption of hydrogen from ZrMn2
hydride was the reason for the discrepancy between the XPS and SXES results.
Valence band and core level spectra of the cubic Laves phase compounds
ZrV2 and ZrCr2 and of their hydrides measured by XPS also are shown in
Fig. 5.21. The intensity of the hydrogen induced peak at 6-7 eV below E r varies
with the hydrogen content of the compounds. Its position does not seem to
correlate with the stability of these hydrides. The limited resolution, which is
caused largely by the fast desorption, does not allow precise statements on
other variations of the valence band. The core levels show interesting shifts for
 Electronic Properties 203

I [ t I //I I I },--'t I I I

I
//'~ <'"~'~ , I/II
// ~1 //~///"~="~ '-r M%% /-
[rMn2 / /
-" \L.
.<---' ,,\ ...,l_.-"
G43 641 639 ~ %%

ZrCr H
~lt t ZrCr2

j,i I~ i i / t J , , . I t ~ ,Z,%.
ZrCr2 i\
~
578 ~
576 ~ 574 "~_-_
~'~., ~ },.~_.
%~.

//~, A /~,
\ IIl'l ZrV~Hs' /
\ A #, / tl ~ Zrv2
.J.,," t t
~_-S---" ~0,,o\ \ / ./ " \ \
ZrVz
I ] I ],;, I I I I
184 182 180 178 8 4 O=EF
BINDING ENERGY (eV) BINDING ENERGY (eV)
Fig. 5.21. Right: x-ray photoelectron spectra of the valence bands of ZrM n 2, ZrCr2 and ZrV2
and their hydrides. The hydrogen-induced bands appear around 7 eV. Left: x-ray photo-
electron spectra of the core levels ofZr, Mn, Cr, and V in the intermetalliccompounds ZrMn2,
ZrCr2, and ZrV2 and their hydrides. The various core-level shifts indicate the bonding of the
corresponding elements with hydrogen [5.11]

the various components according to their contribution to the hydrogen

bonding and the hydrogen concentration. The shifts increase from Mn (almost
zero) through Cr and V to Zr (0.5-1.3 eV) and that of Zr from ZrMn2H3 (0.5 eV)
through ZrCrzH3. 8 to ZrVzHs. 1 (1.3 eV) in the same sequence as the stability of
the hydrides increases. It is remarkable that not only the core levels of the
elements which form stable binary hydrides (Zr, V) but also those of Cr are
shifted. The Zr core-level shift in ZrV2Hs.1 (1.3 eV) is significant larger than in
binary Zr hydrides. Attempts to correlate the core-level shifts to the heat of
hydride formation [5.269] by analogy with the models which correlate the core-
level shifts in binary alloys to the heat of alloy formation [5.202], yield a
characteristic value for the shift of each component when normalized to equal
hydrogen contents.
SXES studies of ZrV2 deuterides show "hydrogen induced bands" in both
the Zr L 0 (4d-2p transition) and V Kp (3d-ls transition) spectra [5.271],
revealing strong Zr-H and V-H bonding, in agreement with the XPS results. A
204 M. Gupta and L. Schlapbach
I I r ]
Ce3d I I i

(Mg K~t) XPS (Mg Ka) /~

Oo

% i
".% * •
CeRuzH~4 .-"
"% .2
>- %
%
z
_ _ I %..,
z f2

.o%
• %

CeRuz ,"
o" "~, OX.
"-'.',...... • ~% ox.

• .o ,,%
%

, 1
I i
fo fl fafo ,, '2...
I I I [
910 90O 89O 880 -15 -I0 -5 0
~BINDING ENERGY (eV) ENERGY REL.TOE F (eV)

Fig. 5.22. Ce 3d core level spectra (XPS) of CeRu2 and its hydride with final state assignment.
The intensity of the fo final state satellite is significantly weaker in the hydride, revealing a
transition from nonmagnetic e-type Ce in CeRu2 to magnetic -/-type Ce in the hydride, i.e. a
hydrogen-induced change in the 4f localisation and hybridization [5.278]
Fig. 5.23. XPS valence band spectra of Pd, ZrPd2, Zr2Pd, and Zr. The Pd d-band becomes
significantly narrower and shifts away from Ev with increasing Zr concentration. In ZrPd 2 the
emission at Ev is very weak [5.15]

correlation between the position of the Z r - H bonding band as measured by

SXES Zr Lo spectra in the hydrides ZrVCuH2. 8, ZrVFeH3.7, ZrVCrH2.s, and
ZrVCoH3. 9 and the thermal stability of these hydrides has been discussed
[5.272]. It is concluded that Zr d-states play the most important role in the
formation of bonding states in this series of hydrides. V d-states also take part,
but to a lesser extent, and Cr, Fe, Co and Cu are thought not to participate in
the metal-H bonding. Susceptibility and specific heat measurements [5.273] of
ZrV 2 and ZrVzH1. a point to a decrease of N(EF) by a factor of 3 in the hydride,
in qualitative agreement with N M R Knight shift measurements [5.274] and
also with the calculated DOS of ZrV2, which predicts a large decrease of N(Ev)
if Ev is shifted upwards.
The susceptibility of TiCrl.s is almost temperature independent (2.6
x 10 -6 emu/g) and grows continuously with hydrogen content in the C-15 as
well as C-14 modifications up to 4.3 x 10 - 6 emu/g for TiCr I .sill.6 reflecting an
increase of Nd(EF) [5.275]. Similarly, an increase of N(EF) upon hydride
formation was also derived from susceptibility and specific heat of Ti0.4Mn0. 6
[-5.276]. TaV 2 exists in the ordered C-15 and in the random bcc modification
and forms with hydrogen an extended solid solution or a dihydride phase,
 Electronic Properties 205

respectively. The magnetic susceptibility decreases linearly with increasing

hydrogen content, however, the decrease in the case of the C-15 modification is
about 4 times stronger [5.277].
An interesting hydrogen induced electronic effect was observed with XPS in
the C-15 compound CeRuz [-5.278]: the strong 4f-conduction-band hybrid-
ization energy, which accounts for the nonmagnetic e-type Ce in the super-
conducting CeRu2, is reduced upon hydride formation from 120meV to
60 meV, accompanied by the reappearance of the magnetic moment of Ce.
Hydrogen induced emission and core-level shifts are also seen (see Table 5.2
and Fig. 5.22).
It was shown [,5.31] that major contributions to the heat of hydride
formation are an exothermic term due to the lowering of occupied metal states
and an endothermic term due to the upwards shift of Ev. If we assume that
within an alloy system ABx the endothermic term does not vary significantly
with the alloy composition, the increase in E v required to accommodate part of
the additional electron determines the hydride stability. An illustrative example
is the Zr-Pd system: Pd, Zr, and PdZr 2 form stable hydrides, whereas PdzZr
does not. Valence band photoelectron spectra show rather high DOS at E v for
Pd, Zr, and PdZr2 (Fig. 5.23). For Pd2Zr, however, the Pd 4d-band is full and
shifted away from EF so that EF falls into a region of low DOS. Accordingly, a
large shift of Ev is required and the hydride is unstable [-5.15].

5.4.5 Amorphous Alloys and Their Hydrides

Hydride forming amorphous or glassy alloys prepared by rapid quenching of

the melt, by coevaporation on cooled substrates [-5.280] or by solid-state
reactions such as rapid interdiffusion or hydrogenation [,5.281] could be
particularly interesting for the study of the electronic properties for the
following reasons: (i) the presence of effects due to the absence of long-range
order, (ii) their possibly quite different behaviour from that of their crystalline
counterparts if interstitial sites with quite different coordination numbers or
much smaller H - H distance could be occupied, and (iii) the fact that they are
experimentally more easily accessible by spectroscopic methods, since gener-
ally they do not disintegrate upon hydride formation.
Band-structure calculations of ordered and disordered crystalline alloys
[-5.194, 196, 197] and cluster calculations [5.282] are thought to yield a good
approximation for amorphous alloys. In most calculations performed so far, at
least one of the components is a transition metal.
Extensive studies of the electronic structure of amorphous alloys by
photoemission and x-ray emission spectroscopy are available [-5.203, 283]. A
comparison of these spectra with the calculated DOS of the corresponding
crystalline alloys suggests that the essential features of the electronic structure
are very similar in amorphous alloys and corresponding ordered close-packed
alloys.
 206 M. Gupta and L. Schlapbach
/1 /J
Fe 21~ ,line Zr 3d lines

Zr76 FeTe÷Hz ~.
• //
715 710 705 18s 1~0 /1 , 1"0
Binding energy leV)

I I I I I ' ~ Fig. 5.24. XPS valence band and core level spectra
of amorphous Zr76Fe24 and its hydride [5.284].
Hydrogen-induced bonding states appear at 6 eV
(I1). The structure 12, which has almost the same
intensity for both samples, is possibly due to
surface contamination. The Fe and Zr core levels
of the hydride are shifted by 0.3 and 0.8eV,
respectively. The increased asymmetry of the Fe
line points to a larger density of Fe states at Ev in
the hydride

Fig. 5.25. Zr L3 and Pd L 3 emission spectra of

amorphous Zr65Pda5 and its hydride compared
I I I I I I I I with calculated DOS for crystalline Zr3Pd I-5.10]
=10 - 8 - 6 - 4 -2 0 2 4
E - E F /eV

Relatively few hydrides of a m o r p h o u s alloys have been studied so far [5.12].

N o cluster calculations nor other electronic structure calculations for hydrides
of a m o r p h o u s or disordered alloys are available. Photoemission spectra of
a m o r p h o u s Fe24Zr76H16o show [-5.284] a broad ( ~ 4 e V ) hydrogen induced
band at 6 eV which develops at the expense of the conduction band emission
(Fig. 5.24). The Fe 2p and Z r 3d core levels are shifted by 0.3 and 0.8eV,
respectively, in the hydride. Furthermore the Fe 2p core level becomes rather
asymmetric indicating an increased Fe D O S at Ev. In hydrided a m o r p h o u s
 Electronic Properties 207

Ni26Zr76 an increased asymmetry of the Ni 2p line and a core-level shift of

both Ni (<0.2eV) and Zr (0.7 eV) were also observed [5.285].
Soft x-ray emission investigations of amorphous Pd35Zr65 (Fig. 5.25) and
Ni33Zr67 reveal [5.10] significant modification of the Zr L3(sd---r2p2/3 ) spectra
as a result of hydride formation (one hydrogen atom per metal atom). A low-
energy shoulder similar to that observed in ZrHx as a result of H - Z r bonding
states is induced about 7 eV below the emission edge. In contrast tothis, hydride
formation has negligible effects on the L 3 spectra of palladium and nickel in
these alloys, again indicating the strong H - Z r bonding. Similar behaviour was
observed in crystalline alloys.
Apart from the small core-level shifts of Fe and Ni, which seem to be absent
in the corresponding crystalline hydrides, photoemission and x-ray emission
did not reveal significant differences between the electronic structure of
hydrides of amorphous alloys and those of their crystalline counterparts. On
the other hand, N M R experiments have suggested that the d-electron DOS at
EF is reduced in amorphous PdZr2 hydride as compared to the hydride of the
crystalline alloy [5.286]. Hydrogen absorption by amorphous Pd-Zr and
Ni-Zr alloys has been shown, by measurements of the specific heat and electrical
resistivity, to reduce both the DOS at E v and also the superconducting
transition temperature [5.287, 288].

5.5 Conclusions

One of the goals of the research efforts on properties of metal hydrides is -

through a microscopic investigation of the bonding properties - to gain a better
insight into the modification of their electronic properties and to establish the
factors which control their stability and their hydrogen absorption capacity.
One would also like to be able to modify these materials - e.g. by alloying - in
order to make them suitable for the different applications.
A good qualitative, and in many examples even quantitative, agreement has
been achieved between band-structure calculations for metal hydrides and
measured spectra: the metal-hydrogen bonding interaction which leads to the
observed additional low-energy structures in the densities of states as well as the
modification of the valence bands upon hydrogen absorption are well
explained by theoretical calculations. The drastic modifications of properties
related to the Fermi energy observed with the hydride formation (electronic
specific heat, the Pauli susceptibility, the Fermi surface geometry, the electrical
resistivity and the superconducting properties), are generally also well accoun-
ted for by theoretical work. The early results concerning the metal-hydrogen
bonding interactions [5.13] and the major contributions to the heat of hydride
formation [5.31] have been confirmed by more recent work [5.32, 35, 36] and
extended to other systems.
The sign of the core-level shift agrees with the sign of the calculated charge
transfer. Variations of the line asymmetry can be described with variations of
208 M, Gupta and L. Schlapbach

the partial DOS at E v. The quantitative calculations of core-level shifts from

first principles is rather difficult and not available up to now. The measured
core-level shifts of binary and ternary hydrides point to a correlation with the
thermochemical stability of the hydrides.
Johansson and Mdlrtensson [5.289] and Steiner and Hiifner [-5.290] have
developed a model based on the equivalent core or Z + 1 approximation to
correlate the thermochemical data and the measured core-level shift of a
metallic element, A, upon the formation of an alloy AxB. The model, which
works so successfully that it was proposed that calorimetric measurements
could be replaced by core-level shift measurements. However, it cannot be
applied directly to correlate the core-level shifts measured upon hydride
formation to corresponding thermodynamic data for the following reasons
[5.190]: As long as the core-level shifts are not measured over a certain range of
hydrogen concentration, further simplifying assumptions have to be made. For
the hydrides of Ti, Zr, Nb, Pd, and Y a rather good agreement was obtained
between measured core-level shifts and those calculated from thermochemical
data. The calculations predict core-level shifts of 0.26, 0.14, and 1.0 eV for the
hydrides VHo.5, HfH z, and TaHo.8, respectively. For hydrides of rare earths,
however, the same approach fails, apparently because the core excitation of rare
earths cannot be described by an adiabatic process. An extension of core-level
shifts of ternary hydrides is possible, but requires their concentration de-
pendence, which is not yet available, or further assumptions and free
parameters.
Since the publication of the Hydrogen in Metals volumes [5.13] theoretical
calculations have gained in numerical accuracy in the density of states and
Fermi energy properties. Self-consistent treatments of the crystal potential are
now included and in some cases, wave functions have been obtained in order to
predict the values of several observables. Accurate total energy calculations
performed on a series of binary hydrides [-5.35, 46] give a good account of the
trends in the experimental values.
Theoretical work on stoichiometric and non-stoichiometric intermetallic
hydrides with complex crystal structure is progressing [5.6, 208, 227, 260-262].
The origin of the lower stability of the technologically important material
MgaNiH 4 as compared to MgH z appears clearly from the theoretical results
which have emphasized the role of the Ni-H bonds in these systems as opposed
to the rather ionic Mg- H interactions in MgHv Theoretical results have also
provided an explanation for the electrical conductivity and for the electronic
origin of the maximum H absorption capacity in the isoelectronic series
MgzFeH6, MgzCoHs, and MgzNiH4. Total energy calculations are however
still lacking for ternary hydrides and most of the intense activity on the
prediction of heats of formation (see Chap. 6) relies on semi-empirical
predictions based on physical concepts of the bonding obtained from
theoretical calculations on binary systems. No quantitative account of the
maximum hydrogen absorption capacity of intermetallic hydrides has yet been
given, and to date most of the discussions rely on arguments such as the number
 Electronic Properties 209

of e m p t y metal d-states a b o v e the Fermi energy of the metallic matrix available

for M - H b o n d i n g [5.224]. We would like h o w e v e r to point out that the
effective m e d i u m theory developed by Norskov and coworkers [5.36] could
provide some insight. The existence of an o p t i m u m host electron density at the
interstitital H sites, which appears to be necessary for h y d r o g e n dissolution,
could provide an explanation for the preferential H-site o c c u p a n c y and the
m a x i m u m H content in c o m p l e x systems.
N o t all of the goals m e n t i o n e d at the beginning of this section have yet been
achieved; theoretical efforts need to be c o n t i n u e d in the field of b o n d i n g
properties and total energies of c o m p l e x intermetallic hydrides. Experimental
developments of in situ i m p l a n a t i o n of p r o t o n s in thin films of FeTi, L a N i 5 etc.
could enable the spectroscopic studies n o t possible to date.

References
5.1 W.M. Mi.iller, J.P. Blackledge, G.G. Libowitz: Metal Hydrides (Academic Press, New
York 1968)
5.2 J. Friedel: Phil. Mag. 43, 153 (1952); Ber. Bunsenges. Phys. Chemie 76, 828 (1972)
5.3 N.N. Engel, J.E. Johnston: In EHydrogOne dans les MOtaux, Congr~s International
Paris, ed. by P. Bastien, Science et Industrie, Paris (1972)
5.4 A.C. Switendick: Hydrogen in Metals I, Topics Appl. Phys., Vol. 28, ed. by G. Alefeld, J.
V61kl (Springer, Berlin, Heidelberg 1978)
5.5 J.H. Weaver, D.J. Peterman, D.T. Peterson: In Electronic Structure and Properties of
Hydrogen in Metals, ed. by P. Jena, C.B. Satterthwaite, NATO Conference Series VI,
Vol. 6 (Plenum, New York 1983) p. 207
5.6 M. Gupta: J. Phys. F 12, 57 (1982); J. Less-Common Met. 88, 221 (1982); 101, 35 (1984)
5.7 L. Schlapbach: Phys. Lett. A91, 303 (1982)
5.8 K. Tanaka, N. Hamasaka, M. Yasuda, Y. Fukai: Proc. Jimis 2, Hydrogen in Metals,
Suppl. Trans. Japan Institute of Metals 21, 65 (1980); Y. Sato, K. Tanaka, M. Yasuda: J.
Less-Common Met. 88, 251 (1982)
5.9 N. Nishimiya, A. Suzuki, T. Sekine, S. Ono, E. Asada: J. Less-Common Met. 88, 263
(1982)
5.10 K. Tanaka, M. Higatani, K. Kai, K. Suzuki: J. Less-Common Met. 88, 317 (1982)
5.11 L. Schlapbach, J. Osterwalder, T. Riesterer: J. Less-Common Met. 103, 295 (1984)
5.12 H. Bernas, A. Traverse, C. Janot: Amorphous Metals and Semiconductors, ed. by P.
Haasen, R. 1. Jaffee (Pergamon, Oxford 1985) p. 435; A. Traverse, H. Bernas: J. Less-
Common Met., 129, 1 (1987)
5.13 G. Alefeld, J. V61kl (eds.): Hydrogen in Metals 1, 11, Topics Appl. Phys., Vols. 28, 29
(Springer, Berlin, Heidelberg 1978)
5.14 T. Sohmura, F.E. Fujita: J. Phys. F10, 743 (1980)
5.15 L. Schlapbach, S. B/ichler, T. Riesterer, S. Hfifner: J. Less-Common Met. 130, 301 (1987)
and references therein
5.16 H. Hohenberg, W. Kohn: Phys. Rev. 136, B 864 (1964); W. Kohn, kJ. Sham: Phys. Rev.
140, A 1133 (1965)
5.17 J.M. Ziman: Solid State Phys. 26, 1 (1971); M.L. Cohen, V. Heine, D. Weaire: Solid
State Phys. 24, 1 (1970); W.A. Harrison: Electronic Structure and the Propertiesof Solids
(Freeman, San Francisco 1980)
5.18 O.K. Andersen: Phys. Rev. B 12, 3060 (1975); The Electronic Structure of Complex
System, ed. by P. Phariseau, W.M. Temmerman, NATO ASI Series, Vol. 113 (Plenum,
New York 1984) p. 11
210 M. Gupta and L. Schlapbach

5.19 E.S. Alekseev, N.I. Kulikov, A.F. Tatarchenko: J. Less-Common. Met. 130, 261 (1987)
5.20 A.C. Switcndick: Ber. Bunsenges. Phys. Chem. 76, 535 (1972)
5.21 G. Lehmann, M. Taut: Phys. Status Solidi (b) 54, 469 (1972); O. Jepsen, O.K. Andersen:
Solid State Commun. 9, 1763 (1971)
5.22 F.M. Mueller, J.W. Garland, M.H. Cohen, K.H. Bennemann: Ann. Phys. (NY) 67, 19
(1971)
5.23 J.C. Slater, G.F. Koster: Phys. Rev. 94, 1448 (1954)
5.24 D.A. Papaconstantopoulos: Handbook of the Band Structure of Elemental Solids
(Plenum, New York 1985)
5.25 J. Friedel: Physics of Metals, ed. by J.M. Ziman, Vol. 1 (Cambridge University Press,
London 1978)
5.26 D.G. Pettifor: J. Phys. F7, 613 (1977); Phys. Rev. Lett. 42, 846 (1979)
5.27 F. Gautier, J. van der Rest, F. Brouers: J. Phys. F5, 1884 (1975)
5.28 L.H. Bennett (ed.): Theory of Alloy Phase Formation, AIME, N.Y. (1980) see
contributions by C.M. Varma, R.E. Watson, L.H. Bennett; C.M. Varma: Solid State
Commun. 31, 295 (1979); R.E. Watson, L.H. Bennett; Phys. Rev. Lett. 43, 1130 (1979)
5.29 V.L. Moruzzi, J.F. Janak, A.R. Williams: Calculated Electronic Properties of Metals
(Pergamon, New York 1978)
5.30 A.R. Williams, C.D. Gelatt, V.L. Moruzzi: Phys. Rev. B25, 6509 (1982)
5.31 C.D. Gelatt, Jr., H. Ehrenreich, J.A. Weiss: Phys. Rev. B 17, 1940 (1978)
5.32 A.R. Williams, J. Kfibler, C.D. Gelatt, Jr.: Phys. Rev. B19, 6094 (1979)
5.33 M. Methfessel, J. Kftbler: J. Phys. F 12, 141 (1982)
5.34 K.-M. Ho, H.-J. Tao, X.-Y. Zhu: Phys. Rev. Lett. 53, 1586 (1984)
5.35 A.C. Switendick: J. Less-Common Met., 130, 249 (1987)
5.36 J.K. N~rskov: Phys. Rev. B26, 2875 (1982); 28, 1138 (1983);
J.K. Norskov, F. Besenbacher: J. Less-Common Met. 130, 475 (1987);
P. Nordlander, J.K. Norskov, F. Besenbacher: J. Phys. F 16, 1161 (1986)
5.37 D.G. Westlake: J. Less-Common Met. 103, 203 (1984)
5.38 R. Griessen, A. Driessen: Phys. Rev. B 30, 4372 (1984); J. Less-Common Met. 103, 245
(1984)
5.39 M. Cyrot, F. Cyrot-Lackmann: J. Phys. F6, 2257 (1976); A. Pasturel, P. Hicter, F.
Cyrot-Lackmann: Solid State Commun. 48, 561 (1983)
5.40 A.R. Miedema: J. Less-Common Met. 32, 117 (1973); K.H.J. Buschow, P.C.P. Bouten,
A.R. Miedema: Rep. Prog. Phys. 45, 937 (1982)
5.41 M.S. Daw, M.I. Baskes: Phys. Rev. Lett. 50, 1285 (1983); S.M. Foiles, M.S. Daw: J. Vac.
Sci. Technol. A3, 1565 (1985)
5.42 W.M. Temmerman, A.J. Pindor: J. Phys. F13, 1869 (1983)
5.43 A. Bansil, R. Prasad, L. Schwarz: Electronic Structure and Properties oJ" Hydrogen in
Metals, ed. by P. Jena, C.B. Satterthwaite, NATO Conf. Series VI, Vol. 6 (Plenum, New
York 1983) p. 249
5.44 P. Soven: Phys. Rev. 156, 809 (1967)
5.45 J.S. Faulkner: Phys. Rev. B13, 2391 (1976)
5.46 D.A. Papaconstantopoulos, P.M. Laufer, A.C. Switendick: Hydrogen in Disorderedand
Amorphous Solids, ed. by G. Bambakidis, R.C. Bowman, NATO AS1 Series, Vol. 136
(Plenum, New York 1986) p. 139; D.A. Papaconstantopoulos, P.M., Laufer: J. Less-
Common Met., 130, 229 (1987)
5.47 M. Manninen, N. Nieminen: J. Phys. F 9, 1333 (1979); F. Perrot, M. Rasolt: Phys. Rev.
B 23, 6534 (1981); M. Manninen, J.M. Puska, R.M. Nieminen, P. Jena: Phys. Rev. B 30,
1065 (1984)
5.48 M. Gupta: The Electronic Structure of Complex Systems, ed. by P. Phariseau, W.M.
Temmerman, NATO ASI Series, Vol. 113 (Plenum, New York 1984) p. 243
5.49 B.M. Klein, W.E. Pickett: J. Less-Common Met. 88, 231 (1982)
5.50 R.P. Gupta: J. Less-Common Met. 88, 299 (1982)
5.51 R.P. Messmer, C.L. Briant: Acta Metall. 30, 457 (1982); C.L. Briant, R.P. Messmer: Acta
Metall. 32, 2043 (1984)
5.52 J.L. Whitten, T.A. Pakkanen: Phys. Rev. B21, 4357 (1980)
 Electronic Properties 211

5.53 M.A. Kahn, J.C. Parlebas, C. Demangeat: Electronic Structure and Properties of
Hydrogen in Metals, ed. by P. Jena, C.B. Satterthwaite, NATO Conf. Series VI, Vol. 6
(Plenum, New York 1983) p. 265
5.54 W.L. McMillan: Phys. Rev. 167, 331 (1968)
5.55 G.D. Gaspari, B.L. Gyorffy: Phys. Rev. Lett. 29, 801 (1972)
5.56 L. Schlapbach: J. Phys. F10, 2477 (1980); L. Schlapbach, C. Pina-Perez, T. Siegrist:
Solid State Commun. 41, 135 (1982)
5.57 R.M. Cotts: In Hydrogen in Metals l, ed. by G. Alefeld, J. V61kl, Topics Appl. Phys., Vol.
28 (Springer, Berlin, Heidelberg 1978) p. 227
5.58 R.G. Barnes: Handbook of the Physics and Chemistry of Rare Earth, ed. by K.A.
Gschneidner, L. Eyring, Vol. 2 (North-Holland, Amsterdam 1979) p. 387
5.59 A. Schenk: Muon Spin Rotation Spectroscopy (Adam Hilger, Bristol 1985)
5.60 F.N. Gygax, A. Hintermann, W. Riiegg, A. Schenck, W. Studer, A.J. van der Wal: J.
Less-Common Met. 101, 97 (1984)
5.61 B. Lengeler: De Haas van Alphen Studies of The Electronic Structure of the Noble
Metals and their Dilute Alloys, Springer Tracts Mod. Phys., Vol. 82 (Springer, Berlin,
Heidelberg 1978) p. 1
5.62 R. Griessen, L.M. Huisman: In Electronic Structure and Properties of Hydrogen in
Metals, ed. by P. Jena, C.B. Satterthwaite, NATO Conf. Series VI, Vol. 6 (Plenum, New
York /983) p. 235
5.63 R. L~isser, B. Lengeler: Phys. Rev. B 18, 637 (1978)
5.64 M. Hasegawa, S. Koike, H. Asano, M. Hirabayashi, S. Nuguchi: Proceedings Jimis 2,
Hydrogen in Metals, Suppl. Trans. Japan Institute of Metals 21, 61 (1980)
5.65 A. Ostrasz, E. Debowska: J. Less-Common Met. 130, 307 (1987); E. Debowska, B.
Rozenfeld: Z. Physik, Chemie N F 145, 71 (1985)
5.66 J.C. Fuggle, F.U. Hillebrecht, R. Zeller, Z. Zolnierek, P.A. Bennett, Ch. Freiburg: Phys.
Rev. B27, 2145 (1982)
5.67 R.E. Watson, M.L. Perlman: In Structure and Bonding, ed. by J.D. Dunitz (Springer,
Berlin, Heidelberg 1975) p. 83
M. Cardona, L. Ley (eds.): Photoemission in Solids 1, II, Topics Appl. Phys., Vols. 26,
77 (Springer, Berlin, Heidelberg 1978, 1979)
5.68 W. Speier, J.C. Fuggle, R. Zeller, B. Ackermann, K. Szot, F.U. Hillebrecht, M.
Campagna: Phys. Rev. B30, 6921 (•984)
5.69 C.F. Hague, R.H. Fairlie, W.M. Temmerman, B.L. Gyorffy, P. Oelhafen, H.J.
G~ntherodt: J. Phys. F 11, L 95 (1981); D.S. Urch: In Electron Spectroscopy, ed. by C.R.
Brundle, A.D. Baker, Vol. 3 (Academic, New York 1979) p. 1
5.70 J. Osterwalder: Z. Physik B61, 113 (1985)
5.71 B. Lengeler, R. Zeller: J. Less-Common Met. 103, 337 (1984)
5.72 J.H. Weaver, R. Rosei, D.T. Peterson: Phys. Rev. B 19, 4855 (1979)
5.73 D.A. Papaconstantopoulos, M. Klein: Phys. Rev. Lett. 39, 574 (1977); D.A. Papacon-
stantoponlos, B.M. Klein, E.N. Economou, L.L. Boyer: Phys. Rev. B 17. 141 (1978);
Phys. Rev. B 18, 2784 (1978)
5.74 M. Gupta, A.J. Freeman: Phys. Rev. B 17, 3029 (1978)
5.75 C.T. Chan, S.G. Louie: Phys. Rev. B27, 3325 (1983)
5.76 D.E. Eastman, J.K. Cashion, A.C. Switendick: Phys. Rev. Lett. 27, 35 (1971)
5.77 L. Schlapbach, J.P. Burger: J. de Phys. 43, L 273 (1982); and in Electronic Structure and
Properties of Hydrogen in Metals, ed. by P. Jena, C.B. Satterthwaite, NATO Conf.
Series VI, Vol. 6 (Plenum, New York 1983) p. 229
5.78 T. Riesterer, J. Osterwalder, L. Schlapbach: Phys. Rev. B 32, 8405 (1985)
5.79 P.A. Bennett, J.C. Fuggle: Phys. Rev. B26, 6030 (1982)
5.80 E. Belin, L. Schlapbach, M. Gupta: J. Phys. F 13, L 193 (1983)
5.81 S. Sinha, S. Badrinarayanan, A.P.B. Sinha: J. Phys. F16, L229 (1986)
5.82 A. Traverse, T. Kachnowski, L. Thom6, H. Bernas: Phys. Rev. B 22, 4355 (1980)
5.83 M. Gupta: Metal Hydrides, ed. by G. Bambakidis, NATO ASI Series, Vol. 76 (Plenum,
New York 1981) p. 255; M. Gupta, J.P. Burger: J. Phys. F10, 2649 (1980)
212 M. Gupta and L. Schlapbach

5.84 L. Schlapbach, T. Riesterer, J. Osterwalder, S. Biichler: unpublished

5.85 D. Horn-Chandesris: Th~se d'Etat, Universit6 Paris-Sud Orsay, unpublished (1983)
No. 2708
5.86 O. Blaschko: Phys. Rev. B 29, 5187 (1984)
5.87 W. Wampler, B. Lengcler: Phys. Rev. B 15, 4614 (1977)
5.88 H.L.M. Bakker, R. Griessen, N.J. Koeman, P. Albers, G.H. Sicking: J. Phys. F 16, 721
(1986)
5.89 C.L. Wiley, G. Cinader, F.Y. Fradin: Bull. Am. Phys. Soc. 21, 404 (1976)
5.90 B. Stritzker, H. Wiihl: In [5.13] Vol. I1, p. 243 and references therein
5.91 M. Gupta: Electronic Structure and Properties of Hydrogen in Metals, ed. by P. Jena,
C.B. Satterthwaite, NATO Conf. Series IV, Vol. 6 (Plenum, New York 1983) p. 321;
Phys. Rev. B 24, 7099 (1981)
5.92 D.S. McLachlan, 1. Papadopoulos, T.B. Doyle: J. Physique Coll. 6, 430 (1978)
5.93 J. Eckert, C.F. Majkzrak, L. Passell, W.B. Daniels: Phys. Rev. B 29, 3700 (1984)
5.94 D.A. Papaconstantopoulos: Metal Hydrides, ed. by G. Bambakidis, NATO ASI Series,
Vol. 76 (Plenum, New York 1981) p. 215
5.95 F. Ducastelle, R. Caudron, P. Costa: J. de Phys. 31, 57 (1970)
5.96 M. Gupta: Phys. Rev. B 25, 1027 (1982); Solid State Commun. 29, 47 (1979)
5.97 N.I. Kulikov, V.N. Borzunov, D.A. Zvonkov: Phys. Status Solidi (b) 86, 83 (1978); N.I.
Kulikov: Phys. Status Solidi (b) 91, 753 (1979)
5.98 A. Fujimori, N. Tsuda: J. Less-Common Met. 88, 269 (1982); Solid State Commun. 41,
491 (1982)
5.99 D.A. Papaconstantopoulos, A.C. Switendick: J. Less-Common Met. 103, 317 (1984)
5.100 K. Bohmhammel, G. Wolf, G. Gross, H. M/igde: J. Low Temp. Phys. 43, 521 (1981)
5.101 C. Korn: Phys. Rev. B 28, 95 (1983); C. Korn, S. Goren: J. Less-Common Met. 104, 113
(1984)
5.102 R.C. Bowman, B.D. Craft, J.S. Cantrell, E.L. Venturini: Phys. Rev. B31, 5604 (1985);
R.C. Bowman, Jr., E.L. Venturini, B.D. Craft, A. Attah, D.B. Sullenger: Phys. Rev. B 27,
1474 (1983)
5.103 K. Gesi, Y. Takagi, T. Takeuchi: J. Phys. Soc. Japan 18, 306 (1965)
5.104 K. Kundasamy, N.A. Surplice: J. Phys. D 18, 1377 (1985)
5.105 J.H. Weaver, D.J. Peterman, D.T. Peterson, A. Franciosi: Phys. Rev. B 23, 1692 (1981)
5.106 B.W. Veal, D.J. Lam, D.G. Westlake: Phys. Rev. BI9, 2856 (1979)
5.107 B.C. Lamartine, T.W. Haas, J.S. Solomon: Appl. Surf. Sci. 4, 537 (1980)
5.108 K. Tanaka, N. Hamasaka, M. Yasuda, Y. Fukai: Solid State Commun. 30, 173 (1979)
5.109 T.A. Sasaki, Y. Baba: Phys. Rev. B31, 791 (1985)
5.110 R.L. Tapping: J. Nucl. Materials 107, 151 (1982)
5.111 D.A. Papaconstantopoulos: Proc. 17th Conf. on Low Temperature Physics Part I,
(North Holland, Amsterdam 1984) p. 129
5.112 T. Springer: In [5.13] Vol. 1, p. 75, and references therein
5.113 P.W. Birkel, T.G. Berlincourt: Phys. Rev. B 2, 4807 (1970)
5.114 B. Stritzker, J.D. Meyer: Z. Physik B38, 77 (1980)
5.115 A.C. Switendick: J. Less-Common Met. 103, 309 (1984)
5.116 E.L. Venturini: In [5.35]
5.117 T. Schober, H. Wenzl: In [5.13], Vol. lI, p. 11
5.118 D.J. Peterman, D.K. Misemer, J.H. Weaver, D.T. Peterson: Phys. Rev. B 27,799 (1983)
5.119 Y. Fukai, S. Kazama, K. Tanaka, M. Matsumoto: Solid State Commun. 19, 507 (1976)
5.120 V.V. Nemoshkalenko, M.M. Antonova, M.M. Kindrat, V.P. Krivitskii, N.N. Shev-
chenko: Inorg. Mat. 19, 675 (1983)
5.121 L.M. Brown, A.P. Stephens: Acta Metall. 33, 827 (1985)
5.122 K. Tanaka, C. Sugiura, S. Nakai, Y. Ohno: Japan. J. Appl. Phys. 20, 41 (1981)
5.123 N. Stump, G. Alefeld, T. Tocchetti: Solid State Commun. 19, 805 (1976)
5.124 R. Sen Gupta, S. Chatterjee: J. Phys. F 12, 1923 (1982)
5.125 D. Ohlendorf, E. Wicke: J. Phys. Chem. Solids 40, 721 (1979)
5.126 B. Nowak, M. Minier: J. Phys. F 14, 1291 (1984)
 Electronic Properties 213

5.127 G.G. Libowitz, A.J. Mealand: Handbook on the Physics and Chemistry of Rare Earth,
ed. by K.A. Gschneidner, Jr., L. Eyring, Vol. 3 (North Holland, Amsterdam 1979) p.
299
5.128 R.R. Arons: Landolt-B6rnstein, Group III, Vol. 12c, Magnetic and other Properties of
Oxides and Related Compounds, Chap. 6.3 "Rare Earth Hydrides" (Springer, Berlin,
Heidelberg 1982) p. 372
5.129 P. Knappe, H. Miiller: J. Less-Common Met. 95, 323 (1983); Z. anorg, allg. Chemie 487,
63 (1982)
5.130 E. Kaldis, M. Tellefsen, R. Bischof: J. Less-Common Met., 130, 242 (1986)
5.131 J.J. Didisheim, K. Yvon, P. Fischer, W. H/ilg, L. Schlapbach: Phys. Lett. 78 A, 111 (1980)
5.132 T. Ito, B.J. Beaudry, K.A. Geschneidner, Jr.: J. Less-Common Met. 88,425 (1982); T. Ito,
B.J. Beaudry, K.A. Gschneidner, Jr., T. Takeshita: Phys. Rev. B27, 2830 (1983)
5.133 P. Klavins, R.N. Shelton, R.G. Barnes, B.J. Beaudry: Phys. Rev. B 29, 5349 (1984)
5.134 I.O. Bashkin, M.E. Kost, E.G. Ponyatovskii: Phys. Status Solidi (a) 83, 461 (1984)
5.135 J. Schefer, P. Fischer, W. H/ilg, J. Osterwalder, L. Schlapbach, J.D. Jorgensen: J. Phys.
C 17, 1575 (1984)
5.136 M. Drulis, Z. Bieganski: Phys. Status Solidi (a) 53, 277 (1979); M. Drulis, B. Stalinski:
Phys. Status Solidi (a) 69, 385 (1982)
5.137 J.H. Weaver, D.T. Peterson, R.L. Beubow: Phys. Rev. B 20, 5301 (1979)
5.138 D.J. Peterman, J.H. Weaver, D.T. Peterson: Phys. Rev. B 23, 3903 (1981)
5.139 L. Schlapbach, H.R. Scherrer: Solid State Commun. 41, 893 (1982)
5.140 L. Schlapbach, J. Osterwalder: Solid State Commun. 42, 271 (1982); L. Schlapbach, J.
Osterwalder, H.C. Siegmann: J. Less-Common Met. 88, 291 (1982)
5.141 A. Fujimori, N. Tsuda: Phys. Status Solidi (b) 114, K 139 (1982)
5.142 D.J. Peterman, J.H. Weaver, M. Croft, D.T. Peterson: Phys. Rev. 27, 808 (1983)
5.143 A. Fujimori, J. Osterwalder: J. Phys. C 17, 2869 (1984)
5.144 A. Fujimori, L. Schlapbach: J. Phys. C17, 341 (1984)
5.145 J.H. Weaver, D.T. Peterson, R.A. Butera, A. Fujimori: Phys. Rev. B32, 3562 (1985)
5.146 A.C. Switendick: J. Less-Common Met. 74, 199 (1980)
5.147 M. Gupta: Solid State Commun. 27, 1355 (1978)
5.148 M. Gupta, J.P. Burger: Phys. Rev. B22, 6074 (1980)
5.149 D.K. Misemer, B.N. Harmon: Phys. Rev. B 26, 5634 (1982)
5.150 N.I. Kulikov: J. Less-Common Met. 88, 307 (1982); N.I. Kulikov, A.D. Zvonkov: Z.
Phys. Chem. 117, 113 (1979)
5.151 D.J. Peterman, B.N. Harmon: Phys. Rev. B20, 5313 (1979); D.J. Peterman, B.N.
Harmon, J. Marchiando, J.H. Weaver: Phys. Rev. B 19, 4867 (1979)
5.152 A. Fujimori, F. Minami, T. Tsuda: Phys. Rev. B 22, 3573 (1980); A Fujimori, N. Tsuda:
J. Phys. C 14, 1427 (1981)
5.153 N.I. Kulikov: J. Less-Common Met. 107, 111 (1985)
5.154 R.A. Alikhanov, V.I. Buzin, N.I. Kulikov, M.E. Kost, W. Sikora, L.S. Smirnov: J. Less-
Common Met. 101,291 (1984)
5.155 R.G. Barnes, B.J. Beaudry, R.B. Creel, D.R. Torgeson, D.G. de Groot: Solid State
Commun. 36, 105 (1980)
5.156 A. Fujimori: Phys. Rev. B27, 3992 (1983)
5.157 H. Shaked, J. Faber, Jr., M.H. Mueller, D.G. Westlake: Phys. Rev. B16, 340 (1977)
5.158 M. Gupta: Solid State Commun. 50, 439 (1984)
5.159 P. Vorderwisch, S. Hautecler, W.D. Teuchert: Solid State Commun. 25, 213 (1978)
5.160 C. Stassis, T. Gould, O.D. McMasters, K.A. Gsehneidner, Jr.: Phys. Rev. B 19, 5746
(1979)
5.161 D.L. Johnson, D K . Finnemore: Phys. Rev. 158, 376 (1967); Z. Bieganski, D. Gonzales
Alvarez, F.W. Klaaijsen: Physics 37, 153 (1967)
5.162 G.G. Libowitz: Ber. Bunsenges., Phys. Chem. 76, 873 (1972)
5.163 J.N. Daou: C.R. Acad Sc. 250, 3165 (1960); J.N. Daou, J. Bonnet: C.R. Acad. Sc. 261,
1675 (1975)
5.164 A. Fujimori, M. lshii, N. Tsuda: Phys. Status Solidi (b) 99, 673 (1980); A. Fujimori, N.
Tsuda: J. Phys. C14, L69 (1981)
214 M. Gupta and L. Schlapbach

5.165 D.R. Torgeson, L.T. Lu, T.T. Phua, R.G. Barnes, D.T. Peterson, E.F.W. Seymour: J.
Less-Common Met. 104, 79 (1984)
5.166 D. Zamir, R.G. Barnes, N. Salibi, R.M. Cotts, T.T. Phua, D.R. Torgeson, D.T. Peterson:
Phys. Rev. B 29, 61 (1984)
5.167 A. Raizman, D. Zamir, R.M. Cotts: Phys. Rev. B31, 3384 (1985)
5.168 L. Schlapbach, H.R. Ott, E. Felder, H. Rudigier, P. Thiry, J. Bonnet, Y. Petroff, J.P.
Burger: J. Less-Common Met. 130, 239 (1987); L. Schlapbach, J.B. Burger, P. Thiry, J.
Bonnet, Y. Petroff: Phys. Rev. Lett. 57, 2219 (1986)
5.169 O. Blaschko, G. Krexner, J.N. Daou, P. Vajda: Phys. Rev. Lett. 55, 2876 (1985)
I.S. Anderson, J.J. Rush, T. Udovic, J.M. Rowe: Phys. Rev. Lett. 57, 2822 (1986)
5.170 P. Vajda, J.N. Daou, J.P. Burger, A. Lucasson: Phys. Rev. B31, 6900 (1985)
5.171 W.E. Pickett, A.J. Freeman, D.D. Koelling: Phys. Rev. B23, 1266 (1981)
5.172 L. Schlapbach: J. Less-Common Met. 111,291 (1985)
5.173 O.J. Zoga], B. Stalinski: Proc. Jimis 2, Hydrogen in Metals, Suppl. Trans. Japan. Inst. of
Metals 21, 77 (1980)
5.174 R. Bischof, E. Kaldis, I. Lacis: J. Less-Common Met. 84, 117 (1983)
5.175 P. Fischer, J. Schefer, K. Tichy, R. Bischof, E. Kaldis: J. Less-Common Met. 94, 151
(1983)
5.176 J. Osterwalder, H.R. Ott, L. Schlaobach, J. Schefer, P. Fischer: J. Less-Common Met.
94, 129 (1983)
5.177 R.R. Arons, A. Heidel, F.R. De Boer: Solid State Commun. 51,159 (1984)
5.178 R.R. Arons, A. Heidel, J.W. Cable: J. Less-Common Met., 130, 205 (1987)
5.179 R. Bischof, E. Kaldis, P. Wachter: J. Mag. Magn. Mater. 31-34, 255 (1983)
5.180 J.C. Fuggle, N. Mhrtensson: J. Electron Spectr. Rel. Phenomena 21, 275 (1980)
5.181 J.H. Weaver, M. Gupta, D.T. Peterson: Solid State Commun. 51,805 (1984)
5.182 J. Felsteiner, M. Heilper, I. Gerter, A.C. Tanner, R. Opher: Phys. Rev. B23, 5156 (1981)
5.183 G. Krasko: In Proc. Int. Symp. on Metal-Hydrogen Systems, ed. by T.N. Veziroglu, Z.
Naturforsch. 36a, 1146 (1981) (Pergamon, Oxford 1982)
5.184 J.W. Ward: Chap. 1 in Handbook on the Physics and Chemistry of the Actinides, ed. by
A.J. Freeman, C. Keller (Elsevier, Amsterdam 1985)
5.185 A.C. Switendick: J. Less-Common Met. 88, 257 (1982)
5.186 J.W. Ward, L.E. Cox, J.L. Smith, G.R. Stewart, J.H. Wood: J. de Phys. Colloque C 4, 40,
15 (1979)
5.187 J.C. Fernandes, M.A. Continentino, A.P. Guimarzes: Solid State Commun. 55, 1011
(1985)
5.188 J.H. Weaver, J.A. Knapp, D.E. Eastman, D.T. Peterson, C.B. Satterthwaite: Phys. Rev.
Lett. 39, 639 (1977)
5.189 J.F. Miller, R.H. Caton, C.B. Satterthwaite: Phys. Rev. B 14, 2795 (1976)
5.190 T. Riesterer: Submitted to Z. Physik B
5.191 L. Schlapbach: Solid State Commun. 38, 117 (1981)
5.192 E. Jensen, D.M. Wieliczka: Phys. Rev. B 30, 7340 (1984)
5.193 B.D. Padalia, W.C. Lang, P.R. Norris, L.M. Watson, D.J. Fabian: Proc. R. Soc. London
A 354, 269 (1977)
5.194 V.L. Moruzzi, P. Oelhafen, A.R. Williams, R. Lapka, H.J. Giintherodt, J. Kiibler: Phys.
Rev. B 27, 2049 (1983)
5.195 V.L. Moruzzi, P. Oelhafen, A.R. Williams: Phys. Rev. B 27, 7194 (1983)
5.196 P. Weinberger, J. Staunton, B.L. Gyorffy: J. Phys. F 12, 2229 (1982)
5.197 A. Bansil, M. Pessa: Physica Scripta T4, 52 (1983)
5.198 M.O. Robbins, L.M. Falicov: Phys. Rev. B29, 1333 (1984)
5.198aE.Z. da Silva, P. Strange, W.M. Temmerman, B.L. Gyorffy: Phys. Rev. B36, 3015
(1987)
5.199 G.M. Stocks, H.W. Winter: The Electronic Structure of Complex Systems, ed. by P.
Phariseau, W.M. Temmerman, NATO ASI Series B: Physics, Vol. 113 (Plenum, New
York 1983) p. 463
 Electronic Properties 215

