i

Faculteit Wetenschappen

Departement Fysica

Structure characterization of triple perovskites and related

systems by transmission electron microscopy

Structuurbepaling van triple perovskieten en gerelateerde

systemen aan de hand van transmissie

elektronenmicroscopie

Proefschrift voorgelegd tot het behalen van de graad van

Doctor in de Wetenschappen

Aan de Universiteit Antwerpen, te verdedigen door

Robert Paria Sena

Promotoren:

Prof. Dr. Joke Hadermann Antwerpen

Prof. Dr. Gustaaf Van Tendeloo February, 2017

ii

DOCTORAL COMMITTEE

Chairman:

Prof. Dr. Paul Scheunders, University of Antwerp, Antwerp, Belgium

Supervisors of the PhD. Work:

Prof. Dr. Joke Hadermann, University of Antwerp, Antwerp, Belgium

Prof. Dr. Gustaaf Van Tendeloo, University of Antwerp, Antwerp, Belgium

Members:

Prof. Dr. Peter Battle, Oxford University, Oxford, England

Prof. Dr. Artem Abakumov, Skolkovo Institute of Science and Technology, Moscow, Russia

Prof. Dr. Michiel Wouters, University of Antwerp, Antwerp, Belgium

iii

Contents

Abstract ........................................................................................................................................... vi

Abstract ......................................................................................................................................... vii

Chapter 1 ....................................................................................................................................... 1

Introduction .................................................................................................................................. 1

1.1. General aspects of perovskites .............................................................................................. 2

1.2. Important parameters and properties of perovskites .............................................................. 3

Transition metals ....................................................................................................................... 3

Tolerance factor ........................................................................................................................ 3

Octahedral tilting in perovskites ............................................................................................... 4

Glazer notation .......................................................................................................................... 4

Jahn-Teller (JT) distortion ........................................................................................................ 5

1.3. Ordering in perovskites ......................................................................................................... 5

1.4. Bronze structures ................................................................................................................... 9

Tetragonal tungsten bronze (TTB) structures ........................................................................... 9

Pseudo-tetragonal tungsten bronze (TTB) structures.............................................................. 10

Bond valence sum method ...................................................................................................... 10

1.5. Magnetic properties of perovskites ...................................................................................... 11

1.5.1. General types of spin ordering .......................................................................................... 11

1.5.2. Relaxor Ferroelectrics and Relaxor Ferromagnets ........................................................... 14

Chapter 2 ........................................................................................................................................ 16

Experimental techniques ................................................................................................................ 16

2.1. General Diffraction .............................................................................................................. 16

2.2. Interaction electron-matter .................................................................................................. 17

2.3. Electron diffraction .............................................................................................................. 18

Schemes of the most relevant perovskite supercells encountered in this thesis ..................... 19

2.4. High angle annular dark field scanning transmission electron microscopy and annular

bright field scanning transmission electron microscopy ............................................................ 23

2.5. Energy dispersive X-ray scanning transmission electron microscopy spectroscopy (EDX-

STEM) ........................................................................................................................................ 24

iv

Chapter 3 ........................................................................................................................................ 26

Investigation of crystal structure and magnetic properties of the perovskite based La3Ni2B’O9

(B’=Nb, Ta, Sb0.5Nb0.5) and La2A’Ni2B’O9 (A’=Ca or Sr, B’=Te or W) in search for relaxor

ferromagnetic AA’2B2B’O9 compounds ........................................................................................ 26

3.1. Introduction ......................................................................................................................... 26

3.2. Synthesis .............................................................................................................................. 29

3.3. SrLa2Ni2TeO9 ...................................................................................................................... 30

3.3.1. Experimental Results ........................................................................................................ 30

3.3.2. Discussion ......................................................................................................................... 42

3.4.1. Experimental results on the structure ............................................................................... 45

3.4.2. Experimental results on the magnetic properties .............................................................. 56

3.4.3. Conclusions ...................................................................................................................... 63

Chapter 4 ........................................................................................................................................ 64

A3Fe2TeO9 (A=Sr, Ba) compounds, the search for relaxor ferromagnets continues in compounds

containing Fe3+

instead of Ni2+

....................................................................................................... 64

4.1. Sr3Fe2TeO9 ........................................................................................................................... 64

4.1.1. Introduction ...................................................................................................................... 64

Synthesis ................................................................................................................................. 66

4.1.2. Experimental results ......................................................................................................... 66

4.1.3. Discussion ......................................................................................................................... 78

4.1.4. Conclusion ........................................................................................................................ 81

4.2. Ba3Fe2TeO9 .......................................................................................................................... 82

4.2.1. Introduction ...................................................................................................................... 82

Synthesis ................................................................................................................................. 83

4.2.2. Experimental results ......................................................................................................... 83

4.2.3. Discussion and conclusion ................................................................................................ 92

Chapter 5 ........................................................................................................................................ 93

In search of a Jahn-Teller distorted Cr(II) oxide with Ba3Cr2TeO9 ............................................... 93

5.1. Introduction ......................................................................................................................... 93

Synthesis ................................................................................................................................. 95

5.2. Solution and refinement of the crystal structure .................................................................. 95

5.3. Magnetic Properties ........................................................................................................... 101

v

5.4. Conclusion ......................................................................................................................... 102

Chapter 6 ...................................................................................................................................... 103

Solution of the crystal structure of K6.4(Nb,Ta)36.3O94 with advanced transmission electron

microscopy ................................................................................................................................... 103

6.1. Introduction ....................................................................................................................... 103

Synthesis ............................................................................................................................... 103

6.2. Solution and refinement of the crystal structure ................................................................ 104

6.3. Discussion .......................................................................................................................... 111

6.4. Conclusion ......................................................................................................................... 113

Chapter 7 ...................................................................................................................................... 114

Conclusions and outlook .............................................................................................................. 114

References ............................................................................................................................. 116

List of Abbreviations ............................................................................................................ 121

Publications related to this thesis .......................................................................................... 122

Publications of which the work was not included in this thesis............................................ 122

Oral presentations at conferences ......................................................................................... 123

Poster presentations at conferences....................................................................................... 123

Acknowledgement ................................................................................................................ 125

vi

Abstract

During my Ph.D. research, I have investigated the structure of specific perovskite based oxides in

order to establish the correlation between the structure and the electrical and magnetic properties.

The final goal of our study is to design new compounds with a potential for applicable properties,

such as for example relaxor ferromagnetism.

The synthesis, X-ray and neutron diffraction studies and the determination of the magnetic

properties are not part of the Ph.D. work; these have been performed by collaborating research

groups at the University of Oxford, New Jersey State University, Bragg Institute, School of

Physical, Environmental and Mathematical Sciences (Australia) and Taras Shevchenko National

University of Kyiv. Within my Ph.D. work itself, the crystal structures have been solved, using a

combination of transmission electron microscopy techniques, including selected area electron

diffraction (SAED) combined with real space imaging using different techniques such as high

angle annular dark field scanning transmission electron microscopy (HAADF-STEM), annular

bright field scanning transmission electron microscopy (ABF-STEM) and energy dispersive X-

ray spectroscopy-STEM (EDX-STEM). Based on these studies, models have been proposed and

refined for different perovskite compounds. With these models we have tried to explain the

variations in the properties of the samples and we have compared them to similar materials. The

disclosed relations have rendered fundamental knowledge, applicable for the optimization of the

properties of the investigated materials as well as of related perovskite materials.

vii

Abstract

Gedurende mijn doctoraatsonderzoek heb ik de structuur van specifieke, perovskiet gebaseerde

oxides onderzocht om de correlatie tussen de structuur en de elektrische en magnetische

eigenschappen te bepalen. Het uiteindelijke doel van onze studie is om nieuwe samenstellingen

met een potentieel voor toepasbare eigenschappen te ontwikkelen zoals relaxor ferromagnetisme.

De synthese, X-stralendiffractie en neutronendiffractiestudies and de studie van de magnetische

eigenschappen zijn is geen onderdeel van het huidige doctoraatswerk; de materialen zijn

gesynthetiseerd door de collaborerende onderzoeksgroepen aan de University of Oxford, New

Jersey State University, Bragg Institute, School of Physical, Environmental and Mathematical

Sciences (Australië) en Taras Shevchenko National University of Kyiv. De bulk diffractie

technieken zijn ook uitgevoerd door onze collaborators. Tijdens het doctoraatsonderzoek zelf, is

de kristalstructuur opgelost door gebruik van een combinatie van transmissie

elektronenmicroscopie technieken, waaronder selected area electron diffraction (SAED)

gecombineerd met reële ruimte beeldvorming met behulp van verschillende technieken zoals high

angle annular dark field scanning transmission electron microscopy (HAADF-STEM), annular

bright field scanning transmission electron microscopy (ABF-STEM) en energy dispersive X-ray

spectroscopy-STEM (EDX-STEM). Modellen gebaseerd op deze studies zijn voorgesteld en

verfijnd voor verschillende perovskiet samenstellingen. Met deze modellen hebben we

geprobeerd de verschillen in de eigenschappen van de stalen te verklaren en we hebben ze

vergeleken met gelijkaardige materialen. De opgehelderde relaties leverden fundamentele kennis

op, toepasbaar voor de optimalisatie van de eigenschappen van de onderzochte materialen zowel

als voor gerelateerde perovskiet materialen.

1

Chapter 1

Introduction

The thesis is divided in seven chapters. In Chapter 1, I will introduce general concepts and basic

ideas on different topics that have been necessary to understand and explain our results, mainly

focusing on perovskite materials and their structural, electrical and magnetic properties. As one

crystal structure which belongs to the pseudo-tetragonal tungsten-bronze structures has also been

studied, some basic knowledge about these structures will also be included.

Since for the characterization of the compounds different techniques have been used, I have

described and discussed those techniques in a general way in Chapter 2. Among these techniques

are: X-ray powder diffraction (X-RPD), neutron powder diffraction (NPD), selected area electron

diffraction (SAED), high angle annular dark field scanning transmission electron microscope

(HAADF-STEM), annular bright field scanning transmission electron microscope (ABF-STEM),

energy dispersive X-ray scanning transmission electron microscope (EDX-STEM), energy

dispersive X-ray spectroscopy (EDS).

In the next three chapters, I will introduce my results together with the results of my

collaborators, it involves different single, double and triple perovskites. In Chapter 3, I will focus

on investigation of crystal structure and magnetic properties of the perovskite based La3Ni2B’O9

(B’=Nb, Ta, Sb0.5Nb0.5) and La2A’Ni2B’O9 (A’=Ca or Sr, B’=Te or W) in search for relaxor

ferromagnetic AA’2B2B’O9 compounds. In Chapter 4, I will include my studies about A3Fe2TeO9

(A=Sr, Ba) compounds, the search for relaxor ferromagnets continues in compounds containing

Fe3+

instead of Ni2+

, finally in the Chapter 5, I will present about my investigation on possible

Jahn-Teller distorted Cr(II) oxide in the triple perovskite Ba3Cr2TeO9 . In the Chapter 6, I am

including the results about the crystal structure solution of the pseudo-tetragonal tungsten bronze

structure of K6.4(Nb,Ta)36.3O94. My contribution to the complete investigations consisted of using

advanced transmission electron microscopy techniques to solve the structures at unit cell level as

well as the microstructures, and linking my results with complementary data of the collaborators.

X-ray and neutron powder diffraction, as well as magnetic property studies were performed by

collaborators from Oxford University (group of Prof. Peter Battle), Rutgers University (group

Prof. Martha Greenblatt) and Taras Shevchenko University of Kyiv (Artem Babaryk). Chapter 7

contains the conclusions and outlook.

2

1.1. General aspects of perovskites

The CaTiO3 compound (see Figure 1.1) was discovered in 1839 by Gustav Rose, who later

named it “perovskite” in honor to Count Lev Aleksevich von Perovski. The CaTiO3 perovskite

belongs to a cubic system with space group Pm-3m1, more recently studies about this perovskite

have revealed that the real structure is orthorhombic. Nowadays, the most important complex

oxides in solid state chemistry and materials science are undoubtedly the perovskite based

materials. The general chemical formula of a perovskite is ABX3, where we distinguish six-site

(BO6) (octahedral) and twelve-site (AO12) configurations in the structure. A cation positions are

basically occupied by alkali, alkali earth or rare earth metals while B cation positions are mainly

occupied by transition metals. The X anion positions are occupied by O2-

, Cl-, F

-, I

- anions.

Because of the huge possibilities of combinations of A, B cations and X anions, the materials

present different physical and chemical properties, for instance: relaxors ferroelectric and

relaxors ferromagnetic, multiferroic, catalytic, superconductivity, antiferromagnetic,

ferroelectricity, and so on. Consequently, they have been intensively used in many technological

applications such as: relaxor materials, piezoelectric materials, magnetic memories, dielectrics,

electrolyte materials, etc.

Of course the perovskite structures have to be consistent with the rule of charge valence

neutrality, and one has to take into account that most of the elements of the periodic table have

more than two oxidation states (entire or fractional values). What is more, the ideal stoichiometry

of a perovskite (ABX3) can be modified by changing one, two or three different types of atoms in

the A or B cation positions, hence because of this flexibility, perovskites compounds can adopt

more complex structures, such as: A2BB’O6, AAˈB2O6, AA’BB’O6, A3B2BO9, A2A’B2B’O9, etc..

In fact it is well known that some double perovskites show multiferroicity, which is a

combination of physical properties. In Chapters 3, and 4 we will present some compounds that

present spin-glass behaviour.

During the last decades, A3B2B’O9 triple perovskites have been intensely studied, because they

present new promising physical properties such as relaxor ferromagnetic2, multiferroicity

3, spin-

glass4, etc. Actually, A’2A’’B’2B’’O9 triple perovskites are getting even more interesting. In this

thesis different single, double, triple perovskites have been investigated in a search for new

perovskites with promising structural and magnetic properties.

Figure 1.1. Crystal structure of CaTiO3 perovskite, showing the octahedral coordination of the Ti.

3

1.2. Important parameters and properties of perovskites

Transition metals

Since we are studying perovskite materials that contain transition metals in the B cation positions,

it is crucial to understand some basic ideas about them. The transition metals belong to the d-

block. Each orbital can be occupied by maximum two electrons according to the Hund rule and

therefore the maximum number of electrons that can be found in a d-energy level is 10. The

elements in the periodic table with atomic number 21-30, 39-48, 72-80, 104-112 belong to 3dn,

4dn, 5d

n, 6d

n, respectively. Therefore it is easily recognized whether they are in a high spin state

(HS) or low spin state (LS) by looking at the number of unpaired electrons. Consequently the

physical properties are strongly related to those states. In Figure 1.2 are shown the different type

of orbitals and they have been ordered according to the energy level (n), moment angular (m),

and spin (s), respectively.

Figure 1.2. Representation of different type of orbitals according to different quantum numbers,

5.

Tolerance factor

In the ideal perovskite SrTiO3 which belongs to a cubic system with space group Pm-3m, the

bond distance of Ca-O is √2 times the bond distance of Ti-O. However, in general the

relationship between ionic radii of the A, B cations and X anion does not obey this relationship.

Goldsmith introduced the tolerance factor (t) in 1926.

𝑡 =1

√2

(𝑟𝐴+𝑟𝑂)

(𝑟𝐵+𝑟𝑂) … (1)

4

Where, 𝑟𝐴, 𝑟𝐵, 𝑟𝑂 are ionic radii of the A, B, and X cations and anion, respectively. From the

equation (1), it is clear that for ideal perovskites the tolerance factor is one; however, most of the

perovskites present distortions in the octahedra. As a consequence the perovskite compounds

have been classified in three different groups according to the tolerance factor. The first group

0.9<t<1.0, belongs to the cubic system; if 0.75<t<0.9, they adopt orthorombic or rhombohedral

structures, for 1<t<1.13 they are classified as hexagonal structures6. It should be noted that the

calculation of the tolerance factor only considers the ionic radii, which is an entirely geometrical

consideration. The tolerance factor gives us an indication about the space group of a certain

material, but there are other aspects that play an important role in the determination of the correct

space group, for instance: Jahn-Teller effects, lone pair effects, degree of covalence, and metal-

metal interactions7.

Octahedral tilting in perovskites

The chemical formula of the ideal perovskites is ABX3, if the A-cation decreases in size, then the

octahedron (BX6, considered rigid) has to rotate in order to minimize the energy of the system

and thus to get the most stable structure. Furthermore, by tilting the octahedron, we basically

modify three aspects: the A-X bond length, the A-site coordination, and of course the symmetry

is reduced from the arystotype to a less symmetric structure. Many authors have studied the

octahedral tilting in perovskites, for example: Glazer (1972), Megaw (1973), Woodward (1997),

Howard and Stokes (1998), but nowadays the tilting of the perovskites is mostly represented by

the Glazer notation, which is considered as the standard notation for tilting in perovskites,7.

Glazer notation

The BX6 octahedron in the aristotype perovskite can be rotated along three different axes a, b, c

which are the main axes of the cubic structure ABX3. The symbols abc in Glazer’s notations

denote different degree of rotation, and the positions of the letters refers to the rotation axis:

a#b

#c

# , the symbol # can be filled by +, 0, -, where the symbols +, 0, - stands for ‘in-phase’’,

‘no’’, ‘out-of-phase’’ rotationsof the octahedron . According to Glazer et al.8, there are 23 tilting

systems. Figure 1.3 shows three different tilting systems, where the octahedron has been tilted in

or out of plane. Starting from cubic perovskite, the symmetry can decrease for different

compositions by tilting the octahedron, or by ordering in the A or B cations, or by Jahn-Teller

distortions7,8

.

5

Figure 1.3. Glazer notation for three different tilt systems, left: a

0a

0a

0, middle: a

-a

-a

-, and right: a

-a

-a

+.

Jahn-Teller (JT) distortion

In perovskite structures with 3d transition metal ions in the B position, the octahedron is

surrounded by six oxygen ions. When this 3d state shows 5-fold degeneracy, then this degeneracy

can be split in two different energy states (eg and t2g) due to the intrinsic crystal field of the

octahedron. Now they show 2-fold and 3-fold degeneracy, respectively; the crystal field in the

octahedron is the energy that is necessary to excite an electron from the t2g state to the eg state.

The energy state eg can be split in two different energy states by a so called Jahn-Teller

distortion. Finally we obtain four different energy states not degenerated, which are basically

different d orbitals, eg (3z2-r

2, x

2-y

2), and t2g (zx, yz, xy), however zx and yz orbitals still belong

to the same energy state9, (see Figure 1.4).

In an ideal perovskite structure, the octahedron presents two axial bonds and four polar bonds

with equal length; when a Jahn-Teller distortion takes place, the axial bonds length become

longer (elongated) or shorter (compression) than polar bond lengths.

Figure 1.4. Scheme of the 3d band of the transition metal ion, where the Jahn-Teller effect and the crystal field are

shown.

1.3. Ordering in perovskites

The large selection of possible A, B cations in perovskites can result in charge ordered structures

in A and/or B cations, or eventually in a stoichiometric deficiency in cations or anions.

6

Rock salt ordering of the B cations occurs when the B and B’alternate along all three of the

directions a, b and c. The cell parameter is twice that of the parent perovskite. The other

possibilities are columnar ordering, when the octahedra or A cations are alternating along one

column, or layered ordering, when the octahedra or A cations are ordered layer by layer, (see

Figure 1.5)11

.

In the so-called “double perovskites”, for example, there is rock salt order between the B and B’

cations. The periodicity is therefore twice that of the ideal perovskite (Pm-3m). Therefore ideal

double perovskites belong to a cubic system with space group Fm-3m, where the atom positions

of the A, B cations are located in (1/2,1/2,1/2), and (0,0,0) positions respectively, and the oxygen

atoms are positioned in (x,0,0), where x≈1/4.10

In “triple perovskites” we also have inequivalent

B and B’ positions on a rock salt ordered lattice, but the composition is A3B2B’O9, with twice as

much B as B’ in the unit cell. Therefore the order over the rock salt positions will not be a

complete order of one element on B and the other on B’, but one or both positions will have a

mix of the two elements, however in different ratios, thus keeping the two positions inequivalent.

Figure 1.5. Crystal structure of: left: rock salt ordered double perovskite, middle: columnar ordering of A ordered

double perovskite, and right: layered ordering of A cation ordered double perovskite.

The ideal perovskite (CaTiO3, space group Pm-3m) presents the highest symmetry, but this

symmetry can be lowered by three different effects: ordering in A or B cations (Ba2MnWO610

,

space group Fm-3m), by tilting the octahedra (La3Ni2SbO92, space group P21/n), or by Jahn-Teller

distortions (La2CuSnO612

, space group P4/mmm). Woodward et al.10

, have intensely investigated

all those effects for ordered double perovskites from a theoretical point of view. The results

pertaining the cation ordering and the octahedral tilts are summarized in Table 1.1. We will only

present the table with rock salt order of the B cations, since this is the one relevant for our

compounds. From a theoretical point of view, the B cation ordering is represented by irreducible

representations 𝑅1+ of the basic Pm-3m, and the octahedral tilting by 𝑀3

+ (associated with in-

phase octahedral tilting) and 𝑅4+ out-of-phase tilting of the octahedra) respectively.

7

Table 1.1. Group-subgroup relationship, and the cation ordering parameters in double ordered perovskites, which

have been obtained by using ISOTROPY software. Table is reproduced from 10

.

Space

group

𝑴𝟑+ 𝑹𝟒

+ Tilts Lattice vector Origin Atomic positions(Wyckoff symbols, coordinates)

A B B' X

Fm-3m (0,0,0) (0,0,0) a0a0a0 (2,0,0),(0,2,0),(0,0,2) (0,0,0) 8c 1

4

1

4

1

4 4a 0,0,0 4b

1

2

1

2

1

2 24e x,0,0 x≈

1

4

P4/mnc (0,0,c) (0,0,0) a0a0c+ (1,1,0),(-1,1,0),(0,0,2) (0,0,0) 4d 01

2

1

4 2a, 0,0,0 2b 00

1

2 4e 0,0,z z≈

1

4

8h x, y, 0

x≈1

4 , y≈

3

4

P42/nnm (0,b,b) (0,0,0) a0b+b+ (0,2,0),(00,2),(2,0,0) (0,0,0) 2a 1

4

3

4

1

4

2b 3

4

1

4

1

4

4c 1

4

1

4

1

4

4f, 0,0,0 4e 001

2 8m x,-x,z

x≈0 z≈1

4

16n x,y,z

x≈0 y≈1

4 z≈0

Pn-3 (a,a,a) (0,0,0) a+a+a+ (2,0,0),(0,2,0),(2,0,0) (0,0,0) 2a 1

4

1

4

1

4

6d 1

4

3

4

3

4

4b, 0,0,0 4c 1

2

1

2

1

2 24h x,y,z

x≈1

4 y≈0 z≈0

Pnnn (a,b,c) (0,0,0) a+b+c+ (2,0,0),(0,2,0),(0,0,2) (0,0,0) 2a 1

4

1

4

1

4

2b 1

4

1

4

1

4

2c 1

4

1

4

1

4

2d 1

4

1

4

1

4

4f, 0,0,0 4e 1

2

1

2

1

2 8m x,y,z

x≈3

4 y≈0 z≈0

8m x,y,z

x≈0 y≈3

4 z≈0

8m x,y,z

x≈0,y≈0,z≈3

4

I4/m (0,0,0) (0,0,c) a0a0c- (1,1,0),(-1,1,0),(0,0,2) (0,0,0) 4d 01

2

1

4 2a, 0,0,0 2b 00

1

2 4e 0,0,z z≈

1

4

8x x,y,0

x≈1

4 , y≈

1

4

C2/m I2/m

(0,0,0) (0,b,b) a0b-b-

a0b-b- (-2,-1,0),(0,1,1),(0,1,-1) (0,1,1),(0,1,1),(2,0,0)

(0,0,0) (0,0,0)

4i x,0,z

x≈1

2 , z≈

1

4

2a, 0,0,0 2d, 0,0,1

2 4i, x,0,z

x≈0, z≈1

4

8j, x,y,z

x≈1

4 , y≈

1

4, z≈0

𝑅3 (0,0,0) (a,a,a) a- a- a- (-1,1,0),(0,-1,1),(2,2,2) (0,0,0) 6c, 0,0,z

z≈1

4

3a, 0,0,0 3b, 0,0,1

2 18f, x,y,z

x≈1

3 , y≈

1

6, z≈

5

16

𝑃1

𝐼1

(0,0,0) (a,b,c) a- b- c-

a- b- c- (1,0,1),(1,1,0),(-1,1,0)

(1,1,0),(-1,1,0),(0,0,2) (0,0,0)

(0,0,0) 4i x,y,z

x≈0 y≈1

2 ,z≈

1

4

2a, 0,0,0 2g, 0,0,1

2 4i, x,y,z

x≈0, y≈0, z≈1

4

4i, x,y,z

x≈1

4,y≈

1

4z≈0

4i, x,y,z

x≈1

4,y≈

3

4,z≈0

C2/c (0,b,0) (0,0,c) a0 b+ c- (-2,0,0),(0,0,2),(0,2,0) ( 1

2,0,

1

2 ) 4e 0,y,

1

4

y≈0

4e 0,y, 1

4

y≈1

2

4c, 1

4

1

40 4d,

1

4

1

4

1

2 8f, x,y,z

x≈1

4 y≈0, z≈0

8f, x,y,z

x≈0,y≈1

4, z≈0

8f, x,y,z

x≈1

4,y≈

1

4,z≈

1

4

P21/c P21/n

(a,0,0) (0,b,b) a+ b- b-

a- a- c+ (0,1,-1),(1,1,0),(2,1,-1) (1,1,0),(-1,1,0),(0,0,2)

(0,0,0) (0,0,0)

4e, x,y,z

x≈0 y≈1

2 ,z≈

1

4

2a, 0,0,0 2b, 0,0,1

2 4e, x,y,z

x≈0 y≈0,z≈1

4

4e, x,y,z

x≈1

4 ,y≈

1

4 ,z≈0

4e, x,y,z

x≈1

4 ,y≈

3

4 ,z≈0

P42/n (a,a,0) (0,0,c) a+ a+ c- (2,0,0),(0,2,0),(0,0,2) (1,0,0) 2a, 1

4,1

4,1

4

2b, 1

4,1

4,3

4

4e, 3

4,1

4, 𝑧

z≈1

4

4c, 0,0,0 4d, 0,0,1

2 8g, x,y,z

x≈1

4,y≈0,z≈

1

4

8g, x,y,z

x≈0 ,y≈1

4 ,z≈0

8g, x,y,z

x≈0,y≈0, z≈1

4

8

In order to understand the interpretation of some parameters of the group-subgroup relationship,

which are shown in the Table 1.1, I will illustrate it with two different examples:

1.- From cubic (Pm-3m) to monoclinic (P21/n)

Considering the lattice vector (1,1,0), (-1,1,0), (0,0,2) of the Table 1.1, the cell parameters of the

monoclinic system with space group P21/n are approximately a=b=√2ap, c= 2ap, where ap is the

cell parameter of the parent perovskite (ap=3.98Å). On the other hand, the lattice vector of an

ideal perovskite (Pm-3m) is (1,0,0), (0,1,0), (0,0,1), and taking the lattice vectors from the Table

1.1, it is possible to calculate directly the transformation matrix, from Pm-3m to P21/n, which will

be symbolized like PPm-3m→P21/n. Thus, for P21/n with lattice vectors (1,1,0), (-1,1,0), (0,0,2), the

transformation matrix is:

𝑃𝑃𝑚−3𝑚→𝑃21/𝑛=(1 −1 01 1 00 0 2

) … (2)

As can be read from the table, this new choice of cell parameters will be necessary because of the

presence of an in-phase tilt around the a-axis ((a,0,0) under M3+

), an antiphase tilt of equal size

around the b and c axes ((0,b,b) under R4+

) and rock salt order of the B cations (as is valid for all

entries in this table).

2.- From cubic (Pm-3m) to rhombohedral (R-3)

Doing a similar analysis, we can easily get the cell parameters and the transformation matrix

from Pm-3m to R-3: a=b=√2ap, c= 2√3ap, (ap=3.98Å), and

𝑃𝑃𝑚−3𝑚→𝑃21/𝑛=(1 −1 00 −1 12 2 2

) … (3)

As can be read from the table, this new choice of cell parameters will be necessary because of the

presence of an antiphase tilt of equal size around the a, b and c axes ((a,a,a) under R4+

) and rock

salt order of the B cations (as is valid for all entries in this table 1.1).

9

1.4. Bronze structures

In general ABX3 perovskite structures can be understood as derived from the ReO3, WO3

structures, or in general AxBX3 with x=0, where the A position is completely empty. The bronze

structure occurs when the A cation positons are partially occupied by alkali elements. Therefore

the general chemical formula for a bronze structure is AxBX3, where the A cation positions are

partially occupied (0<x<1), with A=K, Na, Ba, Pb, Tl or rare elements and B=Re, W, Mo.

Crystallographically there are three different types of bronze structures when looking at the type

of tunnels that contain A cations: cubic or lower symmetry bronze with tetragonal tunnels;

tetragonal tungsten bronze which has pentagonal, tetragonal and triangular tunnels, and

hexagonal tungsten bronze which contains hexagonal and triangular tunnels,7.

Figure 1.6. Crystal structure of the ReO3 compound.

Tetragonal tungsten bronze (TTB) structures

From Figure 1.6 we can clearly see that ReO6 is forming square tunnels, four octahedra form one

square tunnel. When these four linked octahedra are rotated 45° we obtain the tetragonal tungsten

bronze (TTB) structure. The TTB structure is created when the A cation is occupied by K, Ba,

etc. in the chemical formula AxWO3. Sometimes, TTB structures are also described as A-site

deficient or defect perovskites.

10

Figure 1.7. Crystal structure of the tetragonal tungsten bronze structure K0.37WO3, viewed along the [001] zone.

Pseudo-tetragonal tungsten bronze (TTB) structures

Tungsten W-atoms can be combined or interchanged by other atoms like Nb or Ta. For example,

the Ba0.15WO3 compound presents, pentagonal, diamond-shaped and triangular tunnels. If these

tunnels are occupied by large divalent and monovalent cations, one obtains a pseudo-tetragonal

tungsten bronze structure. In this thesis we will determine the crystal structure of the pseudo-

tetragonal tungsten bronze K6.4(Nb,Ta)36.3O94.

Figure 1.8 shows the [001] projection of the crystal structure of Nb18W16O93, which is considered

as a pseudo-tetragonal tungsten bronze. Pentagonal, tetragonal and triangular tunnels are clearly

seen in its framework,7.

Figure 1.8. Crystal structure of a pseudo-tetragonal tungsten bronze structure (Nb18W16O93) along [001]; the space

group is Pbam.

Bond valence sum method

In the solution of the crystal structure of K6.4(Nb,Ta)36.3O94 we used the concept “bond valence

sum method”. This method helps in the calculation of the oxidation or valence state of the atoms.

Suppose we have N atoms in one crystal structure, the valence (𝑉𝑖) of a given atom i should

balance the contribution of the rest of atoms that are surrounding the i atom. Mathematically we

can express it like:

11

𝑉𝑖 = ∑ 𝑣𝑖𝑘𝑁𝑘=1 … (4)

𝑉𝑖: Valence of the atom i,

In most of the literature the equation (5), is used. It is an empirical equation that indicates us the

variation of the length (Rij) of a bond with valence 𝑣𝑖𝑗.

𝑣𝑖𝑗 = 𝑒(𝑅0−𝑅𝑖𝑗)

𝑏 … (5)

𝑅0 and b are bond valence parameters; they have different values for each element of the periodic

table and they are estimated empirically. The application of this equation has been used in the

estimation of the oxidation states of Ta5+

. It was noticed that the bond valence parameter of Ta5+

was close to that of the Nb5+

. Further discussion will be presented in the chapter 5.

1.5. Magnetic properties of perovskites

1.5.1. General types of spin ordering

A magnetic field can be produced by accelerated charged particles, however at an atomic level a

magnetic field is generated in a different way; it is related to two different types of motion of the

electrons inside the atoms. In a very simple way: when an electron is orbiting around the nucleus,

it gets a magnetic moment, and when the electron is rotating around its axis it generates a spin

magnetic moment. In both cases a magnetic field is generated. When an external magnetic field H

is applied the material gets a magnetization M.

M=χH … (6)

where χm is the magnetic susceptibility.

Plotting the equation (6), gives us the magnetization as function of the applied magnetic field,

and the slope of the curve is the magnetic susceptibility. Therefore magnetism in materials may

be classified according to their magnetic susceptibility.

When the external magnetic field interacts with the orbital moment or magnetic moment, it

produces a very small negative magnetization upon the materials (-10-6

A/m to -10-5

A/m), and the

magnetization goes down to zero if the external magnetic field is removed. This effect is called

diamagnetism.

Paramagnetism also exhibits a small magnetization, but in contrast to diamagnetism the

magnetization is positive (+10-6

A/m to +10-3

A/m). When the external magnetic field goes down,

the magnetization decreases to zero.

12

Ferromagnetic materials on the other hand present a spontaneous magnetization, even in the

absence of an external magnetic field. Ferromagnetic materials dramatically change their

magnetization when they reach a critical temperature (Curie temperature. The high alignment or

ordering of the magnetic moment or spin magnetic moment that they present below of this

temperature immediately breaks down above Tc. Their behavior suddenly becomes

paramagnetic.

Antiferromagnetic materials present parallel as well as antiparallel magnetic moments or spin

magnetic moments and therefore the net magnetization is lower. At the Neel temperature

antiferromagnetism also vanishes and the behavior becomes paramagnetic. Figure 1.9

summarizes all these types of magnetism in materials.13

Specifically in perovskite materials, the transition metals in the octahedra can possess unpaired

electrons in the d orbitals and in this case a net spin appears. This can lead to different types of

antiferromagnetic order, as shown in Figure 1.10. Conventionally, the ab-planes are here

considered as the planes, which are stacked in layers along the c-direction. Thus in-plane order

would be order in the ab-plane, and interplane order would be order between the different layers.

Part (a) of this figure shows complete ferromagnetic order, which has a strong total

magnetization. A-type antiferromagnetism (Figure 1.10 (b)) occurs when there is parallel

orientation (ferromagnetic coupling) within the planes, but antiferromagnetic coupling between

consecutive planes along the c-axis. In C-type antiferromagnets (Figure 1.10 (c)) materials the

in-plane coupling is antiferromagnetic and the coupling between consecutive planes along the c-

axis is ferromagnetic. Finally in G-type antiferromagnets (Figure 1.10 (d)), there is

antiferromagnetic coupling both in-plane and between consecutive planes.

Spin-glasses. In a spin glass, the magnetic spins are disordered.

Frustrated magnetic systems. For example, in a system with three unpaired electrons on the

vertices of an equilateral triangle (Figure 1.11), it is not possible for all of them to be

antiferromagnetically ordered relative to each other. If the system tends to antiferromagnetism, it

will therefore be frustrated. This concept can be extended to domains with different orientation of

the magnetic moment (μ ).

13

Figure 1.9. Schematic representation of different types of magnetism in materials. Adapted from

14.

Figure 1.10. Different types of antiferromagnetic structures: A-AFM, C-AFM, G-AFM, also the ferromagnetic

structure has been included. The figure was adapted from15

.

14

Figure 1.11. Schematic representation of a frustrated magnetic system.

1.5.2. Relaxor Ferroelectrics and Relaxor Ferromagnets

In double perovskites A2BB’O6, or AA’B2O6, relaxor ferroelectric materials can been explained

taking into account the evolution of long range polar nano-regions within the sample due to

disorder in A or B cations. Smolenski et al. proposed that ferroelectric relaxors undergo a

dynamic behavior in small regions, which are called polar-nanoregions (PNR) with ferroelectric

behavior. These regions only exist below a certain temperature which is called Burn’s

temperature. Furthermore, there are small regions with a net electric polarization with random

orientation; they are surrounded by chemically ordered regions (COR) that are generally

paramagnetic. The presence of these small polarized regions is a consequence of disorder of the

A or B cations. However, in most of the literature the ferroelectric relaxor behavior is explained

in function of the real (ɛ') and imaginary (ɛ'') part of dielectric permittivity.

According to the relaxation model, ferroelectric relaxor behavior is due to the polar nanoregions

having different polarizations, it means that the relaxation time (τ) is different from PNR to

PNR: they respond in a different manner to an external electric field (�� ). This causes huge values

of the dielectric permittivity (ɛ', ɛ''). Relaxor ferroelectrics present a diffuse phase transition and

strong frequency dependence of the dielectric permittivity, which are not present in ordinary

ferroelectric materials.

Figure 1.12 shows two different plots T vs. ɛ', and T vs. ɛ'', where the maximum of the peaks of

the curves changes with temperature and frequency.

15

Figure 1.12. Real (ɛ') and imaginary (ɛ'') part of the dielectric permittivity as function of temperature (T) for Pb(1-

x)Lax(Zr(1-y)Tiy)(1-x/4)O3 (PLZO), adapted from16

.

The idea of relaxor ferroelectrics can be extended to relaxor ferromagnetic materials. For the first

time, the term “relaxor ferromagnetism” was used by Kimura et al.17

in order to explain the

behavior of Nd0.5Ca0.5MnO3 doped with approximately two percent of Cr3+

(impurity). These

small ferromagnetic impurities can be considered as new domains embedded within the

antiferromagnetic matrix of the undoped Nd0.5Ca0.5MnO3. Several perovskites of the type

A3B2B’O9 present a relaxor ferroelectric behavior. In 2013 Battle et al2 found a true relaxor

ferromagnetic material, La3Ni2SbO9, which is a triple perovskite. They assumed that due to B

cation disorder (NiO6, (Ni,Sb)O6 octahedra) small ferromagnetic domains appear, which even by

neutron powder diffraction could not be detected because the domains are too small,16

. However,

one year later Battle et al18

, using high angle annular dark field transmission electron microscopy

combined with neutron powder diffraction demonstrated the existence of Sb rich domains19

.

Figure 1.13. Plot of real (χ') and imaginary (χ'') magnetic susceptibility vs temperature (T), adapted from

17.

16

Chapter 2

Experimental techniques

I will briefly review the basic concepts of the different techniques that have been used in the

characterization of the materials I have investigated. I will first focus on the interaction between

the electron beam and matter.

2.1. General Diffraction

For the analysis of our materials we used different diffraction techniques such as X-ray powder,

neutron powder and electron diffraction. All those techniques are reciprocal space techniques,

and therefore it is important to introduce some basic knowledge on reciprocal space or Fourier

space.

A crystal is a periodic ordering of atoms on a three dimensional lattice, where the lattice can be

mathematically represented with three basis vectors 𝑎 , �� , and 𝑐 , and the angles (α, β, γ) between

those vectors.

Positions in reciprocal space, or Fourier space, can be described by a new triplet of basis vectors

𝑎 ∗, �� ∗, 𝑐 ∗. They can be easily obtained from the basis vectors of the real space.

First, the volume of a crystal in real space is calculated using the equation (7):

𝑉 = 𝑎 . (�� × 𝑐 ) … (7)

The new basis vectors in reciprocal spaces are defined as:

{

𝑎

∗ =(�� ×𝑐 )

𝑉

�� ∗ =(𝑐 ×�� )

𝑉

𝑐 ∗ =(�� ×�� )

𝑉

… (8)

Miller indices (hkl), are a set of numbers which quantify the intercepts between the planes and

main crystallographic axes, therefore, these numbers identify the orientation of the crystal planes

or surfaces. The reciprocal space also forms a three dimensional ordering of reflections.

Therefore, the reciprocal space is represented by new linear combination of basis reciprocal

vectors and we can express the reciprocal lattice as a linear combination between Miller indices

and basis vectors of the Fourier space20

.

𝑔 = ℎ𝑎 ∗ + 𝑘�� ∗ + 𝑙𝑐 ∗ … (9)

17

Figure 2.1. Representation of diffraction in real space and in reciprocal space,

19.

When the electrons interact with a crystal, the waves are diffracted by particular planes of the

crystal. In the Fourier space those diffracted waves generate reflections, i.e. in reciprocal space

we basically see an arrangement of dots, which are related to the Miller indices hkl. The

relationship between direct space and reciprocal space is described by the Bragg’s equation; it

relates the interplanar distance dhkl, the wavelength (λ), and the angle (θ ) between the reflected

waves and the lattice plane as:

2×dhkl× sin (𝜃)=nλ n=1, 2, … … (10)

In transmission electron microscopy texts, often the notation of the reciprocal vectors as �� is

used. �� and 𝐾 ˈ represent the direction of the incident and the diffracted beam, respectively. The

modulus of the vector �� is the inverse of the wavelength. Ewald devised a mathematical sphere

with radius 1

𝜆 , with center in the point where the incident and diffracted waves coincide. When a

reciprocal lattice point 𝑔 is on the sphere, diffraction will occur; this can be expressed by the

following relationship,20

:

𝐾 ˈ = �� + 𝑔 … (11)

Figure 2.2. Schematic representation of the Ewald sphere,20

.

2.2. Interaction electron-matter

As my own contribution in these studies was the transmission electron microscopy part, I will go

into more detail only on the interaction of matter with electrons. Depending on the accelerated

18

voltage, electrons can interact with the nucleus, with the inner-shell electrons or with the outer

shell electrons. During this interaction the electrons can produce several secondary signals such

as: secondary electrons, Auger electrons, characteristics X-rays, visible lights, backscattered

electrons, absorption, electron-hole pairs, and result in elastically and inelastically transmitted

electrons. They are shown schematically in Figure 2.3. These signals are at the origin of different

characterization techniques in transmission electron microscopy (TEM).

The principle of image formation in electron microscopy is similar to the one in optical

microscopy, however an optical microscope uses visible photons to form the image while an

electron microscope uses electrons. Since the electrons are charged particles, one has to use

electromagnetic lenses rather than glass lenses. The advantage of using accelerated electrons is

the improvement of the spatial resolution. The resolution of the optical microscope is basically

limited by the wavelength of visible light; by using accelerated electrons we can theoretically

reach picometer spatial resolution in an electron microscope. However mainly due to the

aberrations and imperfections of the magnetic lenses this resolution is reduced to about 0.1 nm. In

modern microscopes spherical aberration correctors can even lower the resolution to about 50

pm,20

.

2.3. Electron diffraction

Traditionally, X-ray and neutron diffraction techniques are commonly used to refine the crystal

structure of complex systems, however sometimes these techniques encounter problems in

determining a final solution, especially when densely distributed super reflections or ill-defined

reflections are present in the XRPD or neutron spectrum. In most cases selected area electron

diffraction (SAED) combined with real space imaging can provide a solution and allow one to

determine a final crystal structure.

Figure 2.3. Schematic representation of some of the different signals produced during the interaction electron-

matter.

An area of interest of a sample can be selected by inserting an aperture in the image plane. This

technique is then called selected area electron diffraction or SAED, and is the technique used in

19

this thesis. The aperture is placed below the sample, hence the diffraction pattern only contains

reflections from the area that has been selected.

Figure 2.4 shows how the incoming electrons are scattered over an angle 2θ and produce a

reflection in reciprocal space. From Figure 2.4, it is clear that tan(2𝜃) =𝑅

𝐿 . But since the Bragg

angle θ is very small, L is large and therefore tanθ≈sinθ≈θ and we finally get the expression:

𝑅𝑑 = 𝐿𝜆 … (12)

R: distance from the central beam to whatever spots.

d: interplanar spacing

Lλ: camera constant.

During the acquisition of the SAED patterns the camera length is fixed, therefore Rd must be

constant, as well. In practice the camera length is known, and R can directly be measured on the

pattern, and so the corresponding interplanar distance can be calculated for each of the spots. This

selection of interplanar distances can be compared to a database and lead to a correct indexation

and determination of the crystal structure if the structure is known. For new compounds with

unknown structures, the indexation must be puzzled out such that all reflections can be indexed

with integer numbers, and the solution to this puzzle gives the unit cell parameters.

Figure 2.4. Relation between direct and reciprocal space in an electron diffraction experiment

20.

Schemes of the most relevant perovskite supercells encountered in this thesis

As discussed higher, the symmetry of the perovskites can be changed by tilting of the octahedra,

displacement of the B cation, or by a Jahn-Teller distortion. In general the final structures adopt a

lower symmetry and as a consequence superstructure reflections appear in ED patterns. Basically,

ED patterns will show two types of reflections; the subcell reflections coming from the parent

20

perovskite (ideal perovskite), and the supercell reflections from ordering in A or B cations or the

tilting of the octahedra (in phase (-) or out of the phase (+)).

In the ideal perovskite, the patterns along the [100], [010], [001] zones are equivalent; they will

be represented by <100>. Furthermore, there are other patterns from main zones that are

equivalent, for instance <110>: [110], [-1,1,0], [1,0,1], …[-1,0,-1], and <111>: [111], [1,1,-1],

[1,-1,-1]. We have simulated the ED patterns from the most symmetric perovskite and different

tilted perovskites (lower symmetries). We will first discuss the changes in the electron diffraction

pattern from Pm-3m to Fm-3m. Pm-3m (a0a

0a

0) is the parent perovskite which does not present

superstructure reflections, while Fm-3m (a0a

0a

0) without any octahedral tilt will show

superstructure reflections in the ED patterns because of the ordering at the B cation site (Figure

2.5).

21

Figure 2.5. Simulated electron diffraction patterns for the ideal perovskite Pm-3m (left) and Fm-3m (right) along the

main zone axes.

Figure 2.6 shows the simulated ED patterns of a P21/n (a-a

-c

+) structure, which has been indexed

in ideal perovskite (Pm-3m) (a0a

0a

0). It is clear that there are many reflections that cannot be

indexed in Pm-3m, these super reflections basically appear because of the tilting of the octahedra

and/or ordering of the B cations.

Figure 2.6. Indexation in Pm-3m of the simulated ED patterns in P21/n (a

-a

-c

+). The black reflections come from the

parent perovskite while the red dots come from the tilting of the octahedra and the rock salt order of the B cation

positions.

22

Figure 2.7. Simulated SAED patterns indexed in the correct space group P21/n.

23

Figure 2.8. Simulated ED patterns for perovskite based structure with space group R-3cH.

2.4. High angle annular dark field scanning transmission electron microscopy and annular

bright field scanning transmission electron microscopy

Using conventional TEM, a parallel beam of electrons is interacting with the sample. In scanning

mode (STEM), a focused electron beam is scanned over the sample and in a similar way as in

SEM an image is recorded. Both techniques have their own advantages and disadvantages.

A very particular technique is the high angle annular dark field (HAADF) imaging, where only

inelastically scattered electrons are used to form the image. Atomic resolution can be obtained by

using a focused electron beam with a diameter of less than 100 pm. Because of the interaction of

the electron with the matter, the electrons are scattered over a large range of angles. When the

electrons only interact with the nucleus inelastic scattering results, also known as Rutherford

scattering. Normally this scattering is at high angles, and we can collect the information using a

ring detector with inner radius of about 50mrad and an outer radius of 200mrad. It has been

shown that in HAADF-STEM images the intensity of the atomic columns is approximately

proportional to the square of the average atomic number (I=Z2),

21. Because of this high Z

dependence, the technique is useful only for imaging heavy elements, and the technique is less

sensitive for light elements such as oxygen, particularly in the presence of heavier elements.

In contrast to HAADF-STEM, annular bright field scanning transmission electron microscopy

(ABF-STEM) is a type of phase contrast imaging technique. The intensities in the image are less

dependent of average atomic number, the relationship between I and Z is roughly I=Z1/3

, and as a

consequence we can detect lighter elements but the interpretation of the images is less

straightforward. In an ABF-STEM image, the detector collects elastic as well as thermal diffuse

scattering signals and therefore the image shows phase contrast. Moreover the contrast of the

different elements depends on the thickness of the sample.

HAADF-STEM and ABF-STEM are largely complementary techniques. Figure 2.9 shows an

example of a HAADF-STEM and an ABF-STEM image from the same compound.

24

Figure 2.9. Atomic resolution HAADF-STEM and ABF-STEM image of CeCo3V4O12 along [100]. The insets show

the calculated images. The model of the structure is also indicated on the bottom image, the smallest dots are oxygen,

while among the equally large dots blue is V, red is Ce, green is Co.

2.5. Energy dispersive X-ray scanning transmission electron microscopy spectroscopy

(EDX-STEM)

When the incoming accelerated electrons interact with the electrons of the material, some

electrons are excited from lower energy shells to higher energy shells; eventually also the

electrons of the outer shells can be ionized. As the excited electrons are in an unstable state, they

try to recover a more stable state and fall back into a less energetic shell. The direct consequence

is the generation of X-rays with an energy characteristic of the energy configuration of the

element present in the materials. Energy dispersive X-ray spectroscopy collects those

characteristic X-rays by using a detector. Our aberration corrected electron microscopes are

equipped with a new type of silicon drift detectors (SDD). In order to improve the solid angle of

collection (TitanG2: 0.7sr), four detectors are mounted symmetrically around the sample. The

main advantage of these new detectors is that by combining high resolution HAADF-STEM

imaging with energy dispersive x-ray spectroscopy, it is possible to acquire atomic resolution

25

STEM-EDX maps. Figure 2.10 shows an example of atomic resolution elemental EDX-STEM

mapping. The elemental maps directly show three different types of atom columns, one

containing only Ce, the second one only V, and the last one (Co,V),in the CeCo3V4O12 perovskite.

Figure 2.10. HAADH-STEM image, atomic resolution STEM-EDX elemental maps and the combined map of a

quadrupole perovskite CeCo3V4O12 along the [100] zone axis.

26

Chapter 3

Investigation of crystal structure and magnetic properties of the

perovskite based La3Ni2B’O9 (B’=Nb, Ta, Sb0.5Nb0.5) and

La2A’Ni2B’O9 (A’=Ca or Sr, B’=Te or W) in search for relaxor

ferromagnetic AA’2B2B’O9 compounds

3.1. Introduction

In Chapter 1, the perovskite materials and their flexibility to accept different atoms in the cation

or anion positions were discussed. The basic formulation ABO3, where A is usually a relatively-

large divalent or trivalent cation and B is a smaller cation from the d-block or p-block, must often

be changed to (A2–xA’x)BB'O6 or (A3–xA’x)B2B’O9 to show the presence of more than one type of

cation on either the A site or the B site, or both. When the A sites, 12-coordinate in the aristotype

cubic structure are occupied by more than one type of cation, the different cations, A and A’, are

usually distributed in a disordered manner. However, when the six-coordinate B sites are

occupied by multiple cation species the different cations, B and B’, often occupy the octahedral

sites in an ordered manner11, 22, 23,

. The degree of ordering is largely determined by the difference

in size and charge of the two cations and it in turn often determines the properties of the

compound.

For example, Sr2FeTaO6 has a disordered distribution of Fe3+

and Ta5+

over the B sites and

behaves as a spin-glass below 23 K24

, whereas Sr2FeIrO6 has nearly complete 1:1 checkerboard

ordering of Fe3+

and Ir5+

(see Figure 3.1 (a)) and orders antiferromagnetically at 120 K25

.

Figure 3.1. (a) 1:1 checkerboard cation ordering in the A2BB'O6 perovskite structure. Orange and grey octahedral

are occupied by B and Bˈ cations, respectively. The A cations are represented by green balls, and (b) a 4 × 4 grid

illustrating the cation ordering over the B sites of La3Ni2SbO9; one set of sites is occupied by Ni2+

only, whereas the

other set has a 2:1 concentration ratio of Sb5+

and Ni2+

(represented by the symbol Ni/Sb).

27

This is not the only type of cation ordering observed, but it is the most common. Perhaps

surprisingly, it is even found in compounds where it is apparently incompatible with the

concentration ratio of the two species, for example La3Ni2SbO92. In this case one set of sites in

the checkerboard is occupied entirely by Ni2+

while the other is occupied by a disordered 1:2

distribution of Ni2+

and Sb5+

. Figure 3.1 (b) shows the arrangement of the B-site cations in a

(001) sheet. The presence of antiferromagnetic ordering below 275 K in KNiF326

suggests that

there will be strong antiferromagnetic coupling between nearest-neighbour Ni2+

cations in this

non-frustrated array and, because of the 3:1 imbalance between the number of these cations on

the spin-up and spin-down sublattices, this might be expected to result in ferrimagnetism.

Consistent with this, the magnetisation of La3Ni2SbO9 increases markedly on cooling below 105

K and measurements of the magnetisation at 5 K found a value in excess of 1.5µB per formula

unit.

However, neutron diffraction experiments detected little or no magnetic Bragg scattering at 5 K2.

In order to account for this apparent contradiction it was proposed that ferrimagnetic domains

exist, but that they are too small to give rise to Bragg scattering. Support for this hypothesis was

provided by neutron diffraction experiments carried out in an applied field. The field apparently

brought different domains into alignment, thus increasing the coherence length of the magnetic

order and enhanced Bragg scattering was observed18

. The presence of small magnetic domains

was attributed to variations in the Ni/Sb concentration which might be expected to increase the

relative significance of next-nearest-neighbour super-exchange interactions and hence disrupt the

long-range magnetic structure. Evidence for such variations in the Ni/Sb distribution and the

consequent disruption of the cation ordering was provided by high resolution transmission

electron microscopy18

. Regions were observed without order, and with a higher amount of Sb.

Figure 3.2. Energy dispersive X-ray analyses also showed a strong fluctuation in the Ni:Sb ratio

on a nm scale.

Figure 3.2. HAADF-STEM image of La3Ni2SbO9, showing at the left of the picture an ordered region and in the

middle a disordered region with higher Sb content.18

28

Overall, the magnetic behaviour of La3Ni2SbO9, i.e. the enhanced magnetic susceptibility

associated with local ferrimagnetism but no long-range order detectable in a diffraction

experiment, was deemed to be analogous to the electrical behaviour of relaxor ferroelectrics27

, for

example Pb3MgNb2O928

, and La3Ni2SbO9 was consequently described as a relaxor ferromagnet

and was the first true relaxor ferromagnet.

This incited the search in this thesis for other relaxor ferromagnets and for the characteristic

structural features driving the relaxor ferromagnetism. Based on the general formula

(A3–xA’x)B2B’O9 for the family of compounds to which La3Ni2SbO9 belongs, other systematically

chosen combinations of B and B’ were tested.

The choice of specific systematic cation replacements is based on the comparison between

Sr2NiWO630

and SrLaNiSbO631

. These compounds have the same structure, but different

magnetic transition temperatures and superexchange interaction routes. Possible origins for the

differences are the different occupation of the A site and the valence and electronic configuration

of the B’ cation29,34

, since Sb5+

is d10

while W6+

is a non-magnetic d0.

First, insight was sought into the effect of replacing the d10

diamagnetic cation with a cation with

an empty d shell (Ta5+

, Nb5+

, d0), with the compounds La3Ni2NbO9 and La3Ni2TaO9. Also the

combination of d10

and d0 was investigated, with the compound La3Ni2Nb0.5Sb0.5O9. Then, we

investigated also the effect of replacing a B5+

with a B6+

cation, through the compounds

SrLa2Ni2TeO9 and CaLa2Ni2WO9. With W6+

and Te6+

we again also have the possibility to

compare a d0 with a d

10 cation. The different valency of the B cation necessitates also a partial

substitution of the A cation to keep the charge balance, thus we replaced one third of the La3+

by

either Sr2+

or Ca2+

. One must keep in mind, however, that also this change in A cations can have

a significant impact on the physical properties of the compounds.32,33

Of these, only La3Ni2TaO9 and La3Ni2NbO9 have been made before. La3Ni2TaO9 was reported by

Kato et al.35

to have an orthorhombic perovskite structure, but no further structural details were

given. La3Ni2NbO9 was refined from neutron diffraction data in a P21/n structure with the 2c site

occupied for ~90 % by Ni2+

. No magnetic properties were studied before for either of these

compounds.

In this thesis, we studied SrLa2Ni2TeO9 in most detail. For all other compounds, very similar

results were found for the structure from the TEM analysis, therefore the results for the

compounds will be presented in a summarized manner, easy for comparison, after the detailed

description of SrLaNi2TeO9.

The synthesis, magnetic measurements and X-ray diffraction experiments were performed by the

group of Prof. Dr. P. Battle at the University of Oxford.

29

3.2. Synthesis

The synthesis of the samples was done by the group of Prof. Dr. P. Battle at Oxford University

using the traditional ceramic method. This involves mixing stoichiometric quantities of the

precursors, grounding them together and firing them in an alumina crucible. Afterwards, the

reaction mixtures were pelletised before being heated again with intermediate regrinding. After

several days, the reaction product were cooled down to a certain temperature in the furnace and

then quenched to room temperature. The details per sample can be found in Table 3.1.

Table 3.1. Temperatures and times at different stages during the synthesis of the samples.

1st stage 2nd stage 3rd stage 4th stage

T1(°C) t1(h) T2(°C) t2(h) T3(°C) t3(h) T4(°C) t4(h)

SrLa2Ni2TeO9 800 48 1200-1300 216

La3Ni2TaO9 1250 192 1250 36

La3Ni2NbO9 1100 48 1200 168 1250 120 1300 40

La3Ni2Nb0.5Sb0.5O9 1200 132 1250 72

CaLa2Ni2WO9 700 48 1000-1150 264

30

3.3. SrLa2Ni2TeO9

The content of this section was published as the peer reviewed paper. Robert Paria Sena, Joke

Hadermann, Chun-Mann Chin, Emily C. Hunter, Peter D. Battle, Journal of Solid State

Chemistry, 2016, 243,304-311.

The X-ray and neutron powder diffraction and magnetometry studies of SrLa2Ni2TeO9 were

performed by the group of Prof. Dr. P. Battle at Oxford University.

3.3.1. Experimental Results

The electron diffraction patterns, of which the main zones are shown in Figure 3.3, can be

indexed using cell parameters a ~ b ~ √2ap, c ~ 2ap, β ~ 90 , where ap is the cell parameter of the

parent perovskite.

Figure 3.3. Representative SAED patterns from main zone axes of SrLa2Ni2TeO9.

The reflection conditions derived from the full set of electron diffraction patterns are hkl, 0kl,

hk0: no conditions, h0l: h+l=2n, h00: h=2n, 00l: l=2n. The only space group in agreement with

the reflection conditions is P21/n.

The 0k0:k=2n reflection condition is proven with the tilt shown in the Figure 3.4 (a) and (b).

When tilting around the 0k0 row of reflections, the 0k0: k=2n+1 reflections disappear when the

double diffraction paths are destroyed. To retain visible proof that the zone is effectively [100]

and not [010], the tilted pattern in Figure 3.4 (b) still shows the 0kl: k = 2n, l = 2n+1, even

though this also retains some double diffraction. The 0k0: k = 2n+1 are nevertheless barely

visible on Figure 3.4 (b), with 010 and 030 effectively gone.

The 00l: l ≠ 2n reflections are present in [100], but are due to double diffraction, as can be seen

by their absence on both the [010] pattern and on the [100] Fourier transforms in Figure 3.6.

31

Figure 3.4. SAED patterns of the SrLa2Ni2TeO9 compound along the [100] zone, (a) in zone, and (b) tilted over the

b* axis.

One could argue that the 0kl: k=2n+1, l=2n could also be due to double diffraction, with

reflection conditions 0kl: k=2n, h0l:h+l=2n and overlap of these patterns, however, we have

proven with a similar tilting experiment that this is not the case. Starting from the same zone

[100] in Figure 3.5 (a), where we have indicated one such arguable reflection, we have tilted

around the row of reflections 0-12, 0-24, ... and when all double diffraction paths are destroyed as

in the Figure 3.5 (b), the reflections 012 and 0-1-2 are still clearly present and of the same

intensity or even higher intensity than the nearest surrounding reflections. Therefore, the

reflections 0kl: k=2n+1, l=2n are not due to double diffraction.

Figure 3.5. SAED patterns of SrLa2Ni2TeO9 compound along [100] zone, (a) left: in zone, and (b) right: out of zone.

The crystal was tilted along 012 row of reflections.