5.200 J. Hafner: Helv. Phys. Acta 56, 257 (1983)

5.20t F.U. Hillebrecht, J.C. Fuggle, P.A. Bennet, Z. Zolnierek, Ch. Freiburg: Phys. Rev. B27,
2179 (1983)
5.202 N. M~rtensson, R. Nyholm, H. Cairn, J. Hedman, B. Johansson: Phys. Rev. B 24, 1725
(1981)
5.203 P. Oelhafen: In Glassy Metals II, Topics Appl. Phys., Vol. 53, ed. by H. Beck, H.-J.
Giintherodt (Springer, Berlin, Heidelberg 1983) p. 283
5.204 R. Lapka, P. Oelhafen, U.M. Gubler, H.-J. Gfintherodt: Phys. Rev. B 31, 7734 (1985)
5.205 A. Amamou: Solid State Commun. 33, 1029 (1980)
5.206 A. Percheron-Gurgan, C. Lartigue, J.C. Achard: J. Less-Common Met. 109, 287 (1985);
C. Lartigue, A. Percheron-Gurgan, J.C. Achard, J.L. Soubeyroux: J. Less-Common
Met. 113, 127 (1985)
5.207 S.K. Malik, F. Arlinghaus, W.E. Wallace: Phys. Rev. B 25, 6488 (1982)
5.208 M. Gupta: J. Less-Common Met., 130, 219 (1987)
5.209 S. Cabus, K. Gloos, U. Gottwick, S. Horn, M. Klcmm, J. Kfibler, F. Steglich, R.D.
Parks: Solid State Commun. 51,909 (1984)
5.210 S. Nasu, H.H. Neumann, N. Marzouk, R.S. Craig, W.E. Wallace: J. Phys. Chem. Solids
32, 2779 (1971)
5.211 T. Takeshita, G. Dublon, O.D. McMasters, K.A. Gschneidner, Jr.: In The Rare Earth in
Modern Science and Technology, ed. by G.J. McCarthy, J.J. Rhyne, H.B. Silber (Plenum,
New York 1980) p. 563; T. Takeshita, K.A. Gsehneidner, Jr., D.K. Thome, O.D.
McMaster: Phys. Rev. B21, 5636 (1980)
5.212 D. Ohlendorf, H.E. Flotow: J. Chem. Phys. 73, 2973 (1980)
5.213 J.H. Weaver, A. Franciosi, D.J. Peterman, T. Takeshita, K.A. Gschneidner, Jr.: J. Less-
Common Met. 86, 195 (1982)
5.214 T.K. Hatwar, D.R. Chopra: J. of Electron Spectrosc. Relat. Phenom. 35, 77 (1985)
5.215 P. Thompson, J.J. Reilly, L.M. Corliss, J.M. Hastings, R. Hempelmann: J. Phys. F 16,
675 (1986)
5.216 Y. Chung, T. Takeshita, O.D. McMasters, K.A. Gschneidner, Jr.: J. Less-Common Met.
74, 217 (1980)
5.217 J. Palleau, G. Chouteau: J. de Phys. Lett. 41, L227 (1980)
5.218 K.H.J. Buschow: In Handbook of Physics and Chemistry of Rare Earth, ed. by K.A.
Gschneidner, L. Eyring, Vol. 6 (North Holland, Amsterdam 1984)
5.219 F.N. Gygax, A. Hintermann, W. Ruegg, A. Schenck, W. Studer, L. Schlapbach, F.
Stucki: In Recent Developments in Condensed Matter Physics, Vol. 2, ed. by J.T. De
Vreese et al. (Plenum, New York 1981) p. 1
5.220 N. Marzouk, K.S. Craig, W.E. Wallace: J. Phys. Chem. Sol. 34, 15 (1973)
5.221 G.Y. Adachi, K.I. Niki, Ji Shiokawa: J. Less-Common Met. 81, 345 (1981)
5.222 J.W. Larsen, M.L. Fuller, B.R. Livesay: Proceedings of the Second Miami International
Conference on Alternative Energy Sources, Miami Beach, Florida (1979)
5.223 W.M. Walsh, Jr., L.W. Rupp, Jr., P.H. Schmidt, L.D. Longinotti: AIP Conf. Proc. 29,
686 (1975)
5.224 W.E. Wallace, F. Pourarian: J. Phys. Chem. 86, 4958 (1982); W.E. Wallace: J. Less-
Common Met. 88, 141 (1982)
5.225 K.A. Gschneidner, Jr., T. Takeshita, Y. Chung, O.D. McMasters: J. Phys. F: Met. Phys.
12, L 1 (1982)
5.226 T. Takeshita, K.A. Gschneidner, Jr., J.F. Lakner: J. Less-Common Met. 78, 43 (1981)
5.227 D.A. Papaconstantopoulos, A.C. Switendick: J. Less-Common Met. 88, 273 (1983);
Phys. Rev. B32, 1289 (1985)
5.228 J. Yamashita, S. Asano: Progress of Theoretical Physics 48, 2119 (1972)
5.229 D.A. Papaconstantopoulos: Phys. Rev. B 11, 4801 (J975)
5.230 G. Schadler, P. Weinberger: J. Phys. F 16, 27 (1986)
5.231 G.N. Kamm: Phys. Rev. BI2, 3013 (1975)
5.232 R. Hempelmann, D. Ohlendorf, E. Wicke: Hydrids for Energy Storage, ed. by A.F.
Andresen, A.J. Maeland (Pergamon, Oxford 1978) p. 407
216 M. Gupta and L. Schlapbach

5.233 L. Schlapbach, T. Riesterer: Appl. Phys. A32, 169 (1983)

5.234 J.H. Weaver, D.T. Peterson: Phys. Rev. B 22, 3624 (1980)
5.235 N. Motta, M. De Crescenzi, A. Balzarotti: Phys. Rev. B 27, 4712 (1983)
5.236 L. Schlapbach, A. Seller, F. Stucki, P. Ziircher, P. Fischer, J. Schefer: Z. Physik Chemie
N.F., 117, 205 (1979)
5.237 J.M. Welter, G. Arnold, H. Wenzl: J. Phys. FI3, 1773 (1983)
4.238 P. Thomspon, P.A. Pick, F. Rcidinger, L.M. Corliss, J.M. Hastings, J.J. Reilly: J. Phys.
FS, L75 (1978)
5.239 P. Fischer, W. H/ilg, L. Schlapbach, F. Stucki, A.F. Andresen: Mater. Res. Bull. 13, 931
(1978)
5.240 G. Hilscher, G. Wiesinger, R. Hempelmann: J. Phys. F 11, 2161 (1981)
5.241 R. L~isser, B. Lengeler, G. Arnold: Phys. Rev. B22, 663 (1980)
5.242 L.J. Swartzendruber, L.H. Bennett, R.E. Watson: J. Phys. F6, L331 (1976)
5.243 M. Gupta: Solid State Commun. 42, 501 (1982)
5.244 J. Callaway, R.A. Tawil, C.S. Wang: Phys. Lett. A46, 161 (1973)
5.245 T. Shinohara, M. Fujika: Phys. Rev. BT, 37 (1973)
5.246 R. lngalls: Phys. Rev. 155, 157 (1967)
5.247 K. Schubert, K. Andcrko: Z. Metallkde. 42, 321 (1951)
5.248 M. Cyrot, M. Lavagna: J. de Phys. 40, 763 (1979)
5.249 R. Visnov, F. Ducastelle, G. Treglia: J. Phys. FI2, 441 (1982)
5.250 F. Stucki, L. Schlapbach: J. Less-Common Met. 74, 143 (1980)
5.251 J.J. Reilly, R.H. Wiswall: Inorg. Chem. 18, 2254 (1968)
5.252 Z. Gavra, M.H. Mintz, G. Kimmel, Z. Hadari: Inorg. Chem. 18, 3595 (1979); J.
Genossar, P.S. Rudman: J. Phys. Chem. Solids 42, 611 (1981); S. Ono, H. Hayakawa, A.
Suzuki, K. Nomura, N. Nishimiya, T. Tabata: J. Less-Common Met. 88, 63 (1982)
5.253 K. Yvon, J. Schefer, F. Stucki: lnorg. Chem. 20, 2776 (1981); D. Noreus, L.G. Olsson: J.
Chem. Phys. 78, 2419 (1983)
5.254 B. Darrict, J.L. Soubeyroux, M. Pczat, D. Fruchart: J. Less-Common Met. 103, 153
(1984)
5.255 J.J. Didisheim, P. Zolliker, K. Yvon, P. Fischer, J. Schefer, M. Gubelmann, A.F.
Williams: lnorg. Chem. 23, 1953 (1984)
5.256 J.P. Darnaudery, M. Pezat, B. Darriet, P. Hagenmuller: Mater. Res. Bull. 16, 1237
(1981)
5.257 P. Zolliker, K. Yvon, P. Fischer, J. Schefer: lnorg. Chem. 24, 4177 (1985)
5.258 P. Zolliker, K. Yvon, P. Fischer, J. Schefer: J. Less-Common Met. 129, 211 (1987)
5.259 P. Lindberg, D. Noreus, M.R.A. Blomberg, P.E.M. Siegbahn: J. Chem. Phys. 85, 4530
(1986)
5.260 M. Gupta: J. Less-Common Met. 103, 325 (1984); E. Orgaz, M. Gupta: J. Less-
Common Met. 130, 293 (1987)
5.261 M. Gupta, E. Belin, L. Schlapbach: J. Less-Common Met. 103, 389 (1984)
5.262 E. Belin, M. Gupta, P. Zolliker, K. Yvon: J. Less-Common Met., 130, 267 (1987)
5.263 M. Gupta: Unpublished
5.264 R.O. Moyer, Jr., C. Stanitzki, J. Tanoka, M.I. Kay, K. Kleinberg: J. Solid State Chem. 3,
541 (1971); J.S. Thompson, R.O. Moyer, Jr., R. Lindsay: Inorg. Chem. 14,1866 (1975); R.
Lindsay, R.O. Moyer, Jr., J.S. Thompson, D. Kuhn: Inorg. Chem. 15, 3050 (1976)
5.265 Z. Gavra, G. Kimmel, Y. Gefen, M.H. Mintz: J. Appl. Phys. 57, 4548 (1985)
5.266 D. Noreus, K. Jansson, M. Nygren: Z. Physik. Chemie N.F. 146, 191 (1985)
5.267 B.M. Klein, W.E. Pickett, D.A. Papaconstantopoulos, L.L. Boyer: Phys. Rev. B 27, 6721
(1983)
5.268 T. Jalborg, A.J. Freeman: Phys. Rev. B 22, 2332 (1980)
5.269 T. Riesterer, P. Kofel, J. Osterwalder, L. Schlapbach: J. Less-Common Met. 101,221
(1984); T. Riesterer: Z. Physik B 66, 441 (1987)
5.270 1. Jacob, M. Polak: Mater. Res. Bull. 16, 13~1 (1981)
5.271 Y.M. Yarmoshenko, V.N. Kozhanov, V.M. Cherkashenko, E.Z. Kurmaev, E.P.
Romanov: Solid State Commun. 55, 19 (1985)
 Electronic Properties 217

5.272 S.G. Porutsky, E.A. Zhurakovsky: J. Less-Common Met. 120, 273 (1986)
5.273 C. Geibel, W. Goldacker, H. Keiber, V. Oestreich, H. Rietschel, H. Wfihl: Phys. Rev.
B30, 6363 (1984)
5.274 M. Peretz, J. Barak, D. Zamir: Phys. Rev. B23, 103 (1981)
5.275 J.F. Lynch, J.R. Johnson, R.C. Bowman, Jr.: In Electronic Structure and Properties of
Hydrogen in Metals, ed. by P. Jena, C.B. Satterthwaite, NATO Conf. Series VI, Vol. 6
(Plenum, New York 1983) p. 437
5.276 R. Hempelmann, E. Wicke, G. Hilscher, W. Wiesinger: Ber. Bunsenges. Phys. Chem. 87,
48 (1983)
5.277 J.F. Lynch, R. Lindsay, R.O. Moyer, Jr.: Solid State Commun. 41, 9 (1982)
5.278 J. Osterwalder, T. Riesterer, L. Schlapbach, F. Vaillant, D. Fruchart: Phys. Rcv. B 31,
8311 (1985)
5.279 G. Teisseron, P. Vulliet, L. Schlaphach: J. Less-Common Met., 130, 163 (1987)
5.280 H.-J. Giintherodt, H. Beck (eds.): Glassy Metals 1, II, Topics Appl. Phys., Vols. 46, 53
(Springer, Berlin, Heidelberg 1981, 1983)
5.281 K. Samwer: In Hydrogen in Disordered and Amorphous Solids, ed. by G. Bambakidis,
R.C. Bowman, NATO ASI Series, Series B: Physics, Vol. 136, 173 (1986); J. Less-
Common Met., to be published
5.282 R.H. Fairlie, W.M. Temmerman, B.L. Gyorffy: J. Phys. F 12, 1641 (1982)
5.283 C.F. Hague, J.M. Mariot, E. Belin, E. Beauprez: In Amorphous Metals and Non-
Equilibrium Processing, ed. by M. yon Allmen (Les Editions de Physique, Paris 1984)
p. 331; J.M. Mariot, C.F. Hague, P. Oelhafen, H.-J. Gi.intherodt: J. Phys. F16, 1197
(1986)
5.284 S.M. Fries, H.-G. Wagener, S.J. Campell, U. Gonser, N. Blaes, P. Steiner: J. Phys. F 15,
1179 (1985)
5.285 X. Yu, L. Schlapbach: Phys. Rev. B (1988) in press
5.286 R.C. Bowman, Jr., W.L. Johnson, A.J. Maeland, W.K. Rhim: Phys. Lett. 94 A, 181 (1983)
5.287 M. Kulik, G. yon Minnigerode, K. Samwer: Z. Physik B60, 357 (1985)
5.288 U. Mizutani, S. Ohta, T. Matsuda: J. Phys. Soc. Japan 54, 3406 (1985)
5.289 B. Johansson, N. Mfirtensson: Phys. Rev. B 21, 4427 (1980)
5.290 P. Steiner, S. Hiifner: Acta Metall. 29, 1885 (1981)
6. Heat of Formation Models

Ronald Griessen and Thomas Riesterer

With 23 Figures and 3 Tables

First principles calculations of the cohesive properties of metal-hydrogen

systems are still inaccurate and time consuming. There is thus presently a need
for simple empirical models. Empirical and semi-empirical models for the heat
of formation and heat of solution of binary and ternary hydrides are reviewed.
Many of the proposed empirical correlations are in general based on too small a
number of metal-hydrogen systems and are therefore often only useful (and
valid) for restricted classes of materials.
In this chapter it is shown that semi-empirical models for the heat of alloy
formation have to satisfy several general relations. The two semi-empirical
models proposed so far for the heat of formation and heat of solution of metal
hydrides (models by Miedema, Griessen, and Driessen) are shown to follow
these relations. The semi-empirical band structure model of Griessen and
Driessen is more reliable than the cellular model of Miedema although it
involves a much smaller number of fit parameters. The cellular model suffers
furthermore from the fact that it does not allow phase separation. Both models
show however that the energetics of the metal-hydrogen interaction depend
both on geometric and on electronic factors.
The chapter is completed with the derivation of the thermodynamic
quantities relevant for hydride stability and with a set of tables containing
experimental data for the heats of solution and the heats of formation of binary
and ternary hydrides and calculated values for the heats of formation of ternary
transition metal hydrides.

6.1 Introduction

One of the fundamental characteristics of a metal-hydrogen system is the heat

of formation AH, i.e. the amount of heat absorbed or liberated during hydrogen
absorbtion by the metallic host. AH determines essentially the stabilitity of
metal hydrides because it is directly related to the dissociation pressure p via the
van 't Hoff relation

AH AS
~lnp = (6.1)
RT R'
220 R. Griessen and T. Riesterer

where the entropy of formation AS is approximately constant for all

metal-hydrogen systems as it arises mainly from the entropy loss of gaseous
hydrogen during hydrogen uptake by the metal.
Despite the importance of AH, for example for technical applications, first
principles calculations of the heat of formation of metal hydrides are still
lacking for ternary hydrides AyAByBHx because the theory of cohesive
properties of multicomponent systems is inherently complicated and a priori
calculations are still very time consuming. Until now AH has been calculated
for selected binary hydrides MHx only [6.1-6]. Almost no calculations are
available for AH in ternary metal hydrides although the band structure of some
of them has been calculated (e.g. Ti I _rVrHx [6.7], TiFeH 2 and TiPdH 2 [6.2],
Pda_yAgyH~ and Pdl_yRhyH~ [6.8], FeTiH~ [6.9, 10], Pdl_yAgyH x [6.11],
Ca2RuH 6 and MgEFeH 6 [6.12], and Mg2NiH4 [6.13]).
The difficulty in determining d H from first principles has stimulated a
search for empirical correlations between AH and physical parameters such as
the unit cell volume [6.14], interstitial site size [6.15, 16], elastic modulii
[6.17-20], Debye temperatures [6.21] and electronic specific heat
[6.17, 22, 23]. These empirical correlations are useful for the optimization of
metal-hydrogen systems for specific technical applications. In general, how-
ever, they do not provide us with more insight into the physics of hydride
formation and are difficult to generalize.
This is probably the reason why semi-empirical models were developed to
predict AH for large classes of metal hydrides. The best known example of such
a model is due to Miedema and his collaborators [6.24-29]. In this model each
metal is characterized by two parameters (essentially the work function and the
electron density at the boundary of the Wigner-Seitz cell of the elemental
metal). The heat of formation is given as a universal function of these
characteristic parameters. For the application of this model to ternary hydrides
one uses the rule of reversed stability which relates, in a straightforward
manner, the heat of formation of the hydride of an intermetallic compound to
the heats of formation of the intermetallic compound itself and of the binary
hydrides of the metals constituting the compound. The rule of reversed stability
is found to work well for intermetallics made of a minority metal with high
hydrogen affinity and a majority metal with a low hydrogen affinity (e.g. LaNi 5,
YCos, but not FeTi). For other types of compounds it generally leads to too
negative values for AH. Several workers [6.30-32] have tried to give a
microscopic basis to the model of Miedema, but without too much success.
Recently the problem of the heat of formation of hydrides has been tackled
along two different lines, which as shall be discussed later, have some points in
common.
First, Norskov [6.3] proposed a relatively simple scheme based on the
effective medium theory to calculate the embedding energy of an atom in the
host metal. The host metal is replaced by a homogeneous electron gas of a
density equal to that seen by the impurity atom. Norskov's scheme is especially
useful for rare gases in metals because their core levels are so far from any of the
 Heat of Formation Models 221

Fig. 6.1. Comparison of experimental

100
W
heats of solution at infinitedilution for
binary metal hydrides with values cal-
Ag
Ir culated by Norskov by means of the
50 %0 c r effectivemedium theory [6.3]
Pt Mo
Au MRnhc° Fe
Ni
Mn
E 0
PE{

HI Ta N V
18 -5O
Tc
Zr La
Y
Sc
-100

-150 I I I
-100 - 50 0 50 100

A~loo E M [ k J I m o l H ]

host metal levels that covalent effects can be neglected. For hydrogen in metals
this is however not the case and covalent effects must be explicitly taken into
account. Using an appropriate expression for these effects (derived from the
atomic sphere approximation formalism of Andersen et al. [6.33, 34]) he
obtained AH values which reproduce the general trends in the experimental
data. However, as shown in Fig. 6.1 the theoretical values for AH at infinite
dilution may be offby a factor 2 or 3 or even have the wrong sign (for Ag, Pt, Au,
V, Nb, and Ta). Although considerably simpler than first principles calculations
Norskov's scheme is at present not appropriate for calculations of a large
number of more complicated hydrides as is required for example in the tailoring
of intermetallics for hydrogen storage or other specific applications.
Second, Griessen and Driessen [6.35-38] proposed a semi-empirical model
in which each metal is characterized by one parameter (essentially the difference
between the Fermi energy and the energy of the lowest conduction band of the
host metal). This model reproduces well the heat of formation of binary
hydrides and, when generalized to ternary hydrides, predicts AH values in good
agreement with existing experimental data. Due to its simplicity this model
requires only very modest computing time. It can thus be used effectively in a
screening of attractive materials for specific technical applications.
The purpose of this chapter is to review some of these recent developments
and to provide the reader with up-to-date information on the heat of formation
of metal-hydrogen systems. It is by no means meant to be comprehensive and
the interested reader is referred to the excellent reviews of Buschow et al. [6.29],
Westlake [6.39], Flanagan and Oates [6.40], Wenzl [6.41] and Carter and
Carter [6.42].
This article is organized as follows. In Sect. 6.2 we introduce the thermody-
namic quantities relevant to the stability of metal hydrides. In Sect. 6.3 several
222 R. Griessen and T. Riesterer

empirical correlations between the heats of formation of metal hydrides and

characteristic physical parameters of the host metal are indicated and their
applicability is briefly discussed. The two semi-empirical models mentioned
above are described in Sects. 6.4.2 and 6.4.3. The local semi-empirical band
structure model proposed very recently by one of the authors is briefly
presented in Sect. 6.4.4• Conclusions are given in Sect. 6.5. Tables of the heats
of formation of binary and ternary metal-hydrides are given in Sects. 6.6.1 and
6.6.2. Theoretical values for the heats of formation of ternary hydrides
obtained by means of the Griessen-Driessen model are tabulated in Sect. 6.6.3.

6.2 Thermodynamics of Hydrogen Absorption in Metals

Hydrogen may in principle be dissolved in any metal to form a (dilute)
homogeneous solution. Hydrogen is then in thermodynamic equilibrium with
one metal hydride phase MH x and consequently

½~qt2(P,T) = #H(P, T, x), (6.2)

where #H~ is the chemical potential of pure molecular hydrogen (which is in the
gaseous state at pressures lower than 53 kbar [6.43] at room temperature) and
/~H is the chemical potential of atomic hydrogen in solution in the metal. By
differentiating (6.2) with respect to 1/T at constant concentration x = [H]/[M]
we obtain

1 ~(/zn2/T) 1 ~(#u2/T) ~p
2 /?(l/T) v q 2 ~3p T ~ x

-- O([IH/T) ~(IlI-I/T) ~ ~3p x" (6.3)

a(1/T) p,x + ~p r

Using the general relations

c~V
(6.4)
~n,T -a2G.
~p (3n r =T~ ~, -F"

and

~(#/T) (6.5)
a(l/T) p,n = ~3~np, T = H

we can rewrite (6.3) in the following form

1 1 ~p
(6.6)
 Heat of Formation Models 223

In (6.4-6) G is the Gibbs free energy of the total system, n the number of moles,/7
is the partial molar enthalpy and ~"the partial molar volume. The quantity A/7
~-/~n--½Hn~ is the partial molar heat of solution of hydrogen in a metal per
gram-atom of hydrogen.
There is some experimental evidence that F'n, the partial molar volume of
hydrogen in a metal, is independent of pressure and temperature [6.44-47]. It is
therefore possible to determine A/7 from a plot ofp versus In T once the pressure
and temperature dependence of the enthalpy/Tn~ and ~2 of pure hydrogen are
known. Values of the thermodynamic properties in the temperature range
100 K to 1000 K at pressures up to 1 Mbar are given by Heroines et al. [6.48].
Values for F"n are given in the review articles of Peisl [6.44] and Westlake [6.49].
Bouten and Miedema [6.28] and Griessen and Feenstra [6.38] apply their semi-
empirical models to the calculation of the partial molar volume.
At low pressures (p < 100 bar) the partial molar volume of hydrogen gas is
much larger than VH (which is typically 1.7 cm3/mol H) and H z behaves
approximatly as an ideal gas, i.e. pV'n~-RT. Equation (6.6) then simplifies to

_ 1-o R Olnp
HH-- 2 HHz'~ 2 O(~)~" (6.7)

Since HH2 is nearly independent of pressure in this regime, we have replaced/7n~

by its value at the standard pressure of one atmosphere.
As indicated in the Tables of Sect. 6.6 the absorption of hydrogen in a metal
can be exothermic or endothermic. For the early transition-metal hydrides, for
example one finds large negative values for AH (exothermic). For the late
transition metals as well as for the majority of the simple metals A/7 is positive
(endothermic). In contrast to exothermic systems, the metals with A H > 0
absorb only small quantities of hydrogen at moderate pressures. For such
systems one often carries out experiments at constant pressure and determines
the temperature dependence of the solubility of hydrogen in a given metal.
From these data one can derive the partial heat of solution in the following way.
At constant pressure (6.2) leads to

3(ltH/T). ~O(~T) p = Hn- ~ Hn2 (6.8)

~X p, T

The first derivative in (6.8) can be written as

l OHn OSH. (6.9)

T cgx p,T (~X p,T"

In the limit of infinite dilution (0Hn/Ox)lp, T tends towards a constant. The

entropy SH however, diverges because of the contribution R In Ix/(1 - x ) ] of the
224 R. Griessen and T. Riesterer
mixing entropy term. We then have for x--*0

R
--*-- so that, (6.10)
OX p, T X

AH=-R t?lnx (6.11)

O(1/T) p"

The partial molar heat of solution may thus be obtained from a lnx versus
inverse temperature plot.
Until now we have only considered the situation of a homogeneously
hydrogenated metal in equilibrium with pure hydrogen. However, many
metal-hydrogen systems exhibit miscibility gaps for which two metal-hydride
phases, say a and fl, coexist and are in equilibrium with the surrounding
hydrogen 1. Below a critical temperature the pressure-composition isotherms
have a plateau for hydrogen concentrations between x, and xp. The conditions
for thermodynamic equilibrium between MHx,, MHxo, and H 2 are

#~(p, T,x~)=#~(p, T, xp)= ½#H2(P, T) and (6.12)

#Tu(P, T, x , ) = #~(p, T, xs). (6.13)

The atomic fractions x, and xp are independent variables. The corresponding
Gibbs-Duhem equations are, at constant p and T,

x=-- + =0 and (6.14)

~x,

=°- (6.15)

Consider now the following derivative taken along the coexistence line defined
by the equilibrium conditions (6.12, 13) given above,

d(l~;/T) O(#;/T) 0(/z;/T) dp a(#~/T) dxv

d(1/T) -- 0(l/T) + 3p O(I/T~) + c~xv dOlT)" (6.16)

Using the standard relations (6.4, 5) we obtain

d(#~/T)-H~+ Vj? dp ~(#~/T) dx~

(6.17)
d(1/T) T d(1/T~) + Oxv d(1/r)'
where v = ct or fl and j = H or M.

1 The labeling of the phases ~ and /3 does not refer to specific metal-hydride phases but
indicates any coexisting phases.
 Heat of Formation Models 225

Along the coexistence line we have, as a direct consequence of (6.12, 13)

d(#~/T) _ d(#¢~/T) 1 d(l~nJT) (6.18)

d(1/T) dOlT) 2 d(1/T)
and

d(l~/T) _ d(#~/T)
(6.19)
d(1/T) d(1/T) "
Multiplying (6.17) for v = a a n d j = H by x~ and adding it to (6.17) for v = a and
j = M we obtain

d(p~/T) d(#~/T) =H~+ V~ dp (6.20)

x~ d(1/T~ + dO~T) -7" d(1/T)
with
H~ = x~/~ + lq~u (6.21)
and

= x~,V~ + V~. (6.22)

For the fl phase we have similarly

d(#Pw/T) d(iz~/T) Vp dp
xa d(1/T~ -~ d ( 1 / T ~ - H a + T d(1/T)" (6.23)

Subtraction of (6.20) from (6.23) leads, in combination with (6.18) to

dp _ Hg--H~--I(xI1--x~)I~H2
(6.24)
Td(1/T) ½(x a - x~) VH2--(Vp -- V~) '

This is a general expression which is also valid under high pressure conditions.
It can be simplified by using the fact that Vr~ depends only weakly on the
hydrogen concentrations and that P~-~ P~t. We then obtain

A H ~ , - - * t ~ V'
(H;, - - -1~ 2 ) d dp
ln T (6.25)

where AH,~a is the enthalpy of formation (expressed per mol H) of the hydride
MHxo from the hydrogen saturated metal solid solution of composition MHx,,
i.e.

AH~_~p- H a - H ~ 1 _
x#--x~ 2 HH2 (6.26)
226 R. Griessen a n d T. Riesterer

or equivalently AH~,._,a is the heat of the reaction of the process

M H ~ + ½(xtj-- xe)H 2 ~MH~¢. (6.27)

To gain some insight into the various contributions to the right-hand side of
(6.26) it is useful to consider a simple lattice-gas model for which the chemical
potential of hydrogen in a metal is given by an expression of the type

#h~=RTln ~ x ) + A + a x + P n p + ~ 3R 0 ~+ 3RTln(1-e -°~/r) (6.28)

where A, a, and Vn and the Einstein temperature 0 E of the lattice-gas vibrations

(optical Ph0nons ) are treated as constant parameters. We then have

1
lq~~ = A + ax + VnP + 2 ROE + 3ROE e °r~lT- 1
(6.29)

and

~n~ = --R In i X-x --3Rln(l--e-°~/r)+ 3ROET (6.30)

e °~/r- 1

and from the Gibbs-Duhem equations (6.14, 15)

ax 2
/~t~ = - - - + H- ®M (6.31)
2

~= - R ln(1 - x ) + S~-~, (6.32)

m

where H~t and S~t are the molar enthalpy and entropy of the pure host metal.
When introduced into the expression (6.26) for AH~,._,p these relations lead to

A H ~ p = A + ~ (ax p + x,) + PnP + -~

3 ROE+ 3ROE e oE/1- 1~ 21 Ha2"
- (6.33)

Since in this model x~,+ xtj = 1, we have that AH~,~t~=/-3~~ (x = 1/2). The slope of
a lnp versus lIT plot for x = 1/2 is thus continuous at the critical point.
In the low pressure regime /7n2 is independent of pressure and varies
approximately linearly with temperature between 200 and 1000 K with a slope
0/Tn2/OT=I4.5J/K tool H2. This temperature variation is exactly com-
pensated by the optical phonon term at the temperature T~omp=0.38 0~. As
metal hydrides have typically 0E~--1000 K this implies that the temperature
dependence of the last two terms in (6.33) cancel each other around room
 Heat of Formation Models 227

temperature. For the simple lattice-gas model considered here one thus expects
the Inp versus 1/T plots to be linear over a large interval of temperature. For a
detailed discussion of this point the reader is referred to the articles by Flanagan
[6.50], Flanagan and Lynch [6.51], l'Hcke and Blaurock [6.52], and Vvickeand
Brodowsky [6.53-1.
To conclude this section we consider once more the case of endothermic
metal-hydrogen systems. Recently, several late transition metals have been
loaded with hydrogen at pressures up to 70 kbar [6.47]. Unfortunately in many
cases only one isotherm has been measured so that relation (6.6) cannot be used
for the evaluation of the heat of hydride formation. An estimate of A/1 or AH,_,a
may however be obtained in the following way.
For a single phase metal hydride the equilibrium condition (6.2) leads
directly to

A n = T S H-- ½TSn2 • (6.34)

If only half of the available sites of a given type are occupied then the mixing
entropy term in Sn vanishes. The magnitude of ,gH is therefore general.ly small
and A/1 is readily evaluated from the value of gH2 at a given pressure and
temperature I-6.48, 54].
Similarly for a two-phase metal hydride the equilibrium conditions
(6.12, 13) lead to

AH,~p=TAS,_,a with (6.35)

AS=_~a= Sa-S, 1-
x a - x~ 2 Srh and (6.36)

S,= x,S~ + S~. (6.37)

Within the simple lattice-gas model given above the first term on the right-hand
side of (6.36) vanishes identically so that

AH~a= -- ½TgH~ . (6.38)

At low pressures

'KH2= S-°n2-- R lnp (6.39)

and (6.38) reduces to

21np AH~_,o S°2 (6.40)

 228 R. Griessen and T. Riesterer

T [K 3 Fig. 6.2. Variation of plateau pressure with

1000.500 300 200 150 100 temperature for a representative metal hy-
dride according to relation (6.43) with
~'u= 1.7 cma/mol H and 04 = 850 K, For
each line the standard heat of formation
105
AH°+o is given in kJ/mol H. The values of
the chemical potential of pure hydrogen
used in (6.43)are from Heroineset al. [6.48].
As (6.34) for Sn=0 reduces to the same
lo 4
form as (6.38) the curves shown in this
figure are also valid for a single-phase metal
-£
hydride in which only half of the sites of a
given type are occupied. The numbers
! 1o3 indicated correspond then to the standard
heat of solution AB °
12.

lO;

lO 1

2 4 6 8 10
IO00/T [ K ]

an expression which has often been used (see for example [-6.29]) to get
estimates of the heat of formation of metal hydrides with
S°2 = 130.8 J/K mol H 2.
One c o m m o n feature of (6.34) (with Sn = 0) and (6.38) is that AH or AH,_,a
are always negative. This seems at first sight to be in contradiction with
solubility measurements in many simple metals and late transition metals
(except Pd and Mn) where positive heats of solution (reaction) are found (see
first column in Table 6.6.1). This apparent discrepancy is easily resolved by
noting that Sn = 0 in (6.34) is not possible at infinite dilution. The discrepancy
seems however to persist for concentrated hydrides of Cr, Co, Fe, Mo, and Rh
for which slightly positive AH,._,t~ have been found (see second column in
Table 6.6.1). The explanation is that A/~ and AH~,p refer to equilibrium states
at a given pressure (often in the kbar range) and temperature well above room
temperature. In order to reduce the enthalpies AH or AH~,_,p to a chosen
standard state we use the fact that

cqlnFn
fin-fr°= .oi F'n 1 - ~ l n T p,I dp+ ~eHdT' (6.41)

where ~ is the partial molar specific heat (of hydrogen in a metal) at constant
pressure. It has only been measured for a limited number of metal hydrides. A
 Heat of Formation Models 229

direct evaluation of (6.41) is thus not possible in general. The specific heat (vn
can, however, be estimated by means of a simple Einstein model for the optical
phonons in a metal hydride. Furthermore one expects, on the basis of existing
experimental data [-6.44-47], that V"n depends only weakly on pressure and
temperature. With these approximations we find that the standard enthalpy
AH ° (at a suitable reference state with pressure P0 and temperature To) is related
to the enthalpy AH (at p and T) by means of [-see (6.33)]

A H ° = A H - Vn(p-- P o ) - 3R0~ [nBE(T) -- nBE(T0)] + ½(H.2 - H~2), (6.42)

where t/BE is the Bose-Einstein distribution function

1
naE- eO~/r - 1

The Einstein temperature 0E = h~or/kB appropriate for the sites occupied by

hydrogen (for octahedral sites 0E is typically 600 K while for tetrahedral sites
0E~ 1100 K) is assumed to be volume independent in the derivation of (6.42).
Within the approximations used to simplify (6.24) to (6.25), the relation
(6.42) is valid for both A/7 and AH,~p. A useful form of(6.42) can be obtained by
combining it with (6.38). We then have

AH ~o o = ½/~.~- VH{p--Po)-- 3ROE[-nBE(T)- nBE(T0)]- ±H°

2 H2"
(6.43)

At Po = 1 atm and To = 298 K , / t [][2

° -- - 8.45 kJ/mol H z. For a quick estimate of the
standard heat of formation AH°~p from limited high-pressure data we indicate
in Fig. 6.2 how lnp varies with lIT at constant AH°~tj for a representative
hydride (with V. = 1.7 cm3/mol H, 0E = 850 K) according to (6.43). In the low-
pressure regime (p_<l kbar) (6.43) reduces to the well-known van 'tHoff
relation. At higher pressures significant deviations are observed.
With expressions of the type of (6.43) Driessen et al. [-6.54] determined
AH°~¢ for metal hydrides synthesized under high pressures. Typical values are
d H°~t~ --- + 5.4 kJ/mol H for Mo, + 10 for Rh, + (I 5 ___2) for Co, - (3 _ 2) for Ni,
- - ( 8 + I ) (desorption) and +2.3 (absorption) for Cr, +(10__+1.5) for Fe and
- ( 8 __+1) for Mn.
A compilation ofenthalpies of solution at infinite dilution A/I~ and heats of
formation A H ~ o is given in Sect. 6.6.1 for binary metal hydrides and in
Sect. 6.6.2 for selected ternary metal hydrides. Note that in general the literature
values are not available for standard state conditions. In order to demonstrate
the trends in enthalpies from element to element A/7~ and A H ~ p are indicated
in Figs. 6.3 and 4 as a function of position of the host metal in the periodic table.
Roughly speaking the metals on the left of the group VI metals react
exothermically with hydrogen. The other metals (with the exeption of Pd and
Ni) react endothermically or weakly exothermically with hydrogen.
 230 R. Griessen and T. Riesterer

80
]r
Ag AI
Pb
Cr Ru Cu
Mo Pt
40
Fe Rh AU
CO
Mg
Ni Zn
U
I Na
5 o -K Be O[-Mn
Pd
E
V
Ta
, , -40 Th Hf Nb
Li Ti
pr B

CeTb
-80 Y DyLu
Sc

- 120

IA TI A I l l A ]3Z A ~ A "~;71A"~[I] A ~11" A I B ]II~ IllB T~TE

Fig. 6.3. Heat of solution at infinite dilution for binary metal hydrides. For references see
Table 6.6.1. Note that in general the literature values are not available for standard state
conditions; the values indicated here are thus not necessarily standard heats of solution

80

40
I--'---1
I
Co
"6 Fe Rh
Be MO
E 0
Cr' ~,'~n Ni
Pd
e~ Mg zv
,~ -40 ,~'a U Nb
d 2s K Rb
T Np
Ro TS~ HfTi
-80 - g~
Li aa LaVb
Co AmPr
Gd
Sr ScCcLu Zr
TbDyNd
HOErrTm
120

Fig. 6.4. Heat of formation of concentrated metal hydrides. For references see Table 6.6.1.
Same comment as for Fig. 6.3

F o r all binary hydrides investigated at infinite dilution and at higher

c o n c e n t r a t i o n s the enthalpy of solution decreases with increasing h y d r o g e n
content. As w e shall see this general b e h a v i o u r is in m a n y cases related to the
lattice e x p a n s i o n a c c o m p a n y i n g h y d r o g e n a b s o r p t i o n in a metal.
 Heat of Formation Models 231

6.3 Empirical Correlations

In order to rationalize trends in hydride stability and site occupation, several
empirical correlations have been proposed between the enthalpy of hydride
formation and physical parameters of the host metal. Among these parameters
the unit cell volume or the radius of interstitial holes played an important role.
Lundin et al. [6.15, 16] showed for example that the free energy of formation of
various AB 5 hydrides decreases linearly with increasing tetrahedral hole radius
when the B atoms are partially substituted by larger atoms. Such relations are
most useful for the tailoring of materials with specific characteristics but they do
not shed light on the actual interaction of hydrogen with metals. It is however
intuitively clear that larger interstitial holes will tend to stabilize a hydride (see
"Volume effects" in Sect. 6.4.3).
Since the interstitial hole size varies in most cases in proportion to the unit
cell volume, the stability of hydrides has often been related to the cell (molar)
volume of the host metal. A well-documented example is that of the rare-earth-
Ni 5 and rare-earth-Co5 compounds. As shown in Fig. 6.5, the logarithm of the
plateau pressure at constant temperature varies linearly with the volume of the
unit cell (except for CeCos). The simple correlation suggested by the data in
Fig. 6.5 is however not confirmed by the data in Fig. 6.6. As follows from the
high-pressure measurements of Takeshita et al. [6.55], YNi 5 and LaPt 5 have
similar heats of hydride formation although their unit cell volumes are very
different. This shows that other effects such as band filling also influence the
stability of metal hydrides.
Yei and McLellan [6.22] suggest that there is a correlation between AH~
and 7IV where ~ is the coefficient of the linear electronic specific heat and V the
molar volume of the metal. As shown in Fig. 6.7 such a correlation (which was
proposed by comparing the heats of hydrogen solution of four transition metals
and two noble metals only) is not confirmed by the data presently available for
binary hydrides.
Various authors have also tried to find a correlation between the heat of
solution of binary hydrides and some of the cohesive properties of the host
metal. On the basis of continuum elasticity theory Welch and Pick [6.20]
derived the following expression for A/7~

AaQ~o= ½D-- Es + ED, (6.44)

where D is the dissociation energy of the H 2 molecule (218 kJ/mol H), En is the
energy gained by forming electronic bonding states and E D is the distortion
energy due to the size mismatch between hydrogen and the interstitial hole. The
later is approximately given by

(G/V) (6.45)
ED= a + b(G/V) '
232 R. Griessen and T. Riesterer
6

YbNi 5 GdNi 5
4
SnnNi 5
NdNi 5
2 PrNi 5
SrnCo 5
r, LaNi5
0 NdCo5
C~C°5 PrCo 5

LoCo2 Ni3
-2 LaCo3Ni2
L°C°4 Ni LoNi
LoCo54 AI
-4

I I I I I
80 82 84 86 88 90 92

Fig. 6.5. Variation of plateau pressures at constant temperature with unit cell volume for AB5
intermetallic compounds [6.14, 115]

-lO[YNi5ThNi5
GdNi LoPt,
J SmNisYCo5 ThCo
] - I
TbCo~NdN'5 5

-r -15p c~co~GdCo~
PrNi 5 koNi5 LaNi4Pd
3 SmCo5
E
--} CoNI5
PrCo5

t
_r¢~ - 2 0 CeCo~
<3
NdCo5
LoCu5
LoCo5

-25
I I I I I I I ~/, I
82 84 86 88 90 92 94 96 108 110

Fig. 6.6. Heat of formation of ternary ABsH~ hydrides versus unit cell volume. For references
see Table 6.6.2

where G is the shear modulus of the host metal and V the molar volume. By
fitting the parameters a, b, and E BWelch and Pick were able to reproduce nicely
the heats of solution for binary transition metal hydrides. The fact that a
correlation was also found I-6.18, 19] between A/7o~ and the volume-reduced
 Heat of Formation Models 233
120
Fig. 6.7. -Heat of solution at
infinite dilution for binary
(9 metal hydrides as a function
of the volume-reduced
80
electronic specific heat coef-
AgAI ficient. The transition
metals, actinides and rare
earths are indicated within a
4O circle
Au ~) ®
T @
® ®
E
o Bg
I,---.I @

-40

®
r
g
-80 ®

-120 I I I I I
0 0.1 0.2 0.3 0.4 0.5
n ( E F) I V rL. s t a t e s leV/cm 3 ].J

bulk modulus B/V suggests, however, that the volume-reduced shear modulus
is not really a fundamental characteristic of the ability of a metal to absorb
hydrogen. This conclusion is confirmed by the observation that relations
(6.44, 45) fail badly in predicting the heats of solution for noble and simple
metals. As shown in Fig. 6.8 most of these metals have a low shear modulus and
a relatively large molar volume. This results in small values for the volume-
reduced shear modulus G/V. None of these metals however has a particularly
strong affinity for hydrogen. Another prediction of the relations (6.44, 45) which
is not confirmed by experiment concerns liquid metals. For a liquid G = 0 and,
consequently, one expects that A/7o~= (1/2)D- E R= --82 kJ/mol H. However,
as can be seen in Table 6.6.1, the heat of solution of hydrogen in liquid metals is
to within ~ 15 kJ/mol H the same as that of the corresponding metal in the solid
state. For liquid Co A/7~o=32, for Fe AB~o=33, for Mn A/7~=30, for Ni
A/7~o=24, for Nb A / 7 , = - 3 1 , and for Ti A/Too= - 4 7 kJ/mol H. This casts
some doubt on the soundness of the physical basis for the model of Welch and
Pick.
Villars I"6.56] has shown that most of the tabulated properties of the
elements show systematic trends across the periodic table. Since AH also varies
systematically across the periodic table this opens up a wide variety of potential
empirical correlations between AH and a simple function of tabulated
properties.
234 R. Griessen and T. Riesterer

120
Fig. 6.8. Heat of solution at
infinite dilution for binary metal
® hydrides as a function of the
volume reduced shear modulus.
80 The transition metals, actinides
® and rare earths are indicated
Ag
AI
within a circle
Cu Q @®
40 _ ®
Au
T _MQ
"6
E
®
2 o Be

8
IT
<1

-40

-80

-120 t I I
0 01 0.2 Q3
G/V [ 10 `5 Nr'n- 5 ]

Finally, it is worth noting that correlations between the heat of hydride

formation and some physical parameter of an alloy are only useful if this
parameter can be predicted from a model or from theoretical calculations. For
example, if a correlation between A H and, say the compressibility of the host
compound really did exist, then A H could easily be predicted for new (not yet
investigated) alloys as it is well known that for transition metal alloys the
compressibility is rather accurately given by the arithmetic mean of the
compressibilities of their constituent metals. If this is not the case, empirical
correlations will only be of marginal use in the search for new metal hydrides.

6.4 Semi-empirical Models

As mentioned in the introductory section, two semi-empirical models for the
heat of formation of binary and ternary hydrides have been proposed so far: (i)
the"atomic" cell model of Miedema et al. [6.24-29] and (ii) the "band structure"
model of Griessen and Driessen I-6.35-38]. These two models will be dealt with
at the end of this section. First, however, we shall consider semi-empirical
models from a more general point of view and derive expressions which must be
satisfied by any semi-empirical model for the heat of formation of alloys.
 Heat of Formation Models 235

6.4.1 General Properties of Semi-empirical Models

For historical reasons we shall consider in this section primarily the integral
enthalpy of formation A/~ which represents the total heat involved in the
preparation of one mol formula unit of a given compound (or alloy). The heat
of formation AH given per mol H is related to A n by means of

An(MHx)
AH(MHx) =- (6.46)
X

The heat of solution AH for the absorption of hydrogen in MHx is connected to

A n by the relation

AH(MHx) --- OAH(MHx)

Ox (6.47)

If the integral heat of formation varies linearly with hydrogen concentration

then AH = An. For a metal-hydrogen system such as PdHx, for which a highly
nonlinear behaviour of A n is observed at concentrations x > 0.3, AH and AH
are markedly different from each other [6.53, 57, 58].

a) Binary Alloys
As an introduction to the problem of the heat of formation of hydrides we
consider first the simple case of intermetallic binary alloys A 1-yBy. The purpose
of a semi-empirical model is to express the integral heat of formation of an alloy
A~ _yBy in terms of characteristic parameters A; and Bi (with i= 1..... n) of the
constituent metals. Mathematically this means that

AH(A, _yBy)= f (A, B, y), (6.48)

e.g. that the integral heat of formation A n is a function of the vectors

A =(A x..... A,) and B=(B 1..... B,) which characterize the metals A and B, and
of the fraction y of B atoms in A. The function f has some general properties
which derive directly from the facts

An(A~ _ yBy)= AH(ByA, _y), (6.49)

An(Al_yAy)=O for all y, (6.50)
An(A~Bo)=O, (6.51)
AH(AoB1)=O. (6.52)

From (6.49) it follows that

f (A, B, y) = f (B, A, 1 - y ) ; (6.53)

236 R. Griessen a n d T. Riesterer

from (6.50) that

f(A, A, y)= 0; (6.54)

from (6.51) that

f(A, B, 0) =0; (6.55)

and from (6.52) that

f(A, B, 1)=0. (6.56)

The general conditions (6.53-56) can be satisfied by writing the function f in the
form

f(A,B,y)=y(1--y) ~ ~ (A,--Bi)2Jaij
i=1 j = l

+y(1--y) ~ ~ (Ai--Bi)2k+l(Ai--Bl)2rn+lailkm, (6.57)

i=l k=O
/=1 m=O

where the coefficients aij and auk,, are such that simultaneous permutation of A
with B, and y with 1 -- y leave them unchanged (i. e. these coefficients must obey
the symmetry condition given in (6.53). Simple examples of such functions are

a = h(y(l - y)) (6.58)

a = h((1 - y) Ai + yBi) (6.59)
a = hl(h2(Ai)h3(Bj)+ ha(Ai)h2(B~)), (6.60)

where h, ha, h2, and h3 are arbitrary functions.

In (6.57) we have not included more complicated terms such as

(A i- Bi)XJ(At-Bl) 2j' or

(A i - Bi)Zk+ l ( A t _ Bt) z,,, + l( A r _ Bi,)zg' + 1 ( A r _ n r ) 2 m ' + 1

since in this work we shall mainly be concerned with scalar [A = (A)] and two-
dimensional [A = (A 1, AE)] semi-empirical models.
In a scalar model each metal is characterized by one parameter only.
Equation (6.57) reduces then simply to

f(A,B,y)=y(1-y) ~ (A-B)EJaj(A,B,y). (6.61)

j=l

For transition metals, for example, it is reasonable to choose the number Z of

d-electrons as characteristic parameter. Equation (6.61) can then be viewed as
 Heat of Formation Models 237

an expansion in AZ = Z A- Z B. It is a generalization of an expression proposed

by Williams etal. [6.30] (see also [6.6-1). Within a tight-binding model for the
density of states of transition metal alloys, Pettifor [6.31] has indeed derived an
expression of the form (6.61) for AH of AB compounds where both A and B
belong to the same transition metal series. He obtained

AFI(AB)= a(Z) (AZ) 2 with (6.62)

Z=½(ZA+Zs) and (6.63)

a(Z,= I E I _ 3 Z ( 1 0 _ Z ) ] - W ( 5 _ -Z ) dlnV
~-, (6.64)

where the last term represents the contribution to A/q of volume changes during
alloying (W is the width of the d-band in eV). The numerical parameters give the
heat of formation in units of eV per atom. Pettifor found that expression (6.62)
was in reasonable agreement with experimental values. This indicates that in
the general expression (6.61) the coefficients as decrease rapidly with increasingj
so that the lowest order terms are dominant.
In a two-dimensional vectorial model each metal is characterized by two
parameters. To lowest order in the differences of these parameters we obtain
from (6.57) the following expression

f(A, B, y)=y(l - y ) [(A, - B,) 2a,(A, B, y)

+ (A 2 - B2)2a2(A, B, y)
+ (A, -- B 1)(A2 -- Bz)a,2(A, B, y)]. (6.65)

As we shall see later, Miedema's model resembles expression (6.65) in the case of
alloys consisting of two metals of the same type, for example, two transition
metals. Even for this special category of alloys there are however essential
differences: a term of the form (A 1-- B0 (A 2 -- B2) does not appear in the semi-
empirical model of Miedema. Furthermore the coefficients al and a2 depend
here only on A1, A2, B1, B2, and y while in Miedema's model they depend on
additional parameters such as the molar volume, the valence and the type
(transition or non-transition) of the metals A and B.

b) Binary Metal Hydrides

One possible approach to modelling the binary metal hydrides is based on the
assumption that AHx can be treated as a normal alloy after gaseous hydrogen
has been brought to a suitable condensed state. We then have

A/4(AHx) = AH(AH*) + xAHtr..s, (6.66)

where AH t. . . . is the heat of transformation of gaseous H 2 into condensed

(metallic) hydrogen H*, and A/~ is the integral heat of formation of the alloy
238 R. Griessen and T. Riesterer

AHx made of A metal and condensed (metallic) hydrogen. To make contact

with the treatment of binary intermetallic alloys given above we express
AH(AH*) in terms of AFI(A~ _yH*), i.e.