32

According to Howard et al.41

, the space group P21/n implies cation order in the B positions. To

verify this, we performed HAADF-STEM imaging. Figure 3.6 shows a typical crystal along the

[100] zone, twinned with a [010] zone. The difference between the two orientations is almost

imperceptible on the images, but is clear on the Fourier transforms (the Fourier transform for

[010] is taken of a smaller area that that of [100] because there is no larger domain available; this

should not influence the positions and presence of reflections). In the images, the A (La,Sr)

columns form a straight line in [010] but a barely-noticeable zig-zag in [100]. In the Fourier

transforms, extra reflections are present in [100] compared to [010] (compare to Figure 3.6).

Twinning occurs very frequently throughout the sample. This zig-zag is the deviation of the A/A’

cation from the ideal position along the y-axis, as in agreement with the P21/n model from the

same paper by Howard et al.

Note that the reflections 0kl: k=2n+1, k+l=2n+1 are present on the electron diffraction patterns

of [100], consistent with space group P21/n, while they are absent on the Fourier transform of

[100]. This is probably a consequence of the fact that the brightness of the dots on HAADF-

STEM images increases with the average Z of the projected columns. Consequently, on the

HAADF-STEM image the dots of the purely oxygen columns are invisible. They therefore also

do not contribute to the Fourier transform of the HAADF-STEM image. We have calculated the

theoretical diffraction patterns using the model in Table 3.2, leaving out the oxygen positions,

and these reproduce exactly the patterns seen in the Fourier transforms. The calculated patterns

are shown in the Figure 3.7 (a) and (b).

33

Figure 3.6. Aberration-corrected high-resolution HAADF-STEM image of a SrLa2Ni2TeO9 crystal. The crystal is

twinned and shows the [010] zone at the top, but the [100] zone at the bottom, as can be seen from the Fourier

transforms. Close-ups are added for a clearer view. Arrows on the enlargements at the bottom show the slight shifts

of the rows of A cations for [100] and the absence of such shifts in [010], a horizontal line is added as a guide to the

eye.

Figure 3.7. Calculated electron diffraction patterns (using JEMS software) for crystals with thickness of 50 nm and

for 200 kV, using a 2-beam calculation, for (a) the model proposed in this paper and (b) the same model after

removal of all oxygen atoms.

34

Besides twinning, another type of inhomogeneity within the crystals was found using energy

dispersive X-ray (EDX) analysis. First, STEM-EDX was applied at atomic resolution to confirm

the Ni-Te order in the sample. This is shown in Figure 3.8, where the ordered distribution of Ni

and Te over the two B cation positions is clearly visible. Next, an overview STEM-EDX analysis

was done of several crystals to study the homogeneity of the composition throughout the crystal.

Figure 3.8. Aberration-corrected high-resolution HAADF-STEM and STEM-EDX map of SrLa2Ni2TeO9 compound,

along [010] zone axis.

In Figure 3.9, the results are shown for a representative crystal with a total size of 62 nm by 48

nm. The composition is measured over 60 square areas of 4.5 nm×4.5 nm, in low magnification.

The measurement was performed with the crystal tilted out of zone to avoid channeling effects.

Figure 3.9 shows the measured ratios between the cations versus the serial number of the

measured area. On this graph, the ratio (La+Sr):(Ni+Te) stays close to 1, as required for a

perovskite structure. The ratios for the Ni:Te and La:Sr cations, however, deviate significantly

from area to area, deviating from the stoichiometric 2:1 ratios for both. When averaging over the

whole set of measurements, the composition does agree with the stoichiometric one and is

La2.07(9)Sr0.92(9)Ni2.03(13)Te0.97(13)Ox (oxygen values cannot be reliably measured from STEM-EDX

quantification on powder samples).

35

Figure 3.9. STEM-EDX measurements on sixty 4.5nm×4.5nm areas within one crystal.

The possibility was taken into account that the P21/n space group was still an average space

group, and that locally areas with different symmetry would be present, especially considering

the strong deviations in the local composition. After all, electron diffraction patterns are taken

from areas as large as the smallest available selected area aperture on the TEM instrument used,

i.e. a few hundred nanometers usually. Domains of smaller size with different symmetry could be

present. To test this possibility we carried out HRTEM of several <100>p crystals. We preferred

HRTEM above HAADF-STEM for this specific goal for two reasons: first, the contribution of

the oxygen to the images, and thus to the Fourier transforms of the images, is larger for HRTEM

than HAADF-STEM , and secondly, in practice, Fourier transforms of really small areas (<5 nm)

are much more clear for HRTEM than for HAADF-STEM images. We chose zone <100> to

prove the absence of different symmetries, because according to the paper by Howard et al.10

,

group theory leaves only three possible monoclinic space groups in case of cation order

(disregarding other uncommon distortions lowering the symmetry even more instead of

increasing it), i.e. I2/m (space group number 12, conventional setting C2/m, a~b=√2ap, c=2ap),

C2/c (space group number 15, a~b~c~2ap) and P21/n (space group 14, conventional setting P21/c,

a~b~2ap, c=2ap). There are no domains without cation order in the crystallites, this we can

conclude from our HAADF-STEM images. For these three space groups a difference is present in

the <100>p zones. For I2/m these show only the parent cell reflections, for C2/c one of the three

<100>p zones also shows only this pattern of parent cell reflections. The other two <100>p zones

of C2/c show reflections at either hp/2 kp/2 0 with hp,kp odd or hp kp/2 0 with kp odd. For P21/n

one <100>p shows reflections at hp/2 kp/2 0 with hp,kp odd, two show reflections at hp kp/20 with

kp odd. Since we did not see on any Fourier transform of any domain of the <100>p HRTEM

images a pattern with only the subcell reflections, but all showed patterns corresponding to P21/n,

it is unlikely that domains of different symmetry than P21/n are present. The HRTEM images are

not included in the thesis as this entails many images, and many Fourier transforms.

During the reviewing process of the paper on this compound, we were asked to verify that the ED

patterns and TEM images were not mistakenly from Sr2NiTeO636

. For this, we examined 50

random crystallites by EDX analysis and found no La-free crystallites. Furthermore, the space

36

group of Sr2NiTeO6 is I2/m, and as mentioned before, none of the Fourier transforms

corresponded to the expected pattern for I2/m.

To summarize the TEM results, we are confident the space group is P21/n, with the implied

cation order in the B position, no order in the A position and the accompanying octahedral tilt.

The ratio of A:A’ and of B:B’ varies significantly from area to area.

The analysis of the X-ray diffraction showed that beside the monoclinic, perovskite-like phase

also 1.19(4) wt % of Sr3TeO6 and 0.25(3) wt % of NiO are present. The temperature dependence

of the molar magnetic susceptibility of SrLa2Ni2TeO9 is shown in Figure 3.10 (a), where the

effective magnetic moment per Ni2+

and Weiss constant are 2.21(1) µB, +121(2) K, respectively

and they were fitted from Curie-Weiss law. The data collected under ZFC and FC conditions

differ below 35 K, at which temperature the former show a maximum in χ(T); the data collected

under FC conditions show essentially no temperature dependence below 35 K. The field

dependence of the magnetisation per formula unit is shown in Figure 3.10 (b). M(H) is linear at

150 K and nonlinear at 50 K, but no hysteresis is seen at either temperature. However, hysteresis

is observed at 5 K. The remanent magnetisation is ~ 0.06 µB per f.u. and the coercive field is 2.4

kOe. The ac magnetic susceptibility, see Figure 3.10 (c), is a function of frequency and has both

real and imaginary components below ~ 35 K. The transition temperature is frequency-dependent

with ΔTf/[TfΔ(log ω)] = 0.014, a typical value for a canonical spin-glass.

Figure 3.10. (a) The molar dc magnetic susceptibility and (inset) inverse magnetic susceptibility of SrLa2Ni2TeO9 as

a function of temperature, (b) magnetic field dependence of the magnetisation per formula unit of SrLa2Ni2TeO9 at 5

(blue), 50 (green) and 150 (red) K, (c) temperature and frequency dependence of the real and imaginary components

of the ac magnetic susceptibility of SrLa2Ni2TeO9 collected at 1 (red), 10 (green) and 100 (blue) Hz.

Rietveld analysis of the neutron diffraction data collected at room temperature using D2b

confirmed that SrLa2Ni2TeO9 crystallises in the monoclinic space group P21/n; ~ 2.3(3) wt % of

Sr3TeO637

, and ~ 0.43(3) wt. % of unreacted NiO38

were detected in this analysis. As in the

analysis of the X-ray diffraction data, the nickel and tellurium cations were found to be partially

ordered over the octahedral sites. The Ni/Te distribution was refined at room temperature and

then held constant during the analysis of the data collected at 50 and 5 K. The displacement

parameters of the two six-coordinate sites were constrained to be equal during these analyses and

the Ni:Te ratio was constrained to be 2:1. The oxygen sublattice was assumed to be fully

occupied. The same basic structural model was able to account for all the diffraction patterns. No

37

magnetic Bragg scattering was observed in any of the datasets collected on D2b. The patterns

recorded at room temperature and 5 K are presented, together with the calculated patterns, in

Figure 3.11 (a) and (b).

Figure 3.11. Observed (red crosses) and calculated (green line) neutron powder diffraction profiles recorded on D2b

of SrLa2Ni2TeO9 at (a) room temperature and (b) 5 K. Underneath is a difference curve, purple in colour. Reflection

markers are shown, from top to bottom, for Sr3TeO6, NiO and SrLa2Ni2TeO9.

The refined structural parameters are presented in Tables 3.2 and 3.3 and some selected bond

lengths and angles is listed in the Table 3.4. The parameters and fits resulting from the analysis of

the data collected at 50 K are listed in Tables 3.5, and 3.6 and Figures 3.12 (a) and (b),

respectively. The neutron diffraction patterns collected at 5 K and 50 K on D1b were

superimposable; no additional Bragg scattering was seen at the lower temperature. However,

inspection of the data at low angles revealed the presence of a 100 reflection, forbidden in space

group P21/n, and additional intensity in the 101 and 102 reflections of the perovskite phase; the

strongest magnetic reflection of nickel oxide was also visible38,39

These reflections, which were

apparently too weak to be identified in the data collected on D2b, were not seen in the X-ray

diffraction pattern at room temperature. They were therefore assumed to be magnetic in origin.

None of the collinear antiferromagnetic ordering patterns associated with the perovskite structure

gives rise to this combination of reflections.40

38

Table 3.2. Structural parameters of SrLa2Ni2TeO9 at room temperature.

Atom Site x y z Uiso/Å2 Occupancy

Sr 4e 0.5050(8) 0.5210(5) 0.2492(20) 0.0275(4) 0.3333

La 4e 0.5050(8) 0.5210(5) 0.2492(20) 0.0275(4) 0.6667

Ni1 2c 0.00 0.50 0.00 0.0026(19) 0.83(3)

Te1 2c 0.00 0.50 0.00 0.0026(19) 0.17(3)

Ni2 2d 0.50 0.00 0.00 0.0026(19) 0.50(3)

Te2 2d 0.50 0.00 0.00 0.0026(19) 0.50(3)

O1 4e 0.2322(11) 0.2238(18) -0.0390(8) 0.0157(20) 1.0

O2 4e 0.2885(11) 0.7212(15) -0.0309(8) 0.0086(17) 1.0

O3 4e 0.4345(6) 0.9915(7) 0.2480(18) 0.0144(7) 1.0

Rwp = 4.87 %, Rp = 3.68 %, χ2 = 4.175. Space group P21/n: a = 5.6008(1) Å, b = 5.5872(1) Å, c = 7.9018(2) Å,

β = 90.021(6) º.

Table 3.3. Structural parameters of SrLa2Ni2TeO9 at 5 K.

Atom Site x y z Uiso/Å2 Occupancy

Sr 4e 0.5061(8) 0.5237(5) 0.2503(24) 0.0230(4) 0.3333

La 4e 0.5061(8) 0.5237(5) 0.2503(24) 0.0230(4) 0.6667

Ni1 2c 0.00 0.50 0.00 0.0004(2) 0.83(3)

Te1 2c 0.00 0.50 0.00 0.0004(2) 0.17(3)

Ni2 2d 0.50 0.00 0.00 0.0004(2) 0.50(3)

Te2 2d 0.50 0.00 0.00 0.0004(2) 0.50(3)

O1 4e 0.2312(12) 0.2200(20) -0.0345(12) 0.0115(20) 1.0

O2 4e 0.2898(11) 0.7214(18) -0.0366(12) 0.0052(17) 1.0

O1 4e 0.4325(7) 0.9897(6) 0.2463(16) 0.0095(7) 1.0

Rwp = 5.22 %, Rp = 4.00 %, χ2 = 3.709

Space group P21/n: a = 5.5901(1) Å, b = 5.5803(1) Å, c = 7.8886(2) Å, β = 90.026(5) °.

39

Table 3.4. Selected bond lengths (Å) and bond angles in SrLa2Ni2TeO9 at room temperature and 5 K.

Bond lengths Bond angles

Room

temperature 5 K

Room

temperature 5 K

Sr/La – O1 2.880(13) 2.826(17) O1 – Ni/Te1 – O2 86.3(5) × 2 86.7(5) × 2

Sr/La – O1 2.638(11) 2.665(15) O1 – Ni/Te1 – O3 89.1(2) × 2 89.6(3) × 2

Sr/La – O1 2.508(11) 2.513(14) O2 – Ni/Te1 – O3 89.4(2) ×2 89.5(3) × 2

Sr/La – O2 2.760(13) 2.793(17)

Sr/La – O2 2.479(11) 2.453(15) O1 – Ni/Te2 – O2 88.6(5) × 2 89.0(6) × 2

Sr/La – O2 2.759(11) 2.713(14) O1 – Ni/Te2 – O3 88.4(2) × 2 89.5(3) × 2

Sr/La – O3 2.984(4) 3.009(4) O2 – Ni/Te2 – O3 89.6(2) × 2 89.6(3) × 2

Sr/La – O3 2.659(4) 2.633(4)

Sr/La – O3 2.467(6) 2.460(6) Ni/Te1 – O1 – Ni/Te2 159.7(4) 160.8(5)

Ni/Te1 – O2 – Ni/Te2 159.3(4) 157.4(5)

Ni/Te1 – O1 2.041(8) × 2 2.046(8) × 2 Ni/Te1 – O3 – Ni/Te2 158.8(2) 158.1(2)

Ni/Te1 – O2 2.049(6) × 2 2.058(7) × 2

Ni/Te1 – O3 2.025(14) × 2 2.037(12) × 2

Ni/Te2 – O1 1.977(7) × 2 1.959(8) × 2

Ni/Te2 – O2 1.972(7) × 2 1.970(7) × 2

Ni/Te2 – O3 1.994(14) × 2 1.980(12) × 2

40

Table 3.5. Structural parameters of SrLa2Ni2TeO9 at 50 K.

Atom Site x y z Uiso/Å2 Occupancy

Sr 4e 0.5053(9) 0.5241(5) 0.2497(23) 0.0230(4) 0.3333

La 4e 0.5053(9) 0.5241(5) 0.2497(23) 0.0230(4) 0.6667

Ni1 2c 0.00 0.50 0.00 0.0005(2) 0.83(3)

Te1 2c 0.00 0.50 0.00 0.0005(2) 0.17(3)

Ni2 2d 0.50 0.00 0.00 0.0005(2) 0.50(3)

Te2 2d 0.50 0.00 0.00 0.0005(2) 0.50(3)

O1 4e 0.2309(12) 0.2213(20) -0.0353(13) 0.0142(21) 1.0

O2 4e 0.2902(11) 0.7209(17) -0.0352(11) 0.0029(16) 1.0

O3 4e 0.4322(7) 0.9899(7) 0.2459(15) 0.0101(7) 1.0

Rwp = 5.33 %, Rp = 3.96 %, χ2 = 3.859

Space group P21/n: a = 5.5902(1) Å, b = 5.5804(1) Å, c = 7.8887(2) Å, β = 90.026(5) °

Table 3.6. Selected bond lengths (Å) and angles (°) in SrLa2Ni2TeO9 at 50 K.

Bond lengths Bond angles

Sr/La – O1 2.835(15) O1 – Ni/Te1 – O2 86.5(5) ×2

Sr/La – O1 2.656(13) O1 – Ni/Te1 – O3 89.8(3) ×2

Sr/La – O1 2.518(14) O2 – Ni/Te1 – O3 89.8(3) ×2

Sr/La – O2 2.775(16)

Sr/La – O2 2.458(14) O1 – Ni/Te2 – O2 88.8(5) ×2

Sr/La – O2 2.727(13) O1 – Ni/Te2 – O3 89.4(3) ×2

Sr/La – O3 3.009(4) O2 – Ni/Te2 – O3 89.9(3) ×2

Sr/La – O3 2.631(4)

Sr/La – O3 2.454(6) Ni/Te1 – O1 – Ni/Te2 160.6(6)

Ni/Te1 – O2 – Ni/Te2 157.7(5)

Ni/Te1 – O1 2.040(9) ×2 Ni/Te1 – O3 – Ni/Te2 158.0(2)

Ni/Te1 – O2 2.056(7) ×2

Ni/Te1 – O3 2.041(11) ×2

Ni/Te2 – O1 1.966(9) ×2

Ni/Te2 – O2 1.969(7) ×2

Ni/Te2 – O3 1.977(11) ×2

41

Figure 3.12. (a) Observed (red crosses) and calculated (green line) neutron powder diffraction profiles of

SrLa2Ni2TeO9 at 50 K recorded on D2b. Underneath is a difference curve, purple in colour. The black, blue and

green vertical makers represent the Bragg reflection markers for SrLa2Ni2TeO9, NiO and Sr3TeO6. (b) Observed (red

crosses), calculated (green line) and difference (purple line) neutron powder diffraction profiles of SrLa2Ni2TeO9 at

50 K recorded on D1b across the full measured angular range. Reflection markers for structural SrLa2Ni2TeO9,

structural NiO, magnetic NiO, G-type and C-type magnetic structures of SrLa2Ni2TeO9 and Sr3TeO6 are black, red,

blue, green and orange and pink, respectively.

In order to fit the data quantitatively it was therefore assumed that two magnetic phases were

present, one showing G-type ordering and the other C-type, see Figure 3.13 (a) and (b).40

In the

former, Ni2+

cations couple antiferromagnetically to their six nearest-neighbours (NN) whereas in

the latter they couple antiferromagnetically to four NN in the (001) sheets but ferromagnetically

to the two NN along [001]. The coupling to the twelve next-nearest neighbours (NNN) is entirely

ferromagnetic in the G-type structure whereas in the C-type structure there are 4 and 8

ferromagnetic and antiferromagnetic links, respectively. As a consequence of correlations

between the parameters, it was not possible to refine the ordered magnetic moment and the phase

fractions of the magnetic phases simultaneously. We therefore assumed that in these two

magnetic phases all the 2c and 2d sites are occupied by Ni2+

, and that the magnetic moment is

aligned along [100] with a magnitude of 2 µB. Refinements, see Figure 3.14 (a) and (b), then

showed that at 5 K 17(2) and 19(2) wt. % of the sample ordered as G-type and C-type

antiferromagnets, respectively. The limitations of these assumptions are discussed below.

Figure 3.13. (a) G-type and (b) C-type magnetic structures drawn in the unit cell of SrLa2Ni2TeO9.

42

Figure 3.14. (a) Observed (red crosses), calculated (green line) and difference (purple line) neutron powder

diffraction profiles of SrLa2Ni2TeO9 at 5 K recorded on D1b across (a) the full measured angular range and (b) the

low-angle region. Reflection markers are shown from top to bottom for Sr3TeO6, C-type magnetic SrLa2Ni2TeO9, G-

type magnetic SrLa2Ni2TeO9, magnetic NiO, structural NiO and structural SrLa2Ni2TeO9.

3.3.2. Discussion

Our neutron diffraction data demonstrate that there are a number of potentially-important

structural differences between SrLa2Ni2TeO9 and La3Ni2SbO9. The disordered cation distribution

over the A sites of the former will result in local static disorder. This is likely to be responsible

for the high values of the displacement parameters of the A-site cations and the oxide ions at

both room temperature and 5 K, see Tables 3.2 and 3.3. The displacement parameters of the

anions are also enhanced by the occupational disorder on the B sites. Whereas in La3Ni2SbO9

one of the B sites was found, within experimental error, to be totally occupied by Ni2+

there is a

significant deviation from this ideal ordering pattern in SrLa2Ni2TeO9. We shall return to this

point below in our discussion of the magnetic properties of the tellurate. The reduced degree of

ordering is somewhat surprising given that Ni2+

and Te6+

differ by more in both size and charge

than do Ni2+

and Sb5+

. The mean bond lengths around the two six coordinate sites are both

shorter in the tellurate than in La3Ni2SbO92, which is consistent with the presence of some Te

6+

on the six-coordinate site predominantly occupied by Ni2+

, see Table 3.3.

At first sight, our magnetometry data, along with the neutron diffraction data collected on D2b,

suggest that SrLa2Ni2TeO9 is a spin-glass with Tg = 35 K. The difference in the behaviour of this

compound and La3Ni2SbO9 cannot be attributed only to the additional B-site disorder present in

the former. Even in the presence of this disorder, the NN superexchange interactions in the

structure illustrated in Figure 3.3 are not frustrated, although the introduction of Te-O-Te

linkages will modify the relative strengths of the interactions present. The frustration necessary

for the formation of a spin-glass in this structure is only present when NNN neighbour

interactions are competitive with NN interactions. More specifically, when 180° Ni-O-Ni

interactions between cations ~ 3.95 Å apart are in competition with Ni-O-O-Ni interactions

between cations ~ 5.6 Å apart; linear, 7.9 Å Ni-O-Te-O-Ni interactions might also play a role.

43

Each of these interactions, when acting alone, leads to the adoption of a different magnetic

structure33

and so it is plausible to argue that when they are in competition a frustrated spin-glass

will be formed. The Ni-O-Ni interactions will be less dominant in regions where there is a local

excess of tellurium. The data in Figure 3.9 demonstrate that such regions exist and we therefore

suggest that the inhomogeneous composition of the crystallites is partly responsible for the

formation of a spin-glass phase. We must also attempt to explain why SrLa2Ni2TeO9 is a spin-

glass whereas La3Ni2SbO9 is a relaxor ferromagnet. In the latter, Ni-O-Ni interactions dominate

within ferrimagnetic domains, the boundaries of which are thought to be formed by regions in

which the 1:1 cation ordering is absent (Figure 3.2). In these regions, which were seen to extend

over at least 50 Å18

, NNN interactions will be more competitive. We did not observe any

comparable regions in SrLa2Ni2TeO9 and we therefore postulate that the difference in the

magnetic behaviour of the two compounds stems from the presence of infrequent but extended

disordered regions in La3Ni2SbO9 and frequent but local tellurium-rich regions in LaSr2Ni2TeO9.

The high twinning density in the latter might also be a factor. The presence of extended

disordered regions in the antimonate but not in the tellurate is consistent with the greater

difference in size and charge of Ni2+

and Te6+

, as discussed above.

In the discussion above we have linked the magnetic behaviour to the complexity of the

microstructure of these perovskites. The neutron diffraction data collected on D1b indicate that

there is a further level of complexity that we have not yet considered. Weak magnetic Bragg

scattering was observed at 5 K and 50 K. Were it not for the glass-like behaviour of the magnetic

susceptibility, the observation of magnetic ordering at these temperatures would not be

surprising in view of the ordering temperatures of other Ni2+

-containing perovskites30, 41-43

. In

order to model this scattering we have postulated that some regions of the sample order as either

G-type or C-type antiferromagnets. In order to account for this we must assume that local

regions exist wherein the cation ordering is sufficiently regular over a large enough distance to

ensure the dominance of NN interactions and hence G-type antiferromagnetism. In other regions,

a different cation-ordering pattern, favouring NNN interactions, is established. The former is

likely to be nickel-rich and the latter tellurium-rich. We are thus proposing that two different

types of antiferromagnetic island can exist within the predominantly glass-like crystallites.

Further experimental work will be necessary in order to establish the size of these regions, but

the fact that the magnetic scattering persists above the temperature of the susceptibility

maximum suggests that they are too large to be described as clusters. Martin et al have

previously proposed a related model based on phase separation to account for the behaviour of

Pr0.1Sr0.9MnO344

. The presence of antiferromagnetic regions at relatively high temperatures

explains why the effective magnetic moment, 2.21 µB per Ni2+

cation, derived from the Curie-

Weiss law is significantly lower than the spin-only value for Ni2+

, 2.83 µB. In comparable,

perovskite-related compounds the effective magnetic moment is usually enhanced by a small

second-order orbital contribution to a value in the range 3.0-3.7 µB30,41,42

. The magnitude of the

reduction seen in SrLa2Ni2TeO9 is thus consistent with our conclusion, drawn from the neutron

diffraction data, that ~ 36(3) % of the sample is antiferromagnetic at 5 K. We note that the D1b

data can also be modelled using a unique, non-collinear magnetic structure in which the y and z

44

components of the atomic magnetic moments order in C-type and G-type patterns, respectively.

In this case, again assuming an ordered moment of 2 µB per ordered cation, 25(2) % of the Ni2+

cations would be part of the long-range ordered structure. In view of the composition variations

described above we believe that the two-phase magnetic model is more likely to provide the

correct interpretation of our data.

So, to summarize, we synthesized a polycrystalline sample of SrLa2Ni2TeO9 using a standard

ceramic method and characterized it by neutron diffraction, magnetometry and electron

microscopy. The compound adopts a monoclinic, perovskite-like structure with space group

P21/n and unit cell parameters a = 5.6008(1), b = 5.5872(1), c = 7.9018(2) Å, β = 90.021(6)° at

room temperature. The two crystallographically-distinct B sites are occupied by Ni2+

and Te6+

in

ratios of 83:17 and 50:50.

In contrast to La3Ni2SbO9 , SrLa2Ni2TeO9 does not behave as a relaxor ferromagnet. Both ac and

dc magnetometry suggest that the compound is a spin glass below 35 K. The neutron diffraction

data, on the other hand, show that some regions of the sample are antiferromagnetic. Concerning

the crystallographic structure, the principal difference in the average structures of La3Ni2SbO9

and SrLa2Ni2TeO9 deduced from neutron diffraction data is that the 1:1 cation ordering over the

B and B’ sites is imperfect in the latter compound. Electron microscopy revealed twinning on a

nanoscale and local variations in composition. These defects are thought to be responsible for the

presence of two distinct types of antiferromagnetic ordering, C-type and G-type ordered domains.

As the other compounds showed very similar results for the structural characterization, we will

show only the representative data for those compounds, and not treat each of them in the same

detail as we did SrLa2TeNi2O9. At the end we will draw a general conclusion.

45

3.4. ALa2Ni2B’O9 (A=La, Ca, B’= Ta, Nb, Sb, W)

The X-ray and neutron powder diffraction and magnetometry studies of the ALa2Ni2B’O9

compounds were performed by the group of Prof. Dr. P. Battle at Oxford University.

3.4.1. Experimental results on the structure

As the other compounds showed very similar results to SrLa2TeNi2O9 for the structural

characterization by TEM, we will show only the representative data for those compounds, and

not treat each of them in the same detail as we did SrLa2TeNi2O9. At the end we will draw a

general conclusion including all compounds

The XRPD data of the La3Ni2TaO9, La3Ni2NbO9, La3Ni2Nb0.5Sb0.5O9 and CaLa2Ni2WO9 samples

suggested they were the desired perovskite-like phase, and that their reflections could be

accounted for using space group P21/n with a ~ b ~ √2ap, c ~ 2ap and β ~ 90° where ap is the cell

parameter of the parent perovskite. The crystal structures would thus contain two

crystallographically distinct octahedral sites, 2c and 2d, occupied by Ni2+

and the diamagnetic

cations Ta5+

, Nb5+

, Sb5+

, and W6+

. For CaLa2Ni2WO9 where 1/3 of the 4e site is occupied by Ca2+

instead of La3+

in accordance with the necessary charge balance for adding a cation with valence

6+.

As representative samples, La3Ni2TaO9 and La3Ni2NbO9 were refined further. For these

compounds, XRPD data and NPD data from D2b were analysed simultaneously. The distribution

of Ni2+

, Ta5+

and Nb5+

over the six-coordinate sites was allowed to vary. The cation distribution

determined in these refinements was held constant during refinements of the crystal structure

based on D2b data collected at lower temperatures, see below. The structural models deduced

from the D2b data were then used in the analysis of the low-temperature D1b data.

Simultaneous profile analysis of the XRPD and NPD patterns collected at room temperature gave

a satisfactory account of the structures of all compounds. Small amounts of diamagnetic

impurities were detected in each sample; 0.8(1) wt % of orthorhombic LaNbO4 in La3Ni2NbO9

and 0.4(2) wt % of La3TaO7 in La3Ni2TaO9. No Bragg peaks attributable to unreacted NiO were

observed in either case. The observed and calculated room-temperature XRPD and NPD patterns

of La3Ni2TaO9 and La3Ni2NbO9 are displayed in Figures 3.15 and 3.16. The structural parameters

of La3Ni2TaO9 and La3Ni2NbO9 derived from these analyses are presented in Tables 3.7 and 3.8.

In addition, the particle size of La3Ni2TaO9 and La3Ni2NbO9 samples were separately compared.

In order to achieve a satisfactory fit to the data collected from La3Ni2NbO9 it was necessary to

include a parameter, LX, which models the effect of a small particle size on the shape of the

Bragg peaks. For comparison purposes the parameter was also included in refinements of

La3Ni2TaO9, leading to the conclusion that the Nb:Ta particle size ratio was 1.67(7):1. This is

consistent with the findings reported by Skinner et al using TEM.45

For the compounds La3Ni2Nb0.5Sb0.5O9 and CaLa2Ni2WO9 only XRPD data was gathered and the

results of the Rietveld refinement for those compounds is shown in Figure 3.17. The results are

46

shown in Tables 3.9 and 3.10, respectively. Table 3.11 repeats the cell parameters for easy

comparison in one table.

Figure 3.15. Observed (red crosses), calculated (green line) and difference (purple) XRPD profiles of (a)

La3Ni2TaO9 collected at room temperature. Reflection markers are shown for La3TaO7 (cyan) and La3Ni2TaO9

(black). (b) La3Ni2NbO9 (black).