~ , l
A H ( A 1 - yHy) = l ~ x A/t(AH*) (6.67)

with y = x / ( 1 +x). We can then use the results obtained for binary alloys and
write

AFI(A j _~H*)= f (A, H, y), (6.68)

where the function f is now the same as in (6.48).

Within the framework of a scalar model one obtains from (6.61) and (6.68) to
lowest order the result

x
A/4(Anx) = ~ b(A - H) 2 + x A H , .... (6.69)

and for a two-dimensional model from (6.65)

A/t(AHx) = 1 T x [ a , ( A , - H , ) 2 + az(A2 - H2) 2

+ a,2(Aa --U,)(A 2 --H2) ] + x A n t.... . (6.70)

This relation (with a lz = 0) has a form similar to that of Bouten and M i e d e m a

[6.28] for the integral heat of formation of binary transition metal hydrides.
Another way to set up a semi-empirical model for binary metal hydrides is
simply to write

A/4(AHx) = f (A, x). (6.71)

The function f is not subject to any symmetry condition. We know only that

f(A,0)=0 (6.72)

which implies that

f ( A , x) = x ~ ( A , x). (6.73)

For a scalar model we then have

A f f l ( A H x ) = x ~ ai(x)A'. (6.74)
i=0
 Heat of Formation Models 239

In contrast to the case of binary intermetallic alloys all powers of A may appear.
In the simplest model one writes

A/~(AH:,) = x(c~ + c 2 A ) . (6.75)

For the partial molar heat of solution A/~ this implies that

A/t = aA + fl, (6.76)

where e and fl are constants. Griessen and Driessen [6.35 38] found that such a
simple form did indeed give a very satisfactory description of the heats of
formation of binary hydrides when A represents the difference between the
energy of the Fermi level and the energy of the lowest conduction band of the
host metal. Remarkably enough the parameters ~ and fl were even found to be
the same for transition metals, simple metals, rare earths and actinides.

c) Ternary Metal Hydrides

In analogy with the treatment of binary alloys one could indicate general
properties for the heat of formation of ternary alloys AyABr.Cy c. In the context
of this chapter we shall however restrict ourselves to the important class of
ternary metal hydrides A1 _yBrHx. In the spirit of a semi-empirical model, the
integral heat of hydride formation can be written as

AFI(A, _ yBrHx)= f(A, B, y, x). (6.77)

In contrast to the function f used for binary alloys f(A, A, y, x)4= 0 since

T(A, A, y, x) = AH(AHx) (6.78)

and similarly

f(B, B, y, x)= A~(BHx). (6.79)

Furthermore the function f(A, B, y, x) must also satisfy the following symmetry
relation

f(A,B,y,x)=f(B,A, 1-y,x) (6.80)

for all values of y and x. Since the function f(A, B, y, x) does not have the same
properties as the function f(A, B, y) used for binary intermetallic alloys, the
results given in (6.57, 61,65) cannot be used. It is, however, possible to define a
new function

g(A, B, y, x)= f(A, B, y, x)-(1 - y) f(A, A, y, x ) - y f (B, B, l - y, x)

- AH(A 1_rByHx)-(1 -y)AH(AHx)-yAffI(BHx) (6.81)
240 R. Griessen and T. Riesterer

so that

g(A, A, y, x) = 0 (6.82)
g(A, B, 0, x) = 0 (6.83)
g(A, B, 1, x) = 0 (6.84)

and also

g(A,B, y,x)=g(B,A, 1 -- y,x). (6.85)

This means that the function g(A, B, y, x) can be expressed in the same way as
the function f ( A , B, y) in (6.57) with the only difference that the coefficients aq,
aukm, depend now also on the hydrogen concentration x. As the integral heat of
hydride formation A/~(A1 -yByHx) vanishes for x = 0, all these coefficients must
be proportional to x and we can write aj = x~ i.
In a scalar model

g(d, B, y, x) = xy(1 -- y) ~ (A -- B)2J~JA, B, y, x) (6.86)

j~l

and in the special case where both in (6.61) and (6.86) the dominant term is that
with j = 1 we can express the function g in terms of the heat of formation of the
host alloy, i.e.

(Aj Y, X) (6.87)
g(A'B'y'x)=xA~I(AI-yBY) ,( ,B,y,x)"

Combining (6.77-79, 81, 87) we obtain finally

AH(A t _yByHx) ~ (1 -y)AFI(AHx)+ yAlYt(BHx)+kxA~l(A1 _yBy) (6.88)

with k = ~1/al. As we shall see later, this relation resembles the so-called rule of
reversed stability first proposed by van Mal et al. [6.26] for the description of the
hydrogen absorption in various LaNi4M (with M = Pd, Ag, Cu, Co, Fe, or Cr)
compounds.
In a two-dimensional model one obtains to lowest order in the differences of
characteristic parameters

g(A, B, y, x) = xy(1 - y ) [0q(A 1 -- B,) 2 + ~2(A2 - B2) 2

+ aa 2(A~ - B~)(A 2 - B2)] . (6.89)

Although this resembles the functional dependence for A/4(A 1 _ yBy) [see (6.65)]
it is not expected that (6.88) holds for this two-dimensional model (except of
course if ~i/ai=k for i = 1 , 2, 12, and x and y constant in (6.65)).
 Heat of Formation Models 241

Quite often one of the components of an alloy A ~_ rBr, say component B, is

substituted by another metal C. We obtain then quarternary metal hydrides of
the form A t _rBy~ _,)Cy~H~. For a given value o f y and for a given metal A one
can write

FI(A . _ , B , , _ ~ C , . H x ) = h(B, c , z, x). (6.90)

Proceeding as for the ternary metal hydrides one defines a function

g(B,C,z,x)=h(B,C,z,x)-(1-z)h(B,B,z,x)-zh(C,C,l-z,x) (6.91)

with the properties (6.82-85) (with the variables A, B, y, x replaced by B, C, z, x).

For a scalar model,

g(B, C, z, x) = K(y)xAffI(B~ _.C~) . (6.92)

The function K(y) vanishes identically for y = 0 . For y = l , the relations

(6.91, 92) lead to

AIYI(B~ _ zCzH~) = (1 - z)Afft(BHx) + zAffl(CHx)+ K ( y = 1)xAH(B 1_ zCz).

(6.93)
Now assuming for simplicity that K ( y ) = y K ( y = 1) we obtain

A/~(A, _ yBy~1_ z)C,,zH:,)= (1 - z) A/I(A~ _ yByHx)

+ zA/~(A 1_,.Cyril) + yA ffl(B, _ ~C~H~)
-- y(1 - z) A/7(BHx) -- yzA.O(CH~). (6.94)

This expression shows that it is certainly not correct to estimate the heat of
formation of a quarternary metal hydride A~ _yBy~_,)Cy~Hx from a weighted
average of the AH of the ternary hydrides A~ _rBrHx and AI-rCyH~.

d) Local or Site-Dependent Enthalpies

The basic weakness of the semi-empirical models considered above is that they
do not include structural information. There is no distinction between the
various sites which can be occupied by hydrogen. Even for the simplest class of
metal hydrides, the binary hydrides (e.g. PdHx, NbHx), the semi-empirical
models do not make a distinction between octahedral and tetrahedral sites. In
NbHx, for example, the difference in energy between these sites is not
considerable [6.45, 59-62] but for PdHx where only octahedral sites are
occupied the energy of the tetrahedral sites might well be 50 to 100 k J/tool H
higher than that of the octahedral sites. This clearly indicates that site-
dependent properties (e. g. local electron density, geometry, size) also play a role.
242 R. Griessen and T. Riesterer

In a pure metal larger interstitial sites will tend to be occupied first by hydrogen.
In compounds and alloys the local chemical composition around each site (i.e.
in a first approximation the ratio of A to B atoms in the immediate
neighbourhood of the site) is a further factor which determines the heat of
hydride formation. To circumvent this weakness one can attribute to each site a
local heat of hydride formation. From a microscopic point of view this is a
reasonable assumption as it is well-known that the electronic structure of
transition metals is essentially determined by the overlap of nearest-neighbour
wave functions. A general formulation of such local semi-empirical models has
not been given yet. Attempts have, however, been made to rationalize the
observed site occupancies in the hydrides of intermetallic compounds.
In a series of articles Westlake [6.39, 49] proposed a purely geometric model
which is based on a minimum interstitial hole radius of 0.40 A for occupation by
hydrogen, a minimum hydrogen-hydrogen distance of 2.10 A and the assump-
tion that larger interstitial sites are occupied first. This model predicts correctly
the site occupancy by hydrogen in Friauf-Laves phases with the cubic C15
structures such as ZrCr 2, ZrV2, ZrFe2, TiCrz, HfV2, ZrTiz, and TaV2, Friauf-
Laves phases with the hexagonal CI 4 structure as ErMn 2 and ZrMn2, Haucke
compounds such as LaNi 5, LaNigA1, and LaNi4Mn [6.63], A6B23 compounds
[-6.64, 65], ZrNi [6.66, 67] and the hydride V2H(D) [6.68, 69]. Recently it has
also been applied to FeTi [6.70] and Be-based compounds [6.71]. Especially
interesting are the hydrides of TizNi [6.72] in which it seems that H also
occupies large tetrahedral sites coordinated by four Ni atoms. This is
considered to be clear evidence that hydrogen site preference depends more
heavily on interstitial hole size than on the affinity of the coordinating atoms for
hydrogen. In all other cases however, the geometric model predicts the same site
occupancy as a model based on the hydrogen affinity of the metallic nearest
neighbours since in these systems the larger interstitial sites are usually those
coordinated by the higher number of atoms of the metal with the greater affinity
for hydrogen.
Another model which was proposed to explain the site occupancy by
hydrogen is the so-called imaginary binary hydrides model. This model,
originally proposed by Jacob and Shaltiel [6.73, 74], has been refined in
succesive steps. A good description may be found in [-6.75] and [6.76]. The
basic idea of the model is that the affinity for hydrogen of a given site in an
intermetallic compound AraB, may be characterized by an effective local
enthalpy of formation defined as

AH'tb(A,.B.H~)=_ AH (A,.H.~_~b ) + AH ( B . H ) . (6.95)

where x is the number of hydrogen atoms per formula unit AraBn occupying an
interstitial site coordinated by a A atoms and b B atoms. By means of a semi-
empirical model it is possible to calculate the two heats of formation of the
 Heat of Formation Models 243

binary hydrides on the right-hand side of(6.95). At x = 0 the effective local heats
of formation of the various sites differ markedly from each other (in ZrV2 this
difference is for example typically 40 kJ/mol H). It is then normally argued that
around room temperature only the site with the lowest AH' will be occupied
because of the strong energy dependence of the Boltzmann factor I-6.75, 77].
Within the model of Miedema one finds that AH',/b is a monotonically
increasing function ofx. For some sites (for example 2 Zr + 2 V tetrahedral sites
in ZrV2) this increase is however much faster than for other sites (e.g. 1 Zr + 3 V
site). It is therefore possible that different sites with different hydrogen
occupancy are characterized by the same AH' at a certain concentration x. The
simultaneous occupation of these sites is then expected to occur. This is in
qualitative agreement with experimental structural data on the hydrides (or
deuterides) of intermetallic compounds [6.76]. Recently Riesterer [6.78]
argued that the affinity of a site for hydrogen should be characterized by the
heat of solution A/7',Ib instead of the heat of formation [see (6.46, 47)]. For
ZrVzD1.5, ZrV2D2.6, ZrV2D4.5, and ZrCr3D 3 his results agree well with
experimental data. However for ZrNiD3, ZrTil.9D3.8, and FeTiD1. 9 his
approach leads to the same wrong predictions as those derived from (6.95).
In an attempt to predict quantitatively the occupancy of various sites in
intermetallics, Jacob et al. [6.76] assumed that the number xo/b of hydrogen
atoms in an a/b-site is given by a Boltzmann distribution so that

xa/b exp(--AH'~/JkT)
(6.96)
X Y' e x p ( - AH'a/b/kT )'
sites

where x is the total number of hydrogens per formula unit. They obtained site
occupancies in remarkably good agreement with structural data for FeTiH,
TiCr2Ho. 15, ZrVzH1.5 -4.9, ZrCr2 Ho.1 s -4, ZrMn2H3, and LaNisHo. 15 -6- This
agreement is in fact rather surprising since it follows from Lacher's statistical
treatment [6.79] of hydrogen absorption (see also the works of Kirchheim
I-6.80, 81] and Griessen [6.82] on this point) that the Fermi-Dirac distribution
must be used instead of Boltzmann's distribution function. Furthermore (6.96)
does not take into account that the number of sites per formula unit is different
for each type of site. As pointed out by Jacob et al. it is however possible that the
errors introduced by their simplifying assumptions are to a certain extent self-
compensating.
The problem of the local enthalpies has been tackled recently by Griessen
and Driessen [6.37]. In their work on the pressure-composition isotherms of
disordered alloys they showed that the semi-empirical band-structure model
could be used to estimate the heat of formation for hydrogen absorption at a
given site. This model takes into account both the hydrogen affinity of the
coordinating atoms and their interatomic separation in the alloy. It might be
interesting to apply this model (or an improved version of it) to the problem of
site occupancy in intermetallics.
244 R. Griessen a n d T. Riesterer

6.4.2 The Atomic-Cell Model

In a series of articles, Miedema and coworkers [-6.24, 27] have developed a semi-
empirical model for the intgral enthalpy of formation of intermetallic com-
pounds. We give here a brief summary of this model and describe its
applications to binary and ternary metal hydrides [6.25, 26, 28, 29].

a) Intermetallic Compounds and Alloys

The basis of Miedema's model is that the Wigner-Seitz concept of atomic cells is
still meaningful in alloys or compounds. This means that in an ordered
compound, say AB, the atomic cells of A and B are assumed to resemble
strongly the atomic cells of the pure metals A and B. There are, however,
modifications in the boundary conditions when cells are brought into contact
with each other. According to Miedema et al. the changes in boundary
conditions lead to two energy effects. The first one is related to the mismatch of
the electronic density nw~ of metal A and B at the boundary of their respective
Wigner-Seitz cells. The second effect is related to the difference in the chemical
potential ~b* of electrons in metals A and B. The integral enthalpy of formation
of a compound AyaBr, (with yA+yB= 1) is then

AH(AyaBY")=fgP[-(AdP*)z[ + FQ (Anw/~)
132 - ~
R-]
|J (6.97)

with

f = y~AY~B[1 + 8 (y]y~)Z] (6.98)

A~*=~*-~* (6.99)
Anl/3=(n A
• -ws v-ws!
"~1/3 _(rib
v-ws/
"~1/3 (6.100)
YA(VII) 2 / 3 + YB(v/~l)2 / 3
(6.101)
g= (nwl/3)
(n~sl/3) _- - irt,.,A ~- 1 / 3 +(rib
2 LI,'~ws! / x"ws/
)- 1/37
.J (6.102)
yi(Vfl)2/3 (6.103)
Y~= yA(Vj~) 2/3 + yB(V~') 2/3

(v,ab~/3= V?/3 II + a f (~*-- ~*)~, (6.104)

d

where i and j represent A or B. The parameters y~ represent effective surface

concentrations. If both atoms A and B have the same molar volume (VA= VB)
then y~=yg. The function f is an empirical function derived from a fit to
experimental data on the heat of formation of ordered binary compounds. Since
 Heat of Formation Models 245

V/aldepends both on y~ and the function f, the equations (6.98, 103, 104) must be
solved iteratively, except, of course, if volume effects can be neglected i.e. when
the parameter a is negligible.
The numerical values for t/~* and nws are given in [6.84]. The parameters P,
Q, R, and a depend on the type of metals involved in an alloy. For the alkali
metals a=0.14, for the divalent metals a=0.10, for the noble metals and
trivalent metals a=0.07 and a=0.04 for the other elements. The so-called
hybridization term R / P depends in a complicated way on the position of the
metals A and B in the periodic table. For alloys of two transition metals,
however RIP = 0. For tables of values of these parameters the reader is referred
to 1-6.83, 84].
Except for the cases where AffI(A~,aBy,) is close to zero, the model predicts
values which are only 10-20% different from the experimental values for the
heat of formation of ordered compounds of two transition metals. The error is
somewhat larger for alloys of a transition metal with a non-transition metal.
Despite this remarkable accuracy, Miedema's model has been criticized by
14411iams, Gelatt, and Moruzzi [-6.30, 32] and by Pettifor [6.31, 85]. These
authors argue basically that the success of an empirical model is not necessarily
a confirmation of the validity of the underlying physical concepts which have
been used to set up the model.
In the context of our discussion of the general properties of semi-empirical
models, it should be noted that Miedema's model is in fact a six-dimensional
model as each metal is characterized by qS*, nw~, V, a, R', P'. This high
dimensionality is not immediately apparent because the model is not for-
mulated in a consistent vectorial form. In fact it could be written as follows

f ( A , B, y) = y(1 - y ) F ( y , a i, ~, nws,
' cfii*, P i)
'
x {(q~]- q~n)*2 + k , [(nws
) a ,/3 _(nws
) . ,/a]2 +k2(R,A_R,n)2}. (6.105)

In this form the hybridization t e r m k 2 (R] - R ~ ) 2 (in Miedema's notation R/P) is

explicitly shown to vanish for two metals of the same type, i.e. for two metals
with R~ = R~. It is also apparent that no elastic energy term due to a size
mismatch, i.e. a term proportional to (Va - Vn)2 is included in Miedema's model.

b) Binary Metal Hydrides

The atomic-cell model has been extended to the case of binary transition metal
hydrides by Bouten and Miedema I-6.28] by using the procedure described in
Sect. 6.4.1. They treat a binary metal hydride as a "normal" alloy of a metal with
metallic hydrogen and include in the expression for A/~ the enthalpy AH, .... for
the transformation at room temperature of gaseous hydrogen into "metallic"
hydrogen. The parameters used to characterize metallic hydrogen are
q~*= 5.2 __,e.V.-w~nl/3-- 1.5 d.u., Vn = 1.7 cm3/mol, R / P = 3.9, P = 12.3, Q / P = 9.4, and
AHt .... = 100 kJ/mol H. With these parameters Bouten and Miedema obtained
246 R. Griessen and T. Riesterer
I I I I I J I
Fig. 6.9. Comparison ofexperi-
mental heats of formation for
TiCr'o2 YNi5
fiFe fC°5 • ThCo 5
ternary metal hydrides with
ZC,~AO Z r C ~ TiMnl,5 °LONi5 values calculated by means of
-20 ,-, 2t, • 2 ScFe 2
LoCo 5 LORh 2 M iedema' s semi-empirical
Z £Mn2111 t i yelp2 ii0 • 21~
ThNi • eHfNi
LoNi 2
• eYC°3
YNi 3 model [6.29]. References for
e T i C o eTiNJ experimental data are given in
'5 ZrNi • ° H I C ° Table 6.6.2
E
-40

I'
I

-60

- 80 • ZrV 2

I l I I I I I
-80 -60 -40 -20 0

/ ' , H ~4 E k J / m o l H ' ]

a relatively good agreement between calculated and experimental values for

concentrated hydrides (see Fig. 6.9).
In the limit of low hydrogen concentrations the heat of solution is given by

> l+vy--
.....
/ r rj "

where x =Yn/YM, A~*= qS*-qS*, and A Htr.n ~= 100 kJ/mol H. The heats of
solution at infinite dilution A/7~ calculated by means of this equation (with
x = 0) are compared to experimental values in Fig. 6.10. One sees that in general
the trend within the periodic table is correctly reproduced although the
calculated values may be out by as much as a factor of two.
A difficulty in the application of Miedema's model to binary hydrides is that
the model predicts that dAF1/dx>O since
I 1/2/3 ]
dAI~=2(AH, .... - A H ~ ) 1 + V,n~ (1 + 2aA~b*) (6.107)
dx

and rAH~l<AHt,ao~ for practically all elements. This conclusion is not

confirmed by experiment since for most metal hydrides A/~ decreases with
increasing hydrogen concentration (see Table 6.6.1). This is probably due to the
absence of an elastic energy term in (6.97) and (6.105).
 Heat of Formation Models 247

Fig. 6.10. Comparison of experimental

100 heats of solution at infinite dilution for
binary metal hydrides with the values
%' ir Ag
predicted by the cellular semi-empirical
.50
Cr Ru C u model of Miedema and coworkers
M9 Pt
Fe Co [-6.28, 29]. References for the experi-
Rh
Ni
mental data are given in Table 6.6.1
U
M~
~ o
Pd

H[ Th NbTa V
8-5o
I'1- Ti
Zr LQ
Y
5c
-100

-150 I I J
-100 -50 0 50 100

e) Ternary Metal Hydrides

On the basis of their work on hydrogen absorption in LaNi5 and related
compounds, van Mal et al. [6.26, 86] suggested that the integral heat of hydride
formation can be calculated from

AH(ABnHz,n) = A/4(AH,n) + AH(BnHm)-- A_H(AB,) (6.108)

when n is much larger than unity. The transition metal A is assumed to have a
stronger affinity for hydrogen than the transition metal B. As a consequence
hydrogen is attracted by the A atoms in the A B , compound and isolates the A
atoms from the surrounding B atoms. The breaking of A - B bonds is accounted
for by the last term in (6.108). In compounds A B , with n < 5 , however, not all
A - B bonds are broken up. In successive steps Miedema et al. [6.28] generalized
(6.108) to the case of AB, compounds with n as small as one and proposed that

A H(ABnH x +y) = A ffI(AHx) + A/~(BnH,) -- (1 -- F) A I4(A Bn). (6.109)

The factor (1 - F) represents the extent to which A - B contacts between atomic

cells are broken by hydrogen. In A B 5 compounds it is empirically found that
90% of the A - B contacts are broken while only 40% are broken in A B
compounds. Values o f F for various ternary hydrides are tabulated in [-6.29] for
selected values of x, y, and n. Unfortunately no analytical dependence o f F on x,
y, and n has been proposed by the authors. It is, for example, quite evident that
for small values o f x + y the factor (1 - F) must tend towards zero as all the other
terms in (6.109) vanish for x = y = 0 .
It is worth noting that the so-called rule of reversed stability given in (6.109)
resembles the relation (6.88) derived from a general scalar model. Using the
248 R. Griessen and T. Riesterer

notation of Miedema [-6.29] (6.88) may be written as

• . x + y

(6.110)

The main difference from Miedema's expression (6.109) is that here hydrogen is
shared in the same proportions between A and B atoms, while in (6.109) y/n is
always smaller than x. This basic difference probably arises from the fact that
some structural information is implicitly contained in the atomic-cell model
while no such ingredients are included in the "general" semi-empirical models
discussed in Sect. 6.4.1.
Recently Shilov et al. [6.87] proposed a relation for the heats of formation of
ternary hydrides based essentially on parameters derived from Miedema's
model. However, because the present authors were not able to reproduce the
numerical results of Shilov et al. using the parameter values from the cited
references, it is presently difficult to assess the validity of their astonishingly
accurate approach and its applicability to systems not considered by these
authors.

d) Quaternary Metal Hydrides

The cellular approach has been used by Pasturel et al. [6.88-90] to calculate the
integral heat of hydride formation ofAB, _sCsH2,.. In order to derive a relation
for quaternary hydrides they consider first the case of ternary hydrides and
write
m ~ ttl
Afft(AB,H2,,) = 3mA/~(AH3)+ 3- AH(B,Ha)_ -3 AI~(AB,) (6.111)

in agreement with their assumptions that for 2m < 6 some A-B bonds coexist
with both A H and B - H bonds and that the A and B atoms form binary
hydrides of the form AH 3 and B,H3.
For the quaternary hydride they find that

AffI(ABn-sC~Hzm)= 3m A/t(AH3)+ m
3 (n--s)
n A/~(B.H3 )

mS ~ lql
+ 3 n AH(CnH3)-- 3 AffI(AB"-sC~)+Afflx" (6.112)

The term A/41 is a mixing term due to the substitution orB atoms by C atoms. It
is given by

Affl 1= AITI(ABo_,Cs)-- nnS AId(AB,)-- s AH(AC,). (6.113)

 Heat of Formation Models 249

Although (6.112, 113) have been shown to nicely reproduce the experimental
results for LaNis-type compounds, they erroneously predict that for m = 0, the
hydriding heat of reaction is equal to A/t~ instead of vanishing, One can thus
conclude that (6.112, 113) are probably only valid for 2 < 2 m < 6 .

6.4.3 The Band-Structure Model

a) Binary Metal Hydrides

In a search of a correlation between the standard heat of formation of binary
metal hydrides and the smallest possible number of band-structure parameters
of the host metal Griessen and Driessen [6.351 found that the experimental
values for AH,_,~ (see Fig. 6.11) derived from plateau dissociation pressures
(see Sect. 6.2) or from calorimetric measurements in concentrated hydrides
could be reasonably well reproduced by the following simple relation

A H ~0 o = (c~AE+/~) with (6.114)

AE-=EF-E, and (6.115)

~z=29.62 kJ/eV tool H (6.116)
/~= - 135.0 kJ/mol H. (6.117)

For the standard heat of solution A/-1° at infinite dilution one finds a similar
relation (see Fig. 6.12)

A/~O = n~ (c,®AE+/~®) with (6.118)

c~ = 29.62 kJ/eV tool H (6.119)

/~o = - 122.0 kJ/mol H. (6.120)

In (6.114, 115, 118) E F represents the Fermi energy and n~ the number of
electrons per atom in the lowest s-like conduction band of the host metal. E~ is
the energy for which the integrated density of states of the host metal is equal to
n~/2 electrons per atom. The relative positions of Ev and E, are indicated for
several representative metals in Fig. 6.13. Except for the alkali metals (for which
ns= 1) we have ns = 2 and E, corresponds approximately to the centre of the
lowest conduction band since even for transition metals this band has a
predominantly s-character with respect to the interstitial sites occupied by the
hydrogen atoms [6.1, 91, 92].
A remarkable feature of the correlations shown in Figs. 6.11 and 6.12 is that
irrespective of the "type" of host metal (noble, simple or transition metals, rare-
earths or actinides) the heats of formation of their hydrides are reasonably well
250 R. Griessen and T. Riesterer

40
Fig. 6.11. Correlation between
experimental heats of formation
20 and the characteristic band struc-
leo
IFe ll~h ture energy Ev--E~. EF is the
MOI Be Fermi energy and E~ is essential-
IICO • IA I
NI! ly the centre of the lowest s-band
Cro eT¢
oMn (see also Fig. 6.13) I-6.35-]. For
• Pd the alkali metals n~=l. Their
5 AH values have been multiplied
E
MoIIV
INb sTOIu by 2 to correlate them with the
.7 -4o
other metals for which ns = 2
e~
I
eTi oHf
oTh
-80 ~
Boo
col $Yb iZr
• Gd
Sr LO! $ceErsLu

c~ o • Dy
IK 4Y
-120 ~

I I I I I
1 2 3 4 5

EF-E S levi

oW

80
olr
• Ag
Cu IAI
Cre • • Ru
IMO
40 • Pt
IMn
T • M0
Fe ! ~PIOlRh • AU
o IN1 eZn
E IU
,~, 0 IB=

8
IT
<3
ev Hf
--40 4T ~ Nbl • • To

ITi
eSr
ILOoGd I Zr
eCe
- BO Ye • • Lu
Dy
e$¢
Fig. 6.12. Correlation between ex-
perimental heats of solution at
--120
infinite dilution and the character-
istic band structure energy EF--E~
I_ I I I ~ I I I I [6.351
1 2 3 4 5 6 7 8 9
EF-E s leVI
 Heat of Formation Models 251

Fig. 6.13. Relative position of Ev and

RU
E~ for representative metals. E~ is
defined as the energy for which the
integrated electronic density of
states is equal to I state per atom
3 I-6.116]

-6 -4 ~2 0 2 4 ~B -6 -4 -2 0 2
Es EF E(eV) Es Er

Pd Ag

1
-8 -6 -4 -2 o -8 -6 -4 -2 0
Es E~ Es Ev

described by the same relation (6.114). Furthermore the large difference in

hydriding properties of similar metals such as Be and Mg or Pd and Pt is
correctly described.
Until now no first-principles derivation of (6.114) or (6.118) has been given.
In the following we shall, however, show that the effective medium theory
(including covalent effects) of Norskov [6.3] can be cast in the form of(6.118). In
this theory the standard heat of solution at infinite hydrogen dilution is given by
ZJ/~O horn -
= AEerr (no) + AE hyb- EH2, (6.121)

where Eu2 = -- 2.4 eV is the binding energy in the H 2 molecule (expressed per
atom). The effective embedding energy consists of two terms
horn -
f (no)=AE horn ( n- o ) - -
-
AEef cqotno, (6.122)

where AE h°mis the embedding energy of a hydrogen atom into the free electron
gas of the host metal which in the simplest effective medium approximation is
treated as a homogeneous electron gas of appropriate density rio. The second
term C(totri0 is a correction to the zeroth-order (in electron density variation)
energy AEh°m(ro).The second term in (6.121), AE hrb is due to the hybridization
of the hydrogen 1s-like state with the host metal d-states. Within the atomic-
sphere approximation [6.33, 34]

AdAs ~ 1
AEhyb= --40(1 -- f ) Ca - vO ~, (6.123)
R

where A d (for the host metal d-states) and As (for the hydrogen s-state) are
252 R. Griessen and T. Riesterer

functions of the potentials and of the radii of the metal and the hydrogen atomic
spheres, respectively. Ca is the centre of the d-band and V° is the value of the
effective potential of the host metal at the hydrogen site. In most cases Cd-- V ° is
much larger than the d-band width. The summation is over all lattice sites, the
hydrogen at an interstitial site being taken as origin. The parameter f is the
degree of filling of the d-bands which have an s-character at the hydrogen site. It
is equal to 0.15 for Sc and increases up to 0.85 for Ni. For noble metals, f = 1
because of their full d-band [-6.3].
To simplify the notation we introduce an effective band width W~ffdefined
as
AdA~ o 1 (6.124)
= 40 c 7 #

and rearrange the terms in (6.123) so as to obtain

hom(nO)
A H ~ = [Eeff - -- Weft] At-f W etf - - EHz. (6.125)

The expression in square brackets has the interesting property of being almost
constant for all transition metals, the largest value being - 2.97 eV for La and
the smallest -4.24 eV for Pt. The average value is -3.51 eV.
The second term on the other hand varies strongly across the 3d-transition
metals and is therefore naturally associated with the parameter A E of the semi-
empirical band structure model. As shown in Fig. 6.14 there is an approxi-
mately linear relation between fW, ff and AE.
Finally from (6.125) and Fig. 6.14 it follows that

A/7 ° ~ 27(EF- E~)- 110 (6.126)

ifE v and E s are expressed in eV and A H ° in kJ/mol H. The numerical values are
remarkably close to the values of the parameters e® and fl~ given in
(6.119, 120). For transition metals it is thus possible to give a justification of the
semi-empirical model of Griessen and Driessen. There remains the interesting
question of why this semi-empirical model also works well for non-transition
metals.
The validity of the underlying physical basis of the Griessen-Driessen model
can also be tested by looking at implications of (6.114) or (6.118).
One implication, for example, concerns the expansion of the metal lattice
during hydrogen absorption or, equivalently, the standard partial molar
volume 17. of hydrogen in a metal. From relation (6.4) and the definition of
A/7 =/-/n - ½Hn2 we obtain

1 ~3A/7 T~Prl
~'H= B ~ l T , x + ~3T p,x' (6.127)

where B is the bulk modulus of the metal hydride which is usually approxi-
mately equal to that of the host metal [6.93, 94]. As ~'n is experimentally found
 Heat of Formation Models 253
2.0
Fig. 6.14. Relation between the parameter f
® W~ff in Norskoo's theory [6.3] and the
characteristic band structure energy
1.5
® E v - E~ [6.35]

®®
,,p 1.0

Q5

0
0 2 4 6 8
E F- E s [ eV]

to be only weakly temperature dependent, the last term in (6.127) may safely be
neglected. One then finds if AH"~AH°_~a that

~'n= B
1[(; 8~-fOlnc~']
f-VnV J AH°~t~
_13(30lna 81nfly]
Oln V * OlnVnV ) J (6.128)

if E v - - E s is assumed to vary with the host metal volume as V-"/3. The exponent
n depends on the type of metal under consideration. For simple metals the
nearly-free-electron model predicts that n ~ 2. For transition metals band-
structure calculations lead to values close to n = 5 [-6.95, 96]. If for simplicity the
parameters ~ and fl are taken to be independent of volume then the partial
molar volume is given by

1 n
~'H~ -~ 3 (AH°~tj-fl)" (6.129)

For transition metal hydrides such as VH x and PdHx (6.129) leads to

~l.0cm3/mol H for the former and V'H~I.1 cm3/mol H for the latter
in semi-quantitative agreement with the experimental results
V"n = 1.42 cm3/mol H for VHx and F'n = 1.69 cm3/mol H for PdH x. This is quite
remarkable as the partial molar volume ~n is a derivative of the enthalpy. In
other words the empirical correlation given in (6.114, 118) combined with
theoretical results concerning the volume dependence of electronic band widths
predicts A/7 as well as A~q/a In V.2 This strongly suggests that the physical basis

2 Miedema's model [6.28] also leads to good predictions for the volume dilation of metals
during hydrogen absorption. This is, however, not surprising, as ~'a= 1.7 cm3/mol H is
introduced as a fit parameter in this model.
 254 R. Griessen and T. Riesterer

AVmode I ( crn3/mol H )
0 1 2

6L I I

®
®
,@ []
® -6
E
o,~
v

~3 U

Oil I I I I I I Io
0 1 2 3 4 5 6
A.Vmod~l ( ,~ 3 )
Fig. 6.15. Partial molar volume for hydrogen in various metals and intermetallic compounds
compared to the predictions of the band-structure semi-empirical model [6.35, 38]

of (6.114, 118) is sound. As shown by Griessen and Feenstra [6.38] the

agreement between calculated and measured P'n may be improved by allowing
for volume-dependent parameters ~ and/L From a fit to existing data they
obtain

I (2.69AHO_,p + 326) (6.130)

for transition metal hydrides and

1 (1.67AHO_,t~ + 189) (6.131)

for simple metal hydrides when ~'n is expressed in cm3/mol H and AH=~ 0 in
kJ/mol H. A comparison between VI~values calculated by means of (6.130, 131)
and experimental values is shown in Fig. 6.15.
 Heat of Formation Models 255

b) Ternary Metal Hydrides

The most interesting use of the empirical linear relation between the standard
heat of formation AH and the characteristic band-structure energy parameter
AE is to predict AH for ternary hydrides AyaBr,,H x. As band-structure
calculations are available only for a very limited number of alloys or
compounds Griessen and Driessen used the model of Cyrot and Cyrot-
Lackmann I-6.97] to evaluate AE for intermetallics of two d-band metals. The
basic idea is that within the tight-binding approximation the exact density of
states n(E; YA, Yn) calculated by means of the coherent-potential approximation
formalism for an alloy AyABy,~ has the same moments

#~ = [.E~n(E; YA, YB)dE (6.132)

for v = 0 , 1, and 2 (and even v = 3 in the case of two metals with equal
bandwidths) as the "simple" density-of-states (DOS) n*(E) given by

yAn'A(E)+yBn'B(E)
n*(E; YA,YB)= , (6.133)
YA + Yn

where n;(E) is a scaled DOS function of metal i. The various steps involved in
this scaling are shown schematically in Fig. 6.16 for the special case of an alloy
AB (YA= Yn = 1) when the d-band width of metal A is larger than that of metal B,
i.e. WA > WB.
In the first step the widths of the d-bands of both metals are set equal to their
weighted average

W * - yAWA+ ysWn (6.134)

YA + Yn

Since the number of available Bloch states contained in the d-band is constant,
a change in band width implies a change in height of the DOS curve so that

with (6.135)
ni L W* l = - ~ - ; n , \ W~ I
W*
E ' - e i = ( E - e i ) Wi (6.136)

In (6.135, 136) e, is the centre of the DOS curve of metal i (i = A or B) measured

with respect to the vacuum zero. For the individual metals

~ = (Ev-- E,)i -- Wi +4~, (6.137)

256 R. Griessen and T. Riesterer
wA Fig. 6.16. The various steps involved in the construction
of the appropriate density-of-states function n*(E) for
the special case of an alloy AB [6.35, 97]

WB

. ,',E'. ~_ ~'.
W,~ ~
~. = W ~

W~

where ~bi is the work function of metal i. After having brought the D O S curves
ni(E) to a c o m m o n width W* the band energy parameters AE'~ and work
functions ¢'i are given by

W*
AE'i=AE,~ and (6.138)

q~it _-_ 49i + AEi-- AEiI + ~1 ( W * -- Wi) (6.139)

since ~A and en are assumed to be constant. In both (6.138) and (6.139) AEi
- ( E v - Es)i and A E'i =- (EF-- Es)I. On an absolute energy scale the Fermi energies
E~,, and E~B are different. This leads to an electron transfer until the Fermi
energy is constant throughout the alloy.
In the second step of the construction of n*(E; YA, YB) o n e thus equilibrates
the Fermi energies by requiring to first order that

(4 - 4'A)YAnA(E~a) = ( ' ~ , - - q~ )y.n.(E~,.), (6.140)

where E'v, is the scaled Fermi energy of metal i obtained by setting E = EF, in
(6.136) and 4~ is the work function of the alloy AraBy B under consideration. As
 Heat of Formation Models 257

seen from (6.137, 139, 140) ~b* depends on the bandwidths Wi, the DOS ni(Ev)
and the work functions ~b~of the pure metals A and B and, of course, on the
concentration YA and Y8 of A and B atoms.
Since according to the result of Cyrot and Cyrot-Lackmann [6.97], a good
approximate DOS n*(E;y A, YB)can be obtained by taking a weighted average of
the scaled DOS, n~(E),of the pure constituents one expects that for the heat of
formation of the hydride

yAAE] + ynAE*)
AH°~a(AyABv~Hx) = ~ \-
_

YA + YB
_

.1 (6.141)

with

AE* = AE'~+¢'i-- ¢*. (6.142)

As indicated in Fig. 6.16, AE* corresponds to the band-structure energy

parameter of the scaled DOS of metal i after allowing a transfer of electrons to
bring EF, to a common level.
Values for AH°~p were obtained from (6.114-117) by using experimental
values for ~bl,calculated DOS curves for the band widths W/and the optimized
AE values given in [Ref. 6.35, Table II]. These values are such that when
inserted into (6.114) the experimental values for the heats of formation of the
binary hydrides are reproduced [using the values for e and /3 given in
(6.116, 117)]. In most cases the optimized AE differ by no more than a few tenths
of an eV from the values derived from band-structure calculations.
As shown in Fig. 6.17 the values for AH~o~ o calculated by means of (6.141)
are in remarkably good agreement with experimental values. The average
mean-square deviation is more than twice as small as that in Fig. 6.10 which
shows the predictions of the cellular model of Miedema and co-workers which
involves many more fit parameters. From a practical point of view this means
that the predictive power of the Griessen-Driessen model is sufficiently high for
a meaningful overall search of potential hydrogen storage alloys. Calculated
heats of hydride formation are given in Sect. 6.6.3 for ABs, AB2, AB, A2B, and
AsB intermetallics (A is Sc, Ti, V, Y, Zr, Nb, La, Hf, Ta, Pd or Ni and B is a
transition metal).
It is important to point out to the user of these tables that the results listed
correspond to a situation where each hydrogen atom is surrounded by a
fraction YA/(Ya +Ys) of A atoms and a fraction YB/(YA+ Yn) orB atoms. This is an
implicit assumption made in writing (6.141). This is in general .not true for
intermetallic compounds (in AB2 Laves phases, for example, there are three
types of interstitial sites AzB2, AB 3, and B4). Furthermore in disordered alloys
there is a whole distribution of sites. In dilute disordered alloys most of the
interstitial sites are coordinated by atoms of the most abundant metal. The
heats of formation given in Tables 6.6.3a-e correspond then approximately to
hydrogen concentrations which are high enough so that most hydrogens
 258 R. Griessen and T. Riesterer

AHcalculated ( k J/mol H )
-80 -60 -40 -20 O
I I [ i I I O

6TiCrl a
iTiCr~ o TiCo
ITIMn 2
TiNi YNI TiFe ,
• • 5 • oLaNIB
ZrNi 2 , s• . • oNobQsMoos
@YFez ZrMo ." YCos LoCo 5
YzNiTi/I TIFe
ZrMn YCo3 IScFe -20
YC°2 o ZrCr 2o i ILanh2 ~ ILQCOs
ZlZrMnI8 YNi~'IZrINil 0 oTiMn16
ZrCo I I VosCrQ@~Y2Co 7 • IIITiMnl5 t>
~M0: ' l H% ' T
YCo3o • LoNI~
Nb~0MO~2 TiCo
ITi2Pd . "O
N bo,3T~o.7 ZrNi O TICrl s
• OTi0zV0a
• TaVz
NboTsTIo 25 oZrCo -40 c_
eZrCr z
oZrV2 • Ti2Pd
3
o
1"
eLaNi
• Nb0.zsT[0n5
• Nbo.5 Ti o.s
eTaV~
- 60

eZrV2

oLoTNi3
Z r(L95SO0 os
oZr~Ni -80
oTiZr

__ I I I I I I

Fig. 6.17. Comparison of experimental heats of formation with values predicted from the band-
structure semi-empirical model of Griessen and Driessen [6.35]

occupy sites of composition AyABrB. For example, for a AhBH x disordered

hydride the tabulated values correspond roughly to x--- 4 if one assumes that
the interstitial sites are coordinated by 6 nearest-neighbour atoms.
Another factor which has been neglected is the volume change due to
a l l o y i n g of a metal A with a metal B. This point will be treated in s o m e detail in
the next section.
To conclude this section it is worth considering a few special cases of the
Griessen-Driessen model in the light of our discussion of general semi-empirical
models. The simplest model is the square electron density-of-states model for
which na and nB are constant in the energy interval [e i -(WJ2), e i + (W,./2)] and
v a n i s h everywhere else. For this m o d e l of a d-band metal

h A W A = n B Wn = 10 (6.143)
 Heat of Formation Models 259
80 a 8O 80

40 Ru 4 0 ~ Ru ,40 Ru
Rh Rh
7 Mo Mo Mo
E 0 Tc 0 Tc 0 T¢

Pd Pd
o ~ -40
Nb - 4 0 ~ -40 Nb

i
m
-80 IZr -80 Zr -80 Zr

Y Y
I [ l l
-120
0 Cl 1 -120
0 ¢ I
-120
c
Nb B Pd ~
Fig. 6.18a-c. Heats of formation of the hydrides of Y-, Nb-, and Pd-based alloys calculated by
means of(6A41) with the assumption that each hydrogen is surrounded by a fraction 1 - y of Y,
Nb or Pd atoms. Volume effects are not included

and the integral heat of formation is given by

d/~°(A 1 - rByHx) = (1 - y) A/~°(A Hx) + yA/~°(BHx)

+ ~x(Za - ZB) (WB-- Wa) y(1 - y). (6.144)

This expression has the same form as the general relation (6.89) [see also (6.81)]
except that it does not involve squared terms such as (ZA- Zn) 2 or (We-- WA)2.
In this simple model the two relevant parameters which characterize a metal are
Z, the number of d-electrons and W, the width of the d-band. It is quite
remarkable that e, the position of the centre of the d-band, does not appear in
(6.144). This is due to the fact that when an electron flow is allowed from one
metal to the other (see Fig. 6.16) to bring the Fermi energies of both metals to a
common value, the total energy remains constant. This, however, is a pecularity
of this simple model as can easily be shown by considering another density-of-
states model where both metals A and B have the same d-band width, i.e.
WA= We but different density of states at the Fermi energy. Then

A/4°(A, _ rBrHx) = (1 -- y) A / ~ ° ( A H x ) + yA/4°(BHx)

1
--ctxy(1--y)(~pA--q~B)(nA--n~) ( l _ y ) n A + ynB. (6.145)

This expression is consistent with the general form (6.89) derived in Sect. 6.4.1
for a two-dimensional model, each metal being characterized by the work
function qS~ and the value of the density of states ni at Err As discussed by
Griessen and Driessen this expression explains qualitatively why alloys of Pd
and an early transition metal have a tendency to be poorer hydrogen absorbers
than pure Pd. This is confirmed by the results shown in Fig. 6.18 which have
been obtained from expression (6.141). For Pd-based alloys (as well as for Pt-
260 R. Griessen and T. Riesterer

and Ni-based alloys) the A H ° versus y plots do indeed exhibit a large positive
concavity. Conversely for transition metals of neighbouring columns in the
periodic table ~bA~- ~bnand n A ~- ntj (except for Cr, Mo, and W based alloys) and
the nonlinear term in AH ° is negligible.

e) Volume Effects
Due to the importance attached to correlations between heat of formation and
host metal volume or interstitial site size we consider here in some detail
another implication of (6.114) [or (6.118)].
By combining (6.127) and (6.130) we obtain for transition metals a volume
dependence of the heat of formation given by

OAH°~a = -- 2 . 6 9 A H ° . ~ - 326. (6.146)

O l n V r.x

This means that, as expected, a dilation of the host lattice induces a lowering of
the heat of formation. This prediction cannot easily be checked by direct
mechanical compression of the lattice because of the very high pressures which
would be required for a sizeable volume change. A much easier way is to alloy a
metal A to a metal B with a smaller affinity for hydrogen. At low hydrogen
concentrations the sites coordinated exclusively by A atoms will thus be
occupied first and the heat of solution determined in the dilute regime will
correspond to the A/-1 of the pure host alloy at a volume V,lloy which may be
larger or smaller than Va depending on whether B atoms act as dilation or
contraction eentres in the A matrix.
For Pd based alloys the data compiled in Fig. 6.19 imply that
OAB °
1 - 223kJ/molH
OlnV

in reasonable agreement with the value of - 2 9 9 kJ/mol H obtained by

assuming that (6.146) is also valid at infinite dilution. For Nb alloys the data of
Machlin I-6.98] lead to
aA/-/o
- - - 197kJ/molH
OlnV
and the model gives a value of - 232 kJ/mol H.
Another very interesting class of materials are the hydrides of the LaNi5
type compounds and some substituted quaternary hydrides such as
LaNia_yAlrH ~. As shown in Fig. 6.20, LaNisH ~ is stabilized when Ni is
replaced by a larger atom. The slope of the straight line fitted to the
experimental data is

~AH°--'P = - 231 kJ/mol H,

~lnV
 Heat of Formation Models 261

10 co t~o fl

~o~0

5
., /.////f//,~f,~
k_..J

-o 0 q¢
Q..
40 29 252019 15 10 • C<12

• --.~ C e 4 Ce 12,5

<3.
-5
>,

"o
Q_
v
-10

<3

-15

I I I 1 I
- o.1o -o.o o o.o5 o.,o

cl(Pcll,yMy)-(a{Pd) [J~]
Fig. 6.19. Heat of solution at infinite dilution for Pd-based disordered alloys as a function of
lattice spacing. The numbers indicate the amount (in at.%) of metal alloyed to palladium
[6.98, 117]

which is in agreement with the theoretical value of - 272 kJ/tool H. As pointed

out by Percheron-Gu6gan [-6.90] this behaviour of substituted LaNi s hydrides
cannot be explained by means of Miedema's rule of reversed stability since the
intermetallic compounds LaNi4M (M = Cu, Fe, Ni, and Mn) exhibit similar
heats of alloy formation while their heats of hydride formation are different
especially for Mn. Furthermore, LaNi4A1 forms a very stable hydride although
the intermetallic itself is not unstable.
Further, somewhat indirect evidence for the importance of volume effects in
the heat of formation may be found in the isoelectronic alloys such as Nbl _rVr.
The heat of solution (formation) of the binary hydrides VH~ is almost the same
as that of NbHx. However it is found that the critical point of the N b - H phase
diagram is substantially lowered by the addition of vanadium. As shown by
Griessen and Driessen [6.37] this critical-point lowering may easily be
understood by taking into account that vanadium clusters in a niobium-rich
262 R. Griessen and T. Riesterer

Fig. 6.20. Heat of formation of

-10 LaNi 5_yMrH~as a function of unit cell
\ volume. The dashed line corresponds to
\ the prediction of the band structure
\ semi-empirical model I-6.35,38]
\
'LoNi 5 • \
Le Ni4Cu & •
-5 La Ni4 F'e \
E
LQ Ni4.5 AIO. 5
-20 - Lo Ni3Cu 2 \•
\
\
t [J~ N'I4 AI I \
I _ La Ni4 Mn \
<1 \

-30
_1 I I I I I I I
85 9O
CELL V O L U M E [~3]

alloy have a strong hydrogen affinity because the V-V separation in such a
cluster is larger than in pure vanadium. In other words, V acts as a trapping
centre in Nb in agreement with the work of Futran et al. [6.99] and Pick and
Welch [6.100]. In non-isoelectronic alloys volume effects play in general a
secondary role. In Nbl_yMoy, Fenzl and Peisl [6.101] found for example a
decrease of 123 K in T~for y = 0.05. Such a large variation can be semiquantita-
tively explained by the large difference in hydrogen affinity of Mo and Nb
[6.37].
Another manifestation of the dependence of A H on interatomic spacing is
the preference of hydrogen (or its isotopes) for occupying specific sites of the
metallic host lattice. In fcc metals such as Ni, Pd, and Cu, hydrogen occupies the
octahedral interstitial sites while in the group V bcc metals V, Nb, and Ta it
occupies tetrahedral sites (at low concentrations). These site preferences can be
explained by means of the geometric model of Westlake [6.63]. In Pd, for
example, the tetrahedral hole radius is rh~-0.37A. It remains unoccupied
because rh is smaller than the empirical critical hole radius of 0.40 A. The
octahedral site on the other hand is easily occupied (according to this criterion)
as rh --~0.64/~. In our opinion the existence of a sharp criterion for the minimum
hole size must reflect a strong dependence of the metal-hydrogen interac-
tion potential on interatomic spacing.

6.4.4 The Local Band-Structure Model

Very recently Griessen [6.102] proposed a new model, the so-called local band-
structure model in which the heat of hydrogen solution at infinite dilution A/7o~
in a transition metal is given by
 Heat of Formation Models 263

AHoo=aAEW ~/2 Z. Rj-4+ b with (6.147)

J

a=lS.6(kJ/molH)A4eV -3/2 and (6.148)

b = - 90 kJ/mol H (6.149)

if AE and the d-band width W are given in eV and the distance R~ between a
given hydrogen and the j-nearest neighbour host metal atom is expressed in A.
As in the band-structure model of Griessen and Driessen, A E - E v - - E ~ [see
(6.114)]. This model incorporates both nonlocal effects (through the electronic
band energies AE and W) and local information on the site actually occupied by
a hydrogen atom (through the purely geometric term Zi R~ a, the summation
being over all host metal atoms in the first nearest neighbour shell of a given
hydrogen atom).
As shown in Fig. 6.21 the correlation between the experimental values for
AHo~ and the parameter A E W 1/z ~ , j R f 4 is excellent. The discrepancies for
tungsten and iridium are not serious as the experimental uncertainties in the
AH~ of strongly endothermal metal-hydrogen systems may be considerable.
The remarkable agreement between calculated (AHo~°d) and experimental
(AH~P) values for the heats of solution is also clearly indicated in Fig. 6.22.

o.ii.
8O

40
100

E "~ 50
• Ru
"- 0

,,q
/
Y.
]~h
Moe I F~C°

-40 Ht Ta

-5° .jz'T
-80
e$c
I I I I I I I I I I I -100 t t f [
O 2 ,4 6 8 10 12 -100 -50 0 50 100
AE W'2 ~R[4(eV3~2 ~ TM ) Al~m~°d (kJ/molH)

Fig. 6.21. Correlation between experimental values of the cnthalpy of hydrogen solution at
infinite dilution and the local band-structure parameter A E W 1/2 ~_,R i 4
J

Fig. 6.22. Comparison between experimental A/7~P and values calculated by means of the local
band-structure model (6.147-149)
 264 R. Griessenand T. Riesterer
600
Fig. 6.23. Comparison between experi-
mental B~'~~p and values calculated by
_ • Mo means of the local band-structure model
(6.150 or 151). For ferro- or antiferromag-
ove
netic metals the values used for the bulk
"~ 4 0 0 - modulus B are those calculated by Moruzzi
et al. 1,-6.103](see also I-6.38])
•

oo-
Ti
Zre el-If

0 "L~"' I I
0
esc

200
I
400
I
600
- rood
BV H ( kJImol H)

The model also predicts the correct volume dependence of A/7~ and can
therefore be used to evaluate the molar volume of hydrogen in a transition
metal. This is easily seen by using (6.127) and the fact that for a transition metal
both AE and W scale as V-5/3 so that

B~ZH- aA~qoo = 71.2AEW,/2 E R ; 4 (6.150)

c31nV J
or in terms of A/7~,

B~'ri = 3.83AH~o + 345. (6.151)

A comparison between experimental values (BVr~xp) and calculated values

(BV~"°a) is shown in Fig. 6.23. The simple relation (6.147) thus well reproduces
both AH~o and its volume dependence without requiring additional fit
parameters as was the case in Miedema's model and in the band-structure
model of Griessen and Driessen.
The tendency of hydrogen to occupy certain sites in a given crystal structure
follows directly from (6.147) by noting that AE and W are by definition site
independent so that the heats of solution for hydrogen in two different types of
sites (say type I and type II) are related to each other by

(6.152)
A HI~ + 90 2 R j- 4.
s i t e II

This simple relation correctly predicts that octahedral sites are occupied in
palladium, tetrahedral sites in group V metals and that in the early transition
metals dual occupancy (of the octahedral and tetrahedral sites of the hcp
structure) is possible.
 Heat of Formation Models 265

Other applications of the model, for example, to calculate the binding

energy of hydrogen to a vacancy or the heat of solution of hydrogen in
intermetallic compounds are discussed in [-6.I02, 104].

6.5 Conclusions
On the basis of the work reviewed in this chapter it can be concluded that:
l) The heat of solution (formation) of a metal-hydrogen system depends
both on lattice spacing and electronic structure. For recent articles on this
subject the reader is referred to [,-6.105-109]. Only in special cases (such as
hydrogen in dilute alloys and in isoelectronic alloys) are lattice spacing effects
dominant. It is therefore not expected that empirical correlations involving
only cell volume or hole size will hold for a wide class of materials.
2) Empirical correlations between AH and a physical parameter (for
example, shear modulus, compressibility, electronic specific heat, atomic
volume, interstitial hole volume, etc.) have in general been proposed on the
basis of too small a number of metal hydrogen systems. For some of them it is
shown that the proposed correlations do not even hold for the binary hydrides
investigated up to now.
3) For restricted classes of materials empirical correlations may be useful
for the tailoring of metal hydrides with precisely defined characteristics (a given
plateau pressure at a chosen temperature, for example) [6.110, 111].
4) For binary hydrides the trends in AH through long series of the periodic
system are well reproduced by first-principles calculations. For a fast and
accurate evaluation of AH however, one still requires semi-empirical models.
5) For an overall search for metal hydrides with specific properties semi-
empirical models are most valuable. For ternary metal hydrides the band-
structure semi-empirical model of Griessen and Driessen appears to be more
reliable than the cellular model of Miedema and co-workers (although it
involves far less fit-parameters than Miedema's model). One should keep in
mind however that Miedema's model was designed to describe a much wider
class of alloys than just the metal hydrides.
6) The constancy of the volume dilation upon hydrogen absorption
observed in many transition-metal-hydrogen systems is related to the general
behaviour of the volume dependence of their electronic d-band width.
7) The two models proposed so far to explain the site preference of
hydrogen in intermetallics cannot be used for an evaluation of the actual heat of
solution (formation) associated with the various interstitial sites. The
imaginary-binary-hydrides model of Jacob and Shaltiel is based on an ad hoc
definition of local enthalpies and Westlake's geometrical model does not
involve any thermodynamic parameters at all.
8) The local band-structure model recently proposed by Griessen incorpo-
rates both site-dependent effects and electronic band-structure effects. It can
advantageously be used as long as sufficiently accurate first-principles calcu-
lations are not available.
266 R. Griessen and T. Riesterer

6.6 Tables
Table 6.6.1. Heat of solution and formation of binary metal hydrides. A/-Sr~is the heat of
solution at infinite dilution and A H is the heat of formation of concentrated hydrides
(indicated in parenthesis). Both enthalpies are given in kJ/mol H. For most metals the values
for A/7oo and A H are for the reaction of hydrogen with the solid metal (s) unless specified
explicity by (/) for the liquid state

Element A / ~ [kJ/mol H] A H [kJ/mol HI Ref.

AI + 63 [6.118]
(/) + 59 [6.118]
(/) + 59 (1 000-2 000 K) [6.119, 120]
+26 (298 K) --7/+3 (A1Ho.s) (298 K) [6.54]
- 4 (AIH3) calor (298 K) [6.121]
Ag + 63 (823-1 234 K) [6.120]
(l) + 68 (1234-1923 K) [6.120]
Am --95 (AmH2) (750-1 150 K) [6.122]
Au +32 (966 1 373 K) [6.120]
B --4 (BloH1,) [6.119]
Ba --88 (BaH2) (743-823 K) [6.119]
--86 (BAH2) calor (298 K) [6.119]
Be -- 2 [6.118]
~0 (BeH2) estim (298 K) [6.119]
Bi
Ca - 9 2 (Call2) (873-1053 K) [6.119]
- 8 5 (Call2) (1053-1173 K) [6.119]
- 9 4 (Call2) calor (298 K) [6.119]
Cd
Ce --74 (800-1 100 K) [6.123]
-- 73 (824-1 173 K) --•09 (Cell2) (824-1 173K) [6.1253
--102 (Cell2) (870-1270K) [6.120]
- 103 (Cell2) [6.124]
Co +26 (923-1 768 K) [6.120]
+21 (298 K) + 15 (Coil0.5) (298 K) [6.54]
(0 +32 (1 768-2073 K) [6.120]
Cr +52 (1 000-1 620 K) - 6 (CrH) (298-423 K) [6.120]
+28/--2 (298 K) - 8 / + 2 (CrHo.5) (298 K) [6.54]
Cs (0 --56 (CsH) (518-651 K) [6.119]
Cu +46 (770-1 356K) [6.1201
+55 [6.125]
+49 [6.118]
+43 (1 356-1 823 K) [6.120]
Dy - 7 9 (830-1 250 K) - 1 0 4 (DyH2) (830-1250K) [6.120]
--110 (DyH2) [6.124]
Er -112 (ErH2) [6.124]
Eu
Fe (~x) +29 (280-1 184K) [6.120]
+28 (1 184-1 667 K) [6.1201
(6) +29 (1667-1 811 K) [6.120]
(0 +33 (1 811-2093 K) [6.120]
+21 (298 K) +10 (FeHo.5) (298K) [6.54]
 Heat of Formation Models 267

Table 6.6.1 (continued)

Element A/4~ [kJ/mol H] AH [kJ/mol H] Ref.

Ga
Gd - 6 2 (823-1 173 K) --103 (GdH2) (823-1 173 K) [6.125]
- 7 6 (830-1250 K) -101 (GdH2) (830-1250 K) [6.120]
-- 98 (GdH2) [6.124]
Ge (+221) (1073-1183 K) [6.126]
(/) +14 (1770-1 979 K) [6.120]
Hf (~) - 38 --66 (HfH2) [6.118]
Hg
Ho --113 (HoH2) [6.124]
In
Ir +74 (1666-1 853 K) [6.127]
K N0 [6.118]
(0 --56 (KH) (561-688 K) [6.119]
- 58 (KH) calor (298 K) [6.119]
La - 67 (900-1 050 K) [6.1181
--97 (LaH2) [6.124]
--104 (LaH2) (600-1 150K) [6.120]
Li (/) -51 (980 l180K) [6.120]
(0 --97 (LiH) (<967K) [6.128]
-- 72 (LiH) (> 967 K) [6.128]
--90 (LiH) calor (298 K) [6.119]
Lu - 79(830-1 250 K) -- 102 (Lull2) (830-1250 K) [6.120]
-- 104 (LuHz) [6.124]
--102 (Lull2) (1098-1 223 K) [6.1291
Mg +21 (370-940 K) [6.120]
(/) + 27 (970-1 070 K [6.120]
--37 (MgHz) (713-833 K) [6.1/9]
Mn (~) ( - 2) (413-898 K) [6.120]
(/~) +37 (1 050-1 330K) ~6.J2o]
(0 +30 (1 623-1 718 K) [6.1201
+ 1 (298 K) --8 (MnHo.5) (298 K) [6.541
Mo +46(973-1 773 K) [6.120]
+34/+15 (298 K) +5 (MoHo.s) (298K) [6.54]
Na (/) +2 (380-670 K) - - 53 (Nail) (38(~670 K) [6.120]
- - 5 6 (Nail) calor (298 K) [6.119]
Nb - 3 8 (275-2275 K) [6.12o]
--33 (275 353K) [6.120]
--35 (625-944 K) --44 (NbHo.s) (625-944K) [6.130]
--38 (NbHo.5) [6.41]
(/) -31 [6.118]
Nd - 5 0 (823-1 173 K) -103 (NdH2) (823-1 173 K) [6.125]
-- 106 (NdHz) [6.124]
Ni +16 (623-1 673 K) [6.120]
+ 12 (473-873 K) [6.131]
+17 (1000-1 516K) [6.127]
+10 (298 K) - 3 (NiHo.5) (298 K) [6.54]
(0 + 24 (1763-1 973 K) [6.120]
Np - 7 3 (NpHz~.0 (808-883 K) [6.132]
(P) - 56 (NpH2 +~) (743 K) [6.198]
(r) -61 (NpH2+~,) (875 K) [6.198]
268 R. Griessen and T Riesterer

Table 6.6.1 (continued)

Element A/loo [kJ/mol H] A H [kJ/mol H] Ref.

Os
Pa ( - 36/-48) (PaH1.a) [6.133]
Pb +62 (298-600 K [6.1231
(/) + 58 (600-1 973 K) [6.1231
Pd - 10 (273~420 K) [6.118]
-- 20 (PdHo.6) [6.53,118]
Vm
Pr --68 (823-J 173 K) --106 (PrH2) (823-1 173 K) [6.125]
- 104 (PrH2) [6.124]
Pt +42 (575-1 670 K) [6.1201
+27 (883-1466 K) [6.22]
Pu --77 (Pull 2) [6.122]
-- 86 (PuH2) calor (298 K) [6.134]
Ra (-72) (Rail=) [6.135]
Rb (/) - 5 4 (RbH) (519-623 K) [6.119]
Re
Rh +27 (i 131-1 797K) [6.127]
+ 10 (RhHo.5) (298 K) [6.54]
Ru + 54 (1 275-1 776 K) [6.127]
Sb
Sc -90 '-- 100 (ScH2) [6.1181
Se (+39/+33) [6.1351
Si ~+180 [6.118]
(/) +110 [6.118]
Sm - 70 (775-925 K) - 9 8 (SmH=) (775-1100K) [6.120]
- 100 (SmH2) [6.136]
Sn (,9 + 125 (1 273-1 573 K) [6.123]
Sr --61 - 9 2 (SrH2) [6.137]
-100 (SrH2) (...-1273K) [6.119]
- 8 8 (SrH2) calor (298 K) [6.119]
Ta - 3 6 (313-1 213 K) [6.1201
- 3 2 (623-904 K) [6.130]
--38 (TaHo.5) (573-973K) [6.119]
Tb --78 (830-1250 K) - 104 (TbH2) (830-1 250 K) [6.•20]
--106 (TbH2) [6.124]
Tc -12 (TcHo.5) (298K) [6.54]
Te (--84) [6.135]
Th --40 (573-1 073 K) [6.123]
--73 (ThH2) [6.119]
Ti (a) - 52 (573-1 073 K) [6.138]
(/3) - 5 8 (1 173-1423 K [6.1381
(t) --47 (1928-2073 K [6,123]
--67 (Till2) (723 K) [6.123]
--68 (Till2) calor (298 K) [6.119]
TI
Tm -112 (TmH2) [6.124]
u (~) +7 (<941 K) [6.123]
(~) + 17 (941-1048 K) [6.123]
(~) +4 (1048-1405 K) [6.123]
(v) +8 [6.139]
 Heat of Formation Models 269

Table 6.6.1 (continued)

Element AH~, [kJ/mol H] AH [kJ/mol H] Ref.

U (continued)
(/) +11 (>1405 K) [6.123]
(~) -42 (UH3) (580-933K) [6.120]
(/~) -39 (UH3) (933-1047 K) [6.120]
(~) -42 (UH3) (1 023-1 320 K) [6.120]
V - 2 6 (253-373 K) --35 (VHo.5) (253-373K) [6.140]
- 3 3 (293-328 K) --42 (VHo.s) (283-413 K) [6.120]
-36 (Vho.5) [6.41]
- 29(519-827 K) -35 (VHo.5) (519-827K) [6.130]
W +96 (1 100-3070K) [6.120]
Y --79 [6.118]
-- 114 (YH2) (800-1200 K) [6.120]
- 107 (YH2) (1200-1600 K) [6.120]
-93 (YH2) (>1 173 K) [6.124]
Yb --91 (YbH2) [6.124]
Zn (+ 15) (473-673 K) [6.123]
(/~ +20 [6.141]
Zr (~) - 6 4 (773-1073 K) [6.139]
(~) --52 (700-1 170K) [6.120]
(~) --64 (1 073-1 223 K) [6.120]
q) --51 (2200-2760 K) [6.120]
(~) --94 (ZrH2) [6.118]
(~) - 106 (ZrH2) [6.118]
--82 (ZrH2) calor (298 K) [6.119]

Table6.6.2. Heat of formation of selected ternary metal hydrides. In most cases AH is

determined from the dissociation pressure of plateaus in the pressure-composition isotherms.
As the coexisting phases may be different for each metal hydride we use the symbol AH
instead of AH.~a. Heats of formation which have been determined calorimetrically are
indicated by "calor'. Values for AH estimated by means of relation (6.40) are indicated by
"estim". Values of uncertain origin are given in parenthesis. Intermetallics which do not
form hydrides are given with the comment "no hydride" and those which probably
decompose during hydrogen uptake are indicated by "decomp?"

Hydride AH [kJ/mol H] Ref.

A1HfH 1 --21 estim [6.142]

AITh2H 4 ( < - 49) [6.29]
A1Ti3H=o - 53 [6.143]
BesHfHo. z (--24) [6.144]
Be2HfH1. a ( < -- 27) [6.144]
BezTiH 3 (-- 19- -- 13) [6.144]
BesZrHo. 2 (-24) [6.144]
Be2ZrH 1.5 ( < -- 27) [6.144]
CaNi2H3.4 (decomp?) - 4 2 calor [6.145]
CaNi3H4. 5 - 2 9 calor [6.145]
CaNisH4. 5 -- 19 calor [6.145]
CaNisH 4 - 17 calor [6.146]
CaNisH1 --22 calor [6.147]
270 R . G r i e s s e n a n d T. R i e s t e r e r

Table 6.6.2 (continued)

Hydride AH [kJ/mol H] Ref.

CaNisH 1 - 22 [6.1481
CaNisH,.5 -16 [6.149]
C e C o 2H,,.1 ( < -- 36) [6.150]
CeCo3H3. s - 2 3 estim [6.150]
CeCo3H 3 -- 18 [6.120]
Ce2CoTH7 - 22 [6.151]
CeCo s H2.5 -- 20 [6.152]
CeCosH 3 - 19 [6.86]
C e C o s H 2.s -- 15 [6.120]
C e M gl 2 H 2 - - 36 e s t i m [6.153]
CeNizH3.9 ( < - 36) [6.150]
CeNi3H#. z - - 2 2 calor [6.145]
Ce2NiTH,.2 --23 estim [6.150]
CeNisH 6 -- 18 estim [6.150]
CeNisH 6 -- 7 [6.15]
CoTDy2H2. 6 - 23 [6.151]
C o a D y H 1.o - 24 [6.154]
CoTEr2H -20 [6.151]
C o 3 E r H 1.1 -- 22 [6.155]
CoTGd2H2. 6 - 29 [6.151]
Co3GdH 2 - 27 [6.155]
Co2GdH,.5 - 27 [6.73]
CosGdH2.2 - 15 estim [6.152]
CoHfH3. 2 - 30 estim [6.142]
CoHf2H3. 8 ( < -- 36) [6.142]
CovHo2H 1 -21 [6.151]
Co3HoH 1 - 24 [6.155]
CosLaHs.4 -23 [6.152]
CoaLuH 1 -- 18 [6.156]
CosNdH2.7 --21 [6.86]
CoTNd2H2. 7 -- 36 [6.151]
Co3NdH2 - 32 [6.155]
CosPrH2. 9 -- 17 [6.157]
C%PrH2.9 -- 19 [6.152]
Co7Pr2H2.5 - 32 [6.151]
Co7Pr21-12.5 -- 17 [6.157]
Co3PrH 4 -27 [6.157]
CozPrH 4 (decomp?) ( < -- 33) [6.157]
CosSmH2.5 - 16 [6.152]
CosTbH.. ' -- 14 [6.152]
CoTTb2H2. 7 -24 [6.151]
CoaTbH 1 --25 [6.155]
Co17Th 2 no h y d r i d e at 313 K , 40 b a r [6.86]
CosThH2. 6 - 14 estim [6.152]
CoTTh2H~.5 - 2 2 estim [6.86]
CoThH 4 ( < - 24) [6.86]
Co3ThvH29 ( < -- 24) [6.86]
C o T i H 1.4 - 29 [6.158]
C o T i H 1.s -- 31 [6.159]
CoTiHo.9 - 27 [6.160]
 Heat of Formation Models 271

Table 6.6.2 (continued)

Hydride A H [kJ/mol H] Ref.

CoTiHI.s - 24 calor [6.1451

C03TmH 1.3 - 22- [6.161]
CosYH2. 8 --14 [6.120]
CoTY2H1. 5 - 2 7 estim [6.150]
Co3YHx.o -28 [6.155]
Co2YH3.7 - 2 3 estim [6.1501
CoZrHo.6s -41 [6.162, 142]
CoZr2H4. 8 (decomp) [6.142]
CrNi2H ~ o + 29 (573-773 K) [6.120]
+53 (773-1 073 K) [6.120]
Cra.gTiHz. 5 (C14 struct) -13 [6.163]
Crl.sTiH=o (C15 struct) -35 [6.164]
Crl.sTiHoA (C15 struct) -22 [6.1641
Crl.aTiHo. 2 (C15 struct) -12 [6.164]
CrL~TiHo. 9 (C15 struct) --9 [6.164]
CrLvTiH2. 6 -- 10 [6.165]
CrzZrH~o -23 [6.166]
CrEZrH.." - 19 calor [6.167]
CusLaH2. s --22 [6.87]
Cu2Th no hydride (?) [6.144]
CuTiH1 -38 [6.168]
CuTiH = o (decomp. for x > 0.2) - 38 [6.169]
CuTi2H~o (decomp. for x>0.2) -33 [6.169]
CuZr2H 1 (T> 1000 K disprop?) -60 [6.1701
CuZr2Ho.2 (decomp. at higher conc.?) - 74 [6.171]
DyFe2H2 -29 [6.172]
DyFeaH2 -24 [6.172]
DyRu2H3.1 - 2 8 calor [6.145]
ErFe2HI.5 -29 [6.173]
ErFcaH2.s -22 [6.172]
Er6Fe23H 12 -25 [6.1741
ErMn2H4.9 ( < --27) [6.153]
ErNiH3.x --52 calor [6.175]
ErRuzH... --23 [6.87]
EuNisH~.s (ex. prob. more stable EuNisH~ t.2) - 13 [6.176]
Fe3GdHI.5 -25 [6.172]
Fe~GdH2.7 (decomp. at first desorption) (-15) [6.73, 173]
Fe2Hf no hydride [6.29]
FeHfaH3.1 (< -36) [6.142]
Fe3HoHI,5 -22 [6.172]
Fe23Ho6H12 -26 [6.174]
Fe23Lu6Hs -20 [6.1741
FePd3H~o.a - 15 estim [6.197]
Fe2ScH2 -15 [6.196]
FezScH . -18 [6.87]
Fe3TbHl.s -24 [6.172]
FesThH2.4 no plateau at 313 K [6.177]
FevThzH6A - - 22 estim [6.177]
Fe/Ti no hydride [6.178]
FeTi H ~,o +10 [6.179]
272 R. Griessen and T. Riesterer

Table 6.6.2 (continued)

Hydride A H [kJ/mol H] Ref.

FeTiH t - 14 [6.165]
FeTiHI.6 - 17 [6.165]
FeTiHt.6 - 12 calor [6.179]
Fe2TmH t.6 -- 29 [6.180]
F'eU6H3.2 -- 32 estim [6.29]
Fe23Y6H21.5 ( < -- 36) [6.150]
Fe3YH4. 8 ( < - 36) [6.150]
Fe2YH z (decomp. at first desorption) [6.173]
Ga2PrH=o.3 [6.144]
GdMn2H~.v (-44) [6.73]
GdNisH2. 9 -- 13 estim [6.86]
GdNiaH a -45 [6.73]
GdRhzH2.8 - 25 [6.73]
GdRu2Hz. 7 - 30 [6.73]
HfNiH3.2 --26 estim [6.142]
Hf2PdH1.9 [6.181]
HfPd no hydride [6.181]
HfzPtHo.9 [6.29]
HfPt no hydride [6.29]
HfzRhH2. 2 [6.144]
HfRh no hydride [6.29]
IrzLa no hydride [6.29]
La3NiH z -- 111 [6.182]
La3NiHs.s (probably decomp.) [6.183]
LavNi3H... (decomp.) [6.184]
LaNiH3. s - 52 calor [6.184]
LaNiHx.3 - 88 [6.182]
LaNiH4 -63 calor [6,175]
LaNiEH z --27 [6.184]
LaNisHs. 5 - 16 [6.86]
LaNisHs. 5 -- 15 [6.186]
LaNisHs.s - !6 calor [6.146]
LaNisHs.5 -15 calor [6.167, 185]
LaPtH2.s [6.153]
LaPt z no hydride (?) [6.29]
LaPtsHI.2 -- 13 estim [6.55]
L a R h z H 1.4 ( - 22) [6.73]
LaRuzH4.5 very stable [6.73]
LiPdHo. s - 38 [6.187]
LiPtHo. 7 - 67
- [6.187]
LuNiH2.9 - 50 calor [6.175]
LuPdH... > - 14 estim [6.175]
MgzNiHo.o6 - 13 [6.188]
Mg/NiH4 - 32 [6.165]
Mn2ScH... - 32 [6.87]
Mnl.sTiH2., ~ - 16 [6.102]
Mn2.aZrH2 - 18 [6.190]
Mn2.sZrH.. ' - 16 calor [6.167]
M n / Z r H 3.6 -- 22 [6.189]
M n 2 Z r H .. - 2 0 ealor [6.167]
 Heat of Formation Models 273

Table 6.6.2 ( c o n t i n u e d )

Hydride A H [kJ/mol H] Ref.

M°zZrH ~o - 19 [6.166]
NdNisH 4 - 14 [6,15]
NisPrH 6 - 15 [6.15/86]
NiTPr2H... [6.151]
NisSmH1. 6 - 13 estim [6.86]
N i s S m H 1.6 - 15 estim [6.191]
NiThH3, 5 -26 estim [6.•77]
NiTiHo. 9 - 30 [6.160]
NiTi2H1 - 31 estim [6.192]
NisYH 1 - 12 estim [6.55]
NiTYzH a --20 estim [6.150]
Ni3YH 4 -23 estim [6.150]
NisYbH3 - 13 estim [6.191]
Ni2YbH3.t -26 calor [6.145]
NiYbH2.7 -- 73 calor [6.175]
N i s Z r 2 H 1.~ -- 22 [6.193]
NiloZr7H~ 5 - 24 [6.193]
NiZrH2. s - 33 estim [6.142]
NiZr2H4. 5 (decomp.) -81 estim [6.142]
NiZr2Ho.a5 (decomp. at h i g h e r conc.?) -92 [6.171]
PTi3H o -- 43 [6.194]
P d T i 2 H 1.5 - 46 [6.169]
Pd3U no hydride [6.117]
PdYbH2.7 - 33 calor [6.175]
Pd2Zr no hydride [6.181]
PdZrzH2.7 [6.181]
S b T i 3 H ~.o - 55 [6.194]
SnTi3H=o -51 [6.143]
T a V 2 H ~-o (C15 struct,) - 58 [6.195]
V2ZrH ~ o - 77 [6.166]
V 2ZrH 1.5 - 80 [6.166]
 274 R. Griessen a n d T. Riesterer

-~.o
"0

IH

n:

,.t3 . ~
c~

~.o oo
on~ o o
,.0
Z~
Ill

III

.z:,~

I11

~.o o
;>
[1[
E l . -~ ~ ~.-

o~ oo~ ~H
.~ .oo~ .z:
tl t t I
~o -~ "- "n '~

Ill
II t I III
 Heat of Formation Models 275

ill

Ill
II

b~
III
II I

III

,.d
Ill
I

Z
fll

Hi

III
%

0 Ill

~z
Ill
I I I I I 111

o
o'3
o
rrr
r..)
276 R. Griessen and T. Riesterer

111

II
I I I I

III
I II I I

III
I I II I II

HI
'm
I II I III

,..o
Z
"m
I l

III
'm
I II I Ill

HI
I II II I1~

0 III
I II

u~

II II II I

~H
c~

II II II III
 Heat of Formation Models 277

III
I II I

III

III

III
III I Ill

III

z
III
I

III
t III II II

>-
III
I III II I I I I I

c- >-
o III

~z
III
I I l l l l l II

¢9

HI
I I I I I I I II Ill
278 R. Griessen a n d T. Riesterer

I II I

II I I I II

~r~
t ~111 I I I II

r~

I I I I f

I I I II I

~A
I Itll I II I I

m
 Heat of Formation Models 279

Acknowledgements. We are grateful to Dr. A. Driessen, R. Feenstra, D. G. de Groot, and S.

Bfichler for valuble discussions and their critical and careful reading of the manuscript. One of
the authors (T.R.) would like to acknowledge financial support from the National Energy
Research Foundation (NEFF).

References
6.1 C.D. Gelatt, H. Ehrenreich, J.A. Weiss: Phys. Rev. B 17, 1940 (1978)
6.2 C.D. Gelatt: "Calculated Heats of Formation of Metal and Metal Alloy Hydrides", in
Hydrides for Energy Storage, ed. by A.F. Andresen, J.A. Maeland (Pergamon, Oxford
1978) p. 193
6.3 J.K. N~rskov: Phys. Rev. B26, 2875 (1982); B28, 1138 (1983)
6.4 M. Methfessel, J. Kiibler: J. Phys. F 12, 141 (1982)
6.5 M. Gupta, L. Schlapbach: Chap. 5 this volume and references therein
6.6 A.R. Williams, J. Kiibler, C.D. Gelatt: Phys. Rev. B 19, 6094 (1979)
6.7 A.C. Switendick: "The Changes in Electronic Structure on Hydrogen Alloying and
Hydride Formation" in Topics Appl. Phys., Vol. 28 (Springer, Berlin, Heidelberg 1978)
p. 101
6.8 D.A. Papaconstantopoulos, E.N. Economou, B.M. Klein, L.L. Boyer: Phys. Rev. B 20,
177 (1979)
6.9 M. Gupta: Phys. Lett. 88A, 469 (1982); J. Less-Common Met. 88, 221 (1982)
6.10 D.A. Papaconstantopoulos, A.C. Switendick: J. Less-Common Met. 88, 273 (1982)
6.11 W.M. Temmermann, A.J. Pindor: J. Phys. F 13, 1869 (1983)
6.12 M. Gupta: J. Less-Common Met. 103, 325 (1984)
6.13 M. Gupta, E. Belin, L. Schlapbach: J. Less-Common Met. 103, 389 (1984)
6.14 M.H. Mendelsohn, D.M. Gruen, A.E. Dwight: Nature 269, 45 (1977)
6.15 C.E. Lundin, F.E. Lynch, C.B. Magee: J. Less-Common Met. 56, 19 (1977)
6.16 C.B. Magee, J. Liu, C.E. Lundin: J. Less-Common Met. 78, 119 (1981)
6.17 T. Takeshita, K.A. Gschncidner, D.K. Thome, O.D. McMasters: Phys. Rev. B 21, 5636
(1980)
6.18 H.M. Lee: J. Mater. Sci. 13, 1374 (1978)
6.19 H.M. Lee: J. Mater. Sci. 14, 1002 (1979)
6.20 D.O. Welch, M.A. Pick: Phys. Lett. 99A, 183 (1983)
6.21 I. Jacob, A. Wolf, M.H. Mintz: Solid State Commun. 40, 877 (1981)
6.22 Wei-Ming Yei, R.B. McLellan: J. Less-Common Met. 64, P l l (1979)
6.23 M. Peretz, D. Zamir, D. Shaltiel, J. Shinar: Z. Phys. Chem. N F l l 7 , 221 (1979)
6.24 A.R. Miedema: J. Less-Common Met. 32, 117 (1973)
6.25 A.R. Miedema, K.H.J. Buschow, H.H. van Mal: J. Less-Common Met. 49, 463 (1976)
6.26 H.H. van Mal, K.H.J. Buschow, A.R. Miedema: J. Less-Common Met. 35, 65 (1974)
6.27 A.R. Miedema, F.R. de Boer, R. Boom: Physica 103B, 67 (1981)
6.28 P.C.P. Bouten, A.R. Miedema: J. Less-Common Met. 71, 147 (1980)
6.29 K.H.J. Busehow, P.C.P. Bouten, A.R. Miedema: Rep. Prog. Phys. 45, 937 (1982)
6.30 A.R. Williams, C.D. Gelatt, V.L. Moruzzi: Phys. Rev. Lett. 44, 429 (1980)
6.31 D.G. Pettifor: Phys. Rev. Lett. 42, 846 (1979)
6.32 A.R. Williams, C.D. Gelatt, V.L. Moruzzi: Phys. Rev. B25, 6509 (1982)
6.33 A.R. Mackintosh, O.K. Andersen: "The Electronic Structure of Transition Metals" in
Electrons at the Fermi Surface, ed. by M. Springford (Cambridge Univ. Press, London
1980) p. 149
6.34 O.K. Andersen; H.L. Skriver, H. Nohl, B. Johansson: Pure Appl. Chem. 52, 93 (I979)
6.35 R. Griessen, A. Driessen: Phys. Rev. B30, 4372 (1984)
6.36 R. Griessen, A. Driessen, D.G. de Groot: J. Less-Common Met. 103, 235 (1984)
6.37 R. Griessen, A. Driessen: J. Less-Common Met. 103, 245 (1984)
280 R. Griessen and T Riesterer

6.38 R. Griessen, R. Feenstra: J. Phys. FI5, 1013 (1985)

6.39 D.G. Westlake: "Stabilities, Stoichiometries and Site Occupancies in Hydrides of
lntermetallic Compounds" in Electronic Structure and Properties of Hydrogen in
Metals, ed. by P. Jena, C.B. Satterthwaite (Plenum, New York 1983) p. 85; J. Less-
Common Met. 91, 1 (1983)
6.40 W.A. Oates, T.B. Flanagan: Prog. Solid St. Chem. 13, 193
6.41 H. Wenzl: Int. Met. Rev. 27, 140 (1982)
6.42 G.C. Carter, F.L. Carter: "Metal Hydrides for Hydrogen Storage: A Review of
Theoretical and Experimental Research and Critically Compiled Data" in
Metal-Hydrogen Systems, ed. by T.N. Veziroglu (Pergamon, Oxford 1981) p. 503
6.43 V. Diatschenko, C.W. Chu: Science 212, 1393 (1982)
6.44 H. Peisl: "Lattice Strains due to Hydrogen in Metals" in Topics Appl. Phys., Vol. 28
(Springer, Berlin, Heidelberg 1978) p. 53
6.45 Y. Fukai: Jpn. J. Appl. Phys. 22, 207 (1983)
6.46 A. Fukizawa, Y. Fukai: J. Phys. Soc. Jpn. 52, 2102 (1983)
6.47 E.G. Ponyatowskii, V.E. Antonov, I.T. Belash: Usp. Fiz. Nauk 137, 663 (1982) [English
transl.: Phys. Usp. 25, 596 (1982)]
6.48 H. Hemmes, A. Driessen, R. Griessen: J. Phys. C 19, 3571 (1986)
6.49 D.G. Westlake: J. Less-Common Met. 90, 251 (1983)
6.50 T.B. Flanagan: J. Less-Common Met. 63, 209 (1979)
6.51 T.B. Flanagan, J.F. Lynch: J. Phys. Chem. 79, 444 (J975)
6.52 E. Wicke, J. Blaurock: Ber. Bunsengesell. Phys. Chem. 85, 1091 (1981)
6.53 E. Wicke, H. Brodowsky: "Hydrogen in Palladium and Palladium Alloys" in Topics
Appl. Phys., Vol. 29 (Springer, Berlin, Heidelberg 1978) p. 73
6.54 A. Driessen, H. Hemmes, R. Griessen: Z. Phys. Chem. NFI43, 145 (1985)
6.55 T. Takeshita, K.A. Gschneidner, J.F, Lakner: J. Less-Common Met. 78, P43 (1981)
6.56 P. Villars: J. Less-Common Met. 92, 215 (1983)
6.57 T. Kuji, W.A. Oates, B.S. Bowerman, T.B. Flanagan: J. Phys. F13, 1785 (1983)
6.58 R. Feenstra, R. Griessen, D.G. de Groot: J. Phys. F16, 1933 (1986)
6.59 H. Sugimoto, Y. Fukai: Phys. Rev. B22, 670 (1980)
6.60 H. Sugimoto, Y. Fukai: J. Phys. Soc. Jpn. 50, 3709 (198l)
6.61 H. Sugimoto, Y. Fukai: J. Phys. Soc. Jpn. 51, 2554 (1982)
6.62 Y. Fukai: J. Less-Common Met. 101, 1 (1984)
6.63 D.G. Westlake: J. Less-Common Met. 91, 275 (1983)
6.64 D.G. Westlake: Scripta Met. 16, 1049 (1982)
6.65 D.G. Westlake: J. Mater. Sci. 18, 605 (1983)
6.66 D.G. Westlake: J. Less-Common Met. 75, 177 (1980)
6.67 D.G. Westlake: J. Less-Common Met. 88, 17 (1982)
6.68 D.G. Westlake: Mat. Lett. 1, 107 (1982)
6.69 D.G. Westlake: J. Chem. Phys. 79, 4532 (1983)
6.70 D.G. Westlake: J. Mater. Sci. 19, 316 (1984)
6.71 D.G. Westlake: Mat. Res. Bull. 18, 1409 (1983)
6.72 D.G. Westlake: J. Less-Common Met. 105, 69 (1985)
6.73 I. Jacob, D. Shaltiel: J. Less-Common Met. 65, 117 (1979)
6.74 J. Shinar, I. Jacob, D. Davidov, D. Shaltiel: "Hydrogen Sorption Properties in Binary
and Pseudobinary Intermetallic Compounds" in Hydrides for Energy Storage, ed. by
A.F. Andresen, A.J. Maeland (Pergamon, Oxford 1978) p. 337
6.75 J.J. Didisheim: "Structure et Stabilit6 des Hydrures m&alliques ternaire ZrVzH~ et
ZrMn2Hx", Thesis No. 2013, University of Geneva, Switzerland, 1981
6.76 I. Jacob, J.M. Bloch, D. Shaltiel, D. Davidov: Solid State Commun. 35, 155 (1980)
6.77 J.J. Didisheim, K. Yvon, D. Shaltiel, P. Fischer: Solid State Commun. 31, 47 (1979)
6.78 T. Riesterer: J. Less-Common Met. 103, 219 (1984)
6.79 J.R. Lacher: Proc. R. Soc. (London) A 161, 525 (1937)
6.80 R. Kirchheim, F. Sommer, G. Schluckebier: Acta Metall. 30, 1059 (1982)
6.81 R. Kirchheim: Acta Metall. 30, 1069 (1982); Acta Metall. 29, 845 (1981)
 Heat of Formation Models 281

6.82 R. Griessen: Phys. Rev. B27, 2069 (1983)

6.83 P.C.P. Bouten, A.R. Miedema: J. Less-Common Met. 65, 217 (1979)
6.84 A.K. Niessen, F.R. de Boer, R. Boom, P.F. de Ch~,tel, W.C.M. Mattens, A.R. Miedema:
CALPHAD 7, 51 (1983)
6.85 D.G. Pettifor: "Electron Theory of Metals" in Physical Metallurgy Part, I, ed. by R.W.
Cahn, P. Haasen (North-Holland, Amsterdam 1983) p. 73
6.86 H.H. van Mal: Philips Res. Repts. Suppl. 1 (1976)
6.87 A.L. Shilov, M.E. Kost, N.T. Kuznetsov: J. Less-Common Met. 105, 221 (1985)
6.88 A. Pasturel, C. Chatillon-Colinet,A. Percheron-Gu+gan, J.C. Achard: J. Less-Common
Met. 84, 73 (1982)
6.89 A. Pasturel, F. Liautaud, C. Colinet, C. Allibert, A. Percheron-Gu6gan, J.C. Achard: J.
Less-Common Met. 96, 93 (1984)
6.90 A. Percheron-Gu6gan, C. Lartigue, J.C. Achard: J. Less-Common Met. 109, 287 (1985)
6.91 R. Griessen, L.M. Huisman: "Electronic Structure of Metal Hydrides and Deuterides
from De Haas-Van Alphen Measurements" in Electronic Structure and Properties of
Hydrogen in Metals, ed. by P. Jena, C.B. Satterthwaite (Plenum, New York 1983) p. 235
6.92 H.L.M. Bakker, R. Feenstra, R. Griessen, L.M. Huisman, W.J. Venema: Phys. Rev.
B26, 5321 (1982)
6.93 B.M. Geerken, R. Griessen, L.M. Huisman, E. Walker: Phys. Rev. B26, 1637 (1982)
6.94 D.K. Hsu, R.G. Leisure: Phys. Rev. B20, 1339 (1979)
6.95 W. Joss, R. Griessen, E. Fawcett: "Electron States and Fermi Surfaces of Homoge-
neously Strained Metallic Elements" in Landolt-Bdrnstein New Series', Metals: Vol. J2:
Phonon States, Electron States and Fermi Surfaces, Subvol. b; ed. by K.H. Hellwege,
J.L. Olsen (Springer, Berlin, Heidelberg 1983) p. 1
6.96 V. Heine: "Electronic Structure of Metals" in The Physics of Metals I. Electrons, ed. by
J.M. Ziman (Cambridge Univ. Press, Cambridge 1969) p. 1
6.97 M. Cyrot, F. Cyrot-Lackmann: J. Phys. F 6, 2257 (1976)
6.98 E.S. Machlin: J. Less-Common Met. 64, 1 (1979)
6.99 M. Futran, S.G. Coats, C.K. Hall: J. Chem. Phys. 77, 6223 (1982)
6.100 M.A. Pick, D.O. Welch: Z. Phys. Chem. N F l l 4 , 37 (1979)
6.101 W. Fenzl, J. Peisl: Phys. Rev. Lett. 54, 2064 (1985)
6.102 R. Griessen: Phys. Rev. B (1988)
6.103 V.L. Moruzzi, J.F. Janak, A.R. Williams: Calculated Electronic Properties of Metals
(Pergamon, New York 1978)
6.104 E. Salomons, R. Griessen: to be published
6.105 R.Griessen: "Hydrogen in Disordered Solids: Model and Calculations" in Hydrogen in
Disordered and Amorphous Solids, ed. by G. Bambakidis, R.C. Bowman Jr., NATO ASI
Series B: Physics Vol. 136 (Plenum, New York 1986) p. 153
6.106 R. Feenstra, D.G. de Groot, J.H. Rector, E. Salomons, R. Griessen: J. Phys. F 16, 1953
(1986)
6.i07 R. Feenstra, R. Griessen, D.G. de Groot: J. Phys. F 16, 1933 (1986)
6.108 R. Feenstra, B.M. Geerken, D.G. de Groot, R. Griessen: J. Phys. F 16, 1965 (1986)
6.109 R. Feenstra, R. Griessen: Solid State Commun. 59, 905 (1986)
6.110 J.J. Reilly: Z. Phys. Chem. N F l l 7 , 155 (1979)
6.111 R. Wiswall: "Hydrogen Storage in Metals" in Topics Appl. Phys., Vol. 29 (Springer,
Berlin, Heidelberg 1978) p. 201
6.112 M.J. Puska, R.M. Nieminnen: Phys. Rev. B29, 5382 (1984); 32, 1353 (1985)
6.113 M.J. Puska, R.M. Nieminnen: Physica 127B, 417 (1984)
6.114 K.M. Ho, H.J. Tao, X.Y. Zhu: Phys. Rev. Lett. 53, 1586 (1984)
6.115 G. Busch, L. Schlapbach, A. Sciler: "The plateau pressure of RENi5 and RECos
hydrides" in Hydrides for Energy Storage, ed. by A.F. Andresen, J.A. Maeland
(Pergamon, Oxford 1978) p. 293
6.116 V.L. Moruzzi, J.F. Janak, A.R. Williams: Calculated Electronic Properties of Metals
(Pergamon, New York 1978)
282 R. Griessen and T. Riesterer

6.117 R. Feenstra, D.G. de Groot, R. Griessen, J.P. Burger, A. Menowsky: J. Less-Common

Met. 130, 375 (1987)
6.118 E. Fromm, G. H6rz: Intern. Metals Rev. 25, 269 (1980)
6.119 W.M. Mueller, J.P. Blackledge, G.G. Libowitz: Metal Hydrides (Academic, New York,
London 1968)
6.120 H. Behrens, G. Ebel (eds.): "Gases and Carbon in Metals", in Physics Data Series 5
(Fachinformationszentrum Energie, Physik, Mathematik, Karlsruhe)
6.121 G.C. Sinke, L.C. Walker, F.L. Oetting, D.R. Stull: J. Chem. Phys. 47, 2759 (1967)
6.122 J.W. Ward, J. Less-Common Met. 93, 279 (1983)
6.123 E. Fromm, E. Gebhardt (eds.): Gase und Kohlenstoff in Metallen (Springer, Berlin,
Heidelberg 1976)
6.124 G.G. Libowitz, A.J. Maeland: "Hydrides" in Handbook on the Physies and Chemistry of
Rare Earths Vol. 3, ed. by K.A. Gschneidner, L. Eyring (North-Holland, Amsterdam
1979) p. 299
6.125 H.M. Lee: Metallurgical Trans. 7A, 431 (1976)
6.126 R.C. Frank, J.E. Thomas: J. Phys. Chem. Solids 16, 144 (1960)
6.127 R.B. McLellan, W.A. Oates: Acta Metall. 21, 181 (1973)
6.128 E. Veleckis: J. Nucl. Mater. 79, 20 (1979)
6.129 P.R. Subramanian, J.F. Smith: J. Less-Common Met. 87, 205 (1982)
6.130 E. Veleckis, R.K. Edwards: J. Phys. Chem. 73, 683 (1969)
6.131 C. Papastaikoudis, B. Lengcler, W. Jfiger: J. Phys. F 13, 2257 (1983) and rcfs. therein
6.132 M.H. Mintz, Z. Hadari, M. Bixon: J. Less-Common Met. 48, 183 (1976)
6.133 J.M. Haschke, J.W. Ward, W. Bartscher: J. Less-Common Met. 107, 159 (1985)
6.134 C.M. Smith, A.E. Hodges, J.M. Haschke, F.L. Oetting: J. Chem. Thermodyn. 14, 115
(1982)
6.135 N.A. Galaktionowa: Hydrogen Metal Systems Databook (Ordentlich, Holon Israel
1980)
6.136 H. Uchida, Y.C. Huang, M. Tada, K. Fujita: Ber. Bunsenges. phys. Chemie 114, 51
(1979)
6.137 D.T. Peterson, S.O. Nelson: J. Less-Common Met. 72, 251 (1980)
6.138 M. Nagasaka, T. Yamashina: J. Less-Common Met. 45, 53 (1976)
6.139 W.J. Arnoult, R.B. Mc Lellan: Acta Metall. 21, 1397 (1973)
6.140 K. Papathanassopoulos, H. Wenzl: J. Phys. F 12, 1369 (1982)
6.141 V.G. Mogilatenko, D.F. Chernega, K.I. Vashchenko: Soy. Non-ferrous Mctals Res. 9,
346 (1981)
6.142 R.M. van Essen, K.H.J. Buschow: J. Less-Common Met. 64, 277 (1979)
6.143 P.S. Rudmann, J.J. Reilly, R.H. Wiswall: Ber. Bunsenges. phys. Chemic 82, 611 (1978)
6.144 A.J. Maeland: J. Less-Common Met. 89, 173 (1983)
6.145 H. Oesterreicher, K. Ensslen, E. Bucher: Appl. Phys. 22, 303 (1980)
6.146 J.J. Murray, M.L. Post, J.B. Taylor: J. Less-Common Met. 80, 201 (1981)
6.147 J.J. Murray, M.L. Post, J.B. Taylor: J. Less-Common Met. 90, 65 (1983)
6.148 G.D. Sandrock, J.J. Murray, M.L. Post, J.B. Taylor: Mater. Res. Bull. 17, 887 (1982)
6.149 G.D. Sandrock: "Development of Low Cost Nickel-Rare Earth Hydrides for Hydrogen
Storage" in Hydrogen Energy Systems, ed. by T.N. Veziroglu, W. Seifritz (Pergamon,
Oxford t 978) p. 1625
6.150 R.H. van Essen, K.H.J. Buschow: J. Less-Common Met. 70, 189 (1980)
6.151 A. Goudy, W.E. Wallace, R.S. Craig, T. Takeshita: "Thermodynamics and Kinetics of
Hydrogen Absorption in R2Co7,RCo3,and RFe3 Compounds" in Adv. in Chemistry
Series, Vol. 167 (American Chem. Soc., Washington D.C. 1978) p. 312
6.152 F.A. Kuijpers: Philips Res. Repts. Suppl. 1973 No. 2
6.153 K.H.J. Buschow: "Hydrogen Absorption in lntermetallic Compounds" in Handbook on
the Physics and Chemistry of Rare Earth, Vol. 6, ed. by K.A. Gschneidner, L. Eyring
(North-Holland, Amsterdam 1984) p. 1
6.154 H.A. Kierstead: J. Less-Common Met. 73, 61 (1980)
 Heat of Formation Models 283