Figure 3.16. Observed (red crosses), calculated (green line) and difference (purple), NPD profiles of (a) La3Ni2TaO9

(Reflection markers are shown for La3TaO7 (cyan) and La3Ni2TaO9 (black)) and (b) La3Ni2NbO9 (Reflection markers

are shown for LaNbO4 (green) and La3Ni2NbO9 (black)), collected at room temperature.

Figure 3.17. Observed (red crosses), calculated (green line) and difference (purple) XRPD profiles of (a)

La3Ni2Nb0.5Sb0.5O9, collected at room temperature. Reflection markers is shown for La3Ni2Nb0.5Sb0.5O9 (black) (b)

CaLa2Ni2WO9 (black).

47

Table 3.7. Structural parameters of La3Ni2TaO9 at room temperature.

Atom Site x y z Uiso /Å2 Occupancy

La 4e 0.4924(3) 0.4613(9) 0.2506(2) 0.0224(2) 1.0

Ni/Ta1 2c 0 ½ 0 0.0048(1) Ni: 0.982(1)

Ta: 0.018(1)

Ni/Ta2 2d ½ 0 0 0.0048(1) Ni: 0.351(1)

Ta: 0.649(1)

O1 4e 0.7886(7) 0.7950(6) –0.0374(5) 0.0041(8) 1.0

O2 4e 0.7093(8) 0.2851(7) –0.0469(5) 0.0148(11) 1.0

O3 4e 0.5769(3) 0.0172(3) 0.2518(8) 0.0132(3) 1.0

XRPD: Rwp = 6.46 %, Rp = 4.86 %, χ2 = 2.317, NPD: Rwp = 4.72 %, Rp = 3.58 %, χ

2 = 2.317. Space group P21/n:

a = 5.5902(1) Å, b = 5.6407(1) Å, c = 7.9222(1) Å, β = 90.034(3) °.

Table 3.8. Structural parameters of La3Ni2NbO9 at room temperature.

Atom Site x y z Uiso /Å2 Occupancy

La 4e 0.4921(3) 0.4610(1) 0.2501(3) 0.0195(2) 1.0

Ni/Nb1 2c 0 ½ 0 0.0032(1) Ni: 0.964(7)

Nb: 0.036(7)

Ni/Nb2 2d ½ 0 0 0.0032(1) Ni: 0.369(7)

Nb: 0.631(7)

O1 4e 0.7912(9) 0.7965(7) –0.0366(5) 0.0059(10) 1.0

O2 4e 0.7119(9) 0.2835(7) –0.0472(5) 0.0087(10) 1.0

O3 4e 0.5766(3) 0.0168(3) 0.2499(11) 0.0121(3) 1.0

XRPD: Rwp = 7.62 %, Rp = 5.75 %, χ2 = 2.293, NPD: Rwp = 4.56 %, Rp = 3.51 %, χ

2 = 2.293. Space group P21/n:

a = 5.5865(1) Å, b = 5.6400(1) Å, c = 7.9165(1) Å, β = 90.012(5) °.

Table 3.9. Structural parameters of La3Ni2Nb0.5Sb0.5O9 at room temperature.

Atom Wyckoff

position x y z

Fractional

occupancy 10

2*UISO/ Å

2

La 4e 0.4926(9) 0.4609(2) 0.2512(6) 1.0 1.79(3)

Ni1 2c 0 0.5 0 1.0 0.27(15)

Nb1 2c 0 0.5 0 0 0.27(15)

Sb1 2c 0 0.5 0 0 0.27(15)

Ni2 2d 0.5 0 0 0.333 0.46(7)

Nb2 2d 0.5 0 0 0.333 0.46(7)

Sb2 2d 0.5 0 0 0.333 0.46(7)

O1 4e 0.800(3) 0.791(3) -0.056(3) 1.0 0

O2 4e 0.723(3) 0.270(3) -0.038(3) 1.0 0

O3 4e 0.576(2) 0.019(1) 0.248(3) 1.0 0

48

Table 3.10. Structural parameters of CaLa2Ni2WO9 at room temperature.

Atom Wyckoff

position x y z

Fractional

occupancy 100*UISO/Å

2

Ca 4e 0.4865(5) 0.4650(1) 0.2518(2) 0.333 1.78(3)

La 4e 0.4865(5) 0.4650(1) 0.2518(2) 0.667 1.78(3)

Ni1 2c 0 0.5 0 0.977(2) 0.14(7)

W1 2c 0 0.5 0 0.023(2) 0.14(7)

Ni2 2d 0.5 0 0 0.356(2) 0.10(3)

W2 2d 0.5 0 0 0.644(2) 0.10(3)

O1 4e 0.790(2) 0.790(2) -0.032(2) 1.0 0

O2 4e 0.719(2) 0.277(2) -0.051(2) 1.0 0

O3 4e 0.588(2) 0.021(1) 0.242(2) 1.0 0

Table 3.11. Crystallographic information of the samples.

Space group Cell parameters (Å)

La3Ni2TaO9 P21/n a=5.5887(1) Å, b=5.6411(1) Å,

c=7.9205(1) Å, β=90.027(9)

La3Ni2NbO9 P21/n a=5.5828(1) Å, b=5.6387(1) Å,

c=7.9128(2) Å, β=89.96º

La3Ni2Nb0.5Sb0.5O9 P21/n a=5.5916(1) Å, b=5.6375(1) Å,

c=7.9219(1)Å, β=90.103 (9)

CaLa2Ni2WO9 P21/n a=5.5300(1)Å, b=5.5698(1)Å,

c=7.8367(1)Å, β = 90.062(5)°

Selected area electron diffraction patterns were taken from all compounds from many different

crystals, preferably along continuous tilt series. Representative patterns for the main zones are

shown in Figure 3.18. For all compounds, all SAED patterns could be indexed using space group

P21/n, using the unit cell parameters derived from X-ray powder diffraction (listed in Table 3.11).

Since in literature it was reported that La3Ni2NbO9 is ferroelectric below 60 K46

, the space group

assignment P21/n was explicitly double checked for this compound, since the centrosymmetry of

P21/n does not allow ferroelectricity and Pn would be implied. For this it was necessary to prove

that the reflection condition 0k0: k=2n is valid, proving the 21 screw axis. On the SAED patterns

it could be argued that the 0k0: k=2n+1 are true reflections, not due to double diffraction as we

claim. Therefore the SAED pattern along the [100] zone axis was tilted over the b* axis until the

offensive reflections disappeared (Figure 3.19). Figure 3.20 shows the line profile of the

intensities of the reflections along the b* axis. The weak reflections seen when in [100] zone are

indicated by arrows. In the plots of the tilted patters no weak reflections are present, therefore the

reflection condition: 0k0: k=2n is valid and the space group Pn can be ruled out. Taking all

reflection conditions from SAED and assuming that the monoclinic crystal class determined from

XRD is correct, there can then be only one valid space group, i.e. P21/n.

49

Figures 3.21-3.23 show the HAADF-STEM images of the different compounds for the three main

zones ([100], [010] and [001]). All HAADF-STEM images along [100] and [010] show a clear

alternating order in the B cation positions, in agreement with the rock salt order in the structural

model. The [100] and [010] images can only be discerned by the presence of a slight zigzag of

the projected columns of A cations in the [100] images.

Figure 3.24 shows the STEM-EDX results for the different compounds (one compound per

column). There is no significant difference between the natures of the cation distributions for the

different compounds.

50

Figure 3.18. The representative SAED patterns of the La3Ni2BO9 and CaLa2Ni2WO9 samples, where each column

represents a different sample, from left to right: La3Ni2TaO9,La3Ni2NbO9, La3Ni2Nb0.5Sb0.5O9, CaLa2Ni2WO9.

51

Figure 3.19. Tilting SAED patterns along b*. Left pattern: along [100] zone, and middle/right: out of [100] zone.

Figure 3.20. Line profile of the intensities of reflections along b* axis of the SAED patterns of the Figure 3.19.

52

Figure 3.21. Aberration-corrected high resolution HAADF-STEM images along [100] zone axis, top left:

La3Ni2TaO9, top right: La3Ni2NbO9, bottom left: La3Ni2Nb0.5Sb0.5O9, and bottom right: CaLa2Ni2WO9 .

53

Figure 3.22. Aberration-corrected high resolution HAADF-STEM images of all compounds viewed along [010]

zone axes, top left: La3Ni2TaO9, top right: La3Ni2NbO9, bottom left: La3Ni2Nb0.5Sb0.5O9, and bottom right:

CaLa2Ni2WO9 .

54

Figure 3.23. Aberration-corrected high resolution HAADF-STEM images of all compounds viewed along [110]

zone axis, top left: La3Ni2TaO9, top right: La3Ni2NbO9, bottom left: La3Ni2Nb0.5Sb0.5O9, and bottom right:

CaLa2Ni2WO9 .

55

Figure 3.24. HAADF-STEM image and corresponding atomic resolution STEM-EDX maps of the La2ANi2(B,B’)O9

compounds along the [010] zone. The columns show from left to right: La3Ni2TaO9, La3Ni2NbO9,

La3Ni2Nb0.5Sb0.5O9, CaLa2Ni2WO9. Note that the maps are not on the same scale, but are of those areas indicated by

a white rectangle on the HAADF-STEM images in the first row. Compare to Figure 3.25.

56

The maps can be compared to the visualization of the final refined structures shown in Figure

3.25. The atomic coordinates in the Tables 3.7, 3.8, 3.9, and 3.10 show the same crystal structure,

thus only one model of this structure is shown in Figure 3.25.

Figure 3.25. Crystal structure of triple perovskites La2ANi2(B,B’)O9 close to [010] zone axis. Green and blue

octahedra are dominated by Ni and (B,B’) cations, respectively. The A cations are represented by red spheres.

3.4.2. Experimental results on the magnetic properties

The temperature dependence of the dc molar magnetic susceptibilities of La3Ni2TaO9,

La3Ni2NbO9, La3Ni2Nb0.5Sb0.5O9 and CaLa2Ni2WO9 are shown in Figure 3.26 (a), (c), (e) and (g),

respectively. The effective magnetic moment, µeff, and the Weiss temperature, θ, derived from

fitting the data collected above 200 K to the Curie-Weiss law are listed in Table 3.12, along with

the corresponding parameters of La3Ni2SbO918

. The ZFC susceptibility of each compound

reaches a maximum at a temperature TMAX, see Table 3.12, and below this temperature there is a

clear difference between the ZFC and FC susceptibilities. Note that χmax is an order of magnitude

greater in La3Ni2TaO9 and that hysteresis persists somewhat above TMAX in that compound. M(H)

for each compound is shown in Figure 3.26 (b), (d), (f), (h) and the values of the coercive field

(HC) and remanent magnetisation (MR) observed at 5 K are included in Table 3.12. All

compounds show their susceptibility maximum below 100 K.

57

Figure 3.26. (a) Temperature vs. molar magnetic susceptibility and (inset) inverse magnetic susceptibility of

La3Ni2TaO9, (c) La3N2NbO9, (e) La3Ni2Nb0.5Sb0.5O9 and (g) CaL2Ni2WO9. The magnetisation per formula unit as a

function of magnetic field H is given in the right side column. The inset shows the hysteresis loop at 5 K in an H = ±

2 kOe.

58

Table 3.12. Magnetic parameters of ALa2Ni2BO9 (A=Ca, B = W/Sb/Nb/Ta) at 5K.

Composition µeff / µB θ / K MR / µB HC / kOe TMAX / K

La3Ni2SbO9 2.2(1) 159 0.5 0.5 105

La3Ni2TaO9 2.15(1) +144(2) 0.2 0.6 35

La3Ni2NbO9 2.12(1) +124(2) 0.1 1.3 29

La3Ni2Nb0.5Sb0.5O9 2.08(2) +150(2)

CaLa2Ni2WO9 2.25(1) +121(1)

The deviation from linearity at 150 K is greater in the case of La3Ni2TaO9. The temperature and

frequency dependence of the ac molar susceptibilities of La3Ni2TaO9 and La3Ni2NbO9 are shown

in Figure 3.27 (a) and (b), respectively. In the case of La3Ni2NbO9 the temperature of the

susceptibility maximum is frequency dependent and the susceptibility is complex below this

temperature. The parameter ΔTf/[TfΔ(log ω)] takes a value of 0.0167, indicating that La3Ni2NbO9

can be classified as a canonical spin-glass. The temperature of the maximum in χ’ corresponds

closely to TMAX observed in the dc susceptibility. However, the data in Figure 3.27 (a) show

clearly that the magnetic transition in La3Ni2TaO9 occurs at 85 K, the temperature below which

hysteresis is observed in the dc susceptibility and well above TMAX. The ac susceptibility is again

complex below the transition temperature, which shows no clear frequency dependence.

Figure 3.27. The real and imaginary components of the ac susceptibility of (a) La3Ni2TaO9 and (b) La3Ni2NbO9 as a

function of temperature measured at 1 (red), 10 (green) and 100 (blue) Hz.

No additional Bragg peaks were detected in any of the NPD patterns collected on D2b below

room temperature. The results of our analysis of the data collected on that instrument at 5 K are

shown in Figure 3.28 (a) and (b), and the refined structural parameters of La3Ni2TaO9 and

La3Ni2NbO9 at 5 K are shown in Tables 3.14 and 3.13, respectively. Some selected bond lengths

and angles, along with the corresponding quantities reported by Battle et al for La3Ni2SbO9, are

listed in Tables 3.15 and 3.16. However, the enhanced counting statistics of D1b revealed low-

angle Bragg scattering, assumed to be magnetic in origin, which was not observed in the D2b

59

data. More specifically, a 100 reflection and additional intensity in the 012 reflection were

present in the NPD patterns collected at 5 and 20 K from both La3Ni2TaO9 and La3Ni2NbO9, see

Figures 3.29 (a), (b) and 3.30 (a), (b). These reflections are forbidden in the space group P21/n

but are consistent with the presence of C-type antiferromagnetic ordering47

. In such a structure

the nickel cations couple ferromagnetically within (100) sheets but antiferromagnetically between

these sheets. In order to analyse these data we assumed that a fraction of the Ni2+

cations within

each compound lie in nickel-rich regions within which long-range magnetic ordering can be

established. The fraction of the sample engaged in this ordering is inversely proportional to the

magnitude of the ordered moment of the Ni2+

cations and the two cannot be determined

independently from our data. When it was assumed that each ordered cation has a moment of 2

µB, the fraction of ordered cations refined to be 22(2) and 20(2) % of Ni2+

cations in La3Ni2NbO9

and La3Ni2TaO9, respectively. The moments were found to lie along [010].

Figure 3.28. Observed (red crosses), calculated (green line) and difference (purple line) NPD profiles of (a)

La3Ni2TaO9 and (b) La3Ni2NbO9 at 5 K measured on D2b. The colour of reflection markers and what they represent

are the same as in Figure 3.16 (a) and (b).

Figure 3.29. Observed (red crosses), calculated (green line) and difference (purple line) NPD profiles of La3Ni2TaO9

at 20 K measured on D1b (a) across the whole scanning range and (b) the low angle region. Reflection markers for

structural La3Ni2TaO9, La3TaO7 and magnetic La3Ni2TaO9 are black, cyan and orange respectively.

60

Figure 3.30. Observed (red crosses), calculated (green line) and difference (purple line) NPD profiles of La3Ni2NbO9

at 5 K measured on D1b (a) across the whole scanning range and (b) the low angle region. Reflection markers for

structural La3Ni2NbO9, LaNbO4 and magnetic La3Ni2NbO9 are black, green and blue respectively.

Table 3.13. Structural parameters of La3Ni2NbO9 at 5 K.

Atom Site x y z Uiso /Å2 Occupancy

La 4e 0.4918(5) 0.4588(2) 0.2503(12) 0.0140(3) 1.0

Ni/Nb1 2c 0 ½ 0 0.00182)

Ni: 0.964(7)

Nb: 0.036(7)

Ni/Nb2 2d ½ 0 0 0.0018(2)

Ni: 0.369(7)

Nb: 0.631(7)

O1 4e 0.7920(11) 0.7969(10) –0.0383(7) 0.0012(12) 1.0

O2 4e 0.7119(13) 0.2835(11) –0.0464(8) 0.0084(15) 1.0

O3 4e 0.5786(4) 0.0174(4) 0.2517(14) 0.0086(5) 1.0

Rwp = 4.72 %, Rp = 3.59 %, χ2 = 4.618. Space group P21/n: a = 5.5755(1) Å, b = 5.6383(1) Å, c = 7.9029(2) Å,

β = 90.024(9) °.

Table 3.14. Structural parameters of La3Ni2TaO9 at 5 K.

Atom Site x y z Uiso /Å2 Occupancy

La 4e 0.4919(5) 0.4597(2) 0.2498(12) 0.0143(3) 1.0

Ni/Ta1 2c 0 ½ 0 0.0012(2)

Ni: 0.982(1)

Ta: 0.018(1)

Ni/Ta2 2d ½ 0 0 0.0012(2)

Ni: 0.351(1)

Ta: 0.649(1)

O1 4e 0.7896(11) 0.7978(10) –0.0386(7) 0.0018(13) 1.0

O2 4e 0.7096(12) 0.2835(10) –0.0462(8) 0.0062(15) 1.0

O3 4e 0.5781(5) 0.0180(4) 0.2510(11) 0.0081(5) 1.0

NPD: Rwp = 4.60 %, Rp = 3.66 %, χ2 = 3.350.Space group P21/n: a = 5.5782(2) Å, b = 5.6372(2) Å, c = 7.9067(2) Å,

β = 90.064(6) °.

61

Table 3.15. Selected bond lengths (Å) and angles (°) in La3Ni2BO9 (B = Sb, Nb or Ta) at RT.

La3Ni2SbO9 La3Ni2TaO9 La3Ni2NbO9 La3Ni2Nb0.5Sb0.5O9 CaLa2Ni2WO9

Bond lenghts (Å)

La–O1 2.815(9) 2.745(4) 2.733(4) 2.854(21) 2.702(14)

La–O1 2.687(7) 2.721(4) 2.733(4) 2.654(18) 2.703(12)

La–O1 2.384(8) 2.452(4) 2.446(4) 2.340(23) 2.435(14)

La–O2 2.799(9) 2.831(4) 2.838(5) 2.839(23) 2.891(14)

La–O2 2.478(8) 2.433(4) 2.440(5) 2.557(22) 2.414(12)

La–O2 2.686(8) 2.648(5) 2.632(5) 2.606(21) 2.535(13)

La–O3 2.561(2) 2.549(2) 2.549(2) 2.542(8) 2.536(6)

La–O3 2.448(4) 2.429(2) 2.430(2) 3.174(8) 3.149(6)

2c – O1 2.094(5)×2 2.062(3)×2 2.059(4)×2 2.031(17) 2.005(11)

2c – O2 2.055(6)×2 2.061(4)×2 2.054(4)×2 2.040(17) 2.028(10)

2c – O3 2.071(10)×2 2.016(6)×2 2.028(9)×2 2.043(23) 2.081(15)

2d – O1 2.007(6)×2 2.007(3)×2 2.012(4)×2 2.100(16) 2.000(11)

2d – O2 2.012(6)×2 2.023(4)×2 2.024(4)×2 1.993(17) 2.002(11)

2d – O3 1.993(10)×2 2.042(6)×2 2.026(9)×2 2.013(23) 1.963(15)

2c – O1 – 2d 151.62(1) 154.7(2) 154.3(2) 148.0(12) 157.0(8)

2c – O2 – 2d 155.62(1) 152.9(2) 153.4(2) 159.8(14) 153.7(7)

2c – O3 – 2d 155.12(1) 154.9(1) 155.04(9) 155.2(6) 151.4(6)

Bond angles(°)

O1 –2c–O2 89.7(3)×2 89.8(2)×2 89.2(3)×2 86.4(9)×2 88.7(4)×2

O1– 2c –O3 88.9(2)×2 88.8(1)×2 88.8(1)×2 86.8(8)×2 86.5(6)×2

O2 – 2c –O3 89.4(2)×2 88.0(1)×2 87.7(1)×2 88.9(8)×2 87.5(5)×2

O1 – 2d – O2 87.6(3)×2 88.1(2)×2 87.2(3)×2 84.1(9)×2 86.8(5)×2

O1 – 2d – O3 87.8(2)×2 89.8(1)×2 89.8(1)×2 86.1(7)×2 87.7(6)×2

O2 – 2d – O3 89.2(2)×2 88.9(1)×2 88.9(1)×2 88.7(8)×2 89.8(5)×2

62

Table 3.16. Selected bond lengths (Å) and angles (°) in La3Ni2BO9 (B = Sb, Nb or Ta) at 5 K.

La3Ni2SbO9 La3Ni2NbO9 La3Ni2TaO9

La–O1 2.833(8) 2.732(10) 2.743(10)

La–O1 2.658(7) 2.718(8) 2.712(9)

La–O1 2.401(8) 2.435(9) 2.439(9)

La–O2 2.767(9) 2.826(10) 2.819(10)

La–O2 2.463(8) 2..449(9) 2.438(9)

La–O2 2.693(8) 2.624(9) 2.639(9)

La–O3 2.545(3) 2.535(2) 2.536(3)

La–O3 2.463(4) 2.418(3) 2.421(4)

2c – O1 2.088(5) 2.059(5) 2.070(5)

2c – O2 2.050(6) 2.050(6) 2.060(5)

2c – O3 2.036(10) 2.013(11) 2.020(9)

2d – O1 2.011(5) 2.014(6) 2.001(5)

2d – O2 2.012(5) 2.021(6) 2.014(6)

2d – O3 2.025(10) 2.039(11) 2.034(9)

O1 –2c–O2 89.8(3) 89.0(4) 89.5(3)

O1– 2c –O3 89.0(2) 88.9(2) 88.9(2)

O2 – 2c –O3 89.7(2) 88.1(2) 88.1(2)

O1 – 2d – O2 87.5(3) 87.1(4) 87.4(4)

O1 – 2d – O3 87.6(2) 90.0(2) 89.7(2)

O2 – 2d – O3 89.2(2) 89.2(2) 89.2(2)

2c – O1 – 2d 151.3(3) 153.6(3) 153.8(3)

2c – O2 – 2d 155.7(3) 153.7(4) 153.4(4)

2c – O3 – 2d 154.7(2) 154.4(1) 154.5(1)

63

3.4.3. Conclusions

Despite the differences in their magnetic properties (ferromagnets versus spin glasses), all the

compounds reported in this chapter adopt the same monoclinic structure, even though the identity

of the diamagnetic cation was changed or a mixture of two different diamagnetic cations was

used. No differences were found in the local structures by any advanced transmission electron

microscopy technique. Also the bond distances and bond angles refined from X-ray diffraction

give no clarification for the magnetic differences, since (1) the 2c/2d-O bond lengths in the

La3Ni2B”O9 are all similar (Table 3.17) and (2) the selected bond angles of La3Ni2NbO9 lie in

between those of La3Ni2TaO9 and La3Ni2Nb0.5Sb0.5O9, where the latter two are ferromagnets and

La3Ni2NbO9 is a spin glass. The only difference that was found between La3Ni2TaO9 and

La3Ni2NbO9, the representative samples for resp. spin glass and ferromagnet, is a difference in

particle size, with a difference ratio of 1.7 for La3Ni2NbO9:La3Ni2TaO9 as derived from neutron

powder diffraction.

Table 3.17. Selected bond angles and mean bond distances which are involved in superexchange route in

La3Ni2B”O9.

Compounds 2cO12d/ 2cO22d/ 2cO32d/ 2cO/Å 2dO/Å

La3Ni2TaO9 155.8(14) 155.4(14) 151.3(6) 2.05(3) 2.02(3)

La3Ni2NbO9 152.9(17) 155.1(17) 156.3(7) 2.03(4) 2.03(4)

La3Ni2Sb0.5Nb0.5O9 148.0(12) 159.8(14) 155.2(6) 2.04(3) 2.04(3)

64

Chapter 4

A3Fe2TeO9 (A=Sr, Ba) compounds, the search for relaxor

ferromagnets continues in compounds containing Fe3+

instead of Ni2+

4.1. Sr3Fe2TeO9

The content of this section has been published as the following paper: Structural chemistry and magnetic

properties of the perovskite Sr3Fe2TeO9, Yawei Tang, Emily C. Hunter, Peter D. Battle, Robert Paria

Sena, Joke Hadermann, Maxim Avdeev, J. M. Cadogan, Journal of Solid State Chemistry, 242, 86-95,

October 2016.

The X-ray and neutron powder diffraction, Mössbauer and magnetometry investigation of the

SrLa2Ni2TeO9 were done by the groups of Prof. Dr. P. Battle at Oxford University and Prof. Dr.

M. Avdeev at Bragg Institute, Australian Nuclear Science and Technology Organization,

Australia.

4.1.1. Introduction

We have investigated the Sr3Fe2TeO9 compound, which has been previously studied by

Augsberger et al48

and Ivanov et al49

. They described Sr3Fe2TeO9 as a tetragonal perovskite in

which the Fe3+

and Te6+

cations are partially ordered over the six-coordinate B sites. More

specifically, the octahedral sites are divided into two subsets that occur alternately in 3

dimensions throughout the structure in a checker-board pattern. Ivanov et al49

concluded that one

subset, or sublattice, is occupied by Fe3+

and Te6+

in a ratio of 0.72:0.28 and the Fe:Te ratio at the

other site is thus 0.61:0.39, see Figure 4.1; the corresponding values reported by Augsberger et

al48

were 0.9:0.1 and 0.57:0.43.

65

Figure 4.1. Polyhedral representation of the tetragonal structure proposed for Sr3Fe2TeO9 in the space group I4/m.

Red circles represent Sr2+

, green and khaki octahedra represent sites occupied by a partially ordered distribution of

Fe3+

and Te6+

.

Neutron diffraction data collected by Ivanov et al. suggested that the compound showed long-

range magnetic ordering below ~260 K with the atomic moments on the two sublattices aligned

antiparallel along [001]. The magnetic moments per Fe3+

cation on the two sites refined to values

of ~1.6 and ~1.3 µB at 10 K. These values are very low for a high-spin d5 cation. Furthermore,

despite the mismatch in the moments on the two sites, no sharp increase in the magnetisation was

seen on cooling the sample below 260 K. However, hysteresis was apparent in the magnetic

susceptibility below this temperature.

A maximum was observed in the zero-field cooled (ZFC) susceptibility at ~90 K in a measuring

magnetic field of 100 Oe and at ~60 K in a magnetic field of 30 kOe; the susceptibility was field

dependent throughout the measured temperature range although, inconsistently, the magnetisation

at 300 K was shown to be a linear function of field for 0<H/kOe<50. The field-cooled

susceptibility maintained a negative temperature gradient at all temperatures and in all fields.

Ivanov et al suggested that ferrimagnetic ordering is present, at least over short length scales,

below 260 K and that the susceptibility maximum observed at low temperature marks the

formation of a re-entrant spin glass. In contrast, Augsburger et al reported the onset of

ferromagnetism at ~717 K in their sample and they observed a remanent magnetisation 0.35 µB

per Fe3+

cation at 5 K. They analysed their neutron diffraction data using a model in which

nearest-neighbour Fe3+

cations were coupled antiferromagnetically and found an ordered moment

of ~0.5 µB per Fe3+

cation at 300 K and 0.85 µB at 13 K, the latter lower even than the value

reported by Ivanov et al. The samples were synthesized by a solid-state reaction. Djerdi et al50

have recently prepared a sample by a sol-gel route. It is reported to adopt the simple cubic

perovskite structure with no cation ordering over the B sites and to be ferrimagnetic below 667 K.

The suggestion by Ivanov et al that the length scale of the magnetic ordering might be important

in determining the magnetic properties of Sr3Fe2TeO9 is reminiscent of the behaviour of

66

La3Ni2SbO9, another A3B2B’O9 perovskite, which was described as a relaxor ferromagnet2,51

. The

octahedral sites in this monoclinic compound are divided over two sublattices, as they are in

Sr3Fe2TeO9. However, in the case of La3Ni2SbO9 one sublattice is occupied entirely by Ni2+

cations and 2/3 of the second sublattice is occupied by Sb5+

with the remaining 1/3 being

occupied by Ni2+

; there is no long-range ordering of the two cation species occupying the second

sublattice. The imbalance in the concentration of Ni2+

cations on the two sites leads to the

observed magnetic behaviour, although local variations in the composition cause the formation of

small magnetic domains. Clearly, the net magnetisation is expected to be larger in a compound

that contains a more strongly magnetic cation, and it is also likely that the transition temperature

will increase above the value of 108 K seen in La3Ni2SbO9. Bearing in mind the internal

inconsistencies in the description given by Ivanov et al, the differences between the data

collected by different authors and the potential of Sr3Fe2TeO9 to show enhanced relaxor

properties, we have undertaken our own investigation of this compound.

Synthesis

The synthesis was done at the University of Oxford, by the group of Prof. Dr. Peter Battle. A

polycrystalline sample of Sr3Fe2TeO9 was prepared using the standard ceramic method. SrCO3,

Fe2O3 and TeO2 (purity >99.95%) were weighed out in the appropriate stoichiometric ratio and

ground together in an agate mortar for 30 minutes to give a homogeneous mixture, then it was

fired at 700 °C for 24 h, immediately was quenched to room temperature, reground and pressed

into a pellet that was fired in air at 950 °C for 24 h and subsequently annealed at 1200 °C for 48 h

after further grinding. Finally, the furnace was cooled down to 800 ° C and later it was quenched

to room temperature.

4.1.2. Experimental results

The XRPD investigation confirms that Sr3Fe2TeO9 adopts a perovskite-like structure. The

presence of weak Bragg peaks, most noticeably at angles 2θ ˂ 30°, indicated a deviation away

from the ideal cubic perovskite structure, and, influenced by the earlier work of Ivanov et al50

,

initially all XRPD patterns were indexed in the tetragonal space group I4/m, with the unit-cell

parameters a = 5.5749(1) Å, c = 7.8963(3) Å, that is a~b~√2ap, c~2ap (ap: cell parameter of

parent perovskite). Although the fit between the observed and calculated diffraction patterns was

good (Rwpr = 6.06 %, χ2 = 1.47, see Figure 4.2) some weak reflections, for example at 2θ = 18 °,

could not be indexed in the tetragonal cell. These reflections were extremely weak and the fit was

very good so we initially assumed that they derived from a minor impurity phase, but our electron

microscopy study showed that this assumption was wrong.