6.155 H.A. Kierstead: "Thermodynamics and Structural Properties of the Rare Earth-Co 3
Hydrides" in The Electronic Structure and Properties of Hydrogen in Metals, ed. by P.
Jena, C.B. Satterthwaite (Plenum, New York 1983) p. 103
6.156 H.A. Kierstead: J. Less-Common Met. 96, 133 (1984)
6.157 J. Clinton, H. Bittner, H. Oesterreicher: J. Less-Common Met. 41,187 (1975)
6.158 Y. Osumi, H. Susuki, A. Kato, K. Oguro, M. Nakone: J. Less-Common Met. 74, 271
(1980)
6.159 M. Someno, M. Arita, R. Kinaka, Y. Ichinose: Suppl. Trans. Jap. Inst. Metals 21,325
(1980)
6.160 R. Bureh, N.B. Mason: J. Chem. Soc. Faraday Trans. I 75, 561 (1979)
6.161 H.A. Kierstead: J. Less-Common Met. 86, L5 (1982)
6.162 S.J.C. Irvine, I.R. Harris: "The Effect of Induced Disorder on the Hydrogenation
Behaviour of the Phase ZrCo" in Hydrides for Energy Storage, ed. by A.F. Andresen,
A.J. Maeland (Pergamon, Oxford 1978) p. 431
6.163 J.R. Johnson: J. Less-Common Met. 73, 345 (1980)
6.164 J.F. Lynch, J.R. Johnson, J.J. Reilly: Z. Phys. Chemie NF 117, 229 (1979)
6.165 J.J. Reilly: "Synthesis and Properties ofUseful Metal Hydrides" in Hydrides for Energy
Storage, ed. by A.F. Andresen, A.J. Maeland (Pergamon, Oxford 1978) p. 301
6.166 A. Pebler, E.A. Gulbransen: Trans. Met. Soc. AIME 239, 1593 (1976)
6.167 A.T. Pedziwiatr, R.S. Craig, W.E. Wallace, F. Pourarian: J. Solid State Chem. 46, 336
(1983)
6.168 A.J. Maeland: "Preparation and Properties of TiCuH" in Adv. in Chemistry Series,
Vol. 167 (American Chem. Soc., Washington D.C. 1978) p. 302
6.169 R. Kadel, A. Weiss: Ber. Bunsenges. phys. Chemie 82, 1290 (1978), extracted from Figs.
11 and 13
6.170 R. Kadel, A. Weiss: J. Less-Common Met. 65, 89 (1979)
6.171 A. Pebler, E.A. Gulbransen: Electrochem. Technology 4, 211 (1966)
6.172 C.A. Bechmann, A. Goudy, T. Takcshita, W.E. Wallace, R.S. Craig: lnorg. Chem. 15,
2184 (1976)
6.173 H.A. Kierstead: J. Less-Common Met. 86, L1 (1982)
6.174 H.K. Smith, W.E. Wallace, R.S. Craig: J. Less-Common Met. 94, 89 (1983)
6.175 K. Ensslen, E. Bucher, H. Oesterreicher: J. Less-Common Met. 92, 343 (1983)
6.176 Z. Gavra, J.J. Murray, L.D. Calvert, J.B. Taylor: J. Less-Common Met. 105, 291 (1985)
6.177 K.H.J. Buschow, H.H. van Mal, A.R. Miedema: J. Less-Common Met. 42, 163 (1975)
6.178 J.J. Reilly, R.H. Wiswall: Inorg. Chem. 13, 218 0974)
6.179 H. Wenzl: J. Less-Common Met. 74, 351 (1980)
6.180 H.A. Kierstead: J. Less-Common Met. 85, 213 (1982)
6.181 A.J. Maeland, J.J. Libowitz: J. Less-Common Met. 74, 295 (1980)
6.182 D.H.W. Carstens: J. Less-Common Met. 61, 253 (1978)
6.183 V.I. Mikhewa, M.E. Kost, A.L. Shilov: Russ. J. Inorg. Chem. 23, 657 (1978)
6.184 G. Busch, L. Schlapbach, Th. yon Waldkirch: "Hydrides of Rare Earth-Nickel
Compounds: Structure and Formation Enthalpies" in Hydrides for Energy Storage, ed.
by A.F. Andresen, A.J. Maeland (Pergamon, Oxford 1978) p. 287
6.185 B.S. Bowerman, C.A. Wulff, T.B. Flanagan: Z. Phys. Chemie NF 116, 197 (1979)
6.186 D. Ohlendorf, H.E. Flotow: J. Less-Common Met. 73, 25 (1980)
6.187 B. Naehen, W. Bronger: J. Less-Common Met. 52, 323 (1977)
6.188 S. Ono, K. Imanari, K. Nomura, H. Sato: "Hydrogen Solubility in Mg2Ni" in Hydrogen
Energy Progress V,, ed. by T.N. Veziroglu, J.B. Taylor (Pergamon, New York 1984)
p. 1405
6.189 Y. Machida, T. Yamadaya, M. Asanuma: "Hydride Formation of C14-Type Ti Alloys"
in Hydrides for Energy Storage, ed. by A.F. Andresen, A.J. Maeland (Pergamon, Oxford
1978) p. 329
6.190 F. Pourarian, V.K. Shinha, W.E. Wallace, A.T. Pedziwiatr, R.S. Craig: "Hydrogenation
Entropies of the ZrMn2+y System" in The Electronic Structure and Properties of
Hydrogen in Metals, ed. by P. Jena, C.B. Satterthwaite (Plenum, New York 1983) p. 385
284 R. Griessen and T. Riesterer. Heat of Formation Models

6.191 J.L. Anderson, T.C. Wallace, A.L. Bowman, C.L. Radosewich, M.L. Courtney, Los
Alamos Rept. LA-5320-MS (1973)
6.192 H. Buchner, M.A. Gutjahr, K.-D. Beccu, H. S~iufferer: Z. Metallkde. 63, 497 (1972)
6.193 F.H.M. Spit, J.W. Drijver, W.C. Turkenburg, S. Radelaar: "Thermodynamics and
Kinetics of Hydrogen Absorption in Amorphous NiZr-Alloys" in Metal Hydrides, ed.
by G. Bambakidis (Plenum, New York, London 1981) p. 345
6.194 K.V.S. Rama Ran, M. Mrowietz, A. Weiss: Ber. Bunsenges. phys. Chemie 86, 1135
(1982)
6.195 J.F. Lynch: J. Phys. Chem. Solids 42, 411 (1980)
6.196 K.N. Semenenko, R.A. Sirotina, A.P. Savchenkova, V.V. Burnasheva, M.V. Lototskii,
E.E. Fokina, S.L. Troitskaya, V.N. Fokin: J. Less-Common Met. 106, 349 (1985)
6.197 T.B. Flanagan, S. Majchrzak, B. Baranowski: Philos. Mag. 25, 257 (1972)
6.198 J.W. Ward, W. Bartscher, J. Rebizant: J. Less-Common Met. 130, 431 (1987)

Additional References
Baranowski, B., Hochheimer, H.D., Str6ssner, K., H6nle, W.: "High Pressure X-ray
Investigation of AIH3 and A1 at Room Temperature", J. Less-Common Met. 113, 341
(1985)
Fukai, Y.: "Atomistic and Electronic Approaches to Hydrogen in Metals", Cryst. Latt. Def.
and Amorph. Mat. 11, 85 (1985)
Nordlander, P., Norskov, J.K., Besenbacher, F.: "Trends in Hydrogen Heats of Solution and
Vacancy Trapping Energies in Transition Metals", J. Phys. F 16, 1161 (1986)
Riesterer, T.: "Electronic Structure and Bonding in Metal Hydrides, Studied with Photo-
electron Spectroscopy", Z. Phys. B 66, 441 (1987)
Salomon, E., Griessen, R., de Groot, D.G.: "Surface Tension and Subsurface Sites of Metallic
Nanocrystals determined from H Absorption", Europhys. Lett. (1988)
Watanabe, K., Fukai, Y.: "Calorimetric Studies of the Behaviour of Hydrogen in Vanadium
and Vanadium Alloys", J. Phys. Soc. Japan 54, 3415 (1985)
7. Magnetic Properties,
Miissbauer Effect and Superconductivity

Gerfried Wiesinger and Gfinter Hilscher

With 20 Figures

Investigations of the bulk magnetic properties, of M6ssbauer spectroscopy-

and superconductivity studies on hydrides of intermetallic compounds,
predominately of those containing a rare earth metal, are reviewed. The chapter
begins with some introductory remarks dealing with concepts of magnetism
and of M6ssbauer spectroscopy. Experimental results covering both fields are
then discussed with reference to the compound's stoichiometry. Although
strictly speaking they do not belong to the category of "intermetallic
compounds", hydrides of ternary oxides (oxygen stabilized intermetallics) and
of amorphous alloys are covered too. The present review closes with a section
concerning the hydrogen induced change of the superconducting properties of
crystalline compounds and amorphous alloys.

7.1 Introduction and Scope

When studying the bulk magnetic properties and the hyperfine interactions by
magnetisation- and susceptibility measurements and by M6ssbauer spec-
troscopy respectively, one has two experimental techniques (macroscopic and
microscopic respectively) which complement one another and form an ideal
combination to elucidate a great many physical and chemical properties of
intermetallic hydrides such as the chemical nature of hydrogen and its location,
the formation and the stability of various structural phases and the influence of
hydrogen absorption on electronic, magnetic, and superconducting properties
of the host lattice. Moreover, from studying the hydrides, conclusions may
frequently be drawn upon properties of the parent system, particularly in case
of the electronic structure and the magnetic interaction mechanism. Thus such
studies have attracted considerable attention not only in terms of basic research
but also from the point of view of applications.
Several review papers describing magnetism and/or the M6ssbauer effect in
hydrides already have appeared in the last decade, among which the two
M6ssbauer review articles by Wortmann [7.1] and by Wagner and Wortmann
[7.2] covered predominantely binary metal-hydrogen systems. Selected rare
earth (R)--transition metal hydrides have been considered in the papers by
Wallace [7.3-8], where emphasis was laid on the bulk magnetic properties and
hydrogen was assumed to act generally as an electron aceeptor (anionic model).
Shenoy et al. [7.9, 10] reviewed M6ssbauer results obtained in particular from
286 G. Wiesinger and G. Hilscher

hydrogen-storage materials. In his paper Cohen [7.11] stressed M6ssbauer

studies performed on rare-earth nuclei. In several other reviews the present
topic has only been touched upon, e.g. those of Oesterreicher [7.12], Cohen and
Wernick [7.13], Kuijpers [7.14], and Buschow [7.15]. In the late seventies
Buschow and co-workers [7.16-18] made another attempt to find, apart from
Wallace's suggestions, a general interpretation of the hydrogen induced
changes of the magnetic properties in R-transition metal compounds. Besides
the alteration in the interatomic distances - especially important in case of the
Mn compounds - the importance of differences in the electronegativity for Fe,
Co, and Ni compounds has been emphasized. Finally two comprehensive
reviews from Buschow et al. [7.19] and Buschow [7.203 have appeared, where
alongside formation of ternary hydrides, their thermodynamic and their
chemical properties, magnetic features and M6ssbauer spectroscopy are also
included; however, non-R metal compounds are not considered. Attempts have
been made to interpret several properties of the hydrides in terms of Miedema's
atomic cell model.
As can be seen from the reference list, the number of papers on this subject
has continued to grow in the last few years, and thus it seems justified to review
the topic again. Moreover the present article is arranged in a way differing from
the previous reviews and furthermore considers new materials with outstand-
ing physical properties (R2Fe14B-type compounds and amorphous alloys). We
have deliberately made no separationbetween results obtained from magnetic
and from M6ssbauer measurements since the intentions as well as the outcome
of many investigations often prevent an unambiguous assignment to a certain
topic. The chapter dealing with superconductivity remained rather short since
this subject, although of great importance in binary metal-hydrogen systems
(see e.g. the review by Stritzker and Wz~hl [7.21], has been hardly investigated
for ternary systems.
The present chapter begins with an introductory section discussing some
relevant physical concepts in magnetism with emphasis laid on recent
developments in the theory of 3d-magnetism. Subsequently a few remarks
about basics of M6ssbauer spectroscopy are made, followed by the review of
the experimental work in these fields, where we have tried to stress those kinds
of materials which have received less attention in previous reviews.

7.2 Theoretical Background

7.2.1 Theoretical Concepts in Magnetism

Metallic magnetism covers a wide range of phenomena which are intimately

correlated with the electronic structure but also with the metallurgy. The latter
appears to be particularly important for intermetallics and for the binary and
ternary metal hydrides. Although no general prediction of the effect of
hydrogen absorption on the magnetic properties can be given, attempts to
 Magnetic Properties, M6ssbauer Effect and Superconductivity 287

understand the magnetic properties, however, have led to a deeper insight into
the fundamental structure and mechanisms of many different metal-hydrogen
systems. This understanding is often still far from being complete since even the
high-temperature magnetism of the well-known elements such as Fe, Co, and
Ni remains a controversial topic [7.22-24].
In contrast to the magnetism of the 3d-metals which reflect the old
controversy of localized versus itinerant magnetism, the magnetic properties of
the rare earth (R) elements are successfully described in terms of the RKKY
theory; because of the localized nature of the 4f-electrons there is no overlap
between 4f-wavefunctions on different lattice sites and the magnetic coupling
proceeds indirectly via the spatially non-uniform polarization of the conduc-
tion electrons.
The rare earths R form hydrides RHx with two stoichiometries x = 2 and x
approaching or equalling 3 (see e.g. [7.4, 25] and references therein). Electrical
resistivity measurements revealed that metallic conductivity disappears when x
approaches 3. According to Switendick [7.26] hydrogen forms in the trihydrides
a low lying s-band with the capacity to hold six valence electrons supplied by
one R and three H.
Since this low lying band is filled with electrons - the RH a are semi-
conductors- and due to the lack of conduction electrons the RKKY interaction
cannot be transmitted. This accounts for the suppression of the magnetic
interactions which is generally observed. The discussion as to whether
hydrogen is anionic (H-) or protonic (H +) in binary hydrides, has been settled
by Wallace and Mader [7.27]. By combining susceptibility measurements with
crystal-field calculations in PrH2 they could demonstrate the validity of the
anionic model. This means that H, with its low-lying s-band, acts as an acceptor
in the rare-earth series; this is further supported by heat capacity and inelastic
neutron scattering studies [7.28-31].
The preceding discussion demonstrates that magnetic studies of the rare
earth hydrides are exceedingly valuable and confirms the validity of the RKKY
interaction in the binary rare-earth hydrides which is suppressed upon
hydrogen absorption.
Although the general characteristics of the rare earths can be regarded as
having been explained in terms of a semiempirical model based in fact on
bandstructure calculations of Switendick [7.26], CeHx may serve as an example
to illustrate that the details still pose unsolved physical problems [7.32-34].
Very recently Schlapbach [7.34] reported on the electronic specific heat
coefficient 7 of CeH2.6 (7 = 110 m J/mole K a) to be more than about one order of
magnitude larger than that of 7-Ce (7 = 10 m J/mole K2). This indicates that in
Cell2.6 the 4f-band is located close to the Fermi energy. On the other hand
enormous specific heat values are obtained for high effective mass f-electron
materials such as CeCu2Si z, UBe13, UPt3, CeCu6, NpBe13, UzZn~7, and
UCdl ~ which furthermore exhibit maxima in the resistivity at low temperatures
and large values of the magnetic susceptibility; for obvious reasons these are
called heavy-fermion systems [7.35, 36]. By analogy with these compounds it
288 G. Wiesingerand G. Hilscher

was suggested by Schlapbach et al. [7.34] that Cell2.6, although ordered at low
temperatures should also exhibit heavy-fermion-like behaviour. From crystal
field calculations Osterwalder et al. [7.32] and Schefer et al. [7.33] could
furthermore clarify the striking changes of the magnetic interactions between
the Ce 3 + ions which occur in these hydrides at low temperatures as function of
x: no ordering is observed for CEH2.93 down to 1.3 K, CEH2.43 orders
ferromagnetically (T~=4.2K) and CeH3.oo orders antiferromagnerically
(TN= 1.9 K).
In contrast to the binary 4f-hydrides, for the 3d metals and their
intermetaUics no similar simple and convincing conclusions can be drawn
about changes of magnetism in terms of a protonic or an anionic behaviour.
The only general statement which can be made is that changes in the magnetic
properties in the Fe compounds occur in the opposite direction to those in the
Ni and Co compounds. Particulary in Sc-, Y-, Hf-, and R-Fe intermetallies, the
Fe moment is found to be increased on hydrogen absorption while changes in
the magnetic ordering temperatures can be obtained in either direction
[7.19, 20].
As already mentioned, magnetism in the 3d elements and their intermetal-
lies remains a controversial topic and is by no means a solved problem. The
reason for this controversy is the absence of a general agreement upon the
microscopic nature of the magnetic state above and below the Curie
temperature. Two opposite standpoints have so far been used to explain the
magnetic order as a function of temperature. In a Heisenberg model the
description is in terms of localized moments and the magnetisation disappears
at T~ because of disorder in the local moments due to thermal fluctuations.
Nevertheless the absolute value of the magnetic moments remains almost
constant. The usual result of a mean field theory leads within this type of model
to the well-known Brillouin function which in its argument contains the ratio of
the magnetic energy to the thermal energy. Thus a rise in temperature reduces
the magnetisation due to the thermal disorder of the magnetic moments.
In the Stoner-Wohlfarth itinerant-electron model the magnetic moment
and the magnetisation vector are determined by the unpaired electrons of the
exchange-split spin-up and spin-down bands; this appears to be a realistic
description of the magnetism in metals at 0 K. A given material becomes
magnetic if the Stoner criterion [N(EF)I > 1] is fulfilled [with N(EF) the density
of states at the Fermi level and I the intra-atomic exchange or Hubbard
exchange energy]. In this model the thermal excitations of electron-hole pairs
reduce the exchange splitting and drive the transition towards the paramagne-
tic state. Consequently the magnetisation vanishes only if the absolute value of
the magnetic moment goes to zero which only happens if the exchange splitting
is zero. Unfortunately this yields Curie temperatures which are 5-10 times
larger than those observed experimentally. The controversy concerning the
localized versus the itinerant electron model has been settled recently and all
current theories are now based on the latter model. It is obvious that the order
parameter in the magnetic state has to be related to the magnetisation vector.
 Magnetic Properties, M6ssbauer Effect and Superconductivity 289

On the other hand, the question of how the magnetisation arises from the spins
of mobile electrons and of which forces are responsible for its fluctuations are
only partially understood. An improved theory based on the Stoner-Wohlfarth
model (namely a spin-polarized band theory) must allow for the existence of a
magnetic polarisation whose direction varies from one unit cell to the other.
Then the global magnetization vanishes at the Curie temperature not because
the entities, the absolute values of the magnetic moments which may be called
local moments, are zero, but because they point in random directions. For a
very recent formulation of a first-principles theory of ferromagnetic phase
transitions and the electronic structure of metallic ferromagnets we refer to
Gyorffy et al. [7.23] and Staunton et al. [7.37].
The phenomena of 3d magnetism is also strongly connected with the width
and structure of the 3d-band and the density of states at the Fermi level. In a
semiempirical model Griessen and Feenstra [7.38] recently related the heat of
formation of a metal hydride to the difference between the Fermi energy and the
centre of the s-band of the host metal. They found that the product of the bulk
modulus and the lattice expansion due to hydrogen absorption depends on the
heat of formation of the hydride in a linear way. The bandwidth W is, according
to Heine [7.39], proportional to a- s (where a is the interatomic distance). The
simple band model states that narrow bands with high N(EF) favour the
occurrence of magnetism. When the hydrogen induced volume increase is
considered together with magnetovolume data of the parent compound at least
a first estimate of changes in the magnetic ordering temperature may be
obtained. If the electronic structure is only slightly affected, Tc may be
extrapolated from its pressure dependence. Such a correlation was proposed by
Buschow and Sherwood [7.17] for crystalline, and by Coey et al. [-7.40] for
amorphous Y-Fe compounds and by Hilscher et al. [7.41] for the pseudobi-
naries Ti(Fe, Co).
Realistic electronic band structure calculations for the parent intermetallics
and their hydrides are therefore necessary to gain a deeper insight into the basic
physical behaviour of these systems. Otherwise semiempirical models have to
be used in order to interpret the effect of hydrogen upon magnetism.
Besides LaNi 5, TiFe and the pseudobinaries Ti(Fe, Co) also serve as
relevant examples of intermetallic compounds in which hydrogen absorption
affects not only the bulk 3d magnetism but also the metallurgy, giving rise to a
decomposition at the surface. For TiFe this leads to the formation of Ti and Fe
clusters at the surface accompanied by a growing number of antistructure
atoms in the bulk. These additional contributions to the global magnetisation
further complicate the analysis of the data actually measured. The change of the
magnetic properties upon hydrogenation may thus be of different origins which
need not necessarily be correlated with altered bulk or intrinsic magnetic
properties of the material.
Intermetallic compounds of 3d metals (Mn, Fe, Co, Ni) with rare earth
elements exhibit a large variety of interesting magnetic properties. A large
number of hydrogen absorption studies have also been performed on such
290 G. Wiesingerand G. Hilscher

compounds. The magnetic properties of these intermetallics have been

reviewed by Buschow [7.42, 43], Kirchmayr and Poldy [7.44] and Wallace
[7.45]. These properties are of interest for two major reasons: Firstly their study
helps to shed light on some of the fundamental principles of magnetism such as
the indirect exchange or RKKY interaction, crystal field effects, valence
instabilities, magnetoelastic properties, the occurrence of superconductivity
and magnetic ordering in the same compound. Secondly these properties are of
technical interest due to their potential application in the production of
permanent magnets such as the RCo5 and the Nd-Fe-B based materials.
Furthermore, several of those compounds have been proved to be suitable for
hydrogen storage purposes.
In R-3d intermetallics and their hydrides, where both the R and the 3d
element are magnetic we can distinguish between three main types of magnetic
interactions which are quite different in nature: the magnetic interaction (i)
between the localized 4 f moments, (ii) between the more itinerant 3d moments
and (iii) between the 3d and the 4f moments. Generally it is observed [7.43, 44]
that the magnetic interaction in R-3d intermetaUics, where both the R and the
3d element are magnetic, decreases in the following sequence: 3d-3d, 4f-3d,
4f-4f.
Depending upon the composition and the type of combination of these
elements hydrogen absorption usually leads to a weakening of the magnetic
coupling between the 4f moment and the 3d moment and to substantial
changes in the 3d transition metal moment. In Co and Ni compounds hydrogen
absorption usually results in a reduction of the 3d moment, while the opposite is
observed for R-Fe compounds. The weakening of the 4f-3d exchange
interaction upon hydrogen absorption may be explained by a reduced overlap
of the 3d electron wave functions with the 5d-like ones which are polarized by
the 4f moment and thus transmit the interaction to the 3d electrons. Hydrogen
reaction can furthermore lead to concentration fluctuations of H atoms over a
few atomic distances so that the electron concentration may differ from one R
site to the other and therefore changes of the magnetic coupling strength can be
expected. Moreover resistivity measurements [7.25, 46] show that the residual
resistivity of the hydrides is higher than that of the parent intermetallics which
indicates that the mean free path of the conduction electrons is reduced. This
can lead to a damping of the RKKY conduction electron polarization which
tends to decrease the magnetic coupling strength.
Apart from the pure 4f elements and their binary hydrides, compounds
where the transition metal moment is zero also allow the 4 f - 4 f interaction to
be studied directly. As a first approximation one would expect that changes in
magnetic properties can be explained by analogy with the pure R elements and
their hydrides, in terms of the anionic model, by a reduction of the conduction
electron concentration which consequently reduces the R K K Y interaction.
This is obviously the case for GdRu2 and GdRh2, where the Curie temperatures
of 83 and 73 K are reduced upon hydrogen absorption to 65 and 35 K
respectively [7.47]. However the opposite is found for GdCu2 and GdAg,
 Magnetic Properties, M6ssbauer Effectand Superconductivity 291

where, after hydrogen uptake, the paramagnetic Curie temperatures change

from 7 to 57 K and - 5 7 to + 50 K (i.e. from antiferro- to ferromagnetism)
[7.48]. According to Buschow [7.20] the latter compounds may serve as an
example in which altered conduction electron polarization can cause changes in
both magnitude and sign of the magnetic coupling due to the oscillatory
character of the RKKY interaction,
If magnetism in R-intermetallics is dominated by the 4 f moments the
concept of charge transfer, namely that R atoms donate electrons to the H
atoms seems by analogy with binary R-hydrides to be a reasonable explanation
for changes in magnetism. This is in fact confirmed by numerous isomer shift
results obtained from M6ssbauer experiments on rare earth nuclei.
In the case where 3d magnetism is dominant in R-3d compounds, hydrogen
absorption leads to a reduction of the 3d moment in Ni and Co based
intermetallics, but to an enhancement of the Fe moment. For Mn-intermetallics
both changes from para- to ferromagnetism and in the opposite direction are
observed.
As already mentioned, the 3d-moment formation from itinerant electrons is
a controversial topic and the introduction of hydrogen makes the situation
even less clear. Wallace [7.4, 45] explained the controversial changes of the 3d
moment in Fe, Co, and Ni compounds as well as the occurrence or breakdown
of the magnetic order in Mn-compounds in terms of a depopulation of the
exchange split 3d bands as a consequence of a sizeable charge transfer from the
3d metal to the H atoms. From various M6ssbauer experiments [7.20, 47, 48]
and according to reasonable arguments by Buschow [7.15, 20, 32] such a large
charge transfer appears to be rather unlikely to explain the substantial changes
of the 3d moment. For binary 3d and 4d transition metal hydrides and their
intermetallics Wicke [7.49] proposed a screened-proton model combined with
the concept of d-band filling. In terms of these models recent results of UV and
x-ray photoelectron spectroscopy can also be interpreted [7.50, 51 ]. Since band
structure calculations for these intermetallics and their hydrides are not
available, somewhat speculative semiempirical models are used. In view of the
M6ssbauer results on R-nuclei and the above cited experiments, we agree with
Buschow [7.15,20] that hydrogen exists as a screened proton in these
R-intermetallic hydrides with a screening charge larger than one. The excess
screening charge is due to charge transfer from R-atoms to hydrogen rather
than from the 3d atoms.

7.2.2 Physical Concepts of Miissbauer Spectroscopy

a) Introductory Remarks
Only some brief remarks will be given here. For more detailed discussions we
refer the reader to the numerous textbooks dealing with M6ssbauer spec-
troscopy, among which two most recent ones are included in the reference list
[7.52, 53]. The reason for observing the M6ssbauer effect is based on the fact
292 G. Wiesinger and G. Hilscher

that in case of medium y-energies ( < 100 keV) a certain probability exists for the
recoil-free emission and absorption of y-rays. Because of the range of energies
involved, such a process can only occur if the M6ssbauer nucleus is bound in a
solid. Since transitions between nuclear levels are observed by M6ssbauer
spectroscopy, this technique is sensitive to the influences of the electronic
environment of the probe nucleus which may cause a perturbation to these
levels (electron-nucleus- or hyperfine interactions). When 7-rays are emitted or
absorbed under the above conditions they suffer no energy loss, and the
resulting lines exhibit the natural width. Thus a sufficiently high resolution can
be obtained to monitor the minute hyperfine interactions by periodically
changing the Doppler velocity in the range of mm/s up to cm/s corresponding
to energy changes of the order of a few BeV or even less.
Three of these hyperfine interactions are known: (i) An electrostatic one
(Coulomb-like) between the nuclear charge and the electronic charge at the
nucleus, causing a shift of the nuclear levels. Therefore we commonly observe
deviations from the theoretical resonance frequency (isomer shift). (ii) The
interaction between the nuclear quarupole moment and the electric field
gradient (EFG) at the nuclear site results in a partial (quadrupole-) splitting of
the nuclear levels. (iii) The degeneracy of the nuclear states is lifted completely,
when the nuclear magnetic moments experience a magnetic field (nuclear
Zeeman effect). Consequently in the last two cases we observe a splitting of the
y-ray pattern, i.e. multi-line M6ssbauer spectra are obtained. Further param-
eters which can be derived from the analysis of a spectrum are the width of the
M6ssbauer line which can yield information about diffusion, or relaxation
processes and the M6ssbauer-Lamb (f) factor representing the fraction of
recoil-free y-rays emitted or absorbed, thus giving insight into the dynamic
behaviour of the lattice.

b) Experimental Techniques
Since there exist numerous books on M6ssbauer spectroscopy which include a
discussion of experimental techniques we will only touch upon this topic by
mentioning those problems which arise specifically in the study of hydrides of
intermetallic compounds. In principle, two different kinds of experimental set
up are employed: (i) the sample is prepared as a source, the spectrum is scanned
by using a single line absorber (source experiment); (ii) a single line source is
used in combination with the sample positioned as absorber. In contrast to
binary metal-hydrogen systems [7.2], no source experiment has yet been
reported for the case of ternary hydrides. When spectra are to be recorded by
applying the latter technique in transmission geometry, reasonable measuring
times (some hours up to a few days) are only obtained when the M6ssbauer
isotope is present in the absorber in an abundance of at least a few percent. This
value may, however, be considerably lowered, if enriched isotopes are used.
M6ssbauer absorbers are chiefly prepared by melting the intermetallic host
compound and by subsequently loading the sample with hydrogen gas. An
 Magnetic Properties, M6ssbauer Effect and Superconductivity 293

exception is found in case of hydriding amorphous alloys, where electrolytical

charging is preferred. In order to keep the amount of dissolved hydrogen in the
sample constant during the measuring time, sealing the hydride is highly
recommended (e.g. with SO2) (see Volume II, Chap. 2), which particularly
holds for M6ssbauer studies at or above room temperature.
Among the vast number of intermetallic compounds which have been
studied, attention has been focussed on those materials which have proved
suitable for hydrogen storage purposes (LaNis, TiFe .... ). Consequently 57Fe or
a rare-earth isotope (151Eu, lSSGd, 161Dy, 166Er, 169Tm) are frequently
mentioned in the studies under consideration. 181Ta, which is widely used to
investigate binary hydrides [7.1, 2], is of no relevance in the present case.
Because of the y-energies involved, the absorber thickness has to be in the
range of about 20 ~tm (5~Fe, 14.4 keV) up to 1 mm (166Er, 81 keV). For studying
the parent compounds, powder absorbers are generally prepared by grinding
the material. After hydrogen absorption, however, the solid material has
usually already converted into a fine powder. It can thus be directly used as a
proper absorber material preventing further damage to the sample.

c) Hyperfine Interactions
The isomer shift contains simultaneous information about the difference in the
nuclear radii (fiR) of excited and ground state respectively and the difference in
the charge at the nuclear site in absorber and source respectively. Because 6R
can be of either sign (e.g. < 0 for 57Fe, 155Gd and > 0 for 1198n) a shift of the
resonance line to more positive (negative) energies corresponds in the former
case to a smaller (larger) electron density, whereas the opposite is obtained in
the latter case. The isomer shift depends upon valence state and chemical
bonding. It can be directly influenced by altering the population of the s-like
conduction electrons or indirectly via a shielding of the s-electrons by those
with d-character. Walker et al. [7.543 made an attempt to calibrate the observed
57Fe isomer shifts in (ionic) iron compounds and iron alloys in terms of the total
s-electron density. This method may be of some relevance when the problem of
hydrogen induced charge transfer is considered. Buschow [7.20], however,
reported to have obtained unreasonably large values for the transferred charge
when interpreting the change in isomer shift in terms of the Walker data.
Wagner and Wortmann [7.2] had already pointed out the problem of
asigning the total change in isomer shift after hydrogen absorption to different
origins. If the isomer shift S is considered as a function of hydrogen
concentration n and volume v, the overall change after charging can be
expressed as

dn=\O-T~nv,l,d-n-n + fifth ."

dS/dn is measured by MBssbauer spectroscopy. (~3S/~lnv). displays the pure
volume dependence of S and can be obtained from high pressure data. The term
294 G. Wiesingerand G. Hilscher

d lnv/dn represents the hydrogen induced rise in volume and may be evaluated
from x-ray diffraction data. The change at constant volume, (OS/On)v,describing
the actual effect of hydrogen on the electronic structure of the M6ssbauer atom,
can thus be readily calculated. It turns out that this term is rather small in
magnitude, i.e. of the order of + 0.1 mm/s. An appreciable 3d charge transfer
from Fe to H which has frequently been proposed in order to explain the
hydrogen induced changes in the magnetic moment (see e.g. [7.7]) can by no
means account for it, since this would in fact lead to the wrong sign for dS.
Very recently Gupta [7.55] interpreted the isomer shift of TiFe hydride by
applying the APW method and found excellent agreement with the experi-
mental data. She proposed a depletion of the Fe4s states in order to form a low-
lying metal-hydrogen band which, for the case of binary metal-hydrogen
systems, has already been suggested by Switendick [7.26]. This finding is in
complete accordance with XPS data recently reported by Schlapbach [7.56] for
ZrMn2H 3. He was able to verify experimentally that this hydrogen induced
band lies 6.5 eV below the Fermi level. The shift of the Zr 3d levels towards
higher energies can be regarded as a further indication of a charge transfer from
Zr to H. The Fe 3d states seem to be hardly affected by hydrogen absorption.
Buschow [7.20] questioned the justification of making up the hydrogen
induced change in isomer shift from an electronic and a volume term. He
suggested instead that the model of Miedema and van der Woude [7.57] should
be used, where the differences in electronegativity and in the electron density
at the atomic cell boundary have to be considered. An attempt was made to
explain the isomer shift data for ThvFe3H~, but no firm conclusion has been
reached as yet.
The electric quadrupole interaction occurs if both a nuclear quadrupole
moment and a nonzero electric field gradient (EFG) exist at the nucleus. As a
consequence, the nuclear state can split into several sublevels, leading to the so-
called quadrupole splitting of the Mrssbauer line. Commonly one assumes the
EFG to arise from two sources - from a lattice and an electronic contribution.
The former is due to charges on lattice sites surrounding the Mfssbauer atom in
a non-cubic symmetry; the latter is attributed to an anisotropic distribution of
the conduction electrons of the Mrssbauer atom. In the hydrides the influence
of interstitial H atoms on the EFG is far too small to be completely resolved in a
Mrssbauer spectrum. At most a line broadening or an asymmetric line shape
will be observed.
A nucleus with a spin quantum number I > 0 exhibits a nuclear moment
which interacts with the magnetic field at the nuclear site (effective or hyperfine
field). This commonly yields a splitting of the nuclear states into 21+ 1
sublevels, leading to multi-line hyperfine patterns. Because the presence of a
magnetic field is required for this interaction, it will usually only be observed in
magnetically ordered solids. In paramagnetic systems a magnetic hyperfine
splitting can only occur if an external field of several T is applied tothe sample
or when the relaxation times of the atomic spins are sufficiently long compared
 Magnetic Properties, M6ssbauer Effectand Superconductivity 295

with the nuclear precession time. The magnetic field acting on the nucleus is
found to lie in the range between several T (c~-Fe: Boff=33T at room
temperature) and a few hundred T in case of rare earths. Beff consists of different
terms which originate from the Fermi contact interaction, from polarized d(f)
electrons, from dipolar contributions of the atomic moments and if present also
from a contribution due to the external field. It should be mentioned that the
three hyperfine interactions just described can only in a few fortunate cases be
observed separately. The simultaneous occurrence of at least two of them is
more likely.
From the discussion of the hyperfine interactions it is obvious that a given
phase in the sample under investigation should exhibit specific hyperfine
parameters and consequently can be identified by applying M6ssbauer
spectroscopy. Quantative phase analysis using this technique should, however,
be performed with great caution, since the f-factors in the individual phases will
in general differ from each other and even are completely unknown in some
cases. This holds especially in the case of high 7-energies. Although tho
assumption that the area under a given hyperfine pattern is proportional to the
number of M6ssbauer atoms in the corresponding phase may be applied
frequently, clustering of the M6ssbauer atoms can again lead to an error in the
evaluation of the phase concentration.
Further properties which can be investigated by analyzing a M6ssbauer
spectrum, e.g. hydrogen diffusion or hydrogen induced change of the lattice
dynamics are beyond the scope of the present chapter and are considered in
Chap. 3 of Volume II of this series.

7.3 Magnetism and M6ssbauer Spectroscopy-Experimental

Results and Discussion
7.3.1 General Remarks

The organisation of the following section deviates from that used in earlier
reviews. Since up to now only rare earth-transition metal hydrides have been
considered a subdivision into Mn, Fe, Co, and Ni compounds seemed
meaningful. In the present article, however, Zr and Ti compounds, amorphous
alloys and even oxides (Ti2FeOz) are also included and thus the arrangement
following the stoichiometry may be more favourable. Within a certain
composition the results are again reported in the conventional sequence.
Since in case of magnetism and M6ssbauer effect the difference between
hydrogen and deuterium is insignificant, no distinction has been made between
those two isotopes. This is furthermore justified by the fact that exclusively
deuterides have been used in neutron diffraction studies because the large
incoherent scattering cross section of hydrogen prevents the recording of
reasonable neutron spectra in case of the hydrides.
296 G. Wiesinger and G. Hilscher

7.3.2 Transition-Metal-Rich R-TM Compounds

a) RzTM17
This stoichiometry occurs principally in two possible crystal structures, the
rhombohedral Th2Zn17 (space group R3m, No. 166) and the hexagonal
ThzNi17 structure (space group P63/mmc, No. 194) (see e.g. [7.44]). In case of
T M = F e most compounds show both structure types, the trigonal pre-
dominantely for the light, and the hexagonal for the heavy lanthanides.
Altogether four inequivalent TM lattice sites are present for both types which
for R2Fe ~7 leads to rather complex 57Fe M6ssbauer patterns with up to eight
subspectra, the particular number depending on the direction of the easy axis of
magnetisation [7.58, 59].
Up to now only one study on this kind of composition has been reported:
Zukrowski et al. [7.60] investigated some Fe-rich Dy2(Fel_yAly)17-hydrides
using 161Dy as well as 57Fe M6ssbauer spectroscopy. For the host compounds
only the ThzNi~v type of structure was obtained. The sample with y=0.18,
however, could be converted to the Th2Zn 17 type after hydrogen uptake. The
storage capacity was found to be rather limited: between 4 and 5 H atoms/f.u.
could be dissolved. While upon hydrogenation isomer shift and quadrupole
splitting had increased in both cases, only the Fe-hyperfine field showed a
remarkable increase, the hyperfine field at the Dy nuclei remaining essentially
unaffected.

b) RTMs
In this stoichiometry, except for ThF%, only Co and Ni compounds exist,
crystallizing in the CaCus type of structure (space group P6/mmm, No. 191).
RMn s compounds do not form at all.
TM = Fe, ThFe s
Gubbens and van der Kraan [7.61] examined the host compound by means of
M6ssbauer spectroscopy, x-ray diffraction and neutron depolarisation and
concluded that ThF% has a basal plane magnetic anisotropy of a complex
ferrimagnetic character. Studying the hydrides Gubbens et al. [7.621 obtained a
maximum hydrogen uptake corresponding to the composition ThFesH~.7,
leading to a rise in volume of about 6%. Both magnetisation and Curie
temperature were found to have only slightly increased after hydrogenation. As
in case of the parent compound the hydride exhibited a sharp hyperfine pattern,
however, the easy axis of magnetization was concluded to have turned from a
direction between the a- and the b-axis to a direction along the a-axis. The
increase in isomer shift after charging has been explained by Gubbens et al. on
the basis of a modified Miedema-van der Woude model [7.57].
TM = Co
Only compounds containing light rare earths up to Gd have been examined so
far. The reason lies in the outstanding permanent magnetic properties of these
 Magnetic Properties, M6ssbauer Effect and Superconductivity 297

~
a.
10 , I ' I ' I ' ' I i
I ' I

0.7 ~ ;" "- ~ > ~'~ ~

g

, I , I , I , , I , I , I ,
001120 100 80 60 4~ 1 2 3

Ms (Arn2/kg) HYDROGEN COMPOSITION,X

Fig. 7.1. Hydrogen pressure vs saturation magnetisation (left) and hydrogen pressure vs
composition (right) in absorption (o) and desorption (o) process for NdCosH~ at T = 303.2 K
[7.64]

ferromagnetic materials which are lost in the case of the heavy rare earths,
where ferrimagnetic coupling takes place between the R and the Co moments.
The ferromagnetic behaviour of the light RCos's is preserved after charging
with hydrogen; however, a weakening of the R-Co and the Co-Co exchange
interaction is observed [-7.63-65]. Co as well as R moments were found by
K u i j p e r s [7.14] to decrease in a nonlinear way with increasing hydrogen
content Hx. An extrapolation yielded a value of x--,4.5 for the loss of
ferromagnetism. By studying LaCo s doped with 10% Gd with the 155Gd
MGssbauer effect B a u m i n g e r et al. [7.66] could confirm the presence of different
hydride phases already reported by K u i j p e r s and L o o p s t r a [7.63]. While isomer
shift and electric field gradient changed monotonically with the amount of
absorbed hydrogen, no such variation was obtained for the Gd hyperfine field.
Since the magnetisation strongly depends upon the hydrogen concentration
in the sample, pressure-magnetisation isotherms yield similar information
about the presence of different phases and about phase transformations as
pressure-composition isotherms. Considering this, Y a m a g u c h i et al. [-7.64]
developed a vibrating-sample magnetometer connected directly to a
pressure-composition measuring system. This technique proved to be especi-
ally useful for studying the magnetisation of less stable hydrides as a function of
applied field and hydrogen pressure. The authors could demonstrate the
versatility of this method by examining the different phases (0(, if, fin, ~) in the
system NdCosHx [7.64], GdCosHx and YCosH~ [7.65] and PrCosHx and
LaCosHx I-7.67]. The peculiar phases have been identified by the stepwise
changes in magnetisation observed in the magnetisation-temperature isobars
which are displayed in Fig. 7.1.
For R = Y and R = G d , 59Co NMR spin-echo measurements have been
carried out [7.68, 69]. Some discrepancies occurred in assigning the various
298 G. Wiesingerand G. Hilscher

resonance lines to the two Co sites. Probably the hydrogen atoms are located
preferentially in the Co-only layers, causing a greater effect on the Co01) signal
than on that orginating, from the Co(I) sites.
Magnetic relaxation studies on hydrided (deuterated) light RCos's have
been performed by Herbst and KronmiiIler I-7.70, 71] who measured the
magnetic after-effect of the initial susceptibility. They concluded that the
concentration of hydrogen (deuterium) in the planes containing R atoms, by far
exceeds that in the Co-only planes. This result is in disagreement with the
analysis of the 59Co N M R by Yamaguchi et al. I-7.68]. However, considering the
weakly resolved spectra and the large discrepancies in the resonance frequences
with the investigation of Figiel I-7.69], we tend to favour the interpretation of
Herbst and Kronmtiller.

TM = Ni
Emphasis has so far been laid on pure LaNi 5 and on LaNis-containing
materials. Because of their outstanding hydrogen storage properties they have
received considerable attention particularly for technical application. With this
kind of compound we meet a typical example in which bulk and surface effects
occur simultaneously. In certain cases this has not quite been realized and has
sometimes led to apparently inconsistant results.
A large variety of techniques has been applied in order to elucidate the
complex hydrogen absorption mechanism in this kind of material. These can be
roughly divided into surface sensitive methods (photo emission and related
spectroscopies, and to some extent ESR) and experiments in which only the
bulk properties can be studied (magnetic measurements, x-ray diffraction,
transmission M6ssbauer spectroscopy). Careful susceptibility studies, however,
can also be a useful tool for identifying ferro- or at least superparamagnetic
clusters at the surface if the results obtained from freshly cut samples in high
vacuum at variable temperature are compared with those from samples
exposed to some reactive atmosphere. Such disintegration has indeed been
observed on uncharged LaNi5 if oxygen or water was present I-7.72-76]. When
loaded with hydrogen under medium pressures (< 10 bar) the disintegration
continues which, however, is not the case if the hydrogen pressure exceeds this
value substantially [7.75]. This might be the reason for the unusual interpre-
tation of Palleau and Chouteau I-7.77].
The results from the various magnetic measurements can be summarized as
follows: although by applying spin-polarized energy band calculations Malik
et al. I-7.78] predicted LaNi 5 to be weakly ferromagnetic, several authors
confirmed that in fact this compound behaves as a Stoner-enhanced Pauli
paramagnet (see e.g. I-7.75] and references therein). When hydrogen is absorbed
just once, its susceptibility is lowered by a factor of nearly 4. This has been
attributed by Schlapbach I-7.75] to a reduction of the enhancement factor and
has been corroborated by magnetic measurements [-7.77] as well as by the
observation of the g-shift of the Gd ESR in Gd-doped LaNi5H x I-7.79]. At this
point we want to stress that after just one absorption process we are still dealing
 Magnetic Properties, M6ssbauer Effect and Superconductivity 299

with a pure bulk phenomenon, since the specific surface area of the Ni
segregations may still be assumed to be neglegibly small.
Simple hydrogen absorption has been further examined by means of several
Mfssbauer experiments. By investigating Gd-doped samples with the 155Gd
transition (E~ = 86.5 keV) Bauminger et al. [7.66] could confirm the different
hydride phases which had already been detected by Kuijpers [7.14] by means of
magnetic measurements. Furthermore a large reduction of the s-electron
density at the Gd nuclei has been reported which, in our opinion has
erroneously been interpreted as inconsistent with the ESR results of Walsh et al.
[7.79]. The 57Fe (Er = 14.4 keV) results obtained on Fe-doped LaNi 5 [7.80-82]
show some discrepancies in the interpretation of the room temperature
M6ssbauer spectra which points we believe to the influence of the metallurgy on
the hyperfine parameters. In case of the hydrides [7.83-85] the data again
suggest the presence of Ni clusters, whereas no evidence for the formation of Fe
clusters has been found. While for certain concentrations magnetic ordering
temperatures were found to have been substantially reduced after hydrogen
uptake, the Fe hyperfine field at 4 K was proved to remain almost unchanged.
The small change in the Fe-isomer shift upon hydrogenation points to the
predominance of the L a - H interaction in this compound. The complete
insensitivity of the 119Sn (E~ = 23.8 keV) hyperfine parameters in Sn doped
LaNis to hydrogenation, as reported by Oliver et al. [7.86], once more confirms
this interpretation.
Apart from these studies, heavily cycled material has been investigated in
order to examine the degree of reversibility of the absorption-desorption
process after a large number of cycles [7.75, 79, 87-90]. Repeated cycling yields
a distinct enhancement of the susceptibility which furthermore tends to become
temperature and field dependent. As already mentioned, the bulk susceptibility
is reduced after the absorption of hydrogen. Consequently the increase in ;t has
to be attributed to the formation of ferro- or at least superparamagnetic Ni
precipitates, their specific area increasing with hydrogen content. The segrega-
tion and decomposition of the surface was confirmed by Shaltiel et al. [7.891 by
applying ferromagnetic resonance (FMR). A signal which could be attributed
to pure Ni metal clearly demonstrates the presence of clusters at the surface.
With increasing number of cycles the magnetisation also increases and has been
interpreted by Schlapbaeh [7.75] to consist of two superimposed terms, a linear
one due to the bulk, the another field dependent term originating from the Ni
precipitates at the surface. From a quantitative analysis of the M v s H curve
Schlapbach was able to estimate that the Ni clusters contain about 6000 atoms.
Other M6ssbauer transitions have been successfully applied to demonstrate
the formation of the segregations in repeatedly cycled material. This shows that
their volume must be of appreciable magnitude, since the clusters are visible for
y-rays with an energy of more than 20 keV. Cohen et al. [7.87, 88] used the
151Eu spectroscopy (E~ = 21.6 keV), while Rummel et al. [7.90] performed the
only 61Ni experiment (E 7 = 67.4 keV) known so far in the field of intermetallic
hydrides. Identically treated samples were used and although the energy of
300 G. Wiesinger and G. Hilscher

lO0

99

99
L o Ni~
untreated

96

100

99

98
t.(:a N i ~ ,
9t
H2 activated,

99 1cycle
V
9~

100

99

98

97 Lo Nie.
H2 activated.
96 2 000 cycles

I I I I I
J -30 -f~ -1o -5 o 5 ¢o
-9 "~: "2 0 2 4 6 VELOCITY (mm/$)
Ut nrl$*
Fig. 7.2 Fig. 7.3
Fig. 7.2. 61Ni Mfssbauer spectrum of LaNis samples after various treatments: (a) no hydrogen
exposure, (b) activated in hydrogen, (c) after 1584 thermally induced absorption-desorption
cycles I-7.90]
Fig. 7.3. M6ssbauer spectra of lSIEu in LnTMs: A, a fresh sample; B, after hydriding at
150 bar; C, dehydrided after 10 cycles; D, hydrided after 1500 thermal cycles; E, dehydrided
after 1500 cycles. The absorption lines near 2 mm/s arise from Eu 3+ in the unhydrided host;
those near - 11 mm/s arise from Eu2+ in the hydrided material [7.87]

radiation differs substantially, consistent results have been obtained. While

Rummel et a]. could not detect the second phase by x-ray diffraction, they could
unequivocally ascertain it by the 61Ni M6ssbauer spectra. As can be seen by
inspection from Fig. 7.2, no substantial effect is observed after one cycle.
Repeated cycling (1584 times) leads, however, to the occurrence of a magnetic
hyperfine pattern, its splitting agreeing perfectly with that of pure Ni metal. A
careful quantitative analysis of the absorption areas yielded the result that some
of the Ni clusters were still too small to develop long-range ferromagnetic order
and thus remained superparamagnetic, even at liquid helium temperature.
Cohen et al. [7.87, 88] studied the degradation of thermally cycled Eu-
doped LaNi 5. Eu was found to change its valency upon hydriding from 3 + to
2 + which fortunately is accompanied by large changes in isomer shift (up to
 Magnetic Properties, M6ssbauer Effectand Superconductivity 301

- 13 mm/s). This makes an identification of the respective phases quite easy.