67

Figure 4.2. Observed (red crosses) and calculated (green line) X-ray diffraction patterns for Sr3Fe2TeO9 in space

group I4/m; a difference curve (purple) is shown. Black vertical bars represent the positions of Bragg reflections.

EDX analysis found an average Fe:Te ratio of 2.3(3) across the crystallites studied. This value is

consistent with the target composition. SAED patterns, including several tilt series, were taken

from many different crystals. All the patterns could be indexed in the trigonal system, using the

approximate cell parameters a = b= 5.58 Å, c = 13.66 Å. Figure 4.3 shows the SAED patterns

from the main zone axes ( [100]/[12-1], [12-1], [001], and [-110] ) of Sr3Fe2TeO9. The [100] and

[12-1] SAED patterns are overlapping because of the twinning of nanodomains, which are

smaller than the smallest SAED aperture of the microscope. As a consequence, the [100] electron

diffraction pattern could not be found separately. However, the [100] zone axis pattern can be

seen in pure form as the Fourier transform of the [100] HAADF-STEM image in Figure 4.4.

68

Figure 4.3. Representative SAED patterns of the main zones of Sr3Fe2TeO9.

The reflection conditions 0kl: l=2n, 00l: l=2n (4-index system: h-h0l:l=2n, 000l:l=2n) apparent

in the electron diffraction patterns and Fourier transform leave as possible space groups P3c1

and P-3c1. As is convention, without any contra-indication, we continued with the centro-

symmetric space group, i.e. P-3c1.

In the top left image in Figure 4.4, taken along [100], horizontal rows of uniformly bright dots

alternate with rows in which pairs of weak dots are separated from each other by one bright dot.

In HAADF-STEM images, the brightness of the dots increases with the average atomic number,

Z, of a projected column. Sr, Fe and Te have the atomic numbers 38, 26 and 52, respectively. The

weakest dots thus correspond to columns of Fe, and the brightest to columns of Te; the

homogeneously bright rows are populated by Sr. This assignment is confirmed by the atomic

resolution EDX-STEM maps shown in Figure 4.4.

The contrast in the [001] image shown in Figure 4.5 and the inserted atomic resolution EDX-

STEM map also show evidence of cation order. Here, the structure is viewed down columns in

which Sr atoms alternate with octahedral sites; the 2:1 sequence of weak and strong dots in the

lines of atoms running down the page clearly shows the presence of 2:1 Fe:Te ordering. Figure

4.6 shows a [100] and a [001] view of a trigonal perovskite unit cell with this 2:1 ordering of Fe

and Te over the octahedral sites. The observation of [100]/[12-1] twinning prompted the

collection of more images, taken from more than one sample, in an attempt to assess the extent to

which the microstructure is disrupted. These images showed that the [12-1] domains observed in

69

what is essentially a [100] projection range in size from a few nanometers to ~50 nm in diameter,

see Figure 4.7.

Figure 4.4. Top: HAADF-STEM image of Sr3Fe2TeO9 taken along the zone axis [100] and a Fourier transform of a

pure [100] area. Bottom: HAADF-STEM image and corresponding atomic resolution EDX-STEM map of a [100]

oriented area, showing the positions of the columns of Sr, Fe, Te and O atoms.

Figure 4.5. HAADF-STEM image of Sr3Fe2TeO9 taken along the [001] zone axis. The inset is an atomic resolution

EDX-STEM map showing the positions of the different cations along this projection. There is an overlap of A and B

cations along this projection.

70

Figure 4.6. Polyhedral representation of the crystal structure of Sr3Fe2TeO9 in the trigonal space group P-3c1 viewed

along (a) [100] and (b) [001]; FeO6 and TeO6 octahedral are shown in green and blue, respectively. Red circles

represent the Sr2+

cations.

Figure 4.7. Low-magnification image of a crystallite of Sr3Fe2TeO9, taken with the ACOM (automated orientation

mapping) technique. Electron diffraction patterns show that yellow areas are being viewed along [100] and green

areas along [121]. The crystallite is overlapping with others in the red region and the blue region shows only the

basic perovskite reflections, probably because of the thinness of the crystal in this region.

Furthermore, although the trigonal phase dominates the images, we observed crystallites that

contained both trigonal regions and regions showing a checkerboard-like alternation of cations

over the six-coordinate sites, see Figure 4.8. The contrast variations in the latter regions match

the variation of electron density in the tetragonal structural model described above, see Figure

4.1. Therefore and for conciseness, in Figure 4.8 these areas are labelled as I4/m. Note that the

electron diffraction patterns labelled as the trigonal [12-1] zone in Figure 4.3 could also be the

[100] zone of the I4/m phase as the two cannot be distinguished from their electron diffraction

patterns alone. They can, however, be distinguished from their HAADF-STEM images, see

Figure 4.8.

71

Figure 4.8. HAADF-STEM image of Sr3Fe2TeO9 showing the coexistence of areas with chessboard order (labelled

I4/m [100]), P-3c [100] and P-3c [12-1]. The colour overlays clarify the order of the atoms in the surrounding areas;

Sr, Fe and Te are red, green and blue, respectively. Khaki represents columns with, in this projection, a mix of Fe

and Te.

In the light of the microscopy data presented above, the XRPD data were reanalysed in space

group P-3c1 using the structural model shown in Figure 4.6. The overall quality of the Rietveld

fit achieved using this trigonal structure was essentially the same (Rwpr = 6.04 %, χ2 = 1.46 see

Figure 4.9) as that achieved in tetragonal symmetry, with different weak reflections, for example

at 2θ = 19 °, being unaccounted for.

Figure 4.9. Observed (red crosses) and calculated (green line) X-ray diffraction patterns for Sr3Fe2TeO9 in space

group P-3c1; a difference curve (purple) is shown. Black vertical bars represent the positions of Bragg reflections.

72

Consequently our final analysis treated the sample as a mixture of a trigonal phase and a

tetragonal phase, the latter in space group I4/m. In the trigonal structure there are two

independent oxygen sites (6f and 12g), two A sites (2a and 4d) that accommodate strontium

atoms, and two independent B-sites (4d and 2b) for iron and tellurium, see Table 4.1.

Table 4.1 . Atomic parameters of Sr3Fe2TeO9, refined from neutron diffraction data collected at 300 K.

Atom Site x y z Uiso fractional

occupancy

Phase 1

Sr1 2a 0 0 0.25 0.005(1) 1

Sr2 4d 0.6667 0.3333 0.5834(6) 0.0088(6) 1

Fe 4d 0.6667 0.3333 0.3335(5) 0.0036(3) 1

Te 2b 0 0 0 0.0036(3) 1

O1 6f 0.521(1) 0 0.25 0.014(1) 1

O2 12g 0.1942(6) 0.3308(8) 0.5842(4) 0.0062(5) 1

Phase 2

Sr 4d 0 0.5 0.25 0.026(1) 1

Fe1 2a 0 0 0 0.023(4) 1

Fe2 2b 0 0 0.5 0.003(3) 0.3333

Te 2b 0 0 0.5 0.003(3) 0.6667

O1 4e 0 0 0.242(6) 0.024(3) 1

O2 8h 0.285(3) 0.217(2) 0 0.021(1) 1

Rwp = 4.67 %, Rp = 3.56 %, χ2= 4.568

Phase 1: space group P-3c1, a = 5.5761(3) Å, c = 13.654(1) Å, weight fraction=76 %

Phase 2: space group I4/m, a = 5.5886(5) Å, c = 7.9134(8) Å, weight fraction=24 %

In the refinements it was assumed that the iron and tellurium cations are completely ordered over

the latter two sites and thus the cation arrangement in the trigonal phase can be expressed as

Sr[Fe3+

0.6667](4d)[Te6+

0.3333](2b)O3. In the tetragonal structure, however, there is only one position

(4d) for strontium, and the two distinct B-sites (2a and 2b) have equal multiplicities. It was

assumed that one is fully occupied by iron and the other by a random distribution of iron (33%)

and tellurium (67%) to give an overall 2:1 stoichiometry of iron and tellurium. In this case the

cation arrangement can be described as Sr[Fe3+

0.5](2a)[Fe3+

0.1667Te6+

0.3333](2b)O3. This two-phase

model, with the trigonal and tetragonal phases occurring in a ratio of 73.8(6):26.2(6), accounted

for all the observed peaks. The agreement achieved between the observed and calculated

diffraction patterns in this case (Rwpr = 5.93 %, χ2 = 1.41) can be seen in Figure 4.10.

73

Figure 4.10. Observed (red crosses) and calculated (green line) X-ray powder diffraction patterns for Sr3Fe2TeO9

sample; a difference curve (purple) is shown. Black (red) vertical bars represent the positions of Bragg reflection for

the trigonal (tetragonal) phase.

57

Fe Mössbauer spectra collected from Sr3Fe2TeO9 at room temperature are shown in Figure 4.11

(a) and (b). The quantitative results are summarized in Table 4.2. A detailed discussion of the

Mossbauer results falls outside the scope of this thesis and can be found in the paper published

on this chapter. The prominent results are that there might be slightly different nearest-neighbour

environments around the iron sites.

Figure 4.11. (a)

57Fe Mössbauer spectrum of Sr3Fe2TeO9 obtained at room temperature (295 K) on an extended

velocity scale. The fitted curve employed a distribution of hyperfine field, as described in the text. (b) 57

Fe

Mössbauer spectrum of Sr3Fe2TeO9 obtained at room temperature (295 K) on a reduced velocity scale.

Table 4.2. 57

Fe Mössbauer hyperfine parameters of Sr3Fe2TeO9, obtained by fitting the spectrum shown in Figure

4.11 (b). The errors are estimates.

Isomer shift

(𝛿 mm/s)

0.02 mm/s

Quadrupole splitting

(|Δ|mm/s)

0.03 mm/s

% Area

1.5%

0.37 0.49 43.2

0.41 1.15 46.4

0.27 2.23 10.4

(‘sextet’ contribution)

74

The neutron diffraction data collected at 300 K using wavelengths of 1.6220 and 2.4395 Å were

analysed simultaneously in the same two-phase model used in the final analysis of the XRPD

data. Refinement of the atomic coordinates and atomic displacement parameters associated with

this model did not give an entirely satisfactory account of the data. The discrepancies between

the observed and calculated diffraction profiles were most noticeable at low angles and were

therefore assumed to be magnetic in origin. Furthermore, Ivanov et al reported a value of 260 K

for the magnetic transition temperature in tetragonal Sr3Fe2TeO9 and it is therefore reasonable to

assume that the magnetic scattering observed at 300 K is associated with the trigonal phase.

However, in order to complete the analysis and deduce a detailed model for the magnetic

structure it was necessary to consider the Mössbauer spectra. A calculation based on the data in

Table 4.2 gives a ratio of 70.5:29.5. This ratio is in reasonable agreement with the ratio of

~74:26 deduced from the XRPD data and so we assumed that our partition of the absorption

between the two phases was correct. Hence we deduced that 53 - 58 % of the iron in the trigonal

phase is magnetically ordered at room temperature. The antiferromagnetic structure shown in

Figure 4.14 gives rise to additional intensity in the Bragg peaks that were mis-fitted when

magnetic scattering was neglected in our initial analysis of the neutron data. Each Fe3+

cation is

coupled antiferromagnetically to its neighbouring sites, whether they are nearest neighbours or

next-nearest neighbours that are separated by a TeO6 octahedron. When 53 % of the Fe3+

ions in

the trigonal phase were assumed to contribute to the magnetic scattering, the ordered magnetic

moment refined to 3.60(5) µB per ordered cation; the calculated value of the moment is inversely

proportional to the percentage of the Fe3+

ions assumed to be ordered. A value of ~4.5 µB is

expected in a defect-free, saturated antiferromagnet but the value of the hyperfine field is also

lower than the value of ~55 T typically observed in that case and our model thus gives a

reasonable and self-consistent account of the spectroscopic and diffraction data collected at room

temperature. The inclusion of magnetic scattering in the Rietveld analysis resulted in agreement

factors of Rwpr = 4.67 %, χ2 = 4.57. The concentration ratio of the trigonal and tetragonal phases

refined to be 76:24, in agreement with the ratio determined by X-ray diffraction. The final

observed and calculated diffraction profiles are shown in Figure 4.15 and the refined structural

parameters are listed in Table 3.19; the derived bond lengths and angles are listed in Table 3.21.

75

Figure 4.14. Magnetic structure of the trigonal phase of Sr3Fe2TeO9. Green and blue octahedra represent Fe3+

and

Te6+

sites, respectively, and red circles represent Sr2+

cations. Black arrows represent the ordered magnetic moments

of the Fe3+

cations.

Figure 4.15. Observed (red crosses) and calculated (green line) neutron diffraction patterns for Sr3Fe2TeO9 at 300

K; a difference curve (purple) is shown. Black, red and blue vertical bars represent the positions of Bragg reflections

for the trigonal, tetragonal and magnetic phases, respectively. λ = 1.6215 Å.

The observation of a hyperfine field of ~52 T at 5 K led to assigning a proportionately-enlarged

magnetic moment to the ordered Fe3+

cations in the trigonal phase during the analysis of the

neutron diffraction data collected at 3 K. The tetragonal phase was again assumed to be non-

magnetic; the validity of this assumption is discussed below. When the moment was held fixed at

4.34 µB per ordered cation, the value calculated from the hyperfine field, the fraction of magnetic

Fe3+

in the trigonal structure refined to be 55.6 %, a value that lies within the range deduced from

the room-temperature Mössbauer spectra. The refined structural parameters obtained at 3 K are

listed in Table 4.4 and selected bond lengths are included in Table 4.3. The observed and

calculated diffraction patterns are shown in Figure 4.16.

Figure 4.16. Observed (red crosses) and calculated (green line) neutron diffraction patterns for Sr3Fe2TeO9 at 3 K; a

difference curve (purple) is shown. Black, red and blue vertical bars represent the positions of Bragg reflection

positions for the trigonal, tetragonal and magnetic phases, respectively. λ = 2.4395 Å.

76

Table 4.3. Selected interatomic distances (Å) and bond angles (degrees) in Sr3Fe2TeO9 at 300 K and 3 K.

300 K 3 K

Trigonal phase, space group P -3c1

Sr1 – O1 2.673(6) ×3 2.581(7) ×3

Sr1 – O1 2.903(6) ×3 2.982(7) ×3

Sr1 – O2 2.776(5) ×6 2.781(6) ×6

Sr2 – O1 2.790(7) ×3 2.798(8) ×3

Sr2 – O2 2.627(3) ×3 2.618(4) ×3

Sr2 – O2 2.949(3) ×3 2.945(4) ×3

Sr2 – O2 2.809(8) ×3 2.78(1) ×3

Fe – O1 1.976(4) ×3 1.974(4) ×3

Fe – O2 1.980(5) ×3 1.991(7) ×3

Te – O2 1.974(5) ×6 1.958(6) ×6

O1 – Fe – O1 90.1(2) 90.4(2)

O2 – Fe – O2 90.9(3) 90.3(4)

O1 – Fe – O2 178.6(3) 178.9(3)

88.6(2) 88.9(2)

90.4(2) 90.3(4)

Tetragonal phase, space group I 4/m

Sr – O1 2.795(1) ×4 2.7908(5) ×4

Sr – O2 2.991(4) ×4 2.988(3) ×4

Sr – O2 2.614(4) ×4 2.611(2) ×4

Fe1 – O1 1.92(5) ×2 2.02(3) ×2

Fe1 – O2 1.98(2) ×4 2.05(2) ×4

Fe2/Te – O1 2.04(5) ×2 1.94(3) ×2

Fe2/Te – O2 2.01(2) ×4 1.93(2) ×4

77

Table 4.4. Atomic parameters of Sr3Fe2TeO9, refined from neutron diffraction data collected at 3 K.

Atom Site x y z Uiso fractional

occupancy

Phase 1

Sr1 2a 0 0 0.25 0.0003(14) 1.0

Sr2 4d 0.6667 0.3333 0.5824(7) 0.0041(7) 1.0

Fe 4d 0.6667 0.3333 0.3330(5) 0.0025(2) 1.0

Te 2b 0 0 0 0.0025(2) 1.0

O1 6f 0.536(1) 0 0.25 0.003(1) 1.0

O2 12g 0.1944(8) 0.330(1) 0.5830(5) 0.0058(7) 1.0

Phase 2

Sr 4d 0 0.5 0.25 0.0156(8) 1.0

Fe1 2a 0 0 0 0.0037(6) 1.0

Fe2 2b 0 0 0.5 0.0037(6) 0.3333

Te 2b 0 0 0.5 0.0037(6) 0.6667

O1 4e 0 0 0.255(4) 0.0087(6) 1.0

O2 8h 0.291(2) 0.223(2) 0 0.0087(6) 1.0

Rwp = 6.23 %, Rp =4 .69 %, χ2 = 8.033

Phase 1: space group P -3c1, a = 5.5626(3) Å, c = 13.622(1) Å, weight fraction=76 %

Phase 2: space group I 4/m, a = 5.5810(4) Å, c = 7.9045(7) Å, weight fraction=24 %

The field dependence of the molar magnetic susceptibility of our sample of Sr3Fe2TeO9 is shown

in Figure 4.17. The data collected under ZFC and FC conditions overlie at high temperatures but

the ZFC susceptibility has a maximum at 80 K whereas the FC susceptibility has a negative

temperature gradient throughout the measured temperature range. The field dependence of the

molar magnetisation is shown in Figure 4.18. M(H) is linear at 300 K and no hysteresis is

observed. At 150 K M(H) is no longer linear but hysteresis is only observed in the data collected

at 5 K; the function is not symmetrical about the origin.

Figure 4.17. Field dependence of the molar magnetic susceptibility of Sr3Fe2TeO9.

78

Figure 4.18. Magnetic Field dependence of the molar magnetisation of Sr3Fe2TeO9 at (a) 300 and 150 K and (b) 5

K.

4.1.3. Discussion

It is clear that our sample of Sr3Fe2TeO9 differs from those prepared by Djerdi et al, Ivanov et al

and Augsburger et al, and that their samples are different from each other. This is most clearly

shown by a comparison of the temperature dependence of the magnetic susceptibility measured

in each case and, in the case of Augsburger et al, by a comparison of the X-ray diffraction

patterns. The most obvious differences between our synthesis and those performed by Ivanov et

al and Augsburger et al lie in the use of telluric acid, rather than TeO2, by Augsburger et al and

of SrO, rather than SrCO3, by Ivanov et al. Furthermore, both of these groups used an annealing

temperature of only 950 °C. Our choice of 1200 °C was made when heating the reactants at

950 °C failed to produce a perovskite-like phase. The sol-gel protocol used by Djerdi et al ended

with an anneal at 1300 °C. Their cubic sample differs most markedly from our own in terms of

both crystal structure and physical properties. For example, although magnetic order is present in

our sample at 300 K, there is no evidence of a spontaneous magnetisation at that temperature.

Our initial attempts to analyse our X-ray diffraction pattern were influenced by our awareness of

the earlier studies of this compound and our apparent success in accounting for the data using a

monophasic model in tetragonal symmetry will make us more cautious in the future. The results

of our electron-microscopy study demonstrated that our approach was flawed and they also

revealed the full complexity of the system. The electron diffraction patterns, see Figure 4.3,

indicated that the sample has trigonal symmetry and the HAADF-STEM images and EDX maps

shown in Figures 4.4 and 4.5 reveal the presence of the 2:1 Fe:Te ordering pattern shown in

Figure 4.6. This type of cation ordering over the B sites of a perovskite is relatively unusual,

with many A3B2B’O9 compositions, for example all the compounds in the previous chapter of

this thesis, adopting the 1:1 ordering pattern illustrated in Figure 4.1, even though the trigonal

pattern might be considered a better match for their stoichiometry. Although there are some

exceptions, for example Ba3Bi2TeO952

, the latter pattern is most commonly observed in

79

A2+

3B5+

2B’2+

O9 compounds53,54

. Like Sr3Fe2TeO9, these compounds have only divalent cations

on the A site and two types of cation with a charge difference of +3 on the octahedral sites.

Sr3CaRu2O9 and the high-pressure forms of Ba3CaRu2O9 and Ba3CaIr2O9 are the only such

compounds wherein the majority six-coordinate cation is paramagnetic55-57

. The high-pressure

phase of Ba3NaRu2O9 has a different charge distribution over the octahedral sites but also adopts

the 2:1 ordered structure58

. We note that Ba3Fe2TeO9 is a 6H hexagonal perovskite48,49

because,

in the absence of the large Bi3+

cation, the cation size ratio rA/rB is too large to stabilize a pseudo-

cubic perovskite structure.

Our microscopy study revealed that defects are present in the structure of Sr3Fe2TeO9. More

specifically, the electron diffraction patterns indicated that twinning was present on the

nanoscale in some of the crystallites studied and the HAADF-STEM images showed that, in

addition to the dominant 2:1 cation ordering, some crystallites had regions where 1:1 ordering

was present, see Figure 4.8. This ordering is consistent with the tetragonal structure proposed by

Ivanov et al, but also with the monoclinic structure of La3Ni2SbO92. Our data do not allow us to

distinguish unambiguously between the two symmetries so, in the light of the previous studies of

this compound, we assumed these regions to be tetragonal. We were able to optimize the fit to

our XRPD pattern when we included a trigonal phase and a minority tetragonal phase, thus

demonstrating that some of the 1:1 ordered regions are large enough to give rise to Bragg peaks.

Given that our mixed trigonal/tetragonal sample was prepared at a higher temperature than the

tetragonal sample described by Ivanov et al, further syntheses were carried out with week-long

anneals at 950 and 1200 °C in an attempt to prepare monophasic tetragonal and trigonal samples,

respectively. The resulting XRPD patterns always showed the presence of both phases, with a

trigonal:tetragonal ratio that varied between 70:30 and 80:20.

The agreement between the observed and calculated neutron diffraction profiles shown in Figure

3.44 is good. The Fe-O and Te-O bond lengths determined by neutron diffraction, see Table

3.21, are essentially equal in the trigonal phase at 300 K but the Te-O bond is shorter, as

expected59

at 5 K. The difference in bond lengths around the two, six-coordinate sites is

apparently more marked in the minority tetragonal phase, although these distances are less

precisely determined. In addition to the quality of the fit, two additional pieces of evidence show

that the magnetic hyperfine splitting observed in the Mössbauer spectra collected at room

temperature is associated with the trigonal phase. Firstly, the fraction of the total absorption

contained in the sextet is too large to derive from the iron in the minority tetragonal phase and,

secondly, neutron diffraction data collected at room temperature by Ivanov et al showed no

evidence of magnetic ordering. We put more weight on their data than those of Augsburger et al

because the latter draw attention to the presence of one impurity, albeit diamagnetic, in their

sample and we believe that the near-coincidence of the magnetic ordering temperatures of their

Ba3Fe2TeO9 and Sr3Fe2TeO9 samples suggests that the same magnetic impurity might be present

in both. With this assignment, the phase fractions derived from the diffraction and spectroscopic

data are essentially self-consistent. It appears that some, but not all, of the Fe3+

cations in the

80

trigonal phase are antiferromagnetically ordered at room temperature. We attribute the range of

observed hyperfine fields, see Figure 3.42, and the incomplete ordering to the presence of the

defects described above and illustrated in Figure 3.38. The spectrum collected at 5 K shows an

increased magnetisation but when the atomic moment measured by neutron diffraction was fixed

to be consistent with the internal hyperfine field there was no evidence to suggest that

significantly more spins join the antiferromagnetic phase on cooling below room temperature.

The absence of a paramagnetic component in the spectrum collected at 5 K demonstrates that all

the spins in the sample are either ordered or frozen in a spin glass at that temperature. Unlike

Ivanov et al we see no evidence for a magnetic transition in the tetragonal phase at 260 K, see

Figure 3.45, and hysteresis is clearly present in χ(T) below 80 K. We therefore suggest that all

the spins that are disordered at room temperature undergo a transition to a spin-glass state on

cooling below 80 K. If the tetragonal regions in our sample were to adopt a long-range ordered

structure with atomic moments as small as those reported by Ivanov et al, then the combination

of the low phase-fraction and the weak moments would prevent it from being detected in the

neutron diffraction data collected at 3 K. However, there is no evidence in the low-temperature

Mössbauer spectrum for a component with a weak internal hyperfine field and there is no

evidence in the neutron data for a second magnetic component with a strong moment, see Figure

3.45. The participation of the tetragonal regions in the spin-glass transition is thus consistent

with all the available data. The magnetic structure shown in Figure 3.44 satisfies what must be

the very strong superexchange interactions between Fe3+

cations in neighbouring, corner-sharing

octahedra. We cannot offer a detailed discussion of the magnitudes of the ordered magnetic

moments because of the correlations between the fraction of the total iron in the ordered phase

and the magnitude of the moment. We believe this to be the first observation of long-range

magnetic ordering in a trigonal 2:1 cation-ordered perovskite.

The behaviour of M(H) shown in Figure 3.48 is consistent with our model. The absence of

hysteresis at 300 K confirms that the magnetic Bragg scattering observed in the neutron

diffraction pattern comes from a phase that is antiferromagnetic, rather than ferrimagnetic or

ferromagnetic. The non-linearity of the function at 150 K suggests that significant, intercation

interactions are present, as is to be expected at this temperature in an oxide rich in Fe3+

cations,

and the shifted hysteresis loop recorded at 5 K is typical of a spin glass. The observation of this

type of behaviour in pseudo-cubic perovskites with a partially ordered distribution of equal

numbers of two cation species has been seen previously and attributed to the competition

between nearest-neighbour and next-nearest-neighbour interactions60,61

. In the case of the

minority phase in Sr3Fe2TeO9, the small domain size is also likely to be a factor.

81

4.1.4. Conclusion

Sr3Fe2TeO9 is a complex material. The structure and properties are clearly sensitive to the

synthesis conditions. We have shown for the first time that the compound, when synthesized at

1200 °C, adopts a trigonal perovskite-like structure with the Fe3+

and Te6+

cations ordering in a

2:1 sequence that is compatible with both the chemical composition and the multiplicities of the

crystallographically distinct six-coordinate sites in the trigonal space group. This is not the case

in many other A3B2B’O9 compounds, including the relaxor ferromagnet La3Ni2SbO9. The

difference in the cation ordering pattern is the most likely reason that the magnetic behavior of

the two compounds differs so markedly, with the trigonal phase of Sr3Fe2TeO9 being

antiferromagnetic at room temperature. However, the trigonal structure is disrupted by both

nano-twinning and regions where the cations order in a 1:1 sequence. These disruptions prevent

full antiferromagnetic ordering throughout the sample and the unordered spins form a spin-glass

phase that coexists with the antiferromagnetic phase below 80 K.

82

4.2. Ba3Fe2TeO9

The X-ray and neutron powder diffraction and magnetometry investigation of Ba3Fe2TeO9 were

done by the groups of Prof. Dr. P. Battle at Oxford University and Prof. Dr. M. Avdeev at Bragg

Institute, Australian Nuclear Science and Technology Organization, Australia.

4.2.1. Introduction

The structure of a cubic perovskite ABO3 can be considered to consist of pseudo-close-packed

AO3 layers stacked in a cubic-close-packed (ccp) sequence, with B cations occupying six-

coordinate interstitial sites between the layers; the layers lie perpendicular to the [111] axis of the

cubic unit cell. As the size of the cation A increases relative to that of the cation B, the stacking

sequence of the layers changes from ccp to hexagonal-close-packed (hcp), with the layers now

lying perpendicular to the [001] axis of a hexagonal unit cell. BaMnO3 and BaNiO3 exemplify

the hcp structure62,63

. One significant consequence of the change in stacking sequence is that the

connectivity between BO6 octahedra switches from vertex-sharing to face-sharing, thus

introducing short contacts between B cations. The switch from ccp stacking to hcp stacking is

not always complete. Many compounds exhibit mixed stacking sequences, with the so-called 6H

structure, based on the six-layer cchcch sequence being particularly common. This structure

contains B2O9 dimers, formed by face sharing BO6 octahedra, which are linked together by

single, vertex-sharing BO6 octahedra sites. The six-coordinate sites in many perovskites, both

pseudo-cubic and hexagonal, are occupied by more than one species of cation, leading to the use

of formulae of the form A2BB’O6 and A3B2B’O9 in cases where the distribution of cations is non-

random and a particular cation species has a preference for one of the crystallographically-

distinct sites. This cation ordering can have a marked effect on the physical properties of the

compound.

The variation of cation radius down Group 2 of the Periodic Table is such that it is common for a

strontium-containing perovskite to be pseudo-cubic and for the barium-containing analogue to

have a mixed or hexagonal stacking sequence60,64

. The two compounds are often reported

together in the literature, with attention being focused on the structural change and the resulting

differences in physical properties. Sr3Fe2TeO9 (previous section) and Ba3Fe2TeO9 constitute one

such pair of compounds49,50

. Our study on Sr3Fe2TeO9 revealed a hitherto undetected complexity

in the structural chemistry, including the occurrence of unusual 2:1 cation ordering in the

pseudo-cubic structure. The earliest study65

of Ba3Fe2TeO9 identified a 6H structure with Fe3+

in

the vertex- sharing octahedra and a disordered distribution of Fe3+

and Te6+

in the dimers. The

compound was found to show weak ferromagnetism below 220 K. Gagulin et al66

carried out the

first neutron diffraction study of Ba3Fe2TeO9 and concluded that it is a Seignette-magnet, the

name given to multiferroic materials when they were last the focus of research interest.

Augsburger et al49

found evidence for cation disorder in their sample, which they found to be

83

magnetic below 711 K. In contrast, the sample prepared by Djerdj et al51

was paramagnetic

above 100 K and showed a magnetic susceptibility maximum at 20 K.

There is clearly disagreement over the magnetic properties of Ba3Fe2TeO9. We have therefore

undertaken a study in which low-temperature neutron diffraction and Mössbauer spectroscopy

have been used for the first time to resolve some of the controversial issues. We have also used

high-resolution electron microscopy to look more closely at the distribution of the cations

occupying the dimers within the 6H structure.

Synthesis

The synthesis was performed by the group of Prof. Dr. Peter Battle at the University of Oxford.

A polycrystalline sample of Ba3Fe2TeO9 was prepared using the standard ceramic method.

BaCO3, Fe2O3 and TeO2 (purity >99.95%) were weighed out in the appropriate stoichiometric

ratio and ground together in an agate mortar for 30 minutes to give a homogeneous mixture. The

mixture was then loaded into an alumina crucible and fired at 700 °C for 24 h. It was then

quenched to room temperature, reground and pressed into a pellet that was fired in air at 950 °C

for 24 hours and subsequently annealed at 1200 °C for 48 hours after further grinding. Finally,

the furnace was allowed to cool to 800 °C and then the sample was quenched to room

temperature.

4.2.2. Experimental results

The XRPD pattern of the reaction product suggested that the synthesis had produced a phase-

pure 6H hexagonal perovskite, i.e impurities were not found in the refinement. Influenced by the

earlier work of Augsburger et al 49

, we first indexed and analysed the pattern in the hexagonal

space group P63/mmc. Reasonably good agreement between the observed and calculated

diffraction patterns was achieved with Rwpr=5.40 %, χ2=1.85, see Figure 4.19.

84

Figure 4.19. Observed (red), calculated (green) XRPD patterns of Ba3Fe2TeO9 at room temperature. A difference

curve (purple) is shown and reflection positions are marked.

The unit-cell parameters refined to values of a = 5.76882(2) Å, c = 14.19859(6) Å. This

structural model, see Figure 4.20 (a), contains two independent A sites, 2b (0, 0, ¼) and 4f (⅓, ⅔,

z) which accommodate Ba2+

cations and two B sites, 2a (0, 0, 0) and 4f (⅓, ⅔, z), for Fe3+

and

Te6+

cations, respectively. The 2a B sites lie at the centre of the vertex-sharing octahedra and the

4f B sites lie within the M2O9 dimers. The shared face of the dimers is formed by three O1 ions

on 6h (x, 2x, ¼) sites and the O2 ions on 12k (x, 2x, z) sites connect the dimers to the vertex-

sharing octahedra. Refinements of the distribution of Fe and Te over the crystallographically-

distinct B sites showed that the 2a (0,0,0) sites are largely occupied by Fe (green) and that Fe/Te

(khaki) are randomly distributed over the 4f (1/3, 2/3, z) sites. The refined cation distribution can

be represented by the formula Ba3(Fe0.92Te0.08)2a(Fe0.53Te0.47)4fO9.

Structure refinements were also carried out in the non-centrosymmetric space group P63mc, in

which the two cation sites within the M2O9 dimers are inequivalent. The reflection conditions of

this group are the same as those of P63/mmc and the agreement factors resulting from the

analyses in the former, Rwpr=5.53 %, χ2=1.94, were only slightly worse than those achieved in

centrosymmetric P63/mmc. The essential difference between the two models is that in P63/mmc

the sites in the dimer are occupied in a disordered manner whereas in P63mc the cations adopt an

ordered, or partially ordered, distribution over these sites, this difference between both space

groups are shown in Figure 4.20 (a) and (b).

85

Figure 4.20.(a) P63/mmc and (b) P63mc structure of Ba3Fe2TeO9. Green octahedra are occupied largely by Fe, khaki

octahedra are occupied by a disordered arrangement of Fe and Te and blue octahedra are occupied largely by Te.

Mauve circles represent Ba atoms.

HAADF-STEM was used to figure out the correct space group. Figure 4.21 shows a HAADF-

STEM image taken along the [100] zone axis. The calculated images for both models are

compared to the experimental image in Figure 4.21(a). The brightness of the dots in this type of

image increases with the average atomic number along the column of atoms that the dot

represents. In Figure 4.21(a) the brightest dots are the projected Ba columns (Z=56), the weaker

ones the columns containing Fe and Te (resp. Z=26 and 52). The difference between the two

calculated images lies in the columns of B cations occupying the sites within the dimers. One

such pair, along with a neighbouring Ba column, is encircled on the experimental image, on each

of the calculated images and in Figure 4.22(b).

Figure 4.21. (a) Aberration-corrected high angle annular dark field scanning transmission electron microscopy

(HAADF-STEM) image of Ba3Fe2TeO9 along the [100] zone axis and calculated HAADF-STEM images (b) The

cation arrangement within the structure in space group P63/mmc, viewed along [100]. Green, khaki and mauve

circles represent Fe (2a site), disordered Te/Fe (4f site) and Ba, respectively.

86

In P63/mmc these two sites are equivalent because of the mirror plane perpendicular to the c-

axis, whereas in P63mc they are independent positions (2b of P63mc). Analysis of the X-ray

diffraction data in P63mc results in these 2b positions taking on different Fe:Te ratios which can

be seen in the calculated image as a systematic brightness difference between the two columns;

this would also be apparent in experimental images taken from thicker areas of the crystal. In

P63/mmc, the two columns have the same brightness in the calculated image and would be

expected to do so in the thicker areas of the experimental image; only in very thin areas will the

random occupation of the 4f site by either Te or Fe give a difference in brightness between the

two columns. In the experimental image shown in Figure 4.21 (a) the columns in the thick

regions away from the edge of the crystallite have equal brightness and thus match the image

calculated for P63/mmc.

The difference in the Fe:Te ratio on the crystallographically distinct 2a and 4f B positions of

P63/mmc is also clear in the HAADF-STEM images. The white dots representing columns of Fe

and Te in Figure 4.21 (a) come in groups of three and the middle one is always the darkest. One

such group is identified by the rectangle drawn in Figure 4.21 (b). It can be seen that the outer

two dots correspond to the 4f columns and the darker middle dot corresponds to the 2a column.

The 2a column is thus Fe-rich, in agreement with the cation distribution determined by X-ray

powder diffraction.

In the light of the electron-microscopy results, the neutron diffraction data collected at 300 K

were analysed in the space group P63/mmc. The refined atomic parameters and the agreement

factors are listed in Table 4.5 and the most significant bond lengths and bond angles are

presented in Table 4.6. The cation distribution over the sites within the dimers is refined to be

Ba3(Fe0.93Te0.07)2a(Fe0.53Te0.47)4fO9, which is in good agreement with that determined by XRPD.

The resulting agreement between the observed and calculated profiles can be seen in Figure 4.22;

this fit was obtained without assigning ordered magnetic moments to any of the cations.

Table 4.5. Atomic parameters of Ba3Fe2TeO9, refined from neutron diffraction data collected at 300 K.

Atom Site x y z Uiso fractional

occupancy

Ba1 2b 0 0 0.25 0.0050(5) 1.0

Ba2 4f ⅓ ⅔ 0.0919(1) 0.0062(4) 1.0

Fe1 2a 0 0 0 0.0065(5) 0.933(16)

Fe2 4f ⅓ ⅔ 0.85236(9) 0.0088(4) 0.529(8)

Te1 2a 0 0 0 0.0065(5) 0.067(16)

Te2 4f ⅓ ⅔ 0.85236(9) 0.0088(4) 0.471(8)

O1 6h 0.5153(2) 0.0306(3) 0.25 0.0111(4) 1.0

O2 12k 0.8314(2) 0.6628(3) 0.08198(7) 0.0093(2) 1.0

Space group P 63/mmc, a = 5.76622(6) Å, c = 14.1966(2) Å, Rwp = 4.62 %, Rp = 3.69 %, χ2= 3.754

87

Table 4.6.Selected interatomic distances (Å) and bond angles (degrees) in Ba3Fe2TeO9 at 300 K and 3K

300 K 3 K

Ba1 – O1 2.88715(9) ×6 2.88095(7) ×6

Ba1 – O2 2.9199(13) ×6 2.9126(9) ×6

Ba2 – O1 2.8878(18) ×3 2.8760(14) ×3

Ba2 – O2 2.88663(11) ×6 2.88071(7) ×6

2.9666(19) ×3 2.9656(14) ×3

Fe1/Te1 – O2 2.0470(13) ×6 2.0464(10) ×6

Fe2/Te2 – O1 2.0969(15) ×3 2.0932(11) ×3

Fe2/Te2 – O2 1.8909(16) ×3 1.8851(12) ×3

O1 – Fe2/Te2 – O1 77.27(6) 77.49(5)

O2 – Fe1/Te1 – O2 89.13(5) 89.19(4)

90.87(5) 90.81(4)

O2 – Fe2/Te2 – O2 97.78(6) 97.74(4)

O1 – Fe2/Te2 – O2 165.67(7) 165.85(6)

91.606(34) 91.541(25)

Figure 4.22. Observed (red), calculated (green) neutron diffraction patterns of Ba3Fe2TeO9 at room temperature. A

difference curve (purple) is shown and reflection positions are marked.

The Mössbauer spectrum obtained on an extended velocity scale and shown in Figure 4.23 (a)

confirms that all the iron is trivalent Fe3+

and paramagnetic at room temperature. In Figure 4.23

(b) we show a second spectrum collected on a narrower velocity scale at room temperature. It is

clear that at least three quadrupole-split doublets are required to fit this higher resolution

spectrum and in Table 4.7 we give the fitted Mössbauer parameters for these doublets. The

isomer shifts of these components all lie in a narrow range from 0.37 to 0.45 mm/s, indicating

that the oxidation state of Fe is 3+.

88

Table 4.7. Mössbauer parameters of Ba3Fe2TeO9, determined by fitting the higher-resolution spectrum obtained at

295 K (shown in Figure 4.23 (b)).

Spectral

Component δ (mm/s) | Δ | (mm/s)

HWHM

(mm/s)

Area

(%)

Doublet 1 0.376(6) 0.640(4) 0.31(2) 68.8(5)

Doublet 2 0.442(6) 1.596(8) 0.32(2) 17.6(4)

Doublet 3 0.447(6) 0.973(7) 0.27(2) 13.6(3)

δ is the isomer shift relative to α-Fe; |Δ| is the magnitude of the quadrupole splitting; HWHM is the line half-width

at half maximum intensity; area is the relative subspectral contribution to the total spectrum.

Figure 4.23. (a)

57Fe Mössbauer spectrum of Ba3Fe2TeO9 acquired at room temperature on an extended velocity

scale; (b) 57

Fe Mössbauer spectrum of Ba3Fe2TeO9 at room temperature on a narrower velocity scale. The three

fitted quadrupole doublets are shown in colour and correspond to the parameters given in Table 4.7.

The temperature dependence of the molar magnetic susceptibility of our sample of Ba3Fe2TeO9

is shown in Figure 4.24 (a). The ZFC and FC magnetic susceptibilities are equal above 250 K

and the difference between them is small in the temperature range 90 < T/K <250. They differ

more obviously below ~90 K as they increase rapidly to a maximum at 18 K. A Curie-Weiss fit

to the temperature region 250 <T/K < 300 gives θ = -167 K, and µeff = 6.74 µB per Fe3+

. The

field dependence of the molar magnetisation at 5 K is shown in Figure 4.24 (b); M(H) is

nonlinear and a weak hysteresis is observed in low magnetic fields, as can be seen more clearly

in Figure 4.24(c).

89

Figure 4.24. (a) Temperature dependence of the molar magnetic susceptibility of Ba3Fe2TeO9; (b) Magnetic field

dependence of the magnetisation of Ba3Fe2TeO9 at 5 K for -50 < H/kOe < 50; (c) magnetic field dependence of the

magnetisation of Ba3Fe2TeO9 at 5 K for -5< H/kOe< 5.

90

The neutron diffraction data patterns collected at 3 K using wavelengths of 1.6215 Å and

2.4395Å were analysed simultaneously using the structural model described above. Refinement

of the atomic coordinates and atomic displacement did not give an entirely satisfactory account

of these data. The discrepancies between the observed and calculated diffraction profiles were

most noticeable at low angles and were therefore assumed to be magnetic in origin. The Bragg

scattering associated with the antiferromagnetic structure illustrated in Figure 4.25 accounted for

the intensity mismatch, see Figure 4.25.

Figure 4.25. Magnetic structure of Ba3Fe2TeO9; arrows indicate the direction of ordered spins within each

octahedron. Green octahedra are occupied largely by Fe and khaki octahedra are occupied by a disordered

arrangement of Fe and Te. Mauve circles represent Ba atoms.

Figure 4.26. Observed (red), calculated (green) neutron diffraction patterns of Ba3Fe2TeO9 at 3 K (λ = 1.6215 Å). A

difference curve (purple) is shown and reflection positions are marked.

In this structure, the magnetic moment of each Fe3+

cation is coupled in an antiparallel manner to

those on the nearest-neighbour sites, as a consequence the magnetic moments in the pairs of

face-shared octahedra are antiparallel. This direction along which the spin vectors align could

91

not be determined unambiguously from the neutron data alone. However, the Mössbauer

spectrum collected at 13 K, see Figure 4.27, allowed this issue to be resolved.

Figure 4.27.

57Fe Mössbauer spectrum of Ba3Fe2TeO9 acquired at 13 K and fitted with a single sextet.

The Mössbauer spectrum at 13 K shows clearly that none of the Fe3+

cations in Ba3Fe2TeO9 are

paramagnetic at low temperature. Furthermore, all Fe3+

cations experience essentially the same

hyperfine magnetic field. A single-sextet fit to the 13 K spectrum yields a hyperfine magnetic

field of 47.8(2) T with an isomer shift of 0.514(5) mm/s, significantly larger than the weighted

isomer shift of 0.372(5) mm/s measured at 295 K, reflecting the second-order Doppler shift.

Analyses of the neutron data in which the atomic moments on the 2a and 4f sites were allowed to

vary independently resulted in similar values (3.40(4) and 3.40(6) µB per Fe3+

cation,

respectively) at the two sites only when the spins were aligned along [001] fully consistent with

our Mössbauer result. Table 4.8 lists the refined atomic parameters and agreement factors at 3 K,

and selected bond lengths are included in Table 4.6.

Table 4.8. Atomic parameters of Ba3Fe2TeO9, refined from neutron diffraction data collected at 3 K.

Atom Site x y z Uiso fractional

occupancy

Ba1 2b 0 0 0.25 0.0001(5) 1.0

Ba2 4f ⅓ ⅔ 0.09222(9) 0.0006(4) 1.0

Fe1 2a 0 0 0 0.0013(3) 0.933

Fe2 4f ⅓ ⅔ 0.85213(7) 0.0065(3) 0.529

Te1 2a 0 0 0 0.0013(3) 0.067

Te2 4f ⅓ ⅔ 0.85213(7) 0.0065(3) 0.471

O1 6h 0.5149(1) 0.0298(3) 0.25 0.0074(3) 1.0

O2 12k 0.8312(1) 0.6623(2) 0.08220(5) 0.0053(2) 1.0

Space group P63/mmc, a = 5.75427(4) Å, c = 14.1675(1) Å

Rwp = 5.30 %, Rp = 4.12 %, χ2= 4.884

92

4.2.3. Discussion and conclusion

The Ba3Fe2TeO9 compound adopts a hexagonal structure with space group P63/mmc and a =

5.75427(4) Å, c = 14.1675(1) Å, with partial ordering of the of Fe3+

and Te6+

cations over the

six-coordinate sites. It is antiferromagnetically ordered at low temperature (below ~18 K), and

paramagnetic at room temperature.

In comparison with Sr3Fe2TeO9 we see the expected difference in the structure upon changing

the A cation from Sr to Ba, i.e. a change to the hexagonal stacking sequence. Our structure

agrees with the structure found by Harari et al65

, with Fe3+

in the vertex- sharing octahedra and a

disordered distribution of Fe3+

and Te6+

in the dimers, so partially ordered, in contrast to the

disorder reported by Augsburger et al49

.

Contrary to Harari et al. we found no ferromagnetism but antiferromagnetic order at low

temperature. The room temperature order and magnetic susceptibility maximum at 18 K are

similar to what was reported by Djerdj et al51

.

93

Chapter 5

In search of a Jahn-Teller distorted Cr(II) oxide with Ba3Cr2TeO9

The content of this section was published in the paper Ba3(Cr0.97(1)Te0.03(1))2TeO9: in Search of Jahn-Teller

distorted Cr(II) Oxide, Man-Rong Li, Zheng Deng, Saul H. Lapidus, Peter W. Stephens, Carlo U. Segre,

Mark Croft, Robert Paria Sena, Sun Woo Kim, Joke Hadermann, David Walker, Martha Greenblatt,

Inorg. Chem, 55(20), 10135-1014, 2016 Oct 17.

The X-ray and neutron powder diffraction and magnetometry investigation of the SrLa2Ni2TeO9

were done by the group of Prof. Dr. M. Greenblatt at Rutgers University.

5.1. Introduction

The compound Ba3Cr2TeO9 was actually investigated with another goal than the previous

compounds, i.e. not in search of new relaxor ferromagnets and of knowledge on the link between

their structure and properties, but in search of a Jahn-Teller distorted Cr(II) oxide. However, I

have decided to include this compound in the thesis due to its large similarities with the previous

compound Ba3Fe2TeO9.

The Jahn-Teller (J-T) distortion effect of octahedrally-coordinated high spin (HS) d4, low spin

(LS) d7, or d

9 configuration cations usually induces important properties in transition metal

oxides,67-70

such as colossal magnetoresistance (CMR)71-74

and high-temperature

superconductivity75-78

which derive from the strong coupling between static or dynamic charge,

orbital and magnetic interactions. Strong J-T distortion in Mn3+

(HS, d4, t2g

3eg

1) and Cu

2+ (d

9,

t2g6eg

3) has been extensively studied in CMR manganites

71-74 and high-temperature

superconductor cuprates, respectively.74-78

In contrast, the J-T distortion of intermediate spin (IS)

Co3+

and LS Ni3+

(d7, t2g

6eg

1) in oxides is not common,

79 in some cases they appear as dopants in

other host lattices.79-81

A unique local static J-T distortion was observed in LaCo3+

O3 perovskite

since the Co3+

intermediate-spin (IS, t2g5eg

1) state can be thermally excited from the ground-state

LS configuration (t2g6eg

0)70,82,83

. This IS state can be further stabilized by carrier doping in La1-

xSrxCoO3 which is compatible with a J-T glass state.82

To the best of our knowledge, there is no

J-T distortion of HS octahedral Cr2+

(d4) reported in oxides to date, but it has been observed in

the hydrated state in solution84

, or inorganic-organic hybrid compounds85

. Recently, the J-T-

active Cr2+

has also been stabilized in perovskite fluoride KCrF386

and rock-salt superstructure

sulfide Lu2CrS4.87

Double perovskites A2CrWO6 (A = Ca, Sr, Ba) seem to be possible to host

Cr2+

O6 since W6+

is highly favorable at octahedral sites of perovskites and related compounds,88-

90 even in reducing conditions

91. However, it appears to be Cr

3+/W

5+ over competition with the

highly unstable Cr2+

/W6+

in A2CrWO6.92, 93

Mixed valence Cr2+

/Cr3+

state has been claimed in

94

LaCr0.9Ti0.1O3,95

but the 10% (by mass) La2Ti2O7 second phase makes the proposed composition

of the main phase in doubt. Moreover, the average <(Cr0.9Ti0.1)-O> distance (1.972(3) Å) is

almost identical with the average <Cr-O> value (1.974(2) Å in LaCrO3 at room temperature,

which is unlikely to confirm the presence of any Cr2+

state considering the ionic radius

difference between Cr2+

(HS, 0.80 Å) and Cr3+

(0.615 Å) in octahedral coordination.96

These

findings suggest that the J-T distortion of octahedral Cr2+

is significantly more difficult to

stabilize in oxides than those of octahedral Mn3+

, Co2+/3+

, Cu2+

, or Ni3+

.

J-T distortion-induced low-dimensional A2BB’O6 double perovskites are of great interest but

they are rare,97-99

only a few compounds have been prepared to date, such as the layered

Ln2CuMO6 (Ln = La, Pr, Nd, Sm; M = Sn, Zr)12,100,11

and the quasi-two-dimensional Sr2Cu(Mo1-

xWx)O6101

. La2CuSnO6 can be prepared at ambient pressure, while the other Ln2CuMO6

compounds can only be stabilized at 6-8 GPa. In the monoclinic Ln2CuMO6 structure (P21/m)

with alternative CuO6 and MO6 octahedral layers, the J-T distortion of Cu2+

gives in-plane Cu-O

bond lengths between 1.93(4) and 2.06(4) Å and out-of-plane Cu-O distances of 2.22(3)-2.39(3)

Å. Although the CuO2 layers are similar to the high-temperature cuprate superconductors,

Ln2CuMO6 are not superconducting. A possible way to achieve superconductivity is to flatten

the buckling CuO2 layer with electron doping102

. The tetragonal Sr2Cu(Mo1-xWx)O6 family,

prepared at ambient pressure for 0 ≤ x ≤ 0.6 and high pressure (~4GPa) for 0.7 ≤ x ≤1.0, exhibits

a quasi-two-dimensional S = 1/2 square lattice with possible magnetic frustration. The tetragonal

distortion decreases with increasing x in Sr2Cu(Mo1-xWx)O6, accompanied by lowering the first-

order tetragonal-cubic cooperative J-T ordering temperature. These studies suggest that it might

be possible to stabilize the B-site J-T distortion of A2BB’O6 double perovskite, such as Cr2+

O6 in

A2Cr2+

B’O6 at high pressure. Compared with the A2Cr3+

W5+

O6 (A = Ca, Sr, Ba) series,92-94

the

B-site octahedral Cr2+

O6 might be more stable in A2Cr2+

Te6+

O6, since the Te6+

, once formed in

A2Cr2+

Te6+

O6, would be unlikely to be dynamically reduced to Te4+

, considering the difference

of charge, size, and electron structure between octahedral Te6+

(ionic radius r = 0.56 Å) and Te4+

(r = 0.97 Å with 5s2 lone-pair electron). When the A-site is occupied by large cations in the

perovskites, hexagonal structures may be adopted as exemplified by the well-known six-layered

(6H) perovskites such as the 6H-BaTiO3.103

The general formula of these perovskites can be

written as A3B2B’O9 with two types of 2:1 B-site cation ordering: (i) The BO6 face-sharing

octahedral dimers connect with B’O6 octahedra via shared corners, which crystallize in P63/mmc

(No. 194), such as Ba3Cr2MoO9,104

Ba3Ru2MO9 (M = In, Y),105,106

or in P-62c (No. 190), as

Ba3Cr2WO9.104

Both P63/mmc- and P-62c-type structures have similar polyhedral stacking-the

only difference is between the oxygen sites, i.e., 12k in P63/mmc, and 12i in 𝑃62𝑐 (Figure 5.1

(a)); (ii) the atoms are stacked the same way as in (i), but the B-sites are ordered giving face-

sharing B1O6-B2O6 octahedral dimers (Figure 5.1 (b), space group of P63mc (No. 186)) as in

Ba3Ti2IrO9.107

In this work we obtained a new close-packed 6H-hexagonal perovskite

Ba3(Cr0.97(1)Te0.03(1))2TeO9 in attempts to prepare Ba2CrTeO6 double perovskite with J-T-

distorted Cr2+

and report the high-pressure and high-temperature synthesis, the determination of

the crystal structure, formal oxidation state of cations, and the magnetic properties of this

unusual new phase.

95

Figure 5.1 Comparison of hexagonal perovskite structures of the 6H-perovskites crystalized in (a) P63/mmc or P-

62c and (b) P63mc. In (a), there are two A-sites [A1 at 2b (0, 0, 1/4), A2 at 4f (1/3, 2/3, z)], one face-sharing BO6

octahedral site at 4f (1/3, 2/3, z) connected with the B’O6 octahedral site at 2a (0, 0, 0) via corner sharing. For

P63/mmc, and P-62c, the cations are at the same sites, the only difference is the oxygen sites: two oxygen sites on

mirror planes at 6h and 12k for P63/mmc, and on a mirror plane and a general position at 6h and 12i for P-62c,

respectively; while in (b) the structure stacking is the same but with ordered B1 and B2.

Synthesis

The synthesis was done by the group of Prof. Dr. Martha Greenblatt at Rutgers, the State

University of New Jersey. The Ba3(Cr0.97(1)Te0.03(1))2TeO9 compound was obtained in an attempt

to make Ba2CrTeO6 via a two-step solid state reaction. First, the Ba2TeO5 precursor was

synthesized using BaCO3 (99.98 %, Sigma Aldrich) and TeO2 (99.995 %, Alfa Aesar) as

previously reported.108

Then the mixture of Ba2TeO5, Cr powder (99.99 %, Alfa Aesar), and

Cr2O3(99.97 %, Alfa Aesar) (atomic ratio of Ba/Cr/Te = 2:1:1) was heated at 1773 K and 6 GPa

for 4 h in a LaCrO3 heater lined with Ir-capsule inside a MgO crucible in a Walker-type multi-

anvil press,109

and then quenched to room temperature (RT) by turning off the voltage supply to

the resistance furnace. The pressure was maintained during the temperature quenching and then

released slowly in 8-12 h.

5.2. Solution and refinement of the crystal structure

Microprobe analysis on the as-prepared sample shows a Cr-enriched, Ba-Te-depleted phase with

composition Ba2.14(1)Cr1.18(2)Te0.82(2)Ox and some Cr-free phases: BaTeOx and Ba2TeOx (Figure

5.2). XRPD data of the as-prepared light-brown phase can be well-indexed with a hexagonal

phase (a ≈ 5.72 Å, c ≈ 14.05 Å) and trace unknown impurity, which washes away in dilute HCl

acid solution giving the pure phase shown in Figure 5.3. All the measurements and

characterization in this work subsequent to microprobe analysis were performed on the pure

washed phase. EDX analysis of this washed pure phase gives an average formula of

Ba2.00(9)Cr1.24(5)Te0.68(4)Ox, indicating Cr-rich B/B’-sites and some minor uncertainty about the

composition compared to the microprobe result. The BaTeOx and Ba2TeOx phases seen by the

96

microprobe are washed away by dilute HCl acid and have little XRPD signature, suggesting that

they may be amorphous.

Figure 5.2. Atomic proportions of cations in three Ba-Cr-Te oxides studied by electron microprobe analysis of

phases found in as-prepared synthesis experiment. The major phase (title compound) compares reasonably well with

compositions found by EDX of the acid-washed sample and composition assigned by structural refinement by

SXRPD of the washed sample. To a first order, the title composition is the target Ba2CrTeO6 depleted by about 20%

in the BaTeOx compound found coexisting in the same sample, with little mass balance contribution needed from

the rare Ba2TeOx.

Figure 5.3. Comparison of the XRPD patterns of the as-made (bottom) and washed (top) Ba2CrTeO6 before,

showing that the impurity phase in the as-made sample (highlighted by arrow) can be washed away by diluted HCl

solution.

10 20 30 40 50 60

Inte

nsit

y (

a.

u.)

2Cu K

Washed

As-made

97

Figure 5.4. Precession electron diffraction patterns of the title compound.

Electron diffraction patterns (Figure 3.61) show as only reflection conditions hhl: l = 2n and 00l:

l = 2n (corresponding to hh-2hl:l = 2n, 000l: l= 2n in 4-digit indexation), leading to extinction

symbol P--c. This leaves as possible space groups P63mc (No. 186),107

𝑃62𝑐, (No. 190),104

or

P63/mmc (No. 194),102-106

.

Figure 5.5 (a), (b) shows the HAADF-STEM images in [100] and [111] zones, respectively. As

explained in the introduction, the difference between the P63/mmc and P-62c models is only a

slight difference in oxygen positions. Pure oxygen columns are not visible on HAADF-STEM

images because their Z is too low compared to the other elements that are present. ABF-STEM

could be taken to visualize the oxygen, however, the difference is too small to see anyway, so for

discrimination between those two, simulated annealing was applied below. On the other hand,

the images do allow to eliminate the P63mc model (Figure 5.1 (b)), i.e. there is no agreement

between the experimental images and the images calculated with the P63mc model. In the [100]

zone image (Figure 5.5(a)), for example, the P63mc model will result in a different brightness for

the neighboring Cr and (Cr + Te) columns (indicated by the white circles), while for P63/mmc

these columns will have the same occupation and thus the same brightness. On the experimental

image, these columns indeed have the same brightness. In the [111] zone (Figure 5.5 (b)) the

projected Cr/Te columns (2×2 columns) are also indicated. In the experimental image, these four

columns are again identical, in agreement with the calculated image for the P63/mmc model

where the four columns all contain the same amount Cr and Te but in contradiction with the

P63mc model where the ordering between Cr and Te causes two projected atom columns to be

dark and seemingly absent (purely Cr columns), and two bright (containing Cr and Te).

98

Figure 5.5. HAADF-STEM images of the washed sample along (a) [100] and (b) [111] zones. Neighboring Cr and

(Cr + Te) columns are highlighted by circles. Calculated images are indicated by a rectangular white border.

In [120] and [130] a better fit with P63/mmc than with P63mc can also be seen, but there the

difference is less obvious as shown in Figure 5.6.

Figure 5.6. HAADF-STEM images of the washed sample in (a) [120] and (b) [130] zones. Calculated images using

the P63mc and P63/mmc models are shown as insets.

Simulated annealing in P63mc also gave very poor results, but the other two space groups were

satisfactory. Candidate solutions in the other two space groups agree on the metal sites; the

difference is that 𝑃62𝑐, has two oxygen sites in a general position, whereas P63/mmc has the

corresponding sites on a mirror plane (Figure 5.6). The quality of refinements in both space

groups is identical, and in 𝑃62𝑐, the oxygen in a general position refines to the mirror plane

within statistical uncertainty.

Refinements of the synchrotron x-ray powder diffraction (SXPD) data (Figure 5.7) in P63/mmc

yields a structural formula of Ba3(Cr0.969(5)Te0.031(5))2TeO9 (BCTO), that is, 3.1 ± 0.5% (standard

uncertainty) replacement of Cr by Te, in reasonable agreement with the EDX and microprobe

99

results (Figure 5.2). The standard uncertainties in Rietveld refinements are derived from the

propagation of counting statistics through the least-squares minimization of residuals and are

generally several times smaller than any realistic estimate of the accuracy of the derived

parameter. This is typically a consequence of systematic errors in the crystallographic or line-

shape models. In the present case, the refined χ2 of 1.414 is much lower than generally

encountered, indicating that statistical fluctuations of detected X-ray counts are more significant

than is customarily the case. In turn, this suggests that the standard uncertainty should be taken

seriously, and there truly is a measurable occupancy of Te in the Cr site. This result is consistent

with the EDX results above and X-ray spectroscopy results discussed below.