The spectra in Fig. 7.3 demonstrate convincingly that the absorption-
desorption process is by no means completely reversible. Full reversibility can
only be achieved up to a few cycles (Fig. 7.3 curves A, B, C). After a larger
number of cycles the material can no longer be completely dehydrided since
stable EuH 2 has formed (curve E). On the assumption that La behaves
identically to Eu, Cohen et al. explained the degradation as due to the
formation of a stable hydride phase rather than via oxidation which has been
only found to an extent of a few percent (curve D).
0 liver et al. [7.91 ] again made use of the favourable properties of the 15~Eu
M6ssbauer effect to study pure EuNi 5. Comparing several ternary Eu hydrides
with different stoichiometries Oliver et al. concluded from the almost identical
hyperfine field and isomer shift values (being close to that of EuH2) that - as in
LaNi 5- the absorption process occurs primarily via the formation of the binary
rare earth hydride.
In CeNi 5 Malik et al. I-7.92] examined the substitution of Ni by AI and
obtained an increase in stability of the hydride which on the other hand was
accompanied by a drastic decrease of the absorption capacity. Susceptibility
measurements yielded an increase ofz upon charging which has been attributed
to the precipitation of Ni. The question about a change in valency of Ce from
4 + to 3 + upon hydrogenation remained unsolved. Pedziwiatr et al. [7.93]
reported on a gradual change of the Ce valency from 4 + to 3 + in Cu sub-
stituted CeNis. Hydrogen uptake was found to further increase the Ce 3+ con-
tent in the sample. In CaNi 5 Yagisawa and Yoshikawa [-7.94] again observed
a strong preference for the formation of Ni precipitates even in the parent
compound. The behaviour of the bulk susceptibility upon hydrogen absorp-
tion is almost identical to that which has been described in detail for LaNi 5.

c) R2TM7 (Space Group P63/mmc No. 194)

Magnetisation measurements for Y2Ni7 I-7.95] and La2Ni7 [7.96] and their
hydrides have been reported by Buschow. In Y2Ni7 surprisingly the Curie
temperature rises upon hydriding. The magnetisation, however, decreases to
about one half of its value in the parent compound. On the other hand, pressure
experiments predict such a drop in T~.Unexpectedly La2Ni 7 is antiferromagne-
tic and becomes Pauli paramagnetic upon charging (Fig. 7.4) which Buschow
explained by a reduced 5d density of states at Ev caused by the charge transfer
between La and H. In contrast to La2NiT, no broadening of the x-ray diffraction
peaks after hydriding could be observed in the case of Y2Niv. Thus a phase
separation which could give rise to the enhanced value of T~ can be ruled out.
Buschow et al. I-7.97] and Buschow [-7.98] have also studied the hydrides of
La2Co 7 and of Ce2Co 7. In contrast to the former case an increase in magnetic
moment and Curie temperature has been obtained in the latter after hydrogen
absorption. This was explained by Buschow in terms of a change in the
valency of Ce from 4 + to 3 +.
302 G. Wiesinger and G. Hilscher

Fig. 7.4. Temperature dependence of the

Z8

~4
F La2 Ni 7
magnetisation ( ) of pure La2Ni7 in
different external fields and of the suscepti-
bility(---) of the hydride La2NiTH~[7.96]

". Z2

E
E O0
E.H: 09 r
o.6[j /\ V¢ °6'

~-- o

20 f L°2 Ni7Hx
,o

o 5'o ,o'o ,;o 2oo 25o 3oo

r(g)

d) R2Fe14 B-Type Compounds

Very recently the development of an outstanding hard magnetic material based
on Co-free R-Fe-B alloys was successfully demonstrated by rapidly quenched
ribbons [7.99], by sintered magnets [7.100] and lastly by a ternary diffusion
path [7.101]. Energy products of up to 45 MGOe have already been obtained.
The principle source of the unique hard magnetic properties in N d - F e - B are
the intrinsic properties of the tetragonal compound Nd2Fe14B (space group
P42/mnm, No. 136) which exhibits two Nd and six Fe sublattices [7.102-104]:
the large saturation magnetisation and the high uniaxial crystal anisotropy
(/zoMs = 1.6 T,/zoH a = 7.5 T) at room temperature. Moreover, the metallurgical
possibility for the formation of small crystallites of the NdzFe14B phase gives
rise to a high coercivity tHc. The easy axis ofmagnetisation was determined to
lie along the c-axis at room temperature, whereas below about 120 K a complex
magnetic structure occurs. If another rare earth, with the exception of Eu and
Yb, replaces Nd the crystal structure remains unchanged.
Due to the large number of inequivalent Fe lattice sites rather complicated
M6ssbauer spectra are obtained. In some cases, however, they could reason-
ably well be analyzed by considering the relative site occupancy of the Fe atoms
in the unit cell [7.105-107].
Several of these compounds have already been examined with regard to
their hydrogen absorption properties. For R = Nd up to 4 H atoms/f.u, were
found to be dissolved, leading to a rise not only in the Fe moment [7.108, 109]
but also in the Curie temperature [7.34, 108]. The M6ssbauer spectra reveal
not only an increase in Bert but also the occurrence of a pattern characteristic of
a-Fe upon charging with hydrogen. In cases where free metallic Fe is already
present in the host compound this specific hyperfine pattern seems to exhibit an
enlarged intensity after hydrogen absorption. For R = D y , 161Dy M6ssbauer
 Magnetic Properties, M6ssbauer Effectand Superconductivity 303

Fig. 7.5. Anisotropy

t5 Uo HA/r ; Tsr o Ndl5Fe77B8 field H A vs t e m p e r a t u r e
o..........~..., . N%Fe77%% of NdlsFeTvB8 (O),
Y15 Fez7B8 Nd,sFev7BaHa6 (O),
a n d YlsFeTvBa (zx)
[7.110]

~ rsr

o tO0 750 200 250 7/K

5'0

studies have been carried out [-7.106] which indicate a slight reduction of the Dy
hyperfine field.
The temperature dependence of the anisotropy field HA, measured by the
so-called singular point detection technique, is displayed in Fig. 7.5 for
YI 5Fe77B8, Nd~sFevvBs, and Ndl 5Fe77BsH3.8 [-7.110]. From comparing the
data we deduce that for T > 200 K the contribution of the Nd sublattice to H A is
strongly reduced by hydrogen, leading to a drop of the global anisotropy of
Nd15Fe77B8H36 even below that of Y15FevvBs. At low temperatures, however,
the Nd contribution is estimated to still dominate the Fe sublattice contri-
bution to H A. A further indication for the persistence of the Nd anisotropy at
low temperatures can be seen in the occurrence of the spin reorientation even in
the hydride. It has been suggested that the shift of the reorientation temperature
TSR to lower values originates from the weakened Nd anisotropy upon
hydrogen absorption.
Hydrogen absorption and desorption may provide an alternative to milling
for disintegrating bulk Nd-Fe-B into a fine powder as required for the sintering
process and also for the production of polymer bounded magnets. Since
Schlapbach I-7.34] has proved by means of XPS that Nd2Fe14B shows a
selective oxidation of Nd, it is suggested by analogy to LaNi S that in this ease
the surface decomposes into Nd203 and precipitates of Fe. These precipitations
may even be large enough to be detected by 57Fe M6ssbauer spectroscopy. Up
to now we have only obtained, in agreement with Harris et al. [-7.111],
coercivities comparable in magnitude with those from magnets produced by the
standard milling and sintering technique. A further optimisation of the
parameters during the hydriding process is required if hydrogen absorption is
to supercede conventional methods of material pulverization.

e) R6TM23 (TM = Mn, Fe)

These compounds, particularly the pseudobinary (Fe, Mn)-phases, exhibit
several remarkable physical properties. Upon the absorption of hydrogen still
 304 G. Wiesinger and G. Hilscher

25 20 15 i i r i i

Yh6Mn23
20
I5

tL

~o x
~5
2~:

0 0 - " - - - - -II- - - - I ----- I- I h,. 0

0 100 200 300 4;0 500 0 lO0 200 300 400 500 600
r(K) r(K)
a h

Fig. 7.6. (a) Temperature dependence of the magnetisation of Y6Mn23 before ( , left
hand scale) and after hydrogen absorption ( . . . . , right hand scale). (b) Temperature
dependence of the magnetisation of Th6Mn23 before ( , right hand scale) and after
hydrogen absorption ( - - - , left hand scale) [7.112]

further spectacular features appear (see e.g. [7.20]). All compounds crystallize
in the fcc. Th6Mn23 type of structure (space group Fm3m, No. 225), the unit cell
consists of 116 atoms (1 R and 4 TM sublattices) and lattice constants of about
12 N are commonly observed. Substantial amounts of hydrogen (deuterium)
can be dissolved ranging from about 15 atoms/f.u. (Fe compounds) up to 30
(Th6Mn23 [7.112-114]). Hydrogen pressures between 1 bar [7.112-113] and
130 bar [7.115] applied at temperatures between room temperature and 100°C
have been reported. No attempts are known to separate bulk from surface
phenomena, as was done in case of LaNi 5.
The explanation of the fact that Y6Mn23 apparently loses its magnetic order
upon hydrogenation, while in the isostructural Pauli paramagnetic Th6Mn23
magnetic ordering develops when hydrogen is dissolved (see Fig. 7.6), has been
a matter of acute controversy. Various modifications of magnetic order have
been claimed depending on the specific technique which has been applied i.e.
magnetic measurements [7.112 114, 116, 117], neutron diffraction
[7.118-125], or 57Fe M6ssbauer spectroscopy on Fe-doped samples [7.126].
According to the neutron diffraction results which have been reported most
recently by H a r d m a n - R h y n e et al. for YaMn23D23 [7.121] and Th6Mn23D x
(x=16,30) [7.124], the current situation is the following: cubic Y6Mn23
deuteride undergoes a crystallographic phase transition into a primitive
tetragonal structure (PJmmm) at low temperature. In the parent intermetallic
the Mn moments are ferrimagnetically coupled. Accompanied by the structural
transition a magnetic transition also occurs at around 175 K. At still lower
temperatures weak antiferromagnetic ordering of only some of the Mn
moments is postulated which is in agreement with the interpretation of the
M6ssbauer study by S t e w a r t et al. [7.126].
Th6Mn/3D x only suffers a low-temperature distortion when x=16;
Th6Mn23D3o retains fec. structure down to liquid helium temperature.
 Magnetic Properties, M6ssbauer Effect and Superconductivity 305

x=15.7 Fig. 7.7. (a) Temperature dependenceof the mag-

2c
I x: 121
netization of Ho6Fe23Dx; (b) Curie and compen-
sation temperature vs deuterium concentration
[7.134]

16
/ ~=8.2

700J ~ i
~ x x=1.5 200
=0
60O
,o0 ~.~

500
O0 i I I I I i I p i

tO0 200 300 o ~ ~ ~3 16

Deuteriumatom.s~Eu

Whereas the former shows no long-range magnetic ordering even at 4 K, the

latter exhibits almost ferromagnetism (only 4 out of 92 Mn spins point in the
opposite direction). The difference in structural and magnetic properties of
these two compounds is attributed by Hardman-Rhyne et al. to the altered
arrangement of the deuterium atoms, to a deuterium induced charge transfer,
and to a change in the band structure owing to the specific lattice expansion.
The validity of a critical M n - M n distance for the formation of a magnetic
moment, which has been introduced by Buschow and Sherwood [7.112],
however, has been questioned. For details concerning the earlier investigations
we refer to the reviews of Buschow [,7.19, 20] and Wallace [-7.3-7].
For the remaining R6Mn23 compounds hydrogen absorption commonly
leads to a drastic reduction of the Curie temperature and of the magnetisation
[,,7.112, 127] which has sometimes even been claimed to have led to paramagne-
tism down to 4 K I-7.128, 129]. M6ssbauer studies on the rare earth nuclei
R = Dy, Tm [-7.130, 131], however, gave evidence that the small value of the
hydride magnetisation had to be attributed to an antiferromagnetic arrange-
ment of the R sublattice. From a M6ssbauer study on 57Fe-doped Er6Mn23
Stewart et al. [7.132] concluded that only some of the Mn atoms had lost their
moment after hydrogen absorption. The R moments were reported to have
retained almost their free ion value upon hydrogenation, whereas the magnetic
order of the Mn sublattice was substantially reduced. This suggests a
pronounced weakening of the R-Mn exchange interaction in the R6Mn23
hydrides.
Regarding the Fe compounds, the compensation points which are com-
monly observed for the parent intermetallics are shifted towards lower
temperatures when hydrogen (deuterium) is dissolved [7.133, 134]. This is
displayed in Fig. 7.7 taken from the study of Pedziwiatr et al. [-7.134-], where
both the linear correlation between compensation temperature and absorbed
306 G. Wiesingerand G. Hilscher

deuterium as well as the substantial rise in Tc are also demonstrated. A

reduction and an increase in the magnetisation can be observed below and
above Tcomvrespectively [7.133-135]. From a neutron diffraction study Rhyne
et al. [7.136] found that the R moments in Ho6Fe23D ~ remain close to their free
ion value upon deuteration whereas the Fe moments strongly increase. The
latter finding is confirmed by a recent M6ssbauer study of Pedziwiatr et al.
[7.137]. Similar results have been obtained for Tm6Fe23 hydride by Gubbens et
al. [7.138], while in the Lu case a decrease of the individual Fe hyperfine was
reported by Gubbens et al. [7.139] to have occurred after hydrogen uptake. The
reduction of Bef f w a s unexpectedly accompanied by a rise in the magnetic
moment. This peculiar behaviour has been interpreted by the authors to arise
from a hydrogen induced change in the conduction electron contribution to
Bcrf. At this point we want to recall that in these ferrimagnetic compounds, the
R moment is predominant at low temperatures (T < T~omp) with the Fe moment
dominating at higher temperatures. Thus a reduced magnetisation is not
necessarily due to a diminished R moment. It can equally well be the result of an
increase in the Fe moment. Besides the neutron study just mentioned [-7.136]
this has been ascertained by molecular field calculations, from which an
enhancement of the Fe-Fe interaction and a pronounced weakening of the
R-Fe interaction have been derived.
The pseudobinary system (Y, Er)6Fe23Hx has also been examined by
Pedziwiatr et al. [7.135] and was found to have similar properties to the two
boundary compounds. More attention has been paid to the pseudobinaries
[Y6(Fe, Mn)23] where spectacular magnetic properties can be observed. Two
critical concentrations for the onset of magnetic order have been obtained.
While both end compounds exhibit Curie temperatures of about 500 K, there is
no evidence for any long-range magnetic order in the concentration range
0.4<x<0.7. From an analysis of their high pressure magnetisation data
Hilscher et al. [7.140] suggested that these compounds form a spin glass system.
This proposal was later confirmed by the neutron scattering experiments of Lin
et al. [7.141]. An antiferromagnetic short-range order of atomic moments has
been claimed with a correlation length of about 2.5 lattice constants. These
results were ascertained by Hardman-Rhyne and Rhyne [7.120], but almost
simultaneously questioned by Crowder and James [7.122]. In agreement with
the neutron diffraction results, 57Fe M6ssbauer studies [7.142] revealed the
complete absence of a magnetic moment on Fe for Mn concentrations
exceeding x = 0.3. Oesterreicher and Bittner [7.143] were the first to perform
magnetisation studies on various hydrides of this peculiar system. The general
features of the host compounds remain almost unchanged, yet a shift to lower
Mn concentrations is observed after hydrogen uptake. This has been interpre-
ted as being due to the donation of electrons from hydrogen to the transition
metals. By means of X66Er and 57Fe M6ssbauer spectroscopy Zukrowski et al.
/-7.144] investigated Er6(Fe, Mn)23H x and came to the conclusion that the
behaviour of the 3d magnetisation is almost identical to that of the correspond-
ing Y series.
 Magnetic Properties, M6ssbauer Effect and Superconductivity 307

f) RTM 3 (TM = Fe, Co, Ni)

These compounds crystallize in the rhombohedral PuNia-type structure (space

group R3m, No. 166). Buschow [7.145] examined Ce-based compounds and
found CeCo 3 and CeNi 3 to be Pauli paramagnetic. After hydrogen uptake
CeCo3H 4 had become ferromagnetic (T~ = 80 K) which is a unique behaviour
for R-Co compounds and has been ascribed to a change in the valency of Ce
from 4 + to 3 +. This interpretation is suggested by the large hydrogen induced
volume increase obtained by van Essen and Buschow [7.146]. The change in the
valency of Ce upon hydrogenation can be more dearly observed in CeNi 3. The
low value of the susceptibility and its temperature independence indicate
tetravalent Ce. In the hydride, however, Z exhibits a pronounced temperature
dependence, revealing an effective moment of 2.5 pn/Ce atom which corre-
sponds to the 3 + state. In contrast to this, the ferromagnetic compound YNi 3
(T~=85 K) becomes Pauli paramagnetic after the absorption of about 4 H
atoms/f.u. [7.147].
To date only compounds containing a heavy rare earth have been
examined. In all cases ferrimagnetic behaviour has been detected, featuring
compensation points at which the resultant of the rare earth and the Fe(Co)
moments is equal to or near to zero. By studying the Co compounds Malik et al.
[7.148] found for GdCo 3 an increase in magnetisation upon charging which
they attributed to a decrease in the Co moment. In addition T¢ was observed to
be strongly reduced after the absorption of 4.6 H atoms/f.u. (a typical value for
the maximum capacity of these Co-based compounds) which is also thought to
be a consequence of the diminished Co moment, since T~ should be mainly
determined by the Co-Co interaction.
For R = Dy, Ho [7.148], Er and Tm [7.149] hydrogen absorption leads to
a reduction of magnetisation, Curie temperature and compensation point. In
contrast to the host compounds, which can easily be saturated, the corre-
sponding hydrides showed no sign of saturation in a field of /~oH=2T
(Fig. 7.8). A fanning out of the R moments as a function of temperature or
applied field was given as an explanation for this lack of saturation and the
steep variation in the magnetisation of the hydrides at low temperatures.
The magnetic properties of the parent RFe3's have been described in detail
by Herbst and Croat [7.150] by applying a two-sublattice molecular field
model. In the M6ssbauer studies on the host intermetallics, due to the presence
of a spin reorientation, the main effort was concentrated upon ErFe3
[-7.151-156]. When exposed to hydrogen up to 4 H atoms/f.u, are commonly
absorbed [7.157-161]. The R moment decreases with the amount of hydrogen,
while the Fe moment rises. As in case of the R6Fe 23 hydrides, the compensation
points again decrease with x almost linearly. A weakening of the R-Fe
exchange upon hydrogenation has thus been suggested as a likely explanation.
Direct experimental evidence for this altered exchange interaction was later
obtained by Niarchos et al. [7.160] by using both the 57Fe and the 161Dy
(26 keV) Mrssbauer effect in DyFe 3. While the Dy hyperfine field is reduced
 308 G. Wiesinger and G. Hilscher

Fig. 7.8. Magnetisationvs tempera-

80 H=21 kOe ao ture for ErCo 3 ( ) and
ErCo3H,.2 (---). In the insert the
magnetisation at T=4.2 K is dis-
played as a function of the applied
60 field [7.149]

t\
~'," 40
E
E[C/°3~ -- F --"~ -- -~ .....
__
E'c,°3"~.2
\ \
20

100 200 300

]'err,percttur e ( K )

with growing hydrogen concentration, the Fe hyperfine field increases as long

as x<~2.5, but the changes are only rather small. On the other hand, a
substantial reduction in T~omp has been obtained which showed a linear
dependence on the volume expansion. This peculiar correlation led Niarchos
et al. to the suggestion that the Dy-Fe exchange interaction is increasingly
weakened by the presence of hydrogen in the lattice and moreover that it is
primarily associated with the lattice expansion. Similar conclusions had
already been drawn by Niarchos et al. [7.159] in an earlier 57Fe and 166Er
(81 keV) M6ssbauer study on some ErFea hydrides.
Da Cunha and Vasquez [7.162] and da Cunha et al. [7.163] studied the
influence of hydrogen uptake upon the spin reorientation process in ErFe3H=.
The critical temperature was found to be drastically increased in the hydride. A
value of 210K has been given for x=1.5 which proved to be almost
independent of further hydrogen absorption. These M6ssbauer effect data are
confirmed by magnetisation studies of Malik et al. [7.161] who observed a step
in the magnetisation at a similar temperature.
In the case of the ferromagnetic compounds YFe a [-7.164] and ThFe3
[7.~ 65] an increase of the Fe moment after hydriding was reported, whereas the
Curie temperature remained unchanged in the former (545 K) but was found to
decrease in the latter. The magnetisation anomaly at about 250 K observed in
uncharged ThFe3 [7.166] is again found in the hydride and has been attributed
to the preservation of the coordination and of the Fe-Fe distances after the
charging procedure. The results obtained by Narasimhan [-7.158] for three
(Er, Th)Fe a hydrides resembled the properties of the two end compounds.
Very recently H6chst et al. [-7.167] reported on the investigation of YFe2
and YFe a exposed to increasing doses of hydrogen, by means of synchrotron
 Magnetic Properties, M6ssbauer Effectand Superconductivity 309

radiation; photoemission measurements revealed the absence of high d-like

density of states at EF for large exposures, which explains, according to the
authors, the decrease of the magnetic moment and hyperfine field (B) which is
commonly observed in the hydride for hydrogen concentrations larger than 3.5.
The authors point out that their study concerns predominantly the surface. It
should be mentioned that impurity segregation could have contributed
significantly to the observed effects.

7.3.3 Laves-Phase-Type Compounds

These compounds occur predominantely either in the cubic MgCu 2 structure

(space group Fd3m, No. 227) or in the hexagonal MgZn2 structure (space group
P63/mmc, No. 194).

a) Mn Compounds
RMnz
Both types of structure are present here (see e.g. [7.44]). About 4 H atoms/f.u.
can be absorbed whilst still preserving the crystal structure, thus leading to
remarkable volume expansions of up to 40%. For still larger hydrogen
concentrations the absence of x-ray diffraction peaks indicates at least
microcrystallinity or even an amorphous nature of the sample [7.168]. Buschow
[7.115] and Buschow and Sherwood [7.112] were the first to show that in ScMn 2
and in LuMn 2 a magnetic moment had developed on the Mn atoms after
hydrogen absorption, leading to ferromagnetic order with a T~ of about 200 K.
After cooling without an external magnetic field, the magnetisation vs
temperature curves of LuMnEHx exhibited a pronounced maximum which
disappeared when the sample, prior to the measurement, was cooled in the
presence of a magnetic field. This spin-glass-like behaviour was also obtained
for YMn 2 hydride, whereas ThMn2 was found to remain paramagnetic after
having absorbed hydrogen. As is the case for the 6 : 23 compounds, here too a
nonuniform behaviour is found upon hydrogen uptake when a nonmagnetic
element is present. All the host compounds just mentioned were considered to
be paramagnetic. Recently, however, Nakamura [7.169] found YMn 2 to be an
antiferromagnet with a Mn moment of 2.7#a aligned parallel to a [-111]
direction. This has been confirmed by a set of supplementary experiments: low
temperature x-ray diffraction, magnetic and dilatometric measurements,
neutron diffraction and spin-echo NMR. The N6el temperature was deduced to
lie around 100 K and was found to be accompanied by an unusual large volume
change of about 5%. This peculiar finding has been ascribed by Nakamura to
the collapse of the Mn moments above TN due to the strong spontaneous
volume magnetostriction present in this material.
In GdMn 2 hydrogen uptake was found by Buschow and Sherwood [7.112]
to reduce the magnetisation at low temperatures, whereas for temperatures
310 G. Wiesingerand G. Hilscher

above about 100K an increase was obtained. This unique behaviour is

explained by the authors in terms of an alignement of the Gd moments. This is
thought to be caused by the Mn moments, which are assumed to order close to
the Curie temperature of YMn 2 hydride (Tc=284 K). From magnetisation
studies on YMnEH x and LuMn2H x with varying hydrogen content Buschow
and Sherwood proposed the existence of a critical M n - M n distance for the
formation of a local moment at a Mn site, as has already been suggested for the
R6MnEa hydrides. A value of d = (2.86 + 0.06)A has been determined for both
compounds, similar to that for Th6Mn23H x, but substantially larger than that
of (Y, Lu)6Mn23Hx. This was attributed to the different valence electron
concentration in the respective compounds. However, as already mentioned in
Sect. 7.3.3 the concept of critical distance has been questioned by neutron
diffraction groups (see e.g. [7.124]. 161Dy and 166Er M6ssbauer studies have
been reported for DyMnEH x [7.131] and ErMn2H ~ [7.170] respectively.
Magnetic ordering was found to have disappcarcd after hydrogen absorption.
The magnetic hyperfinc field at 1.5 K of about 500T in ErMnEH4.6 was
explained by intermediate relaxation rates. As has been mentioned earlier, the
isomer shift values in the ternary hydrides are again found to lie close to those of
the binary R hydrides.
TiMn 2

Hempelmann and Hilscher [7.171] found three Ti-Mn intermetallics to bc

stable at low temperature: the C14 Laves phase with a wide range of
homogeneity reaching from TiMn 2 to TiMn~.2, the Q-phase TiMnl. ~7 and the
~b-phase TiMnL08. While all parent compounds exhibit temperature indepen-
dent paramagnetism, the three hydrides are ferromagnetic with Curie tempera-
tures of about 212 K, 87 K, and 155 K respectively. The onset of fcrromagne-
tism was attributcd by Hempclmann and Hilscher to both volume expansion
and an increase in N(EF) upon hydrogen absorption. In the context of the large
homogeneity range of thc Laves phase compound, an incrcas¢ of the lattice
constants was observed upon increasing the Ti concentration. Simultaneously
the hydrogen capacity was found to grow, whereas stoichiometric TiMn 2
absorbs no hydrogen at all. Thus the existence of a critical range of the lattice
dimensions for the absorption of hydrogen has been suggested which is in
agreement with results obtained by Oesterreicher and Bittner [7.172] on
(Ti, Zr)Mn 2. However, this may not be a general result.
M6ssbauer studies performed by Hempelmann et al. [7.173] on TiMnL5
doped with 57Fc yielded an asymmetric quadrupole doublet with a compara-
tively small splitting which, however, was strongly enhanced after hydroge-
nation (Fig. 7.9). From inspection of the figure it is obvious that upon
hydrogenation the shape of the spectrum has changed to complete symmetry
with narrow lines. The asymmetric doublet of the parent compound was
attributed by Hempelmann et al. to a non-random distribution of the excess Ti
atoms on the 2a sitcs, which was later confirmed by neutron diffraction
experiments [7.174]. The presence of only one hyperfine pattern after
 Magnetic Properties, M6ssbauer Effectand Superconductivity 311

Fig. 7.9. M6ssbauer spectra, recorded at room

T " 29S K
temperature for Tio.4oMno.6o(SVlTe)(A), TiFe2
],
(B), and Ti0.+oMno.6o(SVFe)Hz.o8(C) I-7.173]

.97
,i
1.

O
L;3 .95
O'h
t-4 B) ii TIFE2
(r)
Z
c~

J
LJ
Off

.97
c)

40MN.60(STFe)HI.0~

-3 I 1 3

VELOCITY (HM/S)

hydrogenation was explained by the authors as originating from the large

lattice expansion (A V/V= 27%) which should release the space restriction for
the (Ti, Mn) tetrahedra, Thus reasons for lattice distortions are no longer
provided and all Fe atoms are now assumed to experience the same EFG, very
similar to that one in the isostructural TiFe 2. It should be pointed out that an
alternative explanation for the occurrence of narrow line widths after hydrogen
absorption can be given by taking into account motional narrowing effects
caused by rapid fluctuations of the hydrogen atoms.
In this work Hempelmann et al. also made an attempt to determine the type
of ferromagnetism in the ternary Ti-Mn hydrides. They used the so-called
Rhodes-Wohlfarth plot in order to classify ferromagnetic order into localized
or itinerant type. We recall that the crucial parameter in this model is the ratio of
magnetic carriers in the paramagnetic and in the ferromagnetic states, qc/qs as a
function of the Curie temperature. In terms of this analysis the Ti-Mn Laves
phase turned out to exhibit localized moments which is confirmed by the fact
that the Arrott-plots (a 2 vs B/a) are not parallel. In the case of the Q- and the
~-phases however, the rather large values of qc/qs with the corresponding Curie
temperatures fit well into the itinerant branch of the Rhodes-Wohlfarth plot,
which led Hempelmann et al. to the conclusion that these compounds might
exhibit pure band ferromagnetism.
The authors also commented on the onset of ferromagnetism upon
hydrogenation of the Laves phase TiMnl.5, by considering the large increase of
312 G. Wiesinger and G. Hilscher

the temperature-independent term of the paramagnetic susceptibility as well as

of the specific heat coefficient 7 observed after hydrogen absorption. This
finding has been correlated with an increased density of states at the Fermi level
N(Ev) which continues to grow in the hydride when the amount of hydrogen is
further increased. Within the framework of this simple picture the high value of
N(E~) in the hydride was interpreted by Hempelmann et al. as one reason for
the onset of ferromagnetism (Stoner criterion). Since the lattice expansion
caused by hydrogen absorption brings about a narrowing of the d-band and
thus an increase of N(Ev), the interpretation of the YMn/results by Buschow
and Sherwood [7.112] emphasizing this lattice expansion, was adopted by
Hempelmann et al. within their model.

ZrMn 2
In a similar way to TiMn 2, the isostructural ZrMn 2 displays a wide
homogeneity range, shifted however, to a larger Mn content, i.e. from ZrMnl. s
[7.175] up to ZrMn3. s [7.176]. Such compounds exhibit favourable hydrogen
absorption properties and compositions close to ZrMn2. 4 are particularly
useful, since their equilibrium pressure is around 1 bar, whereas in stoichio-
metric ZrMn 2 it is two orders of magnitude lower [7.176].
Several authors [7.176-178] agree that upon hydrogen absorption Pauli
paramagnetic ZrMn 2+x becomes a spin glass; this was also found in the case of
several multi-elemental storage compounds, where Mn had been partly
replaced by Fe, Cr, or Co [7.179]. The reason for this probably lies in the
presence of strong segregations which have been detected by means of XPS
[7.56].
Substantial discrepancies in the magnetic ordering temperatures of the
hydrides can be found in the literature which we attribute primarily to the
different hydrogen pressures used during the charging procedure. Further
reasons may be different stoichiometries and hydrogen concentrations. A
recent neutron diffraction study performed by Didisheim and Fischer [7.180] on
ZrMn2D 3 demonstrated that the magnetically ordered state in this compound
is in fact of great complexity.

b) Fe Compounds
Numerous investigations have been carried out on the RFe2's and their
hydrides; this is due to their comparatively simple crystallographic and
magnetic structure. Thus results from experiments such as neutron diffraction
or M6ssbauer spectroscopy can be interpreted more easily than those of
transition-metal-rich compounds. Since for the a magnetic rare earths,
magnetisation measurements are insuffÉcient to separate the R from the Fe
moment, the above experiments provide an ideal method to examine the
influence on both kinds of atoms on the magnetic moment.
Apart from ScFe2 (C14) all parent compounds display the cubic C15
structure. Stable hydride phases with ~ 2 and ~ 3.5 H(D) atoms/f.u, are known
 Magnetic Properties, M6ssbauer Effectand Superconductivity 313

from x-ray and neutron diffraction to retain the parent structure with lattice
parameters increased by ~ 5 % and --~7% respectively. For a still larger
hydrogen concentration (RFezH4) a rhombohedral distortion [7.181] or even
the distruction of the long-range lattice periodicity [7.97] has been detected.
Thus the M6ssbauer spectra of the hydrides sometimes exhibit a certain line
broadening, predominantely when recorded at low temperatures which
alternatively may be due to intermediate jump rates of the H atoms.
When the partner element of Fe does not carry a magnetic moment (Sc, Y,
Ce, Lu) a substantial increase in magnetisation, s 7Fe hyperfine field and isomer
shift has been observed [7.84, 97, 165, 182-185]. For YFe 2 Buschow and van
Diepen [7.182] found the Curie temperature to have decreased upon hydroge-
nation, while in CeFe2 hydride they observed a rise in Tc [7.165]. The largest
increase in the magnetic moment has been detected in ScFezHx which was also
found to exhibit an appreciable homogeneity range. Gr6ssinger et al. [7.186]
studied the influence of misplaced atoms upon the magnetic properties of non-
stoichiometric parent compounds using pulsed fields and M6ssbauer spec-
troscopy. An easy c-axis was reported and a drastic reduction of the anisotropy
field with increasing Fe content in the sample. A strong increase in the Fe
moment from 1.4 to 2.2/~R and in the 57Fe hyperfine field from 24 to 30 T after
hydriding has been observed by Smit and Buschow [7.184] which, however, was
found to be less pronounced when an Fe excess is present in the sample [7.185].
This peculiar finding has been attributed to a different shielding of the Fe
sublattice from the Sc sublattice by the H atoms in the case of an altered amount
of Fe in "ScFez".
From the point of view of magnetic behaviour, the heavy RFe2 hydrides
have been particularly thoroughly investigated. The question about the
influence of hydrogen upon the magnetic moments can be readily answered,
because in this case decisive neutron diffraction results are now available. No
such reliable conclusions, however, can be drawn from magnetisation or
M6ssbauer measurements. On the one hand, saturation is scarcely achieved
[7.168, 187-189] and thus great discrepancies in the magnetic moments can be
found in the literature; on the other hand particularly the 5~Fe M6ssbauer
spectra exhibit a substantial line broadening and wide hyperfine field distri-
butions after hydrogenation [7.168, 181, 182, 188-192]. This brings about
difficulties especially in determining the magnetic ordering temperature from
the onset of the magnetic hyperfine splitting. One can, however, rely on the
sublattice moments derived from liquid helium temperature spectra.
As a typical example from the neutron diffraction studies of Fish et al.
[7.193] and Rhyne et al. [7.194] we choose ErFezD x (Fig. 7.10), although it has
already been presented in the latest review of Buschow [7.20]. It represents,
however, a case in which hydrogenation actually causes different ordering
temperatures for the sublattice magnetisation. From an inspection of Fig. 7.10
the reduction of the overall Curie temperature on hydrogenation by as much as
50% of the host value can be deduced. The magnitude of the Fe sublattice
magnetisation is sometimes observed to remain constant; in case of
314 G. Wiesingerand G. Hilscher

Fig. 7.10. Magneticmoment of Er and Fe vs

temperature in ErFe2 (tz) and ErFe2D3.5
(o) [7.194]
~ SubtethceMdgnehc Moment
"q% uEr Fe2
• Er Fe2 D25

nx
~do 260 ~do ~ o " 5bo 6oo~'
T(K)

HoFezH(D)~ an increase in the Fe moment with x can be detected. Yet the flat
temperature dependence of the Fe sublattice magnetisation is essentially
unaffected by hydrogen absorption in each case. As already mentioned in the
previous section, the rise in the magnetic moment of Fe upon hydrogen uptake
is only observed for x < 3.5. This has likewise been demonstrated for some RFe2
hydrides by Dunlap et al. [7.190] who obtained a breakdown of the 57Fe
hyperfine field in ErFe2H4.1.
The effect on the rare earths is more pronounced: the 0 K moment in the
hydride is remarkably reduced from its free-ion value in the parent compound.
Moreover the moment declines rapidly with elevated temperature, sometimes
reaching zero well below the overall Curie point. This is in contrast to the
behaviour in the pure intermetallics, where both sublattice moments disorder at
identical temperatures. These features have led to the conclusion that a severe
weakening of the R-Fe as well as of the R - R exchange interaction takes place
upon hydrogen absorption. The hypothesis of a reduced R-Fe exchange is
experimentally verified by the lowering of the compensation points with
hydrogenation [7.188, 192] which has already been referred to for the cases of
the R6Fe23 and RFe 3. The relative insensitivity of the Fe-Fe exchange upon
hydrogen uptake has been explained by a nearest-neighbour direct overlap
exchange which should only be slightly influenced by the presence of hydrogen
atoms. These suggestions, however, are inconsistent with certain results from
magnetic measurements where the simple antiparallel arrangement of R and Fe
spins is assumed to be retained in the hydrides as was been proposed prior to
the neutron diffraction work [7.195]. More probably a "fanning" of the loosely
coupled R moments takes place, arising from random local anisotropies [7.181,
188, 190, 194]. This too would explain the difficulties experienced in saturating
the RFe2 hydrides and the discrepancies in the magnitude of the Er moment as
determined from neutron diffraction and from M6ssbauer spectroscopy
 Magnetic Properties, M6ssbauer Effect and Superconductivity 315

Very recently the question concerning the reduction of the R moment in the
hydrides has be taken up again by de Saxce et al. [7.188]. They examined
ErFe2Hx with variable hydrogen content by means of high field magnetisation
measurements and 167Er pulsed spin echo NMR. The R moment was found to
be substantially reduced only for x > 3. For lower hydrogen concentrations the
ferrimagnetic structure seemed to have been retained. For x > 3.3 no evidence
for saturation at 4.2 K could be obtained even in a field of 16 T. This finding and
the field dependence of the susceptibility has been attributed by de Saxce et al.
to the tendency to ferromagnetic alignement of the Er and Fe sublattices rather
than to the fanning of the R moments.
Cohen et al. [7.168] collected isomer shift data for several DyTM 2 hydrides.
Once more the common trend could be shown: a considerable reduction of the
Dy isomer shift with rising hydrogen content in the sample corresponding to a
decrease in the s-electron density at the nuclear site. This reaches values which
almost coincide with that of the binary Dy hydride.

c) Fe-Containing Pseudobinaries
The group of Shaltiel initiated the work on Zr-based pseudobinaries
[7.196-199]. Various combinations have been examined in order to study the
influence of alloying upon the formation and properties of the hydrides. Several
concentration-dependent C15-C 14 (and C 14-C 15) changes of structure have
been observed which, after hydrogen absorption, were found to have been
preserved.
More recently extensive investigations have been started by the Pittsburgh
group on Zr(Fe, Mn)z like systems, where it has been established that
particularly the hyperstoichiometric compounds can serve as hydrogen storage
materials [7.178, 179, 200-205]. In pure Zr(Fe, Mn)2H x the magnetic order is
enhanced upon hydrogenation in the whole composition range. This result
obtained from bulk magnetic measurements is confirmed by a recent 57Fe
Mrssbauer study, where a substantial increase in B~r~ has been observed
[7.206]. Similar results have been reported on Zr(Fe, Cr)2 [7.177, 179, 196]. In
the case of Zr(Fe, V)2 hydrogen absorption leads to an enhancement of
ferromagnetism in the Fe-rich region only, whereas a suppression of super-
conductivity was found in the V-rich range [7.207] (see also Sect. 7.4). By
contrast, in Zr(Fe, A1)2 ferromagnetism is always strongly suppressed on
hydrogenation [-7.178, 204, 208]. Provided that the hydrides just mentioned
exhibit magnetic ordering spin-glass-like behaviour is commonly observed.
59Co and SSMn N M R and 57Fe Mrssbauer spectroscopy have been applied
by Fujii et al. [7.209] and by Okamoto et al. [-7.210] to examine Y(Fe, Co)2H x
and Y(Fe, Mn)2Hx respectively. Their results are summarized in Fig. 7.11. In
the former case for Co concentrations x > 0.1 a common decrease of Fe and Co
moment upon hydrogen absorption has been quoted. In the latter case a
hydrogen induced enhancement of the Mn moment has been deduced from
N M R frequency shifts. Competing coupling tendencies between the magnetic
 316 G. Wiesinger and G. Hilscher

60O
Y :Fe l_yMoy) 4~ Y(Fel_yH)~
3

400~ o s
500

~ 2t00

2O0
!
lO0

0 , I I
0 02 04 06
y-.,,,..
08 ~0 0 o'2 o'., £6 ~8 ~0
Y ~

I000 4 YF
(el-yCO,y9)

5OO

&

i b i
0 0,5
Y ~
ZO O.5 z0

Fig. 7.11a, b. Magnetisation and Curie temperature vs concentration y of(a) Y(Fel _yCor)2 (0)
and their hydrides (t) 1-7.209]. (b) Y(Fel_yMny)2 (0) and their hydrides (e) [-7.210]

atoms were suggested as an explanation for the spin-glass-like magnetism in the

Mn-rich hydrides. Oesterreicher and Bittner [7.211] reported to have found
some indications of partial amorphicity in this system.
Some R(Fe, Co)2 hydrides have been studied by Pourarian et al.
[7.212,213], while Sankar et al. [7.214] reported on Er(Fe, Mn)2H=. The
renowned arguments which have been applied to the ternary hydrides are again
used on this occasion. The verification of the postulate by Pourarian et al. that
the local moments residing at the Fe sites are retained in the Co-rich range
should be taken as a challenge to M6ssbauer spectroscopists. By means of EPR
studies Drulis et al. [7.215] showed that in hexagonal GdFeAI the magnetic
moment increases on hydrogen absorption up to the free-ion value of Gd 3 ÷
This was ascribed to a complete randomization of the Fe moments due to the
weakened Gd-Fe exchange. Their assumption of an ordered Gd sublattice at
low temperatures seems to be questionable; the breakdown of the Fe-Fe
exchange interaction, however, is strongly supported by M6ssbauer studies on
several RFeA1 hydrides [7.216] which, even at 4.2K, yielded almost no
magnetic hyperfine splitting.
 Magnetic Properties, M6ssbauer Effect and Superconductivity 317

d) Compounds of Co, Ni and Others

In case of the Laves phases containing Co, which are all of the cubic C15
structure, the customary reduction of magnetic ordering after hydrogen
absorption is obtained. This has been demonstrated by Buschow [7.217] for
GdCo2H= and by de Jongh et al. [7.218] for PrCo2H x. Clusters of free Co were
found to give rise to spin-glass behaviour. This was shown by susceptibility
measurements [7.218] and confirmed by Buschow and van der Kraan [7.219] in
a study of 5VFe-doped YCo2H4 and HoCo2H 4. In the M6ssbauer spectra a
hyperfine pattern could be detected which almost coincided with that of Fe
impurities in Co metal. For GdNi z Malik and Wallace [7.220] found a loss of
the long-range atomic order upon hydrogen uptake which was accompanied by
a drastic decrease in both the magnetisation and the Curie temperature. This
trend, which commonly is observed in Ni compounds, has again been observed
by Drulis et al. [7.215] who performed EPR measurements on GdNiA1H~.
As already mentioned in Sect. 2.1, Gd compounds containing a non-
magnetic metal make it possible to study directly the influence of hydrogen
upon the R - R coupling. This has been done by using the 86.5keV 155Gd
M6ssbauer transition in GdCu2Hx [7.48] and in GdM2H =(M = Ru, Rh) [7.47].
Upon hydrogenation the Curie temperature behaves very differently in the two
cases: a strong rise is observed in the former case and a decrease in the latter.
This peculiarity has to be seen in the light of the influence of hydrogen on the
oscillatory RKKY-type interaction present in these compounds. Graaf et al.
collected numerous isomer shift and hyperfine field data for binary Gd
compounds and some of their hydrides [7.221]. Hydrogen absorption com-
monly yields a substantial increase in isomer shift which is consistent with a
reduction in s-electron density at the Gd nuclei. This indicates a charge transfer
from Gd to hydrogen, a behaviour which is generally observed when rare earth

06

mGdh
2

04 o
GdCu 2 H x

GdRu2H x
• GdRh2Hx
2 021
GdCu 2

I Fig. 7.1Z Correlation between 155Gd isomer

0 • o shift (relative to 115Eu :Pd) and hyperfine field
GdNu 2 Gd in various intermetallic compounds GdM2 and
J J I their hydrides [7.48]
o 20
Hhrcr)
318 G. Wiesinger and G. Hilscher

M6ssbauer data are considered. For several G d M 2 compounds and their

hydrides Graaf et ai. also claimed an almost linear correlation between Gd
isomer shift and Gd hyperfine field (Fig. 7.12). The positive conduction electron
contribution to the Gd hyperfine field is seen to be increasingly reduced with the
(s-like) charge density at the nuclear site. This leads eventually to the
semimetallic GdH 2, from which Buschow [7.20] derived the negative core
contribution to Beff.
Very dissimilar hydriding behaviour has been obtained for two Eu
compounds: whereas in EuRh2, Eu is found to change its valency (3 + ~ 2 +)
upon hydrogen absorption [7.222, 2233, it remains divalent in EuMg2Hx
[7.91]. The hyperfine parameters of the hydrides were found to approach those
of EuH2; this has been attributed to significant clustering. Huang et al. [-7.224]
reported that the absorption of hydrogen had destroyed magnetic order in the
weak itinerant ferromagnets Ti(Be, Cu)2 and ZrZnl. 9. This has been explained
by the influence of hydrogen upon the peculiar band structure present in these
materials. By proton N M R studies combined with susceptibility measurements
Zogal et al. [-7.225] recently studied Fe2P-type ThNiAIHx and UNiA1Hx. The
existence of two antiferromagnetic phases has been claimed for the latter.