Figure 5.7. Rietveld refinements of the SXRPD data for BCTO in P63/mmc structure at RT. Note the square-root

intensity scale in the data and model, and that the difference plot is scaled to statistical uncertainty of the data. Tick

marks indicate the positions of allowed target phase and boron nitride (BN) peaks on top and bottom, respectively.

(inset) The crystal structure viewed along [110] direction. Ba atoms are shown in blue spheres; O, violet spheres;

TeO6 ochatedra, light yellow; (Cr0.97(1)Te0.03(1))O6 octahedra, dark blue.

The two crystallographically independent Ba1 and Ba2 are located at 2b (0, 0, 1/4) and 4f (1/3,

2/3, z), Te1 at 2a (0, 0, 0), the mixed (Cr/Te)2 at 4f (1/3, 2/3, z), and two oxygen sites at 6h (x, -

x, 1/4) and 12k (x, -x, z), respectively. Table 5.1 lists the detailed crystallographic parameters

and reliability factors. The relatively large R factors do not indicate a deficiency of the

crystallographic model; rather, they are a consequence of the weak scattering from the small

diluted sample.

Table 5.1. Structural parameters of BCTO refined from the SXPD data collected at RT.a

Atom Site Occ. x y z Uiso (Å2)

b

Ba1 2b 1 0 0 1/4 0.45(2)

Ba2 4f 1 1/3 2/3 0.09974(6) 0.71(2)

Te1 2a 1 0 0 0 0.47(3)

(Cr/Te)2 4f 0.969(1)/0.031(1) 1/3 2/3 0.6622(14) 0.72(5)

O1 6h 1 0.5137(7) -0.5137(7) 1/4 0.13(14)

O2 12k 1 0.8404(5) -0.8404(5) 0.0770(3) 0.31(10) aSpace group P63/mmc (194), Z = 2, a = 5.7249(1) Å, c = 14.0488(1) Å, V = 398.75(1) Å

3, Rp/Rwp = 8.15/10.02%,

χ2 = 1.50.

bThe Uiso values are multiplied by 100.

100

The difference curve in Figure 5.7 shows that statistical noise dominates, and the remaining

features are associated with BN peaks, not the BCTO. As shown in the inset of Figure 5.7, the

crystal structure of BCTO is isostructural with a series of 6H polymorph perovskites, such as

BaRuO3,110

BaCrO3,111

and Ba3Ru2MO6 (M = In, Co, Ni, Fe, Y, La, Sm, Eu, and Lu),105,106

and

also Ba3Fe2TeO9 studied in the previous section of this thesis. It consists of face-shared

(Cr0.969(5)Te0.031(5))O6 octahedral pairs interconnected by corner-sharing TeO6 octahedra to form

the framework. Table 5.2 presents the selected interatomic distances. The average metal-oxygen

distances around the 12-coordinated Ba1 and Ba2 are 2.875(4) and 2.891(5) Å, respectively,

comparable with those in the high-pressure made 6H BaRuO3 (2.884(4) and 2.883(4) Å).110

The

⟨Te1−O⟩ distance (1.927(5) Å) is also in line with the observed values (∼1.92-1.96 Å) in other

Te(VI)-containing double perovskites.112,113

However, the average ⟨(Cr0.97Te0.03)-O⟩ distance

(2.020(5) Å) is somewhat longer than the Cr-O distances in Ba2Cr3+

M5+

O6, such as 1.986(2) Å

for M = Nb114

and 2.006(7) Å for M = Ta,115

indicating the possible presence of Cr2+

considering

the size of octahedral site Cr2+

and Cr3+

(r(Cr2+

) = 0.80 Å (HS)), r(Cr3+

) = 0.615 Å).96

Assigning

the Ba2+

and Te6+

formal oxidation states, from bond valence sums (BVS, Table 2) calculations

in BCTO, Cr displays mixed valence of Cr2+

/Cr3+

with at least 10% of the Cr2+

state according to

charge balance, or possibly more if there is any oxygen defect, which cannot be determined from

X-ray diffraction. Octahedral HS Cr2+

is rarely observed in oxides, and our attempt to prepare

Ba2Cr2+

Te6+

O6 at higher pressure and temperature was unsuccessful but yielded Cr-rich BCTO,

presumably with depletion of Te6+

to form Ba3(Cr2+

0.10(1)Cr3+

0.87(1)Te6+

0.03)2TeO9 in the isolated

reaction system. This off-stoichiometry is probably responsible for the small impurity, which is

dissolvable in dilute acid (Figure 5.3).

Table 5.2. Selected interatomic distances (Å), bond valence sums (BVS), octahedral distortion parameters (Δ), and

bond angles (º) in BCTO at RT.

Ba1O12 Ba2O12

Ba1-O1 × 6 2.866(3) Ba2-O1 × 3 2.767(2)

-O2 × 6 2.901(4) … -O2 × 6 2.881(3)

<Ba1-O> 2.884(4) … -O2 × 3 3.022(4)

BVS 2.38 <Ba2-O> 2.890(3)

BVS 2.43

Te1O6 (Cr/Te)2O6

Te1-O2 × 6 1.917(2) (Cr/Te)2-O1 × 3 1.955(3)

BVS 6 -O2 × 3 2.098(4)

ΔTe1 (× 10-4) 0 <(Cr/Te)2-O> 2.027(4)

BVS 1.8

Δ(Cr/Te)2 (× 10-4) 12.4

O2-Te1-O2 88.7(1) O1-(Cr/Te)2-O1 84.4(1)

91.3(1) O1- (Cr/Te)2-O2 92.4(1)

180.0(1)

172.6(1)

O2- (Cr/Te)2-O2 90.6(1)

101

The three long (2.083(5) Å) and three short (1.956(5) Å) metal-oxygen bonds give an octahedral

distortion parameter (Δ)116

of 9.8×10-4

for (Cr0.97(1)Te0.03(1))O6 in BCTO, which is, however,

much smaller than those of the known JT-distorted octahedral Cr2+

X6 clusters. For example, in

the hybrid and multiferroic [C(NH2)3]Cr2+

[(HCOO)3] (X = HCOO),85

the cooperative JT

distortion (CJTD) results in strong axially anti-ferrodistortion of CrX6 (ΔCr2+

= 55.5×10-4

) with

long and short Cr-X bonds of 2.010 and 2.358 Å, respectively. The antiferrodistortive CJTD of

CrX6 is also observed in the perovskite fluoride KCr2+

F3 (ΔCr2+

=46.2×10-4

, short/long C-F bonds

of 1.986(4)/2.294(4) Å) and rock-salt superstructure sulfide Lu2Cr2+

S4 (ΔCr12+

/Cr22+

=

39.5/106.6×10-4

, short/long Cr-S bonds of 2.599(14)/3.003(12) and 2.383(10)/2.946(13) Å for

Cr1 and Cr2, respectively).86,87

Although Cr2+

is for the first time stabilized in a perovskite oxide

under high pressure, the CJTD is absent in BCTO, since there is only ∼10% Cr2+

; moreover, the

face-sharing (Cr0.97(1)Te0.03(1))O6 octahedral pairs do not favor any axial distortion.

X-ray Absorption Near-Edge Spectroscopy confirms the presence of the Cr2+

state in this

compound and the Te6+

5d10

configuration. Details on these measurements can be found in the

published paper, but are out of the scope of this thesis.

5.3. Magnetic Properties

The per formula unit (f.u.) magnetic susceptibility (χ) for the system at hand is shown in Figure

5.8.

Figure 5.8. The temperature dependence of the magnetic susceptibility (χ solid black line/left scale) of the system

with the assumed Ba3Cr2TeO9 f.u. The important high-temperature (labeled HT) plot of χ is also shown as a dashed

blue line on a much-expanded vertical scale (see right scale). The horizontal T-scale remains the same in the HT

expanded plot. The solid thin red lines are the results of the model fit discussed in the text. The susceptibility data

were collected with H = 1 T magnetic field.

The susceptibility data confirm the Cr-Cr dimer sites observed in the structural refinements,

details on the magnetic measurements can be found in the paper published on this compound.

102

5.4. Conclusion

In summary, octahedral Cr2+

O6 is observed in an oxide prepared at high pressure and

temperature. In the 6H-type hexagonal perovskite structure of Ba3(Cr0.97(1)Te0.03(1))2TeO9,

determined by electron and synchrotron X-ray diffractions, mixed-valent Cr2+

/Cr3+

oxygen

octahedra form, but a Jahn-Teller distortion expected of Cr2+

O6 octahedra is hindered by the

small fraction (∼10%) of Cr2+

and the face-shared arrangement of the (Cr0.97(1)Te0.03(1))O6

octahedral pairs. The presence of Cr2+

state is also indicated by structural analysis and X-ray

absorption near edge spectroscopy analysis, giving Ba3(Cr2+

0.10(1)Cr3+

0.87(1)Te6+

0.03(1))2TeO9.

Magnetic susceptibility (χ) indicate the presence of contributions from a singlet ground state with

excited-state magnetic multiplets, along with a low-temperature Curie tail. A quasi-isolated

magnetic dimer model simulation of the magnetic susceptibility data clearly confirms the Cr-Cr

dimer sites observed in the structural refinements. This work demonstrates that it is difficult, but

possible, to stabilize Cr2+

in oxides at high pressure and temperature. The Jahn-Teller distortion

of Cr2+

O6 is expected in rock salt or layered A2Cr2+

Te6+

O6 double perovskite with smaller A-site

(A = Sr2+

, Ca2+

, and solid solution of them) cations in future work. Comparison with the

structural results of the Sr3Fe2TeO9 – Ba3Fe2TeO9 couple at the start of this chapter, supports the

feasability of this endeavour.

103

Chapter 6

Solution of the crystal structure of K6.4(Nb,Ta)36.3O94 with advanced

transmission electron microscopy

6.1. Introduction

Alkali or alkali earth metal niobates have been intensely investigated as dielectric materials117

,

gas sensors, UV-detectors118

and photocatalysts119

. Among these, “K2Nb8O21” attracts particular

attention, because it can be obtained in many different shapes: nanoribbons117

,

micro/nanowires120-121

, whiskers118

, nanorods121

, tubes119

, and nanobelts120

. Since its first

preparation in 1962122

, the crystal structure of this material has not yet been completely

established. Different groups123-125

report on the investigation of the unit cell parameters based

XRPD or an analysis of TEM/SAED data. Roth et al.123

suggested systematic absences

consistent with Pbam or Pba2 space groups. Irle et al.125

reduced the choice to Pbam only,

referring to original tests on second harmonic generation (SHG) activity. Later investigations by

Li et al.126

suggest that the actual composition of “K2Nb8O21” is K4Nb17O45 based on the

structure solved from crystallographic image processing of high HRTEM images. However, the

proposed composition K4Nb17O45 does not satisfy the charge balance and, moreover, the space

group P21212 proposed by Li et al.126

is not in agreement with other findings, therefore a more

detailed investigation of this crystal structure was done.

In this work we attempt to clarify the structure using a compound synthesized following the

synthesis method for “K2Nb8O21” described in117

, but with addition of Ta. The addition of Ta to

the Nb positions will enhance the clarity of their positions when using HAADF-STEM imaging

of the cation positions. We solved the crystal structure using advanced TEM techniques, and

based on the results of the compositional analysis and structure solution we propose the

K6.4Nb28.2Ta8.1O94 formula. The obtained cell parameters are identical to those described in

literature, but with a different distribution of the cations over different tunnels.

Synthesis

The synthesis was done by Artem Babaryk at the Taras Shevshenko University of Kyiv. The

K6.4Nb28.2Ta8.1O94 compound was synthesized via a two-step route. Initially Nb2O5 was mixed

and ground with 18-fold excess of KCl, then the blend was gradually heated in a ceramic tray to

800 °C until it melted. After 6 hours the melt was cooled down to room temperature. The

solidified melt was leached out with deionized water until a negative probe was obtained for the

presence of Cl- anions with an AgNO3 test solution. The recovered white powder was dried

overnight in an oven at 60 °C. XRPD phase analysis showed the presence of KNb3O8 as a single

entity in the product. At the second step KNb3O8 and Ta2O5 were mixed in a molar ratio 2:1,

104

pressed into a tablet at isostatic pressure of 105 N.m

−2 and fired at 1000 °C in a furnace for 24 h

resulting in a yellow-coloured powder.

6.2. Solution and refinement of the crystal structure

EDX analysis showed a cation ratio K:Nb:Ta = 0.15(2):0.66(2):0.19(2). The cell parameters

determined from XRPD using the Le Bail decomposition (Rp=1.53%, wRp=1.95%, Rexp=2.03%)

are a = 37.4611(3) Å, b = 12.4714(1) Å, c = 3.95427(3) Å (Figure 6.1). The extinction symbol

for the space group was determined using SAED, for which tilt series were collected around

different zone axes.

Figure 6.1. Le Bail decomposition of XRPD profile (see description in the text). Experimental, calculated and

difference curves are drawn with solid blue, red and grey lines, Bragg positions are marked with vertical blue bars.

Figure 6.2 shows the main zones [100], [010] and [001]. All SAED patterns could be completely

indexed using the cell parameters determined from XRPD. The reflection conditions hkl: no

conditions, hk0: no conditions, h0l: h=2n, h00: h=2n are clear from these patterns (and thus part

of the extinction symbol, P.a-), however, there is also a 0kl: k=2n reflection condition, which is

not so obvious. At first sight, most [100] patterns show a 0kl:k=2n reflection condition, but on

close inspection some very weak reflections contradicting this reflection condition can be seen

on the [100] SAED patterns. The presence of these reflections seems to indicate the absence of

reflection conditions for 0kl and thus extinction symbol P-a-. However, calculated SAED

patterns using the cell parameters obtained from XRPD show that these reflections will be

present also for Pba- space groups in spite of the reflection condition 0kl:k=2n, and that they are

in fact reflections originating from the very nearby [11 0] zone. To clarify this, the calculated

[100] SAED pattern for Pba- has been included in (see Figure 6.3). Therefore, the extinction

symbol is Pba-, with possible space groups Pbam and Pba2. To decide between these two space

groups, CBED needs to be performed, however, this is impossible for the current material due to

a high amount of twinning. Therefore, Pbam has been selected for the further considerations,

according to the SHG investigation from Irle et al.125

.

105

Figure 6.2. SAED patterns of the main zones of K6.4Nb28.2Ta8.1O94.

Figure 6.3. Calculated selected area electron diffraction pattern of the [100] zone, using the final model as input.

Overlap occurs at k=2n positions of the reflections of [100] and neighboring zones because of the small reciprocal

a* distance. All reflections with k=2n+1 do not belong to the [100] zone itself, but are also visible here for the same

reason as the overlapping k=2n reflections.

Structure solution from precession electron diffraction or electron diffraction tomography was

not possible in this case because a high amount of twinning occurs at a nanoscale. To determine

the positions of the different atomic columns within the unit cell, we used high resolution

HAADF-STEM (Figure 6.4 (a)) and ABF-STEM images (Figure 6.4 (b)) taken along the [001]

zone axis. On HAADF-STEM images the higher Z the brighter the dot. Therefore, in Figure 6.4

(a) the brighter dots correspond to the projected Nb(Ta)5+

(Z=41(73)) columns and less bright

dots to projected K+(Z=19) columns. In the HAADF-STEM images the oxygen columns cannot

be visible, instead of ABF-STEM image, which is sensitive to the lightweight elements.

106

The size of one unit cell is marked by a white rectangle on Figure 6.4 (a). Within one unit cell,

the Nb(Ta)5+

columns form trigonal, tetragonal and pentagonal tunnels. There is a clear presence

of K+ atomic columns inside all the tetragonal tunnels. Four pentagonal tunnels are occupied by

Nb(Ta)5+

and 8 by K+ columns. This allows building a projected model containing the

coordinates of the heavier ions Nb(Ta)5+

and K+. On an ABF-STEM image the oxygen atoms,

can be directly seen. In Figure 6.4 (b), the darker dots are Nb(Ta)5+

columns, while the fainter

ones are K+ or O

2-. All x and y coordinates of columns with only oxygen atoms can now also be

directly estimated from the image and introduced into the model.

Figure 6.4. (a) HAADF-STEM image of K6.4Nb28.2Ta8.1O94, along [001] zone. A calculated image using the model

from Table 4.1 is included, as well as the schematic representation of the final refined structure model, (b) ABF-

STEM image (low band pass filtered) of K6.4Nb28.2Ta8.1O94 viewed along the [001] zone axis. A simulated ABF-

STEM image using the model from TEM is shown outlined by a black rectangle.

From the combination of HAADF-STEM and ABF-STEM images, the positions of all atomic

columns are now fixed, but only in projection. To complete the model with plausible z-

coordinates, it is sufficient to consider the space group Pbam: the mirror plane perpendicular to

the short c-axis restricts the positions of the atoms to either z=0 or z=1/2. Comparing our model

with other bronzes, we can find several with a similar projected structure, such as Na13Nb35O94128

,

Nb7W10O47129

and Nb16W18O94130

. In analogy to these it is clear that the polyhedra should be

completed with apical oxygen atoms at each Nb(Ta)5+

x-y position, with zO=zNb(Ta)+1/2, and that

the relation between the z-positions of K+ and of Nb(Ta)

5+ should be zK=zNb(Ta)+1/2. The apical

oxygens cannot be seen separately, since they are projected on the same dot as the Nb(Ta)5+

positions. Using a simulated annealing algorithm with antibump and bond valence cost functions

the model was optimized. The joint population of the same crystallographic positions by the Nb

and Ta cations is not an obstacle for this optimization because the R0 and B constants in the s =

exp ((R0 – R)/B) formula (s – the bond valence, R-interatomic distance) are very close for these

cations (R0=1.911, B=0.37 for Nb5+

, R0=1.92, B=0.37 for Ta5+ 131

). This optimized model is

shown in Figure 6.5. The atomic coordinates of the optimized model are given in Table 6.1.

(Note that at this stage of the model the K+ indicated by ° in the table are fully occupied by K

+

only and the ones indicated by * are not yet present.).

107

Table 6.1.Atomic coordinates for K6.4Nb28.2Ta8.1O94.(S.G. Pbam, a=37.461 Å, b=12.471 Å, c = 3.954 Å) from TEM.

Atom Wyckoff position x/a y/b z/c

K1 2d 0 0.5 0.5

K2° 4h 0.0558 0.2 0.5

K3 4h 0.1654 0.0126 0.5

K4° 4h 0.2312 0.3207 0.5

K5* 4h 0.95892 0.11645 0.5

K6* 4h 0.71216 0.89315 0.5

K7* 4h 0.62966 0.11454 0.5

Nb1/Ta1 4g 0.929 0.5563 0

Nb2/Ta2 4g 0.8925 0.3166 0

Nb3/Ta3 4g 0.9786 0.2908 0

Nb4/Ta4 4g 0.8982 0.0586 0

Nb5/Ta5 2a 0 0 0

Nb6/Ta6 4g 0.8576 0.7781 0

Nb7/Ta7 4g 0.2338 0.0665 0

Nb8/Ta8 4g 0.1756 0.4907 0

Nb9/Ta9 4g 0.1947 0.7633 0

O1 4g 0.092 0.285 0

O2 4g 0.9767 0.6228 0

O3 4g 0.0582 0.5982 0

O4 4g 0.8805 0.512 0

O5 4g 0.0695 0.8119 0

O6 4g 0.8809 0.9144 0

O7 4g 0.5022 0.3433 0

O8 4g 0.9452 0.9801 0

O9 4g 0.1884 0.1227 0

O10 4g 0.2064 0.9144 0

O11 4g 0.1404 0.8426 0

O12 4g 0.1681 0.6383 0

O13 4g 0.1738 0.3319 0

O14 4g 0.226 0.4681 0

O15 4g 0.2458 0.713 0

O16 4h 0.8524 0.7908 0.5

O17 4h 0.9268 0.5566 0.5

O18 4h 0.8804 0.3572 0.5

O19 4h 0.9711 0.2678 0.5

O20 4h 0.905 0.0665 0.5

O21 2b 0 0 0.5

O23 4h 0.2352 0.0798 0.5

O24 4h 0.1685 0.4763 0.5

O25 4h 0.1847 0.7781 0.5

° K+ in pentagonal channels, positions partially occupied by Nb(Ta)5+* K+ in trigonal channels, occupation lower than 1.

108

Figure 6.5. The structure model with optimized interatomic distances, projected along [001].

The formula derived from this model, in agreement with charge balance requirements, would be

K9(Nb,Ta)17O47. However, the K:(Nb,Ta) atomic ratio of 0.529 in this formula is significantly

different from the 0.176 ratio determined by EDX analysis. Therefore a more detailed inspection

was made for several positions and the following observations can be made: 1) on the HAADF-

STEM image the K+

columns inside the pentagonal tunnels are not all equally bright, therefore

they are not all of equal composition (Figure 6.4 (a)), and 2) some of the trigonal tunnels show a

clear presence of K+ on the HAADF-STEM, ABF-STEM and STEM-EDX images (Figure 6.6).

On the HAADF-STEM image the K+

columns appear as white dots in the centre of some trigonal

tunnels, on ABF-STEM as black dots. On the STEM-EDX image, the EDX map of the K+

positions (red) is overlaid onto the HAADF-STEM image. In Figure 6.6 the trigonal tunnels with

clear K+ presence are indicated by circles (the clearest images for each technique are shown,

therefore the areas shown are not the same areas). Therefore we propose that the real composition

is K6.4Nb28.2Ta8.1O94, as this is in agreement with the required charge balance and the K:(Nb,Ta)

ratio is exactly as measured by EDX analysis. A fraction of the potassium atoms is located in the

trigonal tunnels, divided over all available tunnels, whereas the extra Nb(Ta)5+

ions are located in

some of those pentagonal tunnels mostly filled with K+. The positions of the K

+ in trigonal

channels have been added to Table 6.1 and are indicated by an asterisk.

Finally, the model as-emerged from HAADF/ABF-STEM analysis was validated by combined

Rietveld refinement from XRPD and NPD data (Figure 6.7 and Table 6.2, schematic model in

Figure 6.8).

Table 6.2.Crystal data from refinement against combined X ray and neutron powder diffraction data.

K6.4Nb28.2Ta8.1O94 β = 90°

Mr = 5822.65 γ = 90°

Orthorhombic, Pbam V = 1850.54(63) Å3

a = 37.4676(90) Å Z = 1

b = 12.4934(30) Å neutron radiation, λ = 1.594 Å

c = 3.95333(15) Å Dx, g cm-3

= 5.228

α = 90°

Table 6.3 Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å2).

109

x y z Uiso*/Ueq Occ. (<1)

K1/Nb1 0.449 (4) 0.344 (13) 0.5 2.42 (2)* 0.1486(55)/0.0132(55)

K2/Nb2 0.059 (2) 0.170 (6) 0.5 2.42 (2)* 0.7279(71)/0.2813(71)

K3 0.167 (3) 0.003 (8) 0.5 2.42 (2)* 0.4351(92)

K4/Nb4 0.226 (3) 0.332 (8) 0.5 2.42 (2)* 0.2884(92)/0.2804(90)

Nb1/Ta1 0.9300 (18) 0.573 (5) 0 2.42 (2)* 0.669(13)/0.331(13)

Nb2 0.8898 (12) 0.323 (3) 0 2.42 (2)* 1.0

Nb3/Ta3 0.9765 (16) 0.297 (6) 0 2.42 (2)* 0.776(13)/0.224(13)

Nb4/Ta4 0.903 (2) 0.067 (5) 0 2.42 (2)* 0.697(12)/0.303(12)

Nb5/Ta5 0 0 0 2.42 (2)* 0.604(17)/0.396(17)

Nb6/Ta6 0.8574 (19) 0.792 (6) 0 2.42 (2)* 0.670(10)/0.330(10)

Nb7/Ta7 0.2354 (19) 0.072 (5) 0 2.42 (2)* 0.705(13)/0.295(13)

Nb8/Ta8 0.1677 (12) 0.495 (4) 0 2.42 (2)* 0.639(11)/0.361(11)

Nb9/Ta9 0.1911 (18) 0.791 (6) 0 2.42 (2)* 0.714(35)/0.286(35)

O1 0.0970 (12) 0.291 (2) 0 2.42 (2)* 1.0

O2 0.9817 (7) 0.6092 (16) 0 2.42 (2)* 1.0

O3 0.0559 (5) 0.5794 (18) 0 2.42 (2)* 1.0

O4 0.8822 (5) 0.5011 (14) 0 2.42 (2)* 1.0

O5 0.0719 (5) 0.7850 (16) 0 2.42 (2)* 1.0

O6 0.8805 (7) 0.9275 (18) 0 2.42 (2)* 1.0

O7 0 0.160 (3) 0 2.42 (2)* 1.0

O8 0.9493 (6) 0.991 (2) 0 2.42 (2)* 1.0

O9 0.1918 (5) 0.1454 (16) 0 2.42 (2)* 1.0

O10 0.2071 (7) 0.934 (2) 0 2.42 (2)* 1.0

O11 0.1407 (6) 0.843 (2) 0 2.42 (2)* 1.0

O12 0.1666 (5) 0.6483 (15) 0 2.42 (2)* 1.0

O13 0.1687 (5) 0.3438 (15) 0 2.42 (2)* 1.0

O14 0.2195 (6) 0.4914 (19) 0 2.42 (2)* 1.0

O15 0.2629 (7) 0.207 (4) 0 2.42 (2)* 1.0

O16 0.8595 (12) 0.788 (3) 0.5 2.42 (2)* 1.0

O17 0.9297 (9) 0.578 (3) 0.5 2.42 (2)* 1.0

O18 0.8922 (11) 0.324 (5) 0.5 2.42 (2)* 1.0

O19 0.9774 (7) 0.3100 (15) 0.5 2.42 (2)* 1.0

O20 0.9001 (5) 0.0617 (19) 0.5 2.42 (2)* 1.0

O21 0 0 0.5 2.42 (2)* 1.0

O23 0.2374 (8) 0.074 (4) 0.5 2.42 (2)* 1.0

O24 0.1703 (7) 0.488 (2) 0.5 2.42 (2)* 1.0

O25 0.1918 (9) 0.784 (2) 0.5 2.42 (2)* 1.0

110

A test on concurrent occupancies over all nine Nb sites showed that they are partially filled with

Ta in a Nb:Ta ratio close to 2:1, except for the pentagonal-bipyramidal Nb2(4g) position which is

fully occupied by Nb. According to the procedure, all atoms of the idealized [Nb26Ta8O94]18-

skeleton were located. Further treatment of the K atoms assuming special K1 (2g) and

additionally filled K5-K7 (4g) sites lead to an abnormally high isotropic displacement parameter,

therefore free refinement of their atomic coordinates and occupancies was allowed. The

refinement result suggests that only three of the K2-K4 have survived and K1 occupies a trigonal

general (4g) position at overall improvement of the refinement. The excess of Nb atoms are

distributed non-uniformly over the K1-K4 sites: K2 and K4 are statistically occupied with 28% of

niobium, while the amount in K1 and K3 is none or negligible. The total amount of Nb is in good

agreement with the EDX-derived bulk composition.

Figure 6.6. Top: HAADF-STEM image, second: ABF-STEM image, and lowest two images: high resolution STEM-

EDX map (overview and close up at same scale as the HAADF/ABF STEM images). In each image trigonal tunnels

containing K+ are indicated by circles. The top three images are shown at the same scale, but are not from the same

area.

111

Figure 6.7. Experimental (XRPD – black filled circles, NPD whole detector – ×, and the NPD center of the detector

– +), calculated (superimposed orange, purple and violet solid lines) and difference (orange, purple and violet dashed

lines) profiles after the combined X-ray/neutron Rietveld refinement (Rp = 1.7%, wRp = 2.6%, GoF = 1.3). Bragg

positions are marked with vertical green bars.

Figure 6.8. Projection of the models along [001] of (top) K6.4Nb28.2Ta8.1O94 based on TEM and refined from

combined X-ray/neutron powder diffraction data and (bottom) Ba0.39Sr0.61Nb2O6 (transformed to Pba2 for ease of

comparison).

6.3. Discussion

As to the tetragonal tungsten bronze (TTB) domain, Roth et al.123

indicated a narrow

concentration region of 80.0<x(Nb2O5)<84.3 mol. % and 77.8<x(Ta2O5)<81.6 mol. % at the

pseudo-binary phase diagrams Nb2O5-KNbO3 and Ta2O5-KTaO3 at 1000 °C. The total amount of

112

M2O5 content (e.g., total mol. % of Nb2O5 and Ta2O5) for the currently reported phase is about

85%, which falls just outside the quantitative limits of the TTB phase field. Lundger and

Sundberg127

have made an attempt to explore the stoichiometric composition KNb7O18 (12.5 mol.

% of K2O and 87.5 mol. % of Nb2O5) indexing its powder diffraction pattern only on the main

reflections in the conventional TTB cell metrics (a = b ~ 12.49 Å, c ~ 3.96 Å), but with the

remark that superstructure reflections were observed. Complementary HRTEM observations

were interpreted as a microstructure of the TTB type, intergrown with H-Nb2O5 and KNb13O33

inclusions, thus giving rise to a compositional inhomogeneity. A phase with similar cell

parameters was designed by Irle et al.125

to exist in the range 79.2<x(Nb2O5)<81.6 mol %.

The structure of K6.4Nb28.2Ta8.1O94 solved in this paper is very similar to that of K4Nb17O45 as

solved by Li et al.126

, however in our case there is a definite presence of K+ ions in the tetragonal

tunnels and in the trigonal tunnels, plus the placement of about/less than one Nb(Ta)5+

per unit

cell in a K+ position in a pentagonal tunnel, making those occupied by a combination of K

+ and

Nb(Ta)5+

in a ratio close to 3:1. In the model by Li et al.126

the trigonal and tetragonal tunnels

were left empty. Our model resembles that of Na13Nb35O94, where the tetragonal tunnels are

occupied by Na+, and several pentagonal tunnels are occupied by a combination of sodium and

niobium in a ratio 3:1, however there the trigonal tunnels are empty. Furthermore, the space

group derived for our compound from SAED is Pbam (corresponding to the one proposed by

Roth et al.123

for “K2Nb8O21”, while Li et al.126

and also Teng et al 132

solved their structure in

P21212. In our case the clear reflection condition h0l:h=2n definitely rules out the possibility of

P21212.

Data mining for a close structured candidate lead to a match with Ba0.39Sr0.61Nb2O6 (space group

P4bm, a = 12.488 Å, c = 3.949 Å, c/a = 0.3162)133

. For the sake of comparison, we transformed

the structure to the maximal subgroup Pba2 with the (3 0 0, 0 1 0, 0 0 1) translation matrix134

(Figure 4.8). Apart from additional filling of the trigonal channels with K atoms as mentioned

before, it is worth to notice that the pentagonal channels host either K or Nb in the present case,

whereas for the reference structure these are occupied by a Sr/Ba mixture.

To explain the present observations, we employed the charge distribution method. Using this

method, the ratio q/Q (formal oxidation number/calculated charge) for K2/Nb2 and K4/Nb4 are

2.134/2.028 and 1.69/1.753. The calculated charge was obtained using the “Charge Distribution

Method”(CHARDI).135

The ratios are close to 1, suggesting that the established partial

occupancies are correct. Remarkably, the absolute values of q and Q are between the formal

charges (oxidation numbers) of K+ and Nb

5+, but also close to the formal charges of Sr

2+ and Ba

2+

that are frequently found in TTB structures136

. It likely indicates that the effective charge on the

Ta atoms might be comparable to that on the Ba atoms, permitting concurrent distributions on

charge compensating positions137

, but such assumption must be confirmed by theoretical

calculations, which are not accessible for the present case due to multiple statistical disorder.

Even though there is Ta in our compound, the identical NbO6 framework suggests it would be

worthwhile to reinvestigate the pure niobate analog using the current advanced TEM techniques,

to clarify the position of the K+ ions in the different tunnels also for the pure Nb-compound. It is

113

noteworthy that in KxLiyWO3 the authors also concluded that the trigonal channels were

occupied, albeit with the Li+ cations.

138

Aside from the similar bronzes mentioned in the results section, there are other bronzes in

literature with very similar parameters, but for which neither the space group nor the structure has

so far been solved, such as Nb8W9O47 (a = 36.69 Å, b = 12.19 Å, c = 3.945 Å)139

and

Pb0.065Nb0.935O2.17F0.47(a = 37.11 Å, b = 12.433 Å, c = 3.947 Å)140

. Possibly they are isostructural

to K6.4Nb28.2Ta8.1O94.

6.4. Conclusion

The compound K6.4Nb28.2Ta8.1O94 was synthesized via two-step solid state route and its structure

was determined using a combination of X-ray, neutron diffraction and imaging TEM techniques.

The space group Pbam with unit cell parameters a = 37.468(9) Å, b = 12.493(3) Å, c =

3.95333(15) Å is established and, contrary to previous models for related compounds such as

K2Nb8O21, a clear presence of K+ is observed in the trigonal and tetragonal tunnels. Arguments

are given that also point to a possible different composition and cation distribution for the related

K2Nb8O21 than reported in literature.

114

Chapter 7

Conclusions and outlook

The main goal of this doctoral thesis was the investigation of the physical and chemical

properties of several promising complex perovskites and to link their crystal structure with those

properties. Of all compounds reported in this thesis, the synthesis and bulk diffraction studies (X-

ray diffraction, neutron diffraction) and measurements of physical properties were done by our

collaborators. The work performed within this Ph.D. thesis is restricted to the investigation by

transmission electron microscopy (TEM) of these samples.

Chapters 1 and 2 were introductions to the investigated materials and experimental techniques. In

Chapter 3, we search for a new relaxor ferromagnet in triple perovskites of the type

La2A’Ni2B’O9, where A’ and B’ cation positions were replaced by different cations (A’=Sr2+

,

Ca2+

or kept La3+

; B’= Te6+

, Ta5+

, Nb5+

, Nb5+

0.5Sb5+

0.5, W6+

). The structural investigation shows

that they all have the same P21/n space group, with similar cell parameters (a~b=√2𝑎𝑝, c=2𝑎𝑝,

where ap is the cell parameter of the parent perovskite). All compounds show rock salt type order

in the B positions. This rock salt type order cannot be complete, however, since the stoichiometry

of these phases with B:B’=2:1 does not match the stoichiometry for pure rock salt order, i.e.

B:B’=1:1. This is presumably the origin of the different magnetic properties these compounds

assume, such as spin glass behaviour (B’=Nb5+

, Te6+

, W6+

)

or ferromagnetic (B’= Ta5+

,

Nb5+

0.5Sb5+

0.5). In the TEM part of this chapter, we attempted to uncover evidence of local

structural consequences of this incapacity to order completely, and structural differences

explaining the differences in magnetic properties. However, using the different possible advanced

TEM techniques, we found no significant difference between the different compounds that might

explain their different magnetic properties. This investigation is explained in detail for the Te5+

containing compound, as was also published, for the other compounds we chose to give a

comparative overview only. The Te6+

compound contains twin domains, as expected for low

symmetry perovskites, and also local variations in composition. These defects are thought to be

responsible for the presence of two distinct types of antiferromagnetic ordering, C-type and G-

type ordered domains, that were in agreement with the neutron diffraction data, while the

magnetometry showed spin glass behaviour below 35K.

So far, the only clear difference between the compounds La3Ni2NbO9 and La3Ni2TaO9 was that

the average particle size in the former 1.7 times larger is.

As outlook for this chapter, it is noteworthy that relaxor ferroelectric compounds often show

diffuse scattering due to the nanodomains. Therefore, it would be interesting to see if the

difference in domain structure which is expected to exist to explain the magnetic difference, can

be solved using other techniques than advanced electron microscopy, such as for example the

study of diffuse scattering in these compounds.

115

In Chapter 4, a similar analysis was carried out for a second group of compounds, now with

composition A3Fe2TeO9 and as A cation either Sr2+

or Ba2+

. The change from Sr2+

to Ba2+

in

A3Fe2TeO9 changes the structure from trigonal P-3c to hexagonal P63/mmc. In Sr3Fe2TeO9,the

Fe3+

and Te6+

cations order in a 2:1 sequence, contrary to the compounds in the previous chapter

(the rock salt order is a 1:1 sequence). So here the composition is effectively compatible with the

multiplicity of the available crystallographic sites in the structure. The trigonal phase of

Sr3Fe2TeO9 is antiferromagnetic at room temperature. However, inside this matrix of the main

phase, there are nano-twins and nano-regions where the cations order in a 1:1 sequence after all

(the same order as the compounds in Chapter 3). This prevents that the sample orders

antiferromagnetically over its full volume, and results in the coexistence of a spin-glass phase

next to the antiferromagnetic phase below 80 K.

When instead of Sr2+

, there is Ba2+

in the A position, the A3Fe2TeO9 structure turns hexagonal

with space group P63/mmc. There was disagreement about the magnetic properties and the cation

order in this compound in literature. We found that the Ba-containing structure has a combination

of face-sharing and edge-sharing BO6 octahedra, with partial ordering of the of Fe3+

and Te6+

cations over the six-coordinate sites. In our case, it is antiferromagnetically ordered at low

temperature (below ~18 K), and paramagnetic at room temperature.

In Chapter 5, we investigate the same structure as in Ba3Fe2TeO9, but with chromium cations

instead of Fe3+

, in order to obtain a Jahn-Teller distortion of the Cr2+

O6 octahedra. So far, no

oxides are known with a Jahn-Teller distortion of high spin octahedral Cr2+

(d4), while it has been

observed in the hydrated state in solution and inorganic-organic hybrid compounds. We did not

successfully obtain the Jahn-Teller distortion in our structure as refined, probably because only

10% of the six-coordinated sites was occupied by chromium with Cr2+

plus the face-shared

arrangement of the (Cr0.97(1)Te0.03(1))O6 octahedral pairs. However, our work demonstrates that it

is in any case feasible to stabilize Cr2+

in oxides at high pressure and temperature.

As outlook for this part, it might be possible to obtain a measurable Jahn-Teller distortion of

Cr2+

O6 in rock salt or layered A2Cr2+

Te6+

O6 double perovskite with a smaller A-site (A = Sr2+

,

Ca2+

, and solid solution of them) cations in future work.

In Chapter 6 we solved the structure of K6.4Nb28.2Ta8.1O94. This sheds light on the compound

known in literature as K2Nb8O21, of which the crystal structure was never successfully refined

and the exact structure was disagreed upon. The atomic composition of this sample was slightly

modified, by introducing partially Ta atoms in the positions of the Nb atoms, obtaining the

K6.4Nb28.2Ta8.1O94 sample. It adopts an orthorhombic structure with space group Pbam with a

framework equivalent to that published in literature, however the occupation of the tunnels of the

framewrok turned out to be different than published. K+ cations are also present in some

tetragonal and trigonal tunnels, which was never reported before for this compound.

116

References

1Patrick M. Woodward, PhD thesis, 1996, Oregon State University.

2Peter D. Battle, Sophie I. Evers, Emily C. Hunter, and Mark Westwood, Inorg. Chem., 2013, 52,

6648-6652. 3Jun-Ming Liu, Ce-Wen Nan, Front. Phys., 2012, 7, 373-374.

4P. D. Battle, T.C. Gibb, C.W. Jones, Journal of Solid State Chemistry, 1989, 78, 281-293.

5http://physics.stackexchange.com/questions/238180/why-do-non-hydrogen-atomic-orbitals-

have-the-same-degeneracy-structure-as-hydrog. 6Mark R. Levy, PhD thesis, 2005, Imperial College.

7Roger M. Mitchell, perovskites: modern and ancient, Almaz Press Inc., 2002, 1, 1-108

8A. M. Glazer, Acta Cryst., 1975. A31, 756-762.

9Kuo Sheng Chao, PhD thesis, 2007, National Chiao Tung University.

10Christopher J. Howard, Brendan J. Kennedyc and Patrick M. Woodward, Acta Cryst., 2003,

B59, 463-471. 11

Graham King and Patrick M. Woodward, J. Mater. Chem., 2010, 20, 5785-5796. 12

M. T. Anderson, K. R. Poeppelmeier, Chemistry of Materials, 1991, 3, 476-482. 13

Roy Thompson, Frank Oldfield, Environmental Magnetism, Springer, 1986. 14

http://www.birmingham.ac.uk/Documents/college-eps/metallurgy/research/Magnetic-Materials-

Background/Magnetic-Materials-Background-4-Classification-of-Magnetic-Materials.pdf. 15

Jun-Shuai Chai, Hao Tian,

Ai-Jie Mao, Li-Juan Deng, Xiao-Yu Kuang, RSC Adv., 2016, 6,

54041-54048. 16

A. Peláiz-Barranco, F. Calderón-Piñar,O. García-Zaldívar and Y. González-Abreu, INTECH,

2012, 85-87. 17

T. Kimura, Y. Tomioka, R. Kumai, Y. Okimoto, and Y. Tokura, Phys. Rev Lett., 1999, 83,

3940-3943. 18

Peter D. Battle, Maxim Avdeev, Joke Hadermann, Journal of Solid State Chemistry, 2014, 220,

163-166. 19

Mark de Graef, Cambridge University Press, 2003,94-101. 20

David B. Williams, C. Barry Carter, Springer, 2009, 2014-2015. 21

Pennycook, S. & Boatner, L. A., Nature, 1988, 336, 565-567. 22

S. Vasala and M. Karppinen, Prog. Solid State Chem., 2015, 43, 1-36. 23

M. Anderson, K. Greenwood, G. Taylor and K. Poeppelmeier, Prog. Solid State Chem., 1993,

22, 197-233. 24

E. J. Cussen, J. F. Vente, D. Battle and T. C. Gibb, J. Mater. Chem.,1997, 7, 459-463. 25

P. D. Battle, G. R. Blake, T. C. Gibb and J. F. Vente, J. Solid State Chem., 1999, 145, 541-548. 26

R. L. Martin, R. S. Nyholm and N. C. Stephenson, Chem. Ind., 1956, 3, 83-85. 27

A. A. Bokov and Z.-G. Ye, J. Mater. Sci., 2006, 41, 31-52. 28

P. K. Davies and M. A. Akbas, J. Phys. Chem. Solids, 2000, 61, 159-166. 29

V. C. Fuertes, M. C. Blanco, D. G. Franco, J. M. De Paoli, R. D. Sánchez and R. E. Carbonio,

Mater. Res. Bull., 2011, 46, 62-69.

117

30D. Iwanaga, Y. Inaguma and M. Itoh, Mater. Res. Bull., 2000, 35, 449-457.

31M. P. Attfield, P. D. Battle, S. K. Bollen, T. C. Gibb and R. J. Whitehead, J. Solid State Chem.,

1992, 100, 37-48. 32

L. M. Rodriguez-Martinez and J. P. Attfield, Phys. Rev. B, 1996, 54, R15622-R15625. 33

J. P. Attfield, Chem. Mater., 1998, 10, 3239-3248. 34

N. S. P. Bhuvanesh and J. Gopalakrishnan, J. Mater. Chem., 1997, 7, 2297-2306. 35

S. Kato, T. Sakamoto, N. Watanabe, M. Sugai and Y. Ohshima, Mater. Lett., 1994, 21, 101-

104. 36

L. Ortega-San Martín, J. P. Chapman, G. Cuello, J. González-Calbet, M. I. Arriortua and T.

Rojo, Z. anorg. allg. Chem., 2005, 631, 2127-2130. 37

B. Stöger, M. Weil and E. Zobetz, Zeitschrift fur Krist., 2010, 225, 125-138. 38

G. A. Slack, J. Appl. Phys., 1960, 31, 1571-1582. 39

W. L. Roth and G. A. Slack, J. Appl. Phys., 1960, 352, 10-12. 40

E. O. Wollan and W. C. Koehler, Phys. Rev., 1955, 100, 545-563. 41

S. Nomura and T. Nakagawa, J. Phys. Soc. Japan, 1966, 21, 1068. 42

S. Nomura and T. Nakagawa, J. Phys. Soc. Japan, 1966, 21, 1679-1684. 43

V. Scatturin, L. Corliss, N. Elliott and J. Hastings, Acta Crystallogr., 1961, 14, 19-26. 44

C. Martin, A. Maignan, M. Hervieu, B. Raveau, Z. Jirák, M. M. Savosta, A. Kurbakov, V.

Trounov, G. André and F. Bourée, Phys. Rev. B - Condens. Matter Mater. Phys., 2000, 62, 6442-

6449. 45

Thitirat Inprasit, SujitraWongkasemjit, Pimpa Limthongkul, Stephen J. Skinner, Materials

Letters, 2016,162, 37-39. 46

K. Dey, A. Indra, D. De, S. Majumdar, and S. Giri, ACS Appl. Mater. Interfaces, 2016, 8,

12901-12907. 47

Robert Paria Sena, Joke Hadermann, Chun-Mann Chin, Emily C. Hunter, Peter D. Battle,

Journal of Solid State Chemistry, 2016, 243, 304-311. 48

M. S. Augsburger, M. C. Viola, J. C. Pedregosa, R. E. Carbonio, J. A. Alonso, J. Mater. Chem.,

2006, 16, 4235-4242. 49

S. A. Ivanov, P. Nordblad, S. G. Eriksson, R. Tellgren, H. Rundlof, Mater. Res. Bull., 2007, 42,

776-789. 50

I. Djerdj, J. Popović, S. Mal, T. Weller, M. Nuskol, Z. Jagličić, Ž. Skoko, D. Pajić, C.

Suchomski, P. Voepel, R. Marschall, B. Kozlevčar, B. M. Smarsly, Crystal Growth and Design,

2016. 51

P. D. Battle, M. Avdeev, J. Hadermann, J. Solid State Chem., 2014, 220, 163. 52

J. H. Park, P. M. Woodward, Int. J. Inorg. Mater., 2000, 2, 153-166. 53

P. K. Davies, H. Wu, A. Y. Borisevich, I. E. Molodetsky, L. Farber, Annu. Rev. Mater.Res.,

2008, 38, 369-401. 54

U. Treiber, S. Kemmlersack, J. Solid State Chem., 1982, 43,51-62. 55

J. T. Rijssenbeek, S. Malo,V. Caignaert, K. R. Poeppelmeier, J. Am. Chem. Soc., 2002, 124,

2090-2091.

118

56J. T. Rijssenbeek, T. Saito, S. Malo, A. T. Masaki, M. Takano, K. R. Poeppelmeier, J. Am.

Chem. Soc., 2005, 127, 675-681. 57

J. G. Zhao, L. X. Yang,Y. Yu, F. Y. Li, R. C. Yu, C. Q. Jin, J. Solid State Chem., 2009, 182,

327-330. 58

A. M. Arevalo-Lopez, M. S. Senn, L. Skedd, J. P. Attfield, Z. Anorg. Allg. Chem., 2014, 640,

1164-1167. 59

R. D. Shannon, Acta Crystallogr., 1976, A32, 519. 60

P. D. Battle, T. C. Gibb, A. J. Herod, S. H. Kim, P. H. Munns, J. Mater. Chem., 1995, 5,865. 61

P. D. Battle, T. C. Gibb, A. J. Herold, J. P. Hodges, J. Mater. Chem., 1995, 5, 75. 62

E. J. Cussen, P. D. Battle, Chemistry of Materials, 2000, 12, 831-838. 63

H. Krischner, B. O. Kolbesen, K. Torkar, Journal of Solid State Chemistry,1971, 3, 349-357. 64

P. C. Donohue, L. Katz, R. Ward, Inorganic Chemistry, 1966, 5, 335-338. 65

D. Harari, J. C. Bemier, P. Poix, Journal of Solid State Chemistry, 1972, 5, 382-390. 66

V. V. Gagulin, N. V. Fadeeva, A. G. Belous, A. V. Titov, K. P. Mitrofanov, M. V. Plotnikova,

S. P. Solov'ev, Yu. N. Venevtsev, Physica Status Solidi A-Applications and Materials, 1978, 48,

183-189. 67

Gehring, G. A.; Gehring, K. A., Rep. Prog. Phys., 1975, 38, 1. 68

Falvello, L. R., J. Chem. Soc., Dalton Trans., 23, 1997, 4463. 69

Goodenough, J. B., Annu. Rev. Mater. Sci.,1998, 28, 1. 70

Tokura, Y.; Nagaosa, N., Science, 2000, 288, 462. 71

Raveau, B.; Hervieu, M.; Maignan, A.; Martin, C. J. Mater. Chem., 2001, 11, 29. 72

Maignan, A.; Martin, C.; Hebert, S.; Hardy, V. J., Mater. Chem., 2007, 17, 5023. 73

Raveau, B.; Maignan, A.; Martin, C.; Hervieu, M. Chem. Mater., 1998, 10, 2641. 74

Cheong, S.-W.; Mori, S.; Chen, C. H.; Uehara, M., Nature, 1999, 399, 560. 75

Bednorz, J. G.; Muller, K. A. Z., Phys. B: Condens. Matter, 1986, 64, 189. 76

Keimer, B.; Kivelson, S. A.; Norman, M. R.; Uchida, S.; Zaanen, J., Nature, 2015, 518, 179. 77

Keller, H.; Bussmann-Holder, A.; Muller, K. A., Mater. Today, 2008, 11, 38. 78

Orenstein, J.; Millis, A. J., Science, 2000, 288, 468. 79

Delmas, C.; Saadoune, I.; Dordor, P., Mol. Cryst. Liq. Cryst. Sci. Technol., Sect. A, 1994, 244,

337. 80

Foglio, M. E.; Barberis, G. E., Braz. J. Phys., 2006, 36, 40. 81

Ivanova, T. A.; Petrashen’, V. E.; Chezhina, N. V.; Yablokov, Y. V., Phys. Solid State, 2002,

44, 1468. 82

Louca, D.; Sarrao, J. L.; Thompson, J. D.; Roder, H.; Kwei, G. H., Phys. Rev. B: Condens.

Matter Mater. Phys., 1999, 60, 10378. 83

Zhang, Y.; Xiang, H. J.; Whangbo, M. H., Phys. Rev. B: Condens. Matter Mater. Phys., 2009,

79, 054432. 84

Kritayakornupong, C. J., Comput. Chem., 2008, 29, 115. 85

Stroppa, A.; Barone, P.; Jain, P.; Perez-Mato, J. M.; Picozzi, S., Adv. Mater., 2013, 25, 2284. 86

Margadonna, S.; Karotsis, G. J., Am. Chem. Soc., 2006, 128, 16436.

119

87Tezuka, K.; Wakeshima, M.; Nozawa, M.; Oshikane, K.; Ohoyama, K.; Shan, Y. J.; Imoto, H.;

Hinatsu, Y., Inorg. Chem., 2015, 54, 9802. 88

King, G.; Thimmaiah, S.; Dwivedi, A.; Woodward, P. M., Chem. Mater., 2007, 19, 6451. 89

King, G.; Wayman, L. M.; Woodward, P. M., J. Solid State Chem., 2009, 182, 1319. 90

De, C.; Kim, T. H.; Kim, K. H.; Sundaresan, A., Phys. Chem. Chem. Phys., 2014, 16, 5407. 91

Retuerto, M.; Li, M. R.; Ignatov, A.; Croft, M.; Ramanujachary, K. V.; Chi, S.; Hodges, J. P.;

Dachraoui, W.; Hadermann, J.; Tran, T. T.; Halasyamani, P. S.; Grams, C. P.; Hemberger, J.;

Greenblatt, M., Inorg. Chem., 2013, 52, 12482. 92

Saad, H.-E.; Musa, M., Comput. Mater. Sci., 2014, 92, 298. 93

Philipp, J. B.; Majewski, P.; Alff, L.; Erb, A.; Gross, R.; Graf, T.; Brandt, M. S.; Simon, J.;

Walther, T.; Mader, W.; Topwal, D.; Sarma, D. D., Phys. Rev. B: Condens. Matter Mater. Phys.,

2003, 68, 144431. 94

Majewski, P.; Geprags, S.; Boger, A.; Opel, M.; Erb, A.; Gross, R.; Vaitheeswaran, G.;

Kanchana, V.; Delin, A.; Wilhelm, F.; Rogalev, A.; Alff, L., Phys. Rev. B: Condens. Matter

Mater. Phys., 2005, 72, 132402. 95

Martinelli, A.; Ferretti, M.; Cimberle, M. R.; Ritter, C., Mater. Res. Bull., 2011, 46, 190. 96

Shannon, R., Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr., 1976, 32,

751. 97

Takada, K.; Sakurai, H.; Takayama-Muromachi, E.; Izumi, F.; Dilanian, R. A.; Sasaki, T.,

Nature, 2003, 422, 53. 98

Ganesanpotti, S.; Tassel, C.; Hayashi, N.; Goto, Y.; Bouilly, G.; Yajima, T.; Kobayashi, Y.;

Kageyama, H., Eur. J. Inorg. Chem., 2014, 2014, 2576. 99

Hosaka, Y.; Ichikawa, N.; Saito, T.; Manuel, P.; Khalyavin, D.; Attfield, J. P.; Shimakawa, Y.,

J. Am. Chem. Soc., 2015, 137, 7468. 100

Azuma, M.; Kaimori, S.; Takano, M., Chem. Mater., 1998, 10, 3124. 101

Vasala, S.; Cheng, J. G.; Yamauchi, H.; Goodenough, J. B.; Karppinen, M., Chem. Mater.,

2012, 24, 2764. 102

Novikov, D. L.; Freeman, A. J.; Poeppelmeier, K. R.; Zhukov, V. P., Phys. C, 1995, 252, 7. 103

Burbank, R. D.; Evans, H. T., Acta Crystallogr., 1948, 1, 330. 104

Shikano, M.; Ishiyama, O.; Inaguma, Y.; Nakamura, T.; Itoh, M., J. Solid State Chem., 1995,

120, 238. 105

Rijssenbeek, J. T.; Huang, Q.; Erwin, R. W.; Zandbergen, H. W.; Cava, R. J., J. Solid State

Chem., 1999, 146, 65. 106

Doi, Y.; Matsuhira, K.; Hinatsu, Y., J. Solid State Chem., 2002, 165, 317. 107

Dey, T.; Mahajan, A. V., Eur. Phys. J. B, 2013, 86, 1. 108

Kwon, D.-K.; Lanagan, M. T.; Shrout, T. R., Mater. Lett., 2007, 61, 1827. 109

Walker, D.; Carpenter, M. A.; Hitch, C. M., Am. Mineral., 1990, 75, 1020. 110

Zhao, J. G.; Yang, L. X.; Yu, Y.; Li, F. Y.; Yu, R. C.; Fang, Z.; Chen, L. C.; Jin, C. Q., J. Solid

State Chem., 2007, 180, 2816. 111

Arevalo-Lopez, A. M.; Reeves, S. J.; Attfield, J. P. Z., Anorg. Allg. Chem., 2014, 640, 2727.

120

112Ivanov, S. A.; Nordblad, P.; Mathieu, R.; Tellgren, R.; Ritter, C., Dalton Trans., 2010, 39,

5490. 113

Mathieu, R.; Ivanov, S. A.; Tellgren, R.; Nordblad, P., Phys. Rev. B: Condens. Matter Mater.

Phys., 2011, 83, 174420. 114

Choy, J.-H.; Hong, S.-T.; Choi, K.-S., J. Chem. Soc., Faraday Trans., 1996, 92, 1051. 115

Choy, J.-H.; Hong, S.-T.; Park, J.-H.; Kim, D.-K., Jpn. J. Appl. Phys., 1993, 32, 4628. 116

Brown, I. D.; Shannon, R. D., Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen.

Crystallogr., 1973, 29, 266. 117

X. Cheng-Yan, Z. Liang, Y. Li, H. Kai, S. Wen-Zhu and Q. Lu-Chang, Ceramic International,

2008, 34, 435. 118

C. Y. Xu, L. Zhen, J. T. Jiang, C. S. Lao and L. Yang, Ceramics International, 2009, 35, 3021. 119

K. Teshima, Y. Niina, K. Yubuta, T. Nakazawa, T. Suzuki, T. Shishido, N. Ishizawa, and S.

Oishi, Japanese Journal of Applied Physics, 2008, 47, 629. 120

B. Yu, B. Cao, H. Cao, X. Zhang, D. Chen, J. Qu and H. Niu, Nanotechnology, 2013, 24,

085704. 121

X. Cheng-Yan, Z. Liang, Y. Rusen and W. Zhong Lin, J. Am. Chem. Soc., 2007, 129, 15444. 122

J. E. Guerchais, Bull. Soc. Chim. France, 1962, 1, 103. 123

R. S. Roth, H. S. Parker, W. S. Brower and D. B. Minor, NASA Techn. Report CR,1974,

134599. 124

P. Appendino, Ann. Chim. (Rome), 1973, 63, 547. 125

E. Irle, R. Blachnik and B. Gather, Thermochimica Acta, 1991, 179, 157. 126

F. H. Li, Journal of Microscopy, 1998, 190, 249. 127

M. Lundberg and M. Sundberg, J. Solid State Chem., 1986, 63, 216. 128

D. C. Craig and N. C. Stephenson, J. Solid State Chem., 1971, 3, 89. 129

F. Krumeich, M. Worle and A. Hussain, J. Solid State Chem., 2000, 149, 428. 130

A. W. Sleight, Acta Chemica Scandinavica, 1966, 20, 1102. 131

Brown and Altermatt, Acta Cryst. B, 1985, 41, 244. 132

C. M. Teng, F. H. Li, D. Y. Yang and Q. Z. Wu, J. Chin. Ceram. Soc., 1986, 4, 484. 133

J. Schefer, D. Schaniel, V. Petricek, T. Woike , A. Cousson and M. Woehlecke, Z. Kristallogr.,

2008, 223, 399. 144

C. Capillas, E. S. Tasci, G. de la Flor, D. Orobengoa, J. M. Perez-Mato and M.I. Aroyo, Z.

Kristallogr., 2011, 226, 186. 135

M. Nespolo, G. Ferraris and H. Ohashi, Acta Cryst. B, 1999, 55, 902. 136

A. Simon and J. Ravez, C. R. Chimie, 2006, 9, 1268. 137

X. Kuang, F. Pan, J. Cao, C. Liang, M. R. Suchomel, F. Porcher and M. Allix, Inorg. Chem.,

2013, 52, 13244. 138

E. Banks and A. Goldstein, Inorg. Chem., 1968, 7, 966. 139

R. S. Roth and J. Waring, Res. Natl. Bur. Stand. A, 1966, 70, 281. 140

O. Savborg and M. Lundberg, Mater. Res. Bull., 1980, 15, 1433.

121

List of Abbreviations

ABF-STEM Annular bright field-scanning transmission electron microscope.

AFM Antiferromagnetic

Z Atomic number

BCTO Ba3Cr2TeO9

µB Borh magneton

TC Curie temperature

ɛ Dielectric permittivity

EDX-STEM Energy dispersive X-ray- scanning transmission electron microscope.

μeff Effective magnetic moment

FFT Fast Fourier transform

FC Field cooling

FM Ferromagnetic

FT Fourier transform

a+b

0c

- Glazer notation

hcp Hexagonal close packed

HR-TEM High resolution-transmission electron microscope

HAADF-STEM High angle annular dark field-scanning transmission electron microscope.

HS High spin state

𝑅1+ Irreducible representations

𝑀3+ In-phase octahedral tilting

J-T Jahn-Teller

LS Low spin state

M Magnetization

H Magnetic field

χ Magnetic susceptibility

NPD Neutron powder diffraction

XRPD X-ray powder diffraction

𝑅4+ Out-of-phase tilting

PED Precession electron diffraction

RT Room temperature

SAED Selected area electron diffraction

SG Spin-glass

S.G. Space group

TTB Tetragonal tungsten bronze