7.3.4 TiFe and Related CsCI-Type Compounds(Space Group Pm3m, No. 221)

Together with LaNi 5, TiFe belongs to the compounds which have most
frequently been investigated with respect to their hydriding behaviour. When,
more than ten years ago, Reilly and Wiswall Jr. [-7.226] first identified the three
different hydride phases, an enormous interest in TiFe hydrides developed. The
main reason of course was the compound's excellent storage properties which
promised useful applications.
Very soon M6ssbauer spectroscopy was included in these investigations: by
comparing transmission with back scattering spectra (using conversion
electrons - CEMS) Ron et al. [7.227] were able to show that, after hydrogena-
tion, Fe precipitates had formed close to the surface. Subsequently Hempel-
mann and co-workers [7.228, 229] studied TiFe during hydrogen treatment by
means of susceptibility measurements. Although almost temperature indepen-
dent to begin with, the susceptibility was found to increase substantially
with rising numbers of charge-discharge cycles. Finally the sample behaved as a
superparamagnet, a fact which has been attributed to Fe-rich precipitates that
form magnetic clusters. Hempelmann and Wicke [7.229] supported this
assumption by applying a simple band model from which they deduced that
charged TiFe should not be stable in a well-ordered structure, but that the
formation of Ti and Fe clusters is favoured.
Nevertheless the finding ofRon et al. was confirmed by Shenoy et al. [7.230]
and by Bliisius and Gonser [7.231] who repeated the comparative
transmission/CEMS study. From the latter work we have taken the surface
emission spectrum of an activated TiFe disc which is displayed in Fig. 7.13; the
ct-Fe hyperfine pattern is clearly visible, whereas it is completely absent in a
 Magnetic Properties, M6ssbauer Effect and Superconductivity 319

Fig. 7.13. Conversion electron M6ssbauer

spectrum of a hydrided TiFe disc recorded at
1 04 room temperature I7.231]

1.03
g
ql

~ 1.02

~_1.01

100

-6 -~ -2 0 2 ~ 6
Velocity [mm/s]

transmission spectrum. This distinctly points to the fact that the clusters form
predominantely at the surface (d< 1000 A).
The group of Schlapbach performed several, chiefly surface-sensitive
experiments on the activation behaviour of TiFe [-7.73, 89, 232-2373 described
also in the comprehensive review of Schlapbach and Riesterer [7.238]. For
uncharged TiFe, neutron scattering studies indicated complete CsCI order
which was found to diminish only insignificantly upon charging. The increase in
the magnetisation after hydrogen absorption was concluded to originate from
ferromagnetic precipitates at the surface, in agreement with the Mrssbauer
results. Furthermore several electron microscopy studies have been carried out
and have yielded somewhat contradictory results. Fon Waldkirch et al. [7.239]
reported Fe precipitates at the surface of TiFe, whereas Schober and Westlake
[7.240] and Khatamian et al. [7.241,242] claimed a decomposition into
different oxides and TiFe 2. These discrepancies have been attributed by Reilly
and Reidinger [7.243] to the different preparation techniques employed.
The different hydride phases detected by Reilly and Wiswall, Jr. [7.226] were
found by Swartzendruber et al. [7.2443 to give rise to specific hyperfine patterns:
the cubic a-phase exhibits a single line spectrum and for the orthorhombic fl-
[7.233,245, 246] and the monoclinic y-phase [7.247] quadrupole split doublets
were obtained. Later on this was confirmed by Schdfer et al. [7.2463. As
commonly observed in Fe compounds, a distinct increase in isomer shift with
hydrogen content is observed which recently has been interpreted by Gupta
[7.55] using the augmented plane wave (APW) method. We recall that the
negative isomer shift of TiFe relative to ct-Fe ( - 0 . 1 4 mm/s) is cancelled in case
of the r-hydride. For the y-phase a further increase in isomer shift (+ 0.27 mm/s)
is observed. Both values are in close agreement with the results derived by
Gupta.
320 G. Wiesinger and G. Hilscher

Mizuno and Morozumi [-7.248] studied TiyFe (0<y_-< 1) alloys and corre-
lated a minimum in the M6ssbauer line intensity with the maximum of the
storage properties for y ~ 1.15 due to a weakening of the Ti-Fe binding forces.
In our opinion the change in intensity is more likely to be due to the disturbed
structure in the nonstoichiometric alloys which should strongly influence the
f-factor.
If Co is alloyed to TiFe, a system is formed which exhibits unique magnetic
properties thus and has been the subject of numerous investigations: while the
boundary compounds TiFe (y=0) and TiCo (y= l) are Pauli paramagnets,
ferromagnetism is observed in an intermediate concentration range
(0.25 < y <0.65) I-7.41,251]. M6ssbauer measurements by Bennett and Swart-
zendruber [7.250], however, showed only a single line spectrum, giving no
evidence for the existence of a localized moment at the iron site. Subsequently
NMR studies on all nuclei present I-7.251-253] were applied to elucidate this
peculiar finding. The Knight shift results were explained by a greater
s-admixture and a greater d-spin moment at the Fe sites than at the Co sites. A
change of the sign of the hyperfine coupling constant was then presumed which
should lead to a cancellation of Bcrf in the intermediate concentration range
[-7.253]. On the other hand Piekart et al. I-7.254] reported on polarized neutron
measurements from which a moment of 0.15 #B for the (Fe, Co)-sublattice was
derived for the Co concentrations y=0.45, 0.50, and 0.55.
More recently, several theoretical [-7.255, 256] as well as experimental
studies [-7.41, 140, 257-261] on Ti(Fe, Co) have demonstrated the existence
of so-called antistructure (AS) atoms, i.e. misplaced Fe(Co) atoms on
the Ti sublattice. Since in such a case an Fe atom is surrounded by 8 Fe(Co)
neighbours a local magnetic moment at these atoms may form. Giner and
Gautier [7.255] were the first to point out the remarkable difference of the
density of states (DOS) between atoms on correct sites and AS atoms; they
showed this by solving the tight binding Hamiltonian using the coherent
potential approximation (CPA). From the position of the Fermi level in the
high-density region of the As-Fe and AS-Co DOS Giner and Gautier
concluded that these atoms could indeed play an important role in the onset of
ferromagnetism as observed by magnetic measurements. By performing
K K R - C P A calculations Schadler and Weinberger [-7.256] showed recently that
Fe-AS atoms are responsible for the occurrence of magnetic order Since the eg
and the tzg-like virtual bound states of the Fe-AS atoms were found to lie very
close to the Fermi energy with an exchange splitting much larger than that of
the Co atoms.
By studying magnetovolume effects of some off-stoichiometric (Fe, Co)
compounds Buis et al. [7.258] concluded that the ferromagnetic behaviour has
to be ascribed to the presence of Fe(Co) atoms in a highly susceptible matrix.
The pressure dependence of the spontaneous magnetic moment and of the
Curie temperature are shown in Fig. 7.14, where the pressure derivatives of T~
and of ao are plotted as a function of T~. The diverging pressure derivative at
decreasing To values obtained for the Co-rich stoichiometric and for the non-
 Magnetic Properties, M/Sssbauer Effect and Superconductivity 321

x ,t0"2 x ;0 -2
-15 (a) -15 (bJ
i •
I
I
u
i
i
13 .c~ i
-10 I
ol
° ~,

~O

\
-ta
_ . o _ _ "-o-. o..~.°.%_ ~ ;,
-o---o--~3o~_.~
L
I I I
50 tO0 5O I00
rc ~K)
Fig. 7.14a, b

Fig. 7.14. Pressure derivatives of the Curie tem-

perature (a) and magnetisation (b) of stoich-
F J j ~ iometric Ti(Fe, Co) compounds (e, Co-rich; o,
12..

a"~f Fe-rich) and off-stoichiometric Ti=(Feo.sCoo.s)l

compounds (D) [7.260]
-=

12,,

! /
b 'i /
Fig. 7.15. M6ssbauer spectrum recorded at 4.2 K
ti.97. for Ti(Feo.6Coo.4); the bars on top of the spectrum
indicate the line positions [7.41]
.7 -5 -3 -I t 3 5
velocity [minis1

stoichiometric compounds clearly point to band-type magnetic behaviour. The

constant values in case of the Fe-rich stoichiometric compounds can be
explained by the presence of local magnetic moments at Fe-AS atoms.
Until recently, however, no direct experimental evidence for AS atoms has
been presented; this is connected with the apparent puzzle regarding a zero
hyperfine field within the ferromagnetic region. By M6ssbauer measurements
with an extended measuring time yielding count rates of about 12 million per
channel Hilscher et al. [7.41] finally succeeded in resolving a weak magnetic
hyperfine pattern which is otherwise buried in the background (see Fig. 7.15).
322 G. Wiesinger and G. Hilscher

Fig. 7.16. Magnetisation vs concentration of

12 Ti(Fel_ yCoy) (e) and of some or-phase hydrides (o);
! \ the broken line represents the data obtained in a field
of 7 T I-7.41]
l0

I
~'6 I
I
¢N

02 04 06 08
Y~

The intensity and splitting are in good agreement with the values predicted
from bulk magnetic measurements for the Fe-AS atoms.
If the pseudobinary compounds are hydrided they behave very similarly to
TiFe [7.41, 261, 2623. Again ct-, fl-, and y-phases are found which exhibit
equivalent M6ssbauer spectra to those observed for TiFe by Swartzendruber et
al. I-7.244]. Hilscher et al. [-7.41] tried to interpret the rise in T¢ upon
hydrogenation as a negative pressure effect by comparing their results with high
pressure magnetisation data [7.140, 263, 2643. This, however, could successfully
be done only for low Co concentrations (small changes of T~). For higher Co
content and larger increase in Tc(~40 K) no agreement could be obtained at all.
Here we want to recall that pressure experiments dearly point to the
presence of localized moments in the Fe-rich regime, whereas for the Co-rich
side of the system they point to itinerant ferromagnetism. The change of To and
o-upon hydrogenation is positive for y < 0.45 (Fe-rich) and negative for y > 0.45
(Co-rich) (see Fig. 7.16). Therefore, not even a qualitative correlation between
the hydrogen induced change of T~and its pressure dependence can be expected
for the Co-rich compounds. However, with the assumption that additional AS
atoms are created upon hydrogenation the magnetic properties of the
Ti(Fe, Co) hydrides could be explained at least qualitatively: the number of
Fe-AS atoms carrying the local moments decreases, while N(EF) increases with
the amount of Co. The occurrence of two critical concentrations for the onset of
ferromagnetism (y = 0.25 and y = 0.65) is suggested to be a consequence of these
two competing phenomena. The authors concluded that despite the presence of
surface segregation the dominant feature responsible for the complex magnetic
behaviour of this system is the presence of local moments which trigger the
long-range magnetic order.
The paramagnetic behaviour of both the fl- and the 7-phase was attributed
by Hilscher et al. I-7.41] to the fact that the formation of Fe-AS atoms is
impossible in these non-cubic compounds. By performing proton spin-lattice
 Magnetic Properties, M6ssbauer Effectand Superconductivity 323

relaxation studies on TiCoH:, Bowman et al. [7.265] very recently emphasized

the sensitivity of the position of EF to hydrogen absorption. Additional studies
of this kind in, combination with band structure calculations would give a
deeper insight into this complex problem.

7.3.5 Compounds of Low Transition Metal Content

MgzNi (CuA12 type of structure) belongs to the group of suitable hydrogen

storage materials. By applying surface-sensitive techniques (AES, XPS, FMR)
Schlapbach et al. [7.266], Stucki and Schlapbaeh I-7.73] and Shaltiel et al. [-7.89]
could demonstrate that after continuous hydriding a decomposition at the
surface takes place. As in LaNis (see Sect. 7.3.2) this gives rise to the formation
ofsuperparamagnetic clusters. The bulk susceptibility, however, decreases after
hydrogen absorption. Very recently Aubertin et al. [7.267] reported on
M6ssbauer studies on isostructural Fe-doped ZrzNi hydrides. The large
hydrogen induced rise in isomer shift ( + 0.58 ram/s) has been attributed largely
to charge transfer effects.
Eu2IrH5 has been assigned by Moyer Jr. and Lindsay [7.268] to the space
group Fm3m (No. 225) and is claimed to order ferromagnetieally below 20 K,
as has recently has been corroborated by an 151Eu M6ssbauer study of Stadnik
and Moyer Jr. [7.269]. Remarkable similarities to binary EuH2 have been
emphasized.
The remaining compounds with 2:1 stoichiometry which have been
investigated with regard to their hydriding properties are all of the cubic Ti2Ni
type of structure (space group Fd3m, No. 227). Hf2Fe was found by Buschow
and van Diepen [7.270] and by Tuscher [7.271] to be converted from a Pauli
paramagnet to a ferromagnet upon hydrogen absorption. Recently Vulliet et al.
[7.272] extended this study to variable hydrogen content and detected the
maximum magnetic moment at a concentration of 3 H atoms/f.u. The
broadened M6ssbauer spectra were interpreted to reflect a disordered magnetic
phase.
Of the Ti-based compounds, Ti2Ni, Ti2Co and the pseudobinaries between
these two and Fe have been examined. Pure Ti2Fe, however, does not form
[7.273, 274]. Nevertheless it can be stabilized by small amounts of oxygen
(between 6% and 14% depending on the exact Ti: Fe ratio) [7.275-278]. The
resulting ternary oxide (q-phase) is believed to play a significant role in the
activation process of the storage compound TiFe [7.238, 279] which for large
scale applications is frequently present as ferrotitanium, where substantial
amounts of oxygen are dissolved.
The oxygen-free compounds Ti2Co and Ti2Ni were reported by Tuscher
[7.274] and by Hiebl et al. I-7.273] to be exchange-enhanced Pauli paramagnets
displaying a complex temperature dependence. After hydrogen uptake a
temperature independent susceptibility was obtained. Isostructural Ti2FeOo. 5
shows a Curie-Weiss-like behaviour [7.275]. After hydrogenation an increase
324 G. Wiesingerand G. Hilscher

in the susceptibility was detected, its magnitude depending on the hydrogen

content in the sample. For a hydrogen content exceeding 2 H atoms/f.u.,
magnetic ordering at low temperatures was inferred from the substantial
broadening of the N M R lines.
Ternary oxides Ti2 _rFe20~ with varying Ti : Fe ratios and oxygen contents
were studied earlier by Mintz et al. [7.276] and were more recently comprehen-
sively reinvestigated by Rupp 1,7.278] and by Rupp and Wiesinger I-7.280]. For a
detailed reference list covering the field of hydrogen uptake by oxygen
stabilized Ti2Fe we refer to the paper of Rupp. From the Mfssbauer spectra, a
distribution of electric field gradients and isomer shifts due to incomplete
occupation of lattice sites was deduced. At 4.2 K magnetic ordering upon
hydrogenation has been observed only in the case ofTizFeOo.zHz.6; paramag-
netism, i.e. a quadrupole split spectrum, has been obtained for all the remaining
hydrides. For/3-Ti (20 at.% Fe) as much as +0.66mm/s was found for the
hydrogen induced change in isomer shift, reflecting both a large increase in
volume and a significant change of the electronic structure after charging.
Additionally Rupp and Wiesinger could demonstrate that segregations of
magnetically ordered Fe in the parent alloys are a consequence of an activation
heat treatment under rough vacuum conditions. The hyperfine pattern
characteristic for ~-Fe which occurred in the transmission spectra indicated
that those segregations had already achieved a volume large enough to be
visible for 14.4 keV y-rays. Very recently Rogl et al. [7.281] reported on a
further ternary oxide, the so called ~-phase (space group P63/mmc, No. 194).
Only an insignificant rise in the susceptibility could be detected by Rupp and
Tuscher I-7.282] after hydrogen uptake.
When reinvestigating the Zr-Fe phase diagram Aubertin et al. studied
hydrides of Zr-rich alloys [7.283] and of the q-phase ZrzFeOo. ~ [7.284].
Regarding the latter, the M6ssbauer spectrum differs from that of the Ti
analogue presented by Rupp and Wiesinger [7.280] which is most probably due
to the different amount of hydrogen in the two samples (2.4% and 2.6%
respectively).
ThTFe3-type compounds readily absorb remarkable amounts of hydrogen
(up to about 30 H atoms/f.u.). By studying LaTNi3 hydride Busch et al. [7.285]
and Fischer et al. I-7.286] found that in fact it was metastable, disintegrating into
LaH3 and LaNis. Both the host compound and the hydride displayed Pauli
paramagnetic behaviour. Roughly ten years ago Buschow et al. I-7.287]
investigated the hydrogen storage behaviour of a large number of Th-3d metal
compounds. Only those with the 7:3 stoichiometry were found to absorb
remarkable quantities of hydrogen, whereas the remaining ones exhibited
rather poor storage capacities. More recently Boltich et al. [7.288] and Malik
et al. [-7.289] reported on the bulk magnetic properties of Th7TM3
(TM = Fe, Co, Ni). The parent compounds were found to be Pauli paramagnet-
ic and below about 2 K superconductivity was detected. Hydrogen absorption
led to a rise in the susceptibility for ThTCo 3, while in ThTNi 3 it decreased.
However, no evidence for magnetic order has been found in either hydride.
 Magnetic Properties, M6ssbauer Effectand Superconductivity 325

i
Th7 Fe3~
0109 200 300 400 500 600
TEMPERATURE
Fig.7.17. MagneticsusceptibilityofThTFe3hydridesprepared at differenthydrogenpressures
[-7.290]. The slowlyhydrogenatedsample(D) is not ferromagnetic.The ferromagnetismof the
violently hydrogenated samples (A,B,C) is caused by a disproportionation reaction. (A
-50bar; B -lObar; C --1 bar, 60°C; D -0.6 bar)

Th7Fe3, on the other hand, was claimed to have turned to ferromagnetism after
hydrogen uptake with a Curie temperature of about 350K. However, a
comprehensive investigation by Sehlapbach et al. [-7.290] led to the result that
ferromagnetism in these hydrides is obtained exclusively when violent charging
conditions are applied. This leads to a disproportionation into Th4H ~5 and a
more Fe-rich compound which both exhibit Curie temperatures in the range
cited and are moreover also known to absorb hydrogen [7.287]. After having
been charged smoothly the resulting hydride was found to remain paramag-
netic at least down to 80 K as displayed in Fig. 7.17. This confirms earlier
M6ssbauer studies of Viccaro et al. [-7.291] who reported an almost vanishing
STFe hyperfine field.

7.3.6 Amorphous Alloys

Although the investigations started with Fe-metalloid (B)-type alloys, Fe-metal

alloys, particularly Zr~_rFey, have received the most attention with regard to
hydrogen absorption. Magnetisation and M6ssbauer data yield a critical value
around a concentration y = 0.4, where a magnetic moment first appears on the
Fe. For the ferromagnetic systems the average Fe moment as well as the Curie
temperature T~ [-7.292] frequently increase with increasing y-values. However,
in FerZr ~_y [7.293-295] these quantities pass though a maximum and then
decline as y approaches i (Fig. 7.18). Other systems, notably sputtered FeyY~ -r
[7.292, 296, 297] show no long-range magnetic order for any value ofy because
of a broad distribution of mainly ferromagnetic exchange with some antifer-
romagnetic interactions which leads to supermagnetic order. These varying
326 G. Wiesingerand G. Hilscher

12
' I i I I i I i I
m76
i
Fig. 7.18. Concentration depen-
- 400 dence of the superconducting tran-
10 o12
sition temperature O', left-hand
350 scale) and of the magnetic ordering
PARAMAGNETIC ~ temperature (m, right-handscale) of
300 ~-FerZrl_ r I-7.295] and of some
0
250 hydrides (n) [7.292]; the numbers
6 c~-eeyZr¢oo-x ~'~ close to the open symbols indicate
200 the hydrogen concentration
~ 4 ~NOR~U~ 15o ~
\ / MAGNEnC 1oo
21-su~.o- '~ , , "
LCONDUC-~ ~t 5o
nines I 4, , ~, , , L , ,
~0 0.2 0.4. 0.6 0.8 1.0
Y

results show clearly the difficulty in extrapolating different alloy systems to

y = 1 in order to speculate about the magnetic state of pure amorphous iron.
In F%Zrl _y below the critical Fe concentration (Yc"~0.45), superconductiv-
ity appears with transition temperatures of the order of 3 K. In the range where
magnetic order occurs, the effective STFe hyperfine field deduced from
M6ssbauer measurements increases monotonically with Fe (unlike To),indicat-
ing an increase in the Fe moment. All these characteristics were claimed to be
consistent with a localized-moment description of magnetism in which an
antiferromagnetic exchange component appears at high Fe concentrations.
The concentration dependence of the electric quadrupole splitting shows a
large drop between the Fe-rich and the Zr-rich region. The isomer shift
increases with the Fe-concentration and is found to obey roughly the behaviour
predicted by the model of Miedema and van der Woude [7.57].
In the case of the Fe concentrations y ~ 0.9, magnetisation studies with
[7.294, 298] and without external pressure were interpreted on the basis of
weak itinerant ferromagnetism. By studying thermomagnetisation curves and
hysteresis loops, Hiroyoshi and Fukamichi [-7.299, 300] obtained a phase
diagram, where the Curie temperature and freezing points exhibit opposite
behaviour. Buschow and Smit [7.301] reported a compositional short-range
order to be present in melt-spun alloys but not in vapour-deposited ones. From
magnetisation measurements Kaul [7.302] suggested the presence of two types
of magnetic electrons: those exhibiting itinerant character, giving rise to
ferromagnetism (single-particle contribution) and Invar anomalies, and those
of a localized nature, being responsible for both the ferromagnetic (spin-wave
contribution) and the micromagnetic behaviour. For the compound Zr 1~Fe89
two contradictary results have been reported. While Boliang et al. [7.303]
claimed a reduction of the magnetisation below 60 K which they attributed to a
change from ferro- to asperomagnetism, nothing of this kind was found by Fries
et al. [-7.304]. The complex conversion electron M6ssbauer spectrum, consist-
ing of an ~-Fe component and an amorphous pattern of unknown composition,
 Magnetic Properties, Mrssbauer Effect and Superconductivity 327

Fig. 7.19. M6ssbauer emission

(upper) and transmission spec-

10
.1 --
trum (lower) of Fe9~Zr8 I-7.304]

1.00 --

1.00 --

0,95

O. 9 0 •
I F%Z8r
I I I I
-6 -4 -2 2 4

VBIoci fy ram/s]

led Fries et al. to the conclusion that these inhomogeneities in the ribbon
surface might be the reason of differing results. In Fig. 7.19, taken from their
work, an emission- (surface sensitive) and a transmission spectrum (bulk
sensitive) which were recorded simultaneously can be compared. The Fe
precipitates at the surface, detected by CEMS, are obvious.
So far, hydrogen absorption studies have mainly been performed on Fe-rich
samples with a composition of or close to y=0.9 [-7.40, 292, 303-308].
Substantial amounts of hydrogen can be absorbed. Generally an increase in
both the Curie temperature and the magnetic Fe moment was reported (see
Fig. 7.18). The magnetisation curves were found to resemble those of a collinear
ferromagnet. Fujimori et al. [7.305, 306] pointed out that the large volume
expansion upon hydrogenation might be mainly responsible for the enhance-
ment of Tc which can be qualitatively explained in terms of the inverse effect of
the large decrease of the Curie temperature under applied pressure. Since great
differences between calculated and experimental values of T~are found, Boliang
et al. [7.303] suggested that, as in many similar cases, hydrogen does not act
simply as a negative pressure, but directly alters the band structure, probably by
forming chemical bonds with Zr, which has already been suggested by Tanaka
et al. [-7.309] for Zr-Ni and Zr--Pd in order to interpret their soft x-ray emission
spectra.
Wronski et al. [7.307] noted the fact that drastic changes in magnetic
ordering are accompanied by only a small change in the isomer shift. This
finding they tentatively attributed to the formation of local deformations about
the interstitial H atoms rather than to a simple charge transfer from hydrogen
to the 3d-orbital of Fe.
In a thorough M6ssbauer study with and without applied magnetic field
Fries et al. I-7.304] obtained the unique result that for FesgZr 11Hso the spins are
328 G. Wiesinger and G. Hilscher

1.00- D i Fig. 7.20. M6ssbauer spectra

of Fe89Zrll and of
Fe89ZrxtHs0 recorded with
098" and without applied field; on
the right-hand side of the
100; spectra the hyperfine field
distribution is displayed
g [7.304]
>.
0.98- o

~S
10o- o

.G

O.97-
o_

O96"F ~ e ]..,r,~.yl° ,°',

-g -4 0 4 ~, 0 150 300
Velocity [ram/s] H[kOel

predominentely oriented perpendicular to the ribbon plane (I2/!1 =0.09),

whereas in the uncharged compound they were found to be more or less
randomly distributed with a slight tendency to align in the ribbon plane
(Fig. 7.20). An applied magnetic field of 200 Oe perpendicular to the 7-ray led to
a nearly perfect alignement in the ribbon plane which is commonly observed in
amorphous ribbons. Lowering the temperature also yielded a substantial rise in
the ratio of the two outermost lines.
Very recently Fries et al. [7.308] published MSssbauer and XPS studies on a
more Fe-rich Zr-Fe hydride. The hydrogen sites were found to resemble those
of binary ZrH 2, and the crystallization temperature was substantially de-
creased upon charging. Furthermore a pronounced charge transfer of Zr 4d-
electrons to hydrogen has been proposed. Compared to Zr-Fe only a few
results have been reported for other glassy Zr transition metal (TM) hydrides
[7.310, 3113. The T M = N i system will be mentioned below in the chapter
dealing with superconducting properties.
The influence of hydrogen uptake on the magnetic properties of amorphous
FerYI_~, has been extensively studied [7.292, 306]. A drastic change was
observed in the Fe-rich region: whereas magnetisation and M6ssbauer
experiments in large applied fields clearly pointed to a non-collinear configu-
ration of the Fe moments, these alloys were found to become excellent soft
ferromagnets after hydrogen absorption.
Among other Fe-based amorphous alloys the system F%Gdl -y was studied
by Forester et al. [7.311]. They gained a sharp hydrogen induced reduction of
compensation temperature and Curie temperature which is commonly ob-
served in crystalline R-Fe compounds and is attributed to the weakening of the
 Magnetic Properties, M6ssbauer Effectand Superconductivity 329

R-Fe exchange upon hydrogen absorption. Berry and Pritchet [7.312] and
Chambron et al. [7.313] reported a strong interaction of the hydrogen atoms
with the Bloeh wails which they observed by anisotropy measurements in
(Fe, Ni)B type metallic glasses.
Finally several rare-earth-rich glassy hydrides have been studied by Robbins
et al. [-7.314] and by Sellmyer et al. [7.315]. Spin-glass behaviour was found for
the host compounds which proved to be substantially influenced by
hydrogenation.

7.4 Superconductivity
The discovery of superconductivity in Th4H15 [7.316] and in PdH [7.317]
together with the speculation that metallic hydrogen as a potential high
temperature superconductor might be responsible for Tovalues as high as 9 and
11 K for P d - H and Pd-D respectively, was a stimulus for the development of a
large variety of theories and experiments in this field. Moreover, large T~values
of 16.6 K, 15.6, and 13.6 K could be attained in Pd-Cu-H, Pd-Ag-H, and
Pd-Au-H respectively, although the parent alloys as well Pd itself show no
superconductivity I-7.318].
Stritzker and 14r~hl[7.21] have already reviewed the effect of hydrogen upon
the superconducting properties in metals and alloys with special emphasis laid
on Th-H, P d - H and related systems. Here hydrogen commonly contributes
constructively to superconductivity, while in transition metals, gaseous impu-
rities reduce the superconducting transition temperature T~.
Before going into more detail we want to recall that the mechanism leading
to superconductivity is based on the attractive interaction between conduction
electrons formed via the lattice vibrations. This electron-phonon interaction
enters via the electron-phonon enhancement factor into the so-called
McMillan formula [7.319]:

0o _)
T~= ] ~ exp \2--/~*(1 + 0.622) / '

where 0o is the Debye temperature and # describes the repulsive electron-

electron interaction. 2 is given by

2 - N(Ev)(IZ>
M Q_o2) '

where N(EF) is the density of states (DOS) at the Fermi level, (12 > is the average
value of the electron-phonon interaction over the Fermi surface, M denotes the
atomic mass and <~o2> is an averaged phonon frequency.
With regard to these relations the high superconducting transition
temperatures experimentally observed may result either from an enlarged
330 G. Wiesinger and G. Hilscher

N(Ev) value caused by soft phonon mode contributions, or from additional new
vibrational modes due to dissolved hydrogen in the host lattice. For the system
Pd-H(D) Stritzker and Wfihl [7.21] concluded that in the case of a suppression
of spin fluctuations, the coupling of conduction electrons to the optical
phonons becomes the main reason for the appearance of high T~ superconduc-
tivity. In Th-H, superconductivity is restricted to the stoichiometric compound
T h 4 H ~ and may also be explained by low-lying optical H-phonon modes
1.,7.21] rather than by describing Th4H15 as an alloy of Th and metallic
hydrogen 1.7.320]. For further discussions regarding superconducting binary
metal hydrides such as A1-H and N b - H we refer the reader to the review of
Stritzker and Wiihl [7.21].
We should, however, briefly mention superconducting amorphous alloys:
Pd-Si [7.321,322], Zr-Pd [7.323] and Zr-Ni [7.324] in which hydrogen
absorption reduces T~. For the last two systems in particular, the well-defined
transition temperature becomes smeared out upon hydrogen uptake. This
probably indicates that inhomogeneities appear in the amorphous hydrides.
The superconducting properties of these amorphous alloys can be discussed on
the same basis as those of their crystalline counterparts.
Turning to crystalline compounds, the pseudobinary cubic (C15) Laves
phases (Hf, Zr)V 2 are high field superconducters with T~ reaching 10.1 K in the
case of Hfo.sZro.sV 2 [7.325]. The transition temperature decreases upon
absorption of hydrogen (deuterium) [7.326-328]. From specific heat and
susceptibility measurements and from x-ray analysis Geibel et al. [7.328]
proposed a phase diagram for ZrV2H~ which for x > 1.3 shows a decomposition
into ZrV2HI.3 and ZrV 2 on lowering the temperature. For this hydride specific
heat measurements proved the absence of superconductivity above 2 K and an
electronic specific heat coefficient which is drastically reduced to 3.2 mJ/K2g-at.
This decrease can be understood by considering recent band structure
calculations [7.329] which were performed under the assumption that hydro-
gen delivers its electrons to the conduction band, thus shifting the Fermi energy
into a valley of the DOS curve. Geibel et al. furthermore presented a reasonable
interpretation of the isotope effect in Hfo.sZro.sV2 in terms of the decompo-
sition mentioned above.
Two examples are now given, in which upon hydrogen uptake super-
conductivity is suppressed, while magnetism occurs: cubic C e R u 2 is a
superconductor with Tc ranging from 4.6 to 6.2 K, while CeRu2D 5 is found to
be magnetically ordered below 2.7 K. There pairs of adjacent [111] planes are
ferromagnetically and antiferromagnetically coupled in the sequence + +
- - . Fruchart et al. 1-7.330] explain the transition from superconductivity to
magnetic order by a hydrogen induced valence change of Ce from 4 + to 3 +.
Osterwalder et al. [7.331] studied the effect of hydrogen absorption of the 4 f
localisation in CeRu 2 by means of XPS. The authors concluded that this
intermetallic becomes a fully trivalent Ce compound upon hydriding and that
the 4f conduction-band hybridisation decreases from 120meV to about
60 meV.
 Magnetic Properties, M6ssbauer Effect and Superconductivity 331

ThTFe 3 is a Pauli paramagnet and is superconducting below 1.86 K [7.288,

291,332] which indicates that the Fe 3d-band is completely filled. The hydride
ThTFeaH30 has been claimed to order ferromagnetically with an Fe moment of
1.4/IB, Th being assumed to carry no magnetic moment. The Curie temperature
was estimated to lie between 350 K and 400 K. Malik et al. [7.332] suggested
hydrogen to be essentially anionic in this case too (see e.g. [7.3]). From this it is
plausible that the Fe 3d-band is depopulated in the hydride and that the Fe
moment destroys superconductivity. A re-investigation by Schlapbach et al.
[7.290], however, showed that the hydride remains paramagnetic when the
charging procedure is performed sufficiently smoothly, in order to avoid
decomposition (see also Sect. 7.3.5 and Fig. 7.17).
The most important group of compounds for technical superconductivity
are those with the A15 type of structure (Cr3Si type) containing about 50
superconducting compounds [7.333]. The interest in the effect of hydrogena-
tion upon the superconducting properties is of both scientific and technolog-
ical nature, the latter in order to obtain a better performance in superconduct-
ing wires. Various authors have studied the influence of hydrogen uptake upon
Tc in Nb3Sn [7.334-337], Nb3Ge [7.338], Ti3Au [7.339], Ti3Sb [7.340] and
Nba(Au, Ge) [7.341]. In the case of A15 compounds it is frequently found that
upon hydrogenation the superconducting transition first increases
(A To= + 0.2 K) as long as the hydrogen concentration x ~<0.03. Subsequently a
decrease is obtained to T¢<1.1 K for x ~ 1 [7.336]. The compounds last
mentioned are an exception, since upon hydrogen absorption Tc remains
almost constant or increases only slightly. By neutron diffraction experiments
Vieland et al. [7.335] investigated the crystal structure of some Nb3Sn hydrides,
and for hydrogen concentrations x > 0.1 confirmed the absence of the cubic-
tetragonal lattice transformation occurring in the parent compound at about
50 K. The reduction of To by dissolving hydrogen in Nb3Sn resembles the effect
of alloying Nb3Sn with Sb or A1, where Tc is altered in a similar way. Although
this fact cannot be taken as direct evidence for the applicability of the rigid band
approximation [7.335], the latter has been used by Rama Rao et al. [7.340] for
the interpretation of the reduction of To in Ti3SbH~. The band structure of
Ti3Sb was assumed to resemble that of NbaSn, where E v lies in a maximum of
the DOS curve of the d-band. Hydrogen absorption then increases the valence
electron concentration by shifting the Fermi energy to a lower N(EF) value.
Since T~ depends on the product N(EF). (I2), a reduction of N(EF) upon
hydrogenation may lead to a fall in T~. A decrease in Tc is also observed for other
Ti-based A15 superconductors in the sequence TisIr, Ti3Pt, Ti3Au. There too
the valence electron concentration rises in going from Ir to Au. Thus T~may be
lowered by hydrogen absorption as well as by alloying.
In this connection we note that Matacotta et al. [7.337] reported an
improvement of the current density of technical Nb3SnH x multifilamentary
wires by low hydrogen admixture (x < 0.05), particularly in high magnetic fields
around 12 T. Ferdeghini et al. [7.342] recently re-investigated the effect of
hydrogen uptake on the superconducting performance of technical NbaSn
332 G. Wiesinger and G. Hilscher

wires. T h e y o b t a i n e d a m a x i m u m in the critical c u r r e n t d e n s i t y for a h y d r o g e n

c o n t e n t of x = 0.03 at a m a g n e t i c field o f / z o H = 12 T. T h e y f u r t h e r m o r e s t a t e d
t h a t the relative increase in p e r f o r m a n c e o f the wire was m o r e p r o n o u n c e d at
h i g h e r t e m p e r a t u r e s with a gain o f J~ u p to 2 1 5 % at 10 K. After h e a t t r e a t i n g a
N b 3 S n wire in a h y d r o g e n a t m o s p h e r e Wilhelm a n d Wohlleben I-7.343] o b s e r v e d
a n increase in b o t h the u p p e r critical field a n d the c u r r e n t density. Since T~ is
slightly reduced, t h e y a t t r i b u t e this e n h a n c e m e n t in the h y d r o g e n - t r e a t e d
s a m p l e s to a n increase in the n o r m a l - s t a t e resistivity.

Acknowledgement. This work was supported by the National Austrian Science Foundation
(Fonds zur F6rderung der wissenschaftlichen Forschung) under grant No. 4431.

References
7.1 G. Wortmann: J. de Phys. 12, C6 333 (1976)
7.2 F.E. Wagner, G. Wortmann: "M6ssbauer Studies of Metal-Hydrogen Systems", in
Hydrogen in Metals I, ed. by G. Alefeld, J. V61kl (Springer, Berlin, Heidelberg 1978)
p. 131
7.3 W.E. Wallace: "Magnetic and Electrical Properties of Rare Earth and Rare Earth
lntermetallic Hydrides", in Hydrides .for Energy Storage, ed. by A.F. Andresen, A.J.
Maeland (Pergamon, Oxford 1978) p. 217
7.4 W.E. Wallace: "Magnetic Properties of Metal Hydrides and Hydrogenated Intermetal-
lic Compounds" in Hydrogen in Metals 1, ed. by G. Alefeld, J. V61kl (Springer, Berlin,
Heidelberg 1978) p. 169
7.5 W.E. Wallace: Z. Phys. Chem. N F 115, 219 (1979)
7.6 W.E. Wallace: "Studies of Rare Earth lntermetallic Compounds and Rare Earth
Hydrides", in The Rare Earths in Modern Science and Technology, ed. by G.J. Mc
Carthy, J.J. Rhyne, H.B. Silber (Plenum, New York 1980) p. 1-12
7.7 W.E. Wallace: "Magnetism of Lanthanide and Actinide Intermetallic Hydrides", in
Metal Hydrides, ed. by G. Bambakidis (Plenum, New York 1981) p. 21
7.8 W.E. Wallace: J, Less-Common Met. 88, 141 (1982)
7.9 G.K. Shenoy, B.D. Dunlap, P.J. Viccaro, D. Niarchos: Hyperf. Inter. 9, 531 (1981)
7.10 G.K. Shenoy, B.D. Dunlap, P.J. Viccaro, D. Niarchos: "Hydrogen Storage Materials"
in M6ssbauer Spectroscopy and Its Chemical Applications, ed. by J.G. Stevens, G.K.
Shenoy, Advances in Chemistry Series No. 194 (1981) p. 501
7.11 R.L. Cohen: J. de Phys. 41, C1-333 (1980)
7.12 H. Oesterreicher: Appl. Phys. 24, 169 (1981)
7.13 R.L. Cohen, J.H. Wernick: Science 214, 1081 (1981)
7.14 F.A. Kuijpers: Philips Res. Repts. Suppl. 1973, No. 2, p. 1
7.15 K.H.J. Buschow: J. Magn. Magn. Mat. 29, 91 (1982)
7.16 K.H.J. Buschow: "Change in Magnetic Properties of Rare Earth-Transition Metal
Compounds upon H2 Absorption", in Hydrides for Energy Storage, ed. by A.F.
Andresen, A.J. Maeland (Pergamon, Oxford 1978) p. 273
7.17 K.H.J. Buschow, R.C. Sherwood: J. A.ppl. Phys. 49, 1480 (1978)
7.18 K.H.J. Buschow, P.F. de Chfitel: Pure & Appl. Chem. 52, 135 (1979)
7.19 K.H.J. Buschow, P.C.P. Bouten, A.R. Miedema: Rept. Prog. Phys. 45, 937 (1982)
7.20 K.H.J. Buschow: "Hydrogen Absorption in Intermetallic Compounds" in Handbook on
the Physics and Chemistry of Rare Earths, ed. by K.A. Gschneidner, Jr., L. Eyring, Vol. 6
(North-Holland, Amsterdam 1984) p. 1
7.21 B. Stritzker, H. Wfihl: "Superconductivity in Metal-Hydrogen Systems", In Hydrogen
in Metals II, ed. by G. Alefeld, J. V61kl (Springer, Berlin, Heidelberg 1978) p. 243
 Magnetic Properties, M/Sssbauer Effect and Superconductivity 333

7.22 E.P. Wohlfarth: "Iron, Cobalt and Nickel" in Ferromagnetic Materials, ed. by E.P.
Wohlfarth, Vol. 1 (North-Holland, Amsterdam 1980) p. 3
7.23 B.L. Gyorffy, A.J. Pindor, J. Staunton, G.M. Stocks, H. Winter: J, Phys. F15, 1337
(1985)
7.24 T. Moriya (ed.): Electron Correlation and Magnetism in Narrow-Band Systems, Springer
Ser. Solid-State Sci., Vol. 29 (Springer, Berlin, Heidelberg 1981)
7.25 G.G. Libowitz, A.J. Maeland: "Hydrides" In Handbook on the Physics and Chemistry of
Rare Earths, Vol. 3, ed. by K.A. Gschneidner, Jr., L. Eyring (North-Holland,
Amsterdam 1979) p. 299
7.26 A.C. Switendick: "The Change in Electronic Properties on Hydrogen Alloying and
Hydride Formation" in Hydrogen in Metals I, ed. by G. Alefeld, J. V61kl (Springer,
Berlin, Heidelberg 1978) p. 101
7.27 W.E. Wallace, K.H. Mader: J. Chem. Phys. 48, 84 (1968)
7.28 W.E. Wallace, C. Deenadas, A.W. Thompson, R.S. Craig: J. Phys. Chem. Sol. 32, 805
(1971)
7.29 A.M. van Diepen, R.S. Craig, W.E. Wallace: J. Phys. Chem. Sol. 32, 1853 (1971)
7.30 Z. Bieganski, B. Stalinski: Phys. Stat. Sol. (a) 2, K 161 (1970)
7.31 K. Knorr, B.E.F. Fender: "Crystal Fields in Pr-Hydrides" in Crystal Field Effects in
Metals and Alloys, ed. by A. Furrer (Plenum, New York 1977) p. 42
7.32 J. Osterwalder, H.R. Ott, L. Schlapbach, J. Schefer, P. Fischer: J. Less-Common Met.
94, 129 (1983)
7.33 J. Schefer, P. Fischer, W. Haelg, J. Osterwalder, L. Schlapbach, J.D. Jorgensen: J. Phys
CI7, 1575 (1984)
7.34 L. Schlapbach: J. Less. Common Met. 111,291 (1985)
7.35 F. Steglich, C.D. Bredl, W. Lieke, U. Rauchschwalbe, G. Sparn: Physica 126B, 82 (1984)
7.36 G.R. Stewart: Rev. Mod. Phys. 56, 755 (1984)
7.37 J. Staunton, B.L. Gyorffy, A.J. Pindor, G.M. Stocks, H. Winter: J. Phys. F15, 1387
(1985)
7.38 R. Griessen, R. Feenstra: J. Phys. F 15, 1013 (1985)
7.39 V. Heine: The Physics of Metals, Vol. 1, Electrons (University Press, Cambridge 1969)
p.l
7.40 J.M.D. Coey, D.H. Ryan, D. Gignoux, A. Lienard, J.P. Rebouillat: J. Appl. Phys. 53,
7804 (1982)
7.41 G. Hilscher, G. Wiesinger, R. Hempelmann: J. Phys. F l l , 2161 (1981)
7.42 K.H.J. Buschow: Rept. Prog. Phys. 40, 1179 (1977)
7.43 K.H.L Buschow: "Rare Earth Compounds" in Ferromagnetic Materials, ed. by E.P.
Wohlfarth, Vol. 1 (North-Holland, Amsterdam 1980) p. 299
7.44 H. Kirchmayr, C. Poldy: "Magnetic Properties of Intermetallie Compounds of Rare
Earth Metals" in Handbook on the Physics and Chemistry of Rare Earths, ed. by K.A.
Gschneidner, L. Eyring, Vol. 2 (North-Holland, Amsterdam 1979) p. 55
7.45 W.E. Wallace: Rare Earth Intermetallics (Academic, New York 1973)
7.46 P. Vajda, J.N. Daou: J. Less-Common Met. 101,269 (1984)
7.47 I. Jacob, E.R. Bauminger, D. Davidov, 1. Felner, S. Ofer, D. Shaltiel: J. Magn. Magn.
Mat. 15-18, 1269 (1980)
7.48 H. de Graaf, R.C. Thiel, K.HJ. Buschow: J. Phys. F 12, 1239 (1982)
7.49 E. Wicke: J. Less-Common Met. 101, 17 (1984)
7.50 P. Bennet, J.C. Fuggle: In Electronic Structure and Properties of Hydrogen in Metals, ed.
by P. Jena, C.B. Satterthwaite (Plenum, New York 1983) p. 223
7.51 L. Schlapbach, J.P. Burger: "XPS/UPS Study of the Electronic Structure of PdHo.6" in
Electronic Structure and Properties of Hydrogen in Metals, ed. by P. Jena, C.B.
Satterthwaite (Plenum, New York 1983) p. 229
7.52 U. Gonser (ed.): Mdssbauer Spectroscopy, Topics Appl. Phys., Vol. 5 (Springer, Berlin,
Heidelberg 1975)
7.53 D. Barb: Grundlagen und Anwendungen der Mgssbauerspektroskopie (Akademie Verlag,
Berlin 1980)
334 G. Wiesinger and G. Hilscher

7.54 L.R. Walker, G.K. Wertheim, V. Jaccarino: Phys. Rev. Lett. 6, 98 (1961)
7.55 M. Gupta: Solid State Commun. 42, 501 (1982)
7.56 L. Schlapbach: Phys. Lett. 91, 303 (1982)
7.57 A.R. Miedema, F. van der Woude: Physica 100B, 145 (1980)
7.58 W. Steiner, R. Haferl: Phys. Stat. Sol. (a) 42, 739 (1977)
7.59 P.C.M. Gubbens, K.H.J. Buschow: J. Phys. F12, 2715 (1982)
7.60 J. Zukrowski, A. Barnasik, K. Krop, R. Radwanski, J. Pszczola, J. Suwalski, Z.
Kucharski, M. Lukasiak: Hyperf. Inter. 15/16, 801 (1983)
7.61 P.C.M. Gubbens, A.M. van der Kraan: J. Magn. Magn. Mat. 9, 349 (1978)
7.62 P.C.M. Gubbens, A.M. van der Kraan, K.H.J. Buschow: J. Appl. Phys. 56, 2547 (1984)
7.63 F.A. Kuijpers, B.O. Loopstra: J. Phys. Chem. Sol. 35, 301 (1974)
7.64 M. Yamaguchi, T. Katamune, T. Ohta: J. Appl. Phys. 53, 2788 (1982)
7.65 M. Yamaguchi, T. Ohta, T. Katayama: J. Magn. Magn. Mat. 31-34, 221 (1983)
M. Yamaguchi, D.K. Ross, T. Groto, T. Ohta: Z. Phys. Chem. NF 145, 101 (1985)
7.66 E.R. Bauminger, D. Davidov, I. Fclner, I. Nowik, S. Ofcr, D. Shaltiel: Physica 86-88B,
201 (1977)
7.67 M. Yamaguchi, T. Katamune, T. Ohta: J. Less-Common Met. 88, 195 (1982)
7.68 M. Yamaguchi, S. Sasaki, T. Ohta: J. Less-Common Met. 73, 201 (1980)
7.69 H. Figiel: J. Less-Common Met. 83, L27 (1982)
7.70 G. Herbst, H. Kronmfiller: Phys. Lett. 70A, 341 (1979)
7.71 G. Herbst, H. Kronmfiller: Z. Phys. Chem. NF 116, 31 (1979)
7.72 F.T. Parker, H. Oesterreicher: J. Less-Common Met. 79, 297 (1981)
7.73 F. Stucki, L. Schlapbach: J. Less-Common Met. 74, 143 (1980)
7.74 L. Schlapbach, A. Seiler, F. Stucki, H.C. Siegmann: J. Less-Common Met. 73, 145 (1980)
7.75 L. Schlapbach: J. Phys. FI0, 2477 (1980)
7.76 J.H. Weaver, A. Franciosi, W.E. Wallace, H.K. Smith: J. Appl. Phys. 51, 5847 (1980)
7.77 J. Palleau, G. Chouteau: J. de Phys. 41, L227 (1980)
7.78 S.K. Malik, F.J. Arlinghaus, W.E. Wallace: Phys. Rev. B25, 6488 (1982)
7.79 W.M. Walsh, Jr., L.W. Rupp, Jr., P.H. Schmidt, L.D. Longinotti: "Magnetic Resonance
of Gd in LaNi 5 and LaNi 5 Hydride" in AlP No. 29, 1975 Conf. Proc. p. 686
7.80 J. Lamloumi, C. Lartique, A. Percheron-Guegan, J.C. Achard, G. Jehanno:"Thermody-
namic and Magnetic Properties of LaNis_~FexHy" in The Rare Earths in Modern
Science and Technology, ed. by G.J. McCarthy, H.B. Silber, J.J. Rhyne (Plenum, New
York 1982) p. 487
7.81 S.J. CampelI, R.K. Day, J.B. Dunlop, A.M. Stewart: J. Magn. Magn. Mat. 31-34, 167
(1983)
7.82 J. Lamloumi, A. Percheron-Guegan, J.C. Achard, G. Jehanno, D. Givord: J. de Phys. 45,
1643 (1984)
7.83 U. Atzmony, D. Dayan, M.P. Darieh Mat. Res. Bull. 16, 793 (1981)
7.84 D. Niarchos, P.J. Viccaro, G.K. Shenoy, B.D. Dunlop, A.T. Aldred: Hyperf. Inter. 9,
563 (1981)
7.85 F.W. Oliver, T. Kebede, K. Thompson, J. Gilchrist: Solid State Commun. 46, 837 (1983)
7.86 F.W. Oliver, W. Morgan, E.C. Hammond, S. Wood, L. May: J. Appl. Phys. 57, 3250
(1985)
7.87 R.L. Cohen, K.W. West, J.H. Wernick: J. Less-Common Met. 70, 229 (1980)
7.88 R.L. Cohen, K.W. West, J.H. Wernick: J. Less-Common Met. 73, 273 (1980)
7.89 D. Shaltiel, Th. v. Waldkirch, F. Stucki, L. Schlapbach: J. Phys. FI1, 471 (1981)
7.90 H. Rummel, R.L. Cohen, P. Guetlich, K.W. West: Appl. Phys. Lett. 40, 477 (1982)
7.91 F.W. Oliver, K.W. West, R.L. Cohen, K.H.J. Buschow: J. Phys. F8, 701 (1978)
7.92 S.K. Malik, E.B. Boltich, W.E. Wallace: J. Sol. State Chem. 33, 263 (1980)
7.93 A.T. Pedziwiatr, F. Pourarian, W.E. Wallace: J. Appl. Phys. 55, 1987 (1984)
7.94 K. Yagisawa, A. Yoshikawa: Z. Phys. Chem. NFI17, 79 (1979)
7.95 K.H.J. Buschow: J. Less-Common Met. 97, 185 (1984)
7.96 K.H.J. Buschow: J. Magn. Magn. Mat. 40, 224 (1983)
7.97 K.H.J. Buschow, P.H. Smit, R.M. van Essen: J. Magn. Magn. Mat. 15-18, 1261 (1980)
 Magnetic Properties, M6ssbauer Effect and Superconductivity 335

7.98 K.H.J. Buschow: J. Less-Common Met. 72, 257 (1980)

7.99 J.J. Croat, J.F. Herbst, R.W. Lee, F.E. Pinkerton: J. Appl. Phys. 55, 2078 (1984)
7.100 M. Sagawa, S. Fujimura, N. Togawa, H. Yamamoto, Y. Matsuura: J. Appl. Phys. 55,
2083 (1984)
7.101 H.H. Stadelmaier, N.A. E1 Masry, S.R. Stallard: J. Appl. Phys. 57, 4149 (1985)
7.102 J.F. Herbst, J.J. Croat, F.E. Pinkerton, W.B. Yelon: Phys. Rev. B 29, 4176 (1984)
7.103 D. Givord, H.S. Li, J.M. Moreau: Solid State Commum. 50, 497 (1984)
7.104 C.B. Shoemaker, D.P. Shoemaker, R. Fruchart: Acta Cryst. C40, 1665 (1984)
7.105 H. Onodera, Y. Yamaguchi, H. Yamamoto, M. Sagawa, Y. Matsuura, H. Yamamoto: J.
Magn. Magn. Mat. 46, 151 (1984)
7.106 J.M. Friedt, A. Vasquez, J.P. Sauchez, P. l'Heritier, R. Fruchart: J. Phys. F 16, 651 (1986)
7.107 R. Gr6ssinger, G. Hilscher, H. Kirchmayr, H. Sassik, R. Strnat, G. Wiesinger: Physica
130B, 307 (1985)
7.108 P. l'Heritier, P. Chaudouet, R. Madar, A. Rouault, J.-P. Senateur, R. Fruchart: C. R.
Acad. Sc. Paris 299, Serie II, 849 (1984)
7.109 K. Oesterreicher, H. Oesterreicher: Phys. Stat. Sol. (a) 85, K61 (1984)
7.110 G. Wiesingcr, G. Hilscher, R. Gr6ssinger: J. Less-Common Met. 131,409 (1987)
7.111 I.R. Harris, C. Noble, T. Bailey: J. Less-Common Met. 106, L1 (1985)
7.112 K.H.J. Buschow, R.C. Sherwood: J. Appl. Phys. 48, 4643 (1977)
7.113 E.B. Boltich, W.E. Wallace, F. Pourarian, S.K. Malik: J. Phys. Chem. 86, 524 (1982)
7.114 S.K. Malik, T. Takeshita, W.E. Wallace: Solid State Commun. 23, 599 (1977)
7.115 P.C.M. Gubbens, A.M. van der Kraan, K.H.J. Buschow: J. Magn. Magn. Mat. 30, 383
(1983)
7.116 K.H.J. Buschow: Solid State Commun. 21, 1031 (1977)
7.117 E.B. Boltich, W.E. Wallace, F. Pourarian, S.K. Malik: J. Magn. Magn. Mat. 25, 295
(1982)
7.118 M. Commandre, D. Fruehart, A. Rouault, D. Sauvage, C.B. Shoemaker, D.P.
Shoemaker: J. de Phys. 40, L639 (1979)
7.119 C. Crowder, B. Kebe, W.J. James, W. Yelon: "Magnetic and Structural Properties of
Y6Mn23D23" in The Rare Earths in Modern Science and Technology, ed. by G.J.
McCarthy, H.B. Silber, J.J. Rhyne, Vol. 3 (Plenum, New York 1982) p. 473
7.120 K. Hardman-Rhyne, J.J. Rhyne: J. Less-Common Met. 94, 23 (1983)
7.121 K. Hardman-Rhyne, J.J, Rhyne, E. Prince, C. Crowder, W.J. James: Phys. Rev. B 29, 416
(1984)
7.122 C. Crowder, W.J. James: J. Less-Common Met. 95, 1 (1983)
7.123 K. Hardman, J.J. Rhyne, H.K. Smith, W.E. Wallace: J. Less-Common Met. 74, 97
(1980)
7.124 K. Hardman-Rhyne, H.K. Smith, W.E. Wallace: J. Less-Common Met. 96, 201 (1984)
7.125 A. Delapalme, J. Deportes, R. Lemaire, K. H ardman, W.J. James: J. AppL Phys. 50, 1987
(1979)
7.126 G.A. Stewart, J. Zukrowski, G. Wortmann: J. Magn. Magn. Mat. 25, 77 (1981)
7.127 K.H.J. Buschow: Solid State Commun. 40, 207 (1981)
7.128 F. Pourarian, E.B. Boltich, W.E. Wallace, S.K. Malik: J. Less-Common Met. 74, 153
(1980)
7.129 F. Pourarian, E.B. Boltich, W.E. Wallace, R.S. Craig, S.K. Malik: J. Magn. Magn. Mat.
21, 128 (1980)
7.130 K.H.J. Buschow, P.C.M. Gubbens, W. Ras, A.M. van der Kraan: J. Appl. Phys. 53, 8329
(1982)
7.131 P.C.M. Gubbens, W. Ras, A.M. van der Kraan, K.H.J. Buschow: Phys. Stat. Sol. (b) 117,
277 (1983)
7.132 G.A. Stewart, J. Zukrowski, G. Wortmann: Solid State Commun. 39, 1017 (1981)
7.133 E.B. Boltich, F. Pourarian, W.E. Wallace, H.K. Smith, S.K. Malik: Solid State
Commun. 40, 117 (1981)
7.134 A.T. Pedziwiatr, H.K. Smith, W.E. Wallace: J. Sol. State Chem. 47, 41 (1983)
336 G. Wiesinger and G. Hilscher

7.135 A.T. Pedziwiatr, E.B. Boltich, W.E. Wallace, R.S. Craig: J. Sol. State Chem. 46, 342
(1983)
7.136 J.J. Rhyne, K. Hardman-Rhyne, H.K. Smith, W.E. Wallace: J. Less-Common Met. 94,
95 (1983)
7.137 A.T. Pedziwiatr, Z.M. Stadnik, J. Zukrowski, H.K. Smith, W.E. Wallace: Solid State
Commun. 55, 455 (1985)
7.138 P.C.M. Gubbens, A.M. van der Kraan, K.H.J. Buschow: J. Phys. F14, 235 (1984)
7.139 P.C.M. Gubbens, A.M. van der Kraan, K.H.J. Buschow: Solid State Commun. 37, 635
(1981)
7.140 G. Hilscher, N. Buis, J.J.M. Franse'. Physica B 91, 170 (1977)
7.141 C. Lin, P. v. Blanckenhagen,G. Hilscher, G. Wiesinger: J. Magn. Magn. Mat. 31-34, 199
(1983)
7.142 G.J. Long, K. Hardman, W.J. James: Solid State Commun. 34, 253 (1980)
7.143 H. Oesterreichcr, H. Bittner: Phys-Stat. Sol. (a) 41, KI01 (1977)
7.144 J. Zukrowski, G.A. Stewart, G. Kalkowski, G. Wortmann, G. Wiesinger: "M6ssbauer
Studies of the Er6(Fe a_yMny)23H~System", in 1981 Conf. Cryst. Electric Field Effects
Proc. p. 149 (Plenum, New York 1982)
7.145 K.H.J. Buschow: J. Less-Common Met. 72, 257 (1980)
7.146 R.M. van Essen, K.H.J. Buschow: J. Less-Common Met. 70, 189 (1980)
7.147 K.H.J. Buschow, R.M. van Essen: Solid State Commun. 32, 1241 (1979)
7.148 S.K. Malik, W.E. Wallace, T. Takeshita: Solid State Commun. 28, 977 (1978)
7.149 S.K. Malik, E.B. Boltich, W.E. Wallace: Solid State Commun. 37, 329 (1981)
7.150 J.F. Herbst, J.J. Croat: J. Appl. Phys. 53, 4304 (1982)
7.151 S.K. Arif, D.St.P. Bunbury, G.J. Bowden: J. Phys. FS, 1785 (1975)
7.152 A.M. van der Kraan, P.C.M. Gubbens, K.H.J. Buschow: Phys. Stat. Sol. (a) 31, 495
(1975)
7.153 R.L. Davies, R.K. Day, J.B. Dunlop: J. Phys. F7, 1885 (1977)
7.154 G.J. Bowden, R.K. Day: J. Phys. F7, 181 (1977)
7.155 G.J. Bowden, R.K. Day: J. Phys. F7, 191 (1977)
7.156 G. Wiesinger, R. Haferl, H.R. Kirchmayr: Mikrochim. Acta, Supp]. 9, 177 (1981)
7.157 S.K. Malik, T. Takeshita, W.E. Wallace: Magn. Lett. 1, 33 (1976)
7.158 K.S.V.L. Narasimhan: Magnetisation Measurements on Er a ~ThxFe3 and Some of
Their Hydrides" in The Rare Earths in Modern Science and Technology, ed. by G.J.
McCarthy, J.J. Rhyne (Plenum, New York 1978) p. 81
7.159 D. Niarchos, P.J. Viccaro, B.D. Dunlap, G.K. Shenoy: J. Appl. Phys. 50, 7690 (1979)
7.160 D. Niarchos, P.J. Viccaro, B.D. Dunlap, G.K. Shenoy, A.T. Aldred: J. Less-Common
Met. 73, 283 (1980)
7.161 S.K. Malik, F. Pourarian, W.E. Wallace: J. Magn. Magn. Mat. 40, 27 (1983)
7.162 J.B.M. da Cunha, A. Vasquez: Hyperf. Inter. 9, 547 (1981)
7.163 J.B.M. da Cunha, P.J. Viccaro, A. Vasquez: Hyperf. Inter. 12, 119 (1982)
7.164 K.H.J. Buschow: Solid State Commun. 19, 421 (1976)
7.165 A.M. van Diepen, K.HJ. Buschow: Solid State Commun. 22, 113 (1977)
7.166 A.M. van der Kraan, J.N.J. van der Velden, J.H.F. van Apeldoorn, P.C.M. Gubbens,
K.H.J. Buschow: Phys. Stat. Sol. (a} 35, 137 (1976)
7.167 H, H6chst, E. Colavita, K.H.J. Buschow: Phys. Rev. B31, 6167 (1985)
7.168 R.L. Cohen, K.W. West, F. Oliver, K.H.J. Buschow: Phys. Rev. B21, 941 (1980)
7.169 Y. Nakamura: J. Magn. Magn. Mat. 31-34, 829 (1983)
7.170 P.J. Viccaro, G.K. Shenoy, D. Niarchos, B.D. Dunlap: J. Less-Common Met. 73, 265
(1980)
7.171 R. Hempelmann, G. Hilscher: J. Less-Common Met. 74, 103 (1980)
7.172 H. Oesterreicher, H. Bittner: Mat. Rcs. Bull. 13, 83 (1978)
7.173 R. Hempelmann, E. Wicke, G. Hilschcr, G. Wiesinger: Bet. Bunsenges. Phys. Chem. 87,
48 (1983)
7.174 D. Fruchart, J.L. Soubeyroux, R. Hempelmann: J. Less-Common Met. 99, 307 (1984)
7.175 R.M. van Essen, K.H.J. Buschow: Mat. Res. Bull. 15, 1149 (1980)
 Magnetic Properties, M6ssbauer Effect and Superconductivity 337

7.176 F. Pourarian, H. Fujii, W.E. Wallace, V.K. Sinha, H.K. Smith: J. Phys. Chem. 85, 3105
(1981)
7.177 I. Jacob, D. Davidov, D. Shattiel: J. Magn. Magn. Mat. 20, 226 (1980)
7.178 H. Fujii, F. Pourarian, W.E. Wallace: J. Less-Common Met. 88, 187 (1982)
7.179 S. Hirosawa, F. Pourarian, W.E. Wallace: J. Magn. Magn. Mat. 43, 187 (1984)
7.180 J.J. Didisheim, P. Fischer: J. Less. Common Met. 103, 267 (1984)
7.181 P.J. Viccaro, G.K. Shenoy, B.D. Dunlap, D.G. Westlake, J.F. Miller: J. de Phys. 40,
C2-198 (t979)
7.182 K.H.J. Buschow, A.M. van Diepen: Solid State Commun. 19, 79 (1976)
7.183 D. Niarchos, P.J. Viccaro, G.K. Shenoy, B.D. Dunlap, A.T. Aldred: Hyperf. Inter. 9, 563
(1981)
7.184 P.H. Smit, K.H.J. Buschow: Phys. Rev. B21, 3839 (1980)
7.185 P.H. Smit, H.C. Donkersloot, K.H.J. Buschow: J. Appl. Phys. 53, 2640 (1982)
7.186 R. Gr6ssinger, R. Haferl, G. Hilscher, G. Wiesinger, K.H.J. Buschow, P.H. Smit: Inst.
Phys. Conf. Ser. No. 55, Chap. 4, 295 (1980)
7.187 Y. Berthicr, T. de Saxce, D. Fruchart, P. Vulliet: Physica 130B, 520 (1985)
7.188 T. de Saxce, Y. Berthier, D. Fruchart: J. Less-Common Met. 107, 35 (1985)
7.189 F. Pourarian, W.E. Wallace, A. Elattar, J.F. Lakner: J. Less-Common Met. 74,161 (I 980)
7.190 B.D. Dunlap, G.K. Shenoy, J.M. Friedt, P.J. Viccaro, D. Niarchos, H. Kierstead, A.T.
Aldred, D.G. Westlake: J. Appl. Phys. 50, 7682 (1979)
7.191 P.J. Viccaro, J.M. Friedt, D. Niarchos, D. Dunlap, G.K. Shenoy, A.T. Aldred, D.G.
Wcstlake: J. Appl. Phys. 50, 2051 (1979)
7.192 G.K. Shenoy, B. Schuttler, P.J. Viccaro, D. Niarchos: J. Less-Common Met. 94, 37
(1983)
7.193 G.E. Fish, J.J. Rhyne, S.G. Sankar, W.E. Wallace: J. Appl. Phys. 50, 2003 (1979)
7.194 J.J. Rhyne, G.E. Fish, S.G. Sankar, W.E. Wallace: J. de Phys. 40, C5-209 (1979)
7.195 K.H.J. Buschow: Physica 86--88B, 79 (1977)
7.196 D. Shaltiel, I. Jacob, D. Davidov: J. Less-Common Met. 53, 117 (1977)
7.197 I. Jacob, D. Shaltiel, D. Davidov, I. Miloslavski: Solid State Commun. 23, 669 (1977)
7.198 I. Jacob, D. Shaltiel: Solid State Commun. 27, 175 (1978)
7.199 I. Jacob, J.M. Bloch, D. Shaltiel, D. Davidov: Solid State Commun. 35, 155 (1980)
7.200 H. Fujii, V.K. Sinha, F. Pourarian, W.E. Wallace: J. Less-Common Met. 88, 44 (1982)
7.201 F. Pourarian, W.E. Wallace: J. Less-Common Met. 91, 223 (1983)
7.202 V.K. Sinha, W.E. Wallace: J. Less-Common Met. 91, 229 (1983)
7.203 V.K. Sinha, W.E. Wallace: J. Less-Common Met. 91, 239 (1983)
7.204 D. Rambabu, R. Nagarajan, S.K. Malik, R. Vijayaraghavan, F. Pourarian, W.E.
Wallace: J. Magn. Magn. Mat. 31-34, 759 (1983)
7.205 V.K. Sinha, F. Pouarian, W.E. Wallace: J. Less-Common Met. 87, 283 (1982)
7.206 G. Wicsinger: Hyperf. Inter. 28, 545 (1986)
7.207 H. Fujii, T. Okamoto, W.E. Wallace, F. Pourarian, T. Morisaki: J. Magn. Magn. Mat.
46, 245 (1985)
7.208 F. Fujii, F. Pourarian, W.E. Wallace: J. Magn. Magn. Mat. 27, 215 (1982)
7.209 H. Fujii, J. Fujimoto, S. Takeda, T. Hihara, T. Okamoto: J. Magn. Magn. Mat. 31-34,
223 (1983)
7.210 T. Okamoto, H. Fujii, S. Takeda, T. Hihara: J. Less-Common Met. 88, 181 (1982)
7.211 H. Oesterreicher, H. Bittner: J. Magn. Magn. Mat. 15-18, 1264 (1980)
7.212 F. Pourarian, W.E. Wallace, S.K. Malik: J. Less-Common Met. 85, 95 (1982)
7.213 F. Pourarian, W.E. Wallace, S.K. Malik: J. Magn. Magn. Mat. 25, 299 (1982)
7.214 S.G. Sankar, D.M. Gualtieri, W.E. Wallace: "Low Temperature Magnetic Properties of
the Hydrides and Deuterides of Er(Fe, Mn)2", in The Rare Earths in Modern Science and
Technology, ed. by G.J. Mc Carthy, J.J. Rhyne (Plenum, New York 1978) p. 69
7.215 H. Drulis, W. Petrynski, B. Stalinski: J. Less-Common Met. 101, 229 (1984)
7.216 G. Wiesinger: J. Less-Common Met. 130, 181 (1987)
7.217 K.H.J. Buschow: J. Less-Common Met. 51, 173 (1977)
338 G. Wiesinger and G. Hilscher

7.218 L.J. de Jongh, J. Bartolome, F.J.A.M. Greidanus, H.J.M. de Groot, H.L. Stipdonk,
K.H.J. Buschow: J. Magn. Magn. Mat. 25, 207 (1981)
7.219 K.H.J. Buschow, A.M. van der Kraan: J. Less-Common Met. 91, 203 (1983)
7.220 S.K. Malik, W.E. Wallace: Solid State Commun. 24, 283 (1977)
7.221 H. de Graaf, R.C. Thiel, K.H.J. Buschow: J. Phys. F 12, 2079 (1982)
7.222 K.H.J. Buschow, R.L. Cohen, K.W. West: J. Appl. Phys. 48, 5289 (1977)
7.223 R.L. Cohen, K.W. West, K.H.J. Buschow: Solid State Commun. 25, 293 (1978)
7.224 S.Z. Huang, M.K. Wu, R.L. Meng, C.W. Chu, J.L. Smith: Solid State Commun. 38, 1151
(1981)
7.225 0 3 . Zogal, D.J. Lam, A. Zygmunt, H. Drulis, W. Petrynski, S. Stalinski: Phys. Rev. B 29,
4837 (1984)
7.226 J.J. Reilly, R.H. Wiswall, Jr.: Inorg. Chem. 13, 218 (1974)
7.227 M. Ron, R.S. Oswald, M. Ohring, G.M. Rothberg, M.R. Polcari: Bull. Amer. Phys. Soc.
21, 273 (1976)
7.228 R. Hempelmann, D. Ohlendorf, E. Wicke: "Irreversible Change in Magnetic Properties
of TiFe by Hydrogenation", in Hydrides for Energy Storage, ed. by A.F. Andresen, A.J.
Maeland (Pergamon, Oxford 1978) p. 407
7.229 R. Hempelmann, E. Wicke: Ber. Bunsenges. Phys. Chem. 81,425 (1977)
7.230 G.K. Shenoy, D. Niarchos, P.J. Viccaro, B.D. Dunlap, A.T. Aldred, G.D. Sandrock: J.
Less-Common Met. 73, 171 (1980)
7.231 A. Bl~isius, U. Gonser: Appl. Phys. 22, 331 (1980)
7.232 L. Schlapbach, A. Seiler, F. Stucki: Mat. Res. Bull. 13, 697 (1978)
7.233 G. Busch, L. Schlapbach, F. Stucki, P. Fischer, A.F. Andresen: Int. J. Hydrogen Energy
4, 29 (1979)
7.234 L. Schlapbach, A. Seiler, F. Stucki, P. Zuercher, P. Fischer, J. Schefer: Z. Phys. Chem.
N F 117, 205 (1979)
7.235 F. Stucki, L. Schlapbach, A. Seiler: Ber. Bunsenges. Phys. Chem. 84, 1067 (1980)
7.236 L. Schlapbach, F. Stucki, A. Seiler, H.C. Siegmann: Surf. Sci. 106, 157 (1981)
7.237 F. Stucki: J. Less-Common Met. 83, L37 (1982)
7.238 L. Sehlapbach, T. Riesterer: Appl. Phys. A 32, 169 (1983)
7.239 Th. v. Waldkirch, R. Wessicken, H.-U. Nissen, L. Schlapbach: "A TEM and SAD Study
of TiFe and LaNi 5 Powder Particles: Evidence for Fe and Ni Precipitates", in 1982
Third Intern. Congress on Hydrogen and Materials Proc., p. 293, Paris
7.240 T. Schober, D.G. Westlake: Scripta Metall. 15, 913 (1981)
7.241 D. Khatamian, G.C. Weatherly, F.D. Manchester, C.B. Alcock: J. Less-Common Met.
89, 71 (1983)
7.242 D. Khatamian, G.C. Weatherly, F.D. Manchester: Acta Metall. 31, 1771 (1983)
7.243 J.J. Reilly, F. Reidinger: J. Less-Common Met. 85, 145 (1982)
7.244 L.J. Swartzendruber, L.H. Bennett, R.E. Watson: J. Phys. F 6, L331 (1976)
7.245 P. Thompson, M.A. Pick, F. Reidinger, L.M. Corliss, J.M. Hastings, J.J. Reilly: J. Phys.
FS, L75 (1978)
7.246 W. Sch/ifer, E. Lebsanft, A. Bl~isius: Z. Phys. Chem. NF 115, 201 (1979)
7.247 P. Thompson, J.J. Reilly, F. Reidinger, J.M. Hastings, L.M. Corliss: J. Phys. F9, L61
(1979)
7.248 T. Mizuno, T. Morozumi: J. Less-Common Met. 84, 237 (1982)
7.249 Y. Asada, H. Nose: J. Phys. Soc. Jap. 35, 409 (1973)
7.250 L.H. Bennett, LJ. Swartzendruber: Phys. Lett. 24A, 359 (1967)
7.251 L.J. Swartzendruber, L.H. Bennett: J. Appl. Phys. 39, 2215 (1968)
7.252 L.H. Bennett, L.J. Swartzendruber, R.E. Watson: Phys. Rev. 165, 500 (1968)
7.253 J.C. Swartz, L.J. Swartzendruber, L.H. Bennett, R.E. Watson: Phys. Rev. B 1,146(1970)
7.254 S.J. Pickart, R. Nathans, F. Menzinger: J. Appl. Phys. 39, 2221 (1968)
7.255 J. Giner, F. Gautier: J. de Phys. 38, C7-301 (1977)
7.256 G. Schadler, P. Weinberger: J. Phys. F. 16, 42 (1986)
7.257 P.E. Brommer, J.J.M. Franse, H. Hoelscher: "The Occurence of Ferromagnetism in the
Ti(Fel_xCo~) System", in Inst. Phys. Conf. Ser. 1981, No. 55, Chap. 4, p. 279
 Magnetic Properties, M6ssbauer Effect and Superconductivity 339

7.258 N. Buis, P.E. Brommer, P. Disveld, M.S. Schalkwijk, J.J.M. Franse: J. Magn. Magn.
Mat. 15--18, 291 (1980)
7.259 H. Hoelscher, J.C.P. Klaasse, J.J.M. Franse, P.E. Brommer: "Thermal Expansion and
Specific Heat in TiFexCol_x", in Inst. Phys. Conf. Ser. 1981, No. 55, Chap. 4, p. 283
7.260 N. Buis, P. Disveld, P.E. Brommer, J.J.M. Franse: J. Phys. F l l , 217 (1981)
7.261 G. Hilscher, G. Wiesinger, E. Lebsanft: J. Magn. Magn. Mat. 15--18, 1273 (1980)
7.262 C.J. Koizumi, W.N. Cathey: J. Phys. F 10, 497 (1980)
7.263 J. Beille, D. Bloch, F. Towfiq: Solid State Commun. 25, 57 (1978)
7.264 J. Beille, D. Bloch, F. Towfiq, J. Voiron: J. Magn. Magn. Mat. 10, 265 (1979)
7.265 R.C. Bowman, Jr., B.D. Craft, W.E. Tadlock, E.L. Venturini, J.S. Cantrell: .1. Appl. Phys.
57, 3036 (1985)
7.266 L. Schlapbach, D. Shaltiel, P. Oelhafen: Mat. Res. Bull. 14, 1235 (1979)
7.267 F. Aubertin, S.J. Campbell, U. Gonser: Hyperf. Inter. 28, 997 (1986)
7.268 R.O. Moyer, Jr., R. Lindsay: J. Less-Common Met. 70, P57 (1980)
7.269 Z.M. Stadnik, R.O. Moyer, Jr.: J. Less-Common Met. 98, 159 (1984)
7.270 K.H.J. Buschow, A.M. van Diepen: Solid State Commun. 31, 469 (1979)
7.271 E. Tuscher: Monatsh. Chem. 110, 1275 (1979)
7.272 P. Vulliet, G. Teisseron, J.L. Oddou, C. Jeandey, A. Yaouanc: J. Less-Common Met.
104, 13 (1984)
7.273 K. Hiebl, E. Tuscher, H. Bittner: Mouatsh. Chem. 110, 869 (1979)
7.274 E. Tuscher: Monatsh. Chem. 111,535 (1980)
7.275 K. Hiebl, E. Tuscher, H. Bittner: Monatsh. Chem. 110, 9 (1979)
7.276 M.H. Mintz, Z. Hadari, M.P. Dariel: J. Less-Common Met. 74, 287 (1980)
7.277 C. Stioui, D. Fruchart, A. Rouault, R. Fruchart, E. Roudaut, J. Rebiere: Mat. Res. Bull.
16, 869 (1981)
7.278 B. Rupp: J. Less-Common Met. 104, 51 (1984)
7.279 A. Venkert, M.P. Dariel, M. Talianker: J. Less-Common Met. 103, 361 (1984)
7.280 B. Rupp, G. Wiesinger: J. Less-Common Met. 104, 65 (1984)
7.281 P. Rogl, B. Rupp, G. Wiesinger, J. Schefer, P. Fischer: J. Less-Common Met. 113, 103
(1985)
7.282 B. Rupp, E. Tuscher: J. Less-Common Met. 104, L9 (1984)
7.283 F. Aubertin, U. Gonser, S.J. Campbell: J. Less-Common Met. 101,437 (1984)
7.284 F. Aubertin, U. Gonser, S.J. Campbell: J. Phys. F 14, 2213 (1984)
7.285 G. Busch, L. Schlapbach, Th. v. Waldkirch: J. Less-Common Met. 60, 83 (1978)
7.286 P. Fischer, W. Haelg, L. Schlapbach, K. Yvon: J. Less-Common Met, 60, 1 (1978)
7.287 K.H.J. Buschow, H.H, van Mal, A.R. Miedema: J. Less-Common Met. 42, 163 (1975)
7.288 E.B. Boltich, S.K. Malik, W.E. Wallace: J. Less-Common Met. 74, 111 (1980)
7.289 S.K. Malik, E.B. Boltich, W.E. Wallace: Solid State Commun. 33, 921 (1980)
7.290 L. Schlapbach, C. Pina-Perez, T. Siegrist: Solid State Commun. 41, 135 (1982)
7.291 P.J. Viccaro, G.K. Shenoy, B.D. Dunlap, D.G. Westlake, S.K. Malik, W.E. Wallace: J.
de Phys. 40, C2-157 (1979)
7.292 J.M.D. Coey, D.H. Ryan, Yu. Boliang: J. Appl. Phys. 55, 1800 (1984)
7.293 T. Masumoto, S. Ohnuma, K. Shirakawa, M. Nose, K. Kobayashi: J. de Phys. 41,
C8 686 (1980)
7.294 K. Shirakawa, T. Kaneko, M. Nose, S. Ohnuma, H. Fujimori, T. Masumoto: J. Appl.
Phys. 52, 1829 (1981)
7.295 K.M. Unruh, C.L. Chien: Phys. Rev. B 30, 4968 (1984)
7.296 J.M.D. Coey, D. Givord, A. Lienard, J.P. Rebouillat: J. Phys. F 11, 2707 (1981)
7.297 J. Chappert, J.M.D. Coey, A. Lienard, J.P. Rebouillat: J. Phys. F 11, 2727 (1981)
7.298 K. Shirakawa, K. Fukamichi, T. Kaneko, T. Masumoto: Physica lI9B, 192 (1983)
7.299 H. Hiroyoshi, K. Fukamichi: Phys. Lett. 85A, 242 (1981)
7.300 H. Hiryoshi, K. Fukamichi: J. Appl. Phys. 53, 2226 (1982)
7.301 K.H.J. Buschow, P.H. Smit: J. Magn. Magn. Mat. 23, 85 (1981)
7.302 S.N. Kaul: Phys. Rev. B27, 6923 (1983)
340 G. Wiesinger and G. Hilscher

7.303 Yu. Boliang, D.H. Ryan, J.M.D. Coey, Z. Altounian, J.O. Strocm-Olsen, F. Razavi: J.
Phys. F13, L217 (1983)
7.304 S.M. Fries, H.-G. Wagner, U. Gonscr, L. Schlapbach, R. Montiel-Montoya: J. Magn.
Magn. Mat. 45, 331 (1984)
7.305 H. Fujimori, K. Nakanishi, K. Shirakawa, T. Masumoto, T. Kaneko, N.S. Kazama:
"Enhancement of Amorphous Fe-Zr Ferromagnetism by Hydrogen Absorption", in
1981 4th Int. Conf. on Rapidly Quenched Metals Proc., p. 1629 (Japan. Inst. of Metals,
Sendai 1982)
7.306 H. Fujimori, K. Nakanishi, H. Hiroyoshi, N.S. Kazama: J. Appl. Phys. 53, 7792 (1982);
D.H. Ryan, J.M. Cadogan, E.J. Devlin, J.M.D. Cocy: Z. Phys. Chem. N F 145, 367 (1985)
7.307 Z.S. Wronski, A.H. Morrish, A.M. Stewart: Phys. Lett. 101A, 294 (1984)
7.308 S.M. Fries, H.-G. Wagner, S.J. Campbe]l, U. Gonser, N. Blaes, P. Steiner: J. Phys. F 15,
1179 (1985)
7.309 K. Tanaka, M. Higatani, K. Kai, K. Suzuki: J. Less-Common Met. 88, 317 (1982)
7.310 K. Samwer, X.L. Yeh, W.L. Johnson: J. Non-Cryst. Sol. 61 & 62, 631 (1984)
7.311 D.W. Forester, P. Lubitz, J.H. Schelleng, C. Vittoria: J. Non-Cryst. Sol. 61 & 62, 685
(1984)
7.312 B.S. Berry, W.C. Pritchet: J. Appl. Phys. 52, 1865 (1981)
7.313 W. Chambron, F. Lancon, A. Chamberod: Scr. Met. 18, 29 (1984)
7.314 C.G. Robbins, Z.D. Chen, J.G. Zhao, M.J.O'Shea, D.J. Sellmyer: J. Appl. Phys. 53, 7798
(1982)
7.315 D.J. Sellmyer, C.G. Robbins, M.J.O'Shea: J. Non-Cryst. Sol. 61 & 62, 655 (1984)
7.316 C.B. Satterthwaite, I.L. Toepke: Phys. Rev. Lett. 25, 741 (1970)
7.317 F. Skoskiewicz: Phys. Stat. Sol. (a) 11, K123 (1972)
7.318 B. Stritzker: Z. Physik 268, 261 (1974)
7.319 W.L. McMillan: Phys. Rev. 167, 331 (1968)
7.320 W. Buckel: Z. Phys. Chem. N F I I 6 , 135 (1979)
7.321 B. Stritzker, H.L. Luo: Solid State Commun. 29, 811 (1979)
7.322 A. Traverse, H. Bernas, J. Chaumont, X.-J. Fan, L. Mendoza-Zelis: Phys. Rev. B 30,
6413 (1984)
7.323 K. Kai, M. Ikebe, K. Suzuki, T. Nomoto: J. Less-Common Met. 89, 229 (1983)
7.324 E. Babic, B. Leontic, J. Lukatela, M. Miljak, M.G. Scott: "Transport Properties of some
Hydrogen Doped Zr-Ni Metallic Glasses", in 1981 4th Intern. Conf. Rapidly Quenched
Metals Proc., p. 1617 (Japan. inst. of Metals, Sendai 1982)
7.325 K. lnoue, K. Tachikawa: "High Field Superconducting Properties of Laves Phases in
V Hf and V Hf-Zr Alloys", in 1971, 12th Intern. Conf. Low Temp. Physics Proc., p. 483
(Acad. Press of Japan, Kyoto 1971)
7.326 P. Duffer, D.M. Gualtieri, V.U.S. Rao: Phys. Rev. Lett. 37, 1410 (1976)
7.327 V.U.S. Rao, D.M. Gualtieri, S. Krishnamurthy, A. Patkin, P. Duffer: Phys. Lett. 67A,
223 (1978)
7.328 C. Geibel, W. Goldacker, H. Keiber, V. Oestreich, H. Rietschel, H. Wtihl: Phys. Rev.
B 30, 6363 (1984)
7.329 B.M. Klein, W.E. Pickett, D.A. Papaconstantopoulos, L.L. Boyer: Phys. Rev. B 27, 6721
(1983)
7.330 D. Fruchart, F. Vaillant, A. Rouault, A. Benoit, J. Flouquet: J. Less-Common Met. 101,
285 (1984)
7.331 J. Osterwalder, T. Riesterer, L. Schlapbach, F. Vaillant, D. Fruchart: Phys. Rev. B 31,
8311 (1985)
7.332 S.K. Malik, W.E. Wallace, T. Takeshita: Solid State Commun. 28, 359 (1978)
7.333 J. Muller: Pept. Progr. Phys. 43, 641 (1980)
7.334 P.R. Sahm: Phys. Lett. 26A, 459 (1968)
7.335 L.J. Vieland, A.W. Wicklund, J.G. White: Phys. Rev. B 11, 3311 (1975)
7.336 C. N61scher, P. Mfiller, H. Adrian, M. Lehmann, G. Saemann-Ischenko: Z. Phys. B41,
291 (1981)
 Magnetic Properties, M6ssbauer Effect and Superconductivity 341

7.337 F.C. Matacotta, C. Ferdeghini, M. Ferretti, G. Bruzzone: IEEE Trans. Mag. MAG-19,
897 (1983)
7.338 W.A. Lanford, P.H. Schmidt, J.M. Rowell, J.M. Poate, R.C. Dynes, P.D. Dernier: Appl.
Phys. Lett. 32, 339 (1978)
7.339 3.B. Vetrano, G.L. Guthrie, H.E. Kissinger: Phys. Lett. 26A, 45 (1967)
7.340 K.V.S. Rama Rao, H. Sturm, B. Etschner, A. Weiss: Phys. Lett. 93A, 492 (1983)
7.341 J. Elton, H. Oesterreicher: J. Less-Common Met. 90, L37 (1983)
7.342 C. Ferdeghini, M. Ferretti, F.C. Matacotta, C. Rizzuto, A. Siri: Cryogenics 25, 208
(1985)
7.343 M. Wilhelm, K. Wohlleben: IEEE Trans. Magn. MAG-17, 1625 (1981)
List of Tables

2.1 Crucibles 21
3.1 Thermodynamic data for LaNis-H 63
3.2 Thermodynamic data for FeTi-H 65
3.3 Thermodynamic data for ZrMn2.8 67
4.1 Ternary metal hydrides and deuterides studied
by neutron diffraction: metal atom substructure,
space group, lattice parameter and H(D) interstice 94-102
4.A Structure data for selected ternary metal hydrides
and deuterides studied by neutron diffraction:
space group, lattice parameters and atomic parameters la9-129
4.B Magnetic structures 130-131
5.1 Valence band and core level data for binary
metal hydrides 152-155
5.2 Valence band and core level data for ternary
metal hydrides 178-179
6.6.1 Heat of solution and formation of binary
metal hydrides 266-269
6.6.2 Heat of formation of selected ternary metal hydrides 269-273
6.6.3 Heat of formation of some ternary metal
hydrides calculated with the band-structure
semi-empirical model 274-278
Subject Index

A 15 structure 176, 331 Bonding 105

AB2 Laves phase compounds, hydrides Bonding band Tables 5.1, 2; 140, 150
electronic properties 201 Brillouin zone 142, 187
heat of formation 67, 202; Tables 3.3, Bulk modulus 233
6.6.2
magnetic properties 204, 309 C 14 structure 201,310, 312
peT curves 38 C 15 structure 66, 201,312
phase diagram 15 Calorimeter, -metric 58, 63
preparation 37 Catalysis 4
structure, lattice parameters Tables 4.1, Cathodic charging 40
4.A Capacity 87, 105, 208
substitutions 34 reduction by impurities 42
thermodynamic properties 66, 205 Characterization 29
Activation 38, 60, 319; TAP 64* of hydrides 37
Amorphous alloys and hydrides 7, 28, ofintermetallics 29
107, 316, 325 Charge transfer Chap. 5; 143, 149, 181,
amorphisation 62 291,317
electronic properties 177, 205, 207 Chemicalpotential 13, 53, 222, 226
magnetic properties 207, 325 Coercitivity 302
models 63 Coherent potential approximation CPA
preparation 28, 205 146, 166, 255
resistivity 207 Cohesion, cohesive energy 3, 144, 231
site occupancy 107 Compensation temperature 305,307
structural information 88 Complex hydrides 88
thermodynamic properties 62 Compton scattering 149
Anionic model 114, 139, 290, 324 Configurational entropy 69, 71, 73, 115
Anisotropy 303 Congruently 23
Annealing 24, 60, 67 Contamination see also Impurity 32
Antibonding band 158 by crucible 21
Antiferromagnetic 175, 301, Chap. 7 Conversion electron M6ssbauer
Antistructure atoms 185, 190, 289,320 spectroscopy CEMS 318
Asperomagnetism 326 Corelevel spectra, shift Chap. 5;
Atomic coordinates 87, 91 Tables 5.1, 2; 158, 181,203, 208
Augmented plane wave APW 142, 180 Correlation effects 107
Augmented spherical wave ASW 143 Critical temperature, point 2, 224, 261
Crucible 20, 27, 33
Band structure 8; Chap. 5; 205, 249, 255 ceramic crucible 20
Band width 140, 150, 257; Table 5.1 cold crucible 20
Binary hydrides 106 corrosion 20
electronic properties 150 refractory metal crucible 20, 33
phase diagram 169 Crystal field 288
structure 88 Crystallinity 89, 92, 309

* TAP 64: Topics in Applied Physics, Vol. 64 (Hydrogen in Intermetallic Cwnpounds II).
346 Subject Index

Curie temperature T c 93, 117; Table 4.B; Equilibrium 50, 55, 80, 227
Chap. 7 partial 80
Czochralski technique 27 pressure 2
single condensed phase 50
Debye temperature 173, 329 three condensed phases 56
Decohesion 145 two condensed phases 50
Decomposition 7, 28, 68, 289 Exchange energy 288
Decrepitation 27, 37, 39
Degradation 300 Fermi energy Ev 140, t45, 256; Chap. 5
De Haas van Alphcn 149, 185 Fermi surface FS 147, 160, 173
Density of states DOS, total, partial Ferrimagnetic 306
Chap. 5; 142, 150, 180, 255 Ferromagnetic Chap. 7, 175
Deoxidation 33, 35 Ferromagnetic resonance 299
Desintegration 303 FeTi,FeTi hydrides
Deuterium, deuteride 91,295 electronic properties 185, 318
Differential thermal analysis DTA 58 heat of formation 65; Tables 3.2, 6.6.2
Diffractometer 90 magnetic properties 185, 318
Diffusion 7, 23; TAP 64 order parameter 186
diffusion path 110 pcT curves 38
Dingle temperature 160 phase diagram 15, 185
Disorder 7, 87, 90, 108,166, 177, 194 preparation 27, 37
Disproportionation s e e also resistivity 186
Decomposition 7, 37, 40, 62 single crystals 27
Dissociation 2, 231 structure, lattice parameters Chap. 4;
Tables 4.1, 4.A; 185
Einstein oscillator 55, 77 substitutions 41
Einstein temperature 59, 162, 226, 229 thermodynamic properties 64
Effective medium theory 145, 220, 251 Floating zone technique 27
Elastic energy 245 Fractals 7
Electrical resistivity 148, 174 Free energy, formation, mixing, Gibbs 51,
Electrochemicalcharging 40, 159 57, 223
Electron energy loss spectroscopy EELS Friauf-phase 242
156, 166
Electron field gradient EFG 294, 311
Electron paramagnetic resonance EPR Gap 171,200
316 Geometrical model 69, 88, 113,242
Electron-phonon interaction 147, 162, Getter 21, 24
165, 168,173,329 Glassy s e e Amorphous 28
18-Electron rule 116
Electronic specific heat Chaps. 5, 7; 147, Hard magnetic material 302
151,176, 231,287 Haucke-phase 242
Electronic structure Chap. 5 Heat capacity 59, 64
binary hydrides 150 Heat of formation, solution Chaps. 3, 6;
effect of hydrogen 140 2, 112, 144, 221,230
ternary hydrides 177 band-structure model, local 249, 258,
Ellingham diagram 16, 17 262
Embedding energy 251 effective medium theory 221
Embrittlement 3,145 Griessen-Driessen model 234, 249
Empirical correlations 88,231 local heat of formation 242
Energy bands 142, 150, 170, t87, 196 Miedema model 234, 237, 244
Energy distribution curve 149 partial 223
Enthalpy 54, 225 semi-empirical models 234
Entropy 54 Shilov relation 248
hydride formation 54 table, calculated values ternary
mixing 224 hydrides Table 6.6.3
 Subject Index 347

tabulated values binary hydrides Jahn-Teller effect 163

Table 6.6.1 Jellium model 141,146
tabulated values ternary hydrides
Tables 6.6.2, 3,1-3 Kierstead model 73
volume effects 260 Knight shift 148, 168, 174, 183
Heavy electron, -fermion 8, 175, 287 Korringa-Kohn-Rostocker KKR •42,320
High pressure magnetisation 306
High pressure technology 7 LaNis,LaNi 5 hydride
Homogeneity range of intermetallics 14, electronic properties 177, 182
310 electronic specific heat 181,183
Hybridisation 175, 177, 182, 205, 251 heat of formation 63, 185; Tables 3./,
Hydride carbides 88 6.6.2
Hydrogen bronze 88 magnetic properties 181,183, 298
Hydrogen site occupancy 91 pcTcurves 35, 38
Hydrogenation see Preparation of phase diagram 14
hydrides 37 preparation 24, 38
Hyperfine interactions, -field 161,292, resistivity 184
313 single crystal 27
Hysteresis 44, 59, 81 structure, lattice parameters 108, 177,
182; Tables 4.1, 4.A
Icosahedral 7 substitutions 24, 35, 42, 64, 184
Imaginary binary hydride model 70, 116, thermodynamic properties 63
242 Lattice
Implantation 6, 156, 158 contraction 103
Impurity/purity 18, 19, 35, 41 distortion 162, 186
by crucible 21, 22 dynamics 76
effect on capacity 42 expansion 103
effect on stability 41 parameter 91; Tables 4.1, 4.A
hydrogen 141,201 symmetry 92
of hydrogen gas 37 Lattice gas 3, 6, 78, 79, 226
of the intermetallic compounds 17, 35, Laves phase 6, 20/, 242, 257, 309, 315
41 Line broadening 60, 90, 301
of the metallic constituents 17, 18, 19,41 Linear ordering 6, 107
phase 92 Liquidus 23
segregation 309 Local band structure model 262, 264
Inhomogeneity see also Homogeneity 89 Local density functional approximation
Interaction LDF 141
attractive 116 Local heat of formation model 116
energy 69
H - H 140 Magnetic
magnetic 288 coupling 290
repulsive 114 moment 93, 288
lnteratomic distance 109 order 117; Table 4.B; 175
Interface 2 phase diagram 175
Interstices, interstitial sites Table 4.1; 257 structure Table 4.B
Interstitial hole size 113,231 susceptibility 31,147, 148
Inverse photoemission BIS 149, 158 Magnetism, concepts 286
Ion implantation 156, 158 Maximum hydrogen content 105
Isomer shift IS 190, 293,299, 317 Mean field theory 288
Isotherms seepcT curves Melting
Isotope 1,3, 295 arc melting 20
isotope effect 76, 118, 160, 162 Levitation melting 20, 21
isotope mixture 91 melting in crucibles 20
isotope ratio IR 77 melting techniques 19, 21
Itinerant 288 Vacuum induction melting VIM 32
348 Subject Index

Metal atom substructure Table 4.1 Pairing 6

Metal-semiconductor transition 8, 173 Pairwiseinteraction 6, 69
Metallography 29, 89 Partial heat of solution 223
Matastable 5 Partial molar properties 52
Mg based compounds, hydrides Pauli susceptibility 147, 151,185, 298
electronic properties 193,197 Peritectic 24, 27
heat of formation 68 Phase diagrams of binary intermetallics
magnetic properties 200, 323 13
pcT curves 38 La-Ni 14
phase diagram 14 Mg-Ni 14
preparation 20, 28, 37 of ternary intermetallics 16
resistivity 200 Ti-Cr 15
structure, lattice parameters Chap. 4; Ti-Fe 15
Tables 4.1, 4.A; 194 Phase separation 72
substitutions 20 Phase transition 78, 109, 297,304
thermodynamic properties 68 Phonons s e e a l s o Electron-phonon
Microcrystallinity 309 coupling 162, 165
Micrographs 30, 34 Photoelectron spectroscopy XPS, UPS
Microprobe 29, 89 Chap. 5; 149, 158, 167, 170, 177
Microtwinning 90 Plateau
Miedemamodel 70, 146, 244 pressure see also Pressure-composition
Minimum H-atom distance 110 isotherms 41, 53, 60, 80
Mischmetal 19 slope 7, 44, 62
Miscibility gap 224 Point group symmetry 93
Mixed valence 8 Poisoning of the surface vy SO2, CO, air
Molar volume, partial 223, 253, 254 39, 89
M6ssbauer spectroscopy 148, 190; Positron annihilation 149
Chap. 7; 291 Precipitations 25, 27
Mott transition 174 Preparation of hydrides Chap. 2; 37, 89
Muffin tin MT 142 electrochemically 40
Multi-step isotherms 75 gas phase 37
Multiple plateau 73, 78 ion implantation 6
Multiple sublattice model 75 via organometallic solutions 40
Muon spin rotation/~SR 148, 183 Preparation ofintermetallics Chap. 2; 89
chemical preparation 37
Narrow band phenomena 8 gas phase 28
Ntel temperature TN 93, 117; Table 4.B; melt, industrial scale 32
309 melt, laboratory scale 17
Neutron scattering, diffraction 60, 77, 90, thin films, evaporation 28
93, 182, 306, 312 Pressure-composition isotherms, pcT
polarized 91 curves 2, 21, 57
Nonstoichiometric hydrides 146 for characterization 29
Nuclear density map 92 dynamic 81
Nuclear magnetic resonance NMR 148, effect of annealing 25
161,163, 174, 297 effect of substitutions 41
for FeTi, LaNi 5, Mg2Ni, TiCr 2 38
Octahedral 106, 107, 169, 170, 182, 264 of industrially preparaed alloys 35
Optic mode 76 measurements 57
Optical spectroscopy t 50 multi-step isotherms 73, 75
Order, ordering 6, 80, 115, 169, 319 slope 7, 44
linear ordering 6 Pressure-magnetisation isotherms 297
Order-disorder transition 2, 78, 79, 166 Price 18
Orthorhombic distortion 186 Probe, H as a probe 7
Oxide formation 16 Profile fitting 92
Oxygen stabilized compounds 5, 111,323 Protonicmodel 139
 SuNect Index 349

Pseudopotential 143 Spin glass 306, 309, 312, 316

Pulverization 303 Spin-lattice relaxation 161
Purification Square-planar 106
of constituent metals 19 Square-pyramidal 106, 107
intermetallic compounds 30 Stability 112
Purity s e e Impurity Stoichiometry
of the intermetallic compounds 15
Stoner criterion 288
Quasicrystals 2, 7 Stoner enhancement factor 181,183, 298
S t o n e r - W o h l f a r t h model 2 8 8
Storage, storage capacity 4, 68, 87,177
Raman spectra 173 Strain 90
Rapid solidification 28 modulation 7
Relaxation rate 161,168, 294 removal 59, 60
Reorientation temperature 303 Structure Chap. 4
Repulsion 114, 117 binary hydrides 88
Resistivity, electrical 148, 174, 290 magnetic Table 4.B
Reviews 4, 88, 115, 139, 221,285 refinement 91
Rictveld fitting method 91 ternary hydrides Tables 4.1, 4.A
Rigid band model 139, 161,166, 201 Sublattice substructure 3, 71, 74;
Rocking curves 31 Table 4.1
Ruderman-Kittel-Kasuya-Yosida Sublattice magnetisation 313
RKKY 287, 290, 317 Supercell 147
Rule of reversed stability 112, 240, 247, Superconductivity 8, 162; Chaps. 5, 7;
261 147, 166, 329
Superlattice 7
Superparamegnetism 299, 323, 325
Schrfdinger equation 141 Superstructure 80
Screening Surface 3, TAP 64
core hole 149 area, specific 39
hydrogen impurity 139 surface site 1
Sealing s e e Poisoning 39
Segregation 13, 22, 89,299
Semiconducting 112, 168, 173 Ternary hydride
Semi-empirical models 234 electronic properties 177
Shear modulus 232 structure Table 4.1
Sievert's law 64 thermodynamic properties Chaps. 3, 6;
Single crystal 27, 31 Tables 3.1-3, 6.6.2-3
growth, FeTi, LaNi s 27 Ternary intermetalliccompounds 16
Site energy 69 Tetragonal distortion 162, 169
Slater-Koster SK 143 Tetrahedral 106, 107,264
Slope of plateau 7,44 Thermodynamic properties of alloy
Solid-state reaction 28 formation 12; Chaps. 3, 6
Solidification 20, 22 Thermodynamic properties of hydrides
rapid 28 Chap. 3; 50
unidirectional 26 binary hydrides 237; Table 6.61
Solidus 23 integral 50
Solubility of hydrogen, solution 3,13,223 multiple plateau 73
Space group 92; Table 4.1 partial molar 52
Specific Heat plateau properties 53
electronic s e e Electronic quarternary hydrides 241
lattice 59 ternary hydrides 239; Tables 6.6.2-3
partial molar 228 Thin films
Specific surface area 39 characterization 32
Spin fluctuation 330 preparation 28
350 Subject Index

Tight binding approximation 255 Westlake model 69, 113, 145

Time-of-flight method 90 Work function 163, 256
Total Energy 143, 160, 167, 208 Wyckoff positions 95

Unit cell volume 42, 231 X-ray absorption near edge spectroscopy
XANES 150,159,167,186
Vacancy 265 X-ray absorption spectroscopy XAS 150
Valence band, -spectra Chap. 5; 149, 158 X-ray emission spectroscopy, soft SXES
Valence change 8, 118,307, 330 50; Chap. 5; 164, 177, 200
Van't Hoff relation, plot 2, 58, 219, 229 X-ray photoelectron spectroscopy s e e
Volatility 17 Photoelectron spectroscopy Chap. 5; 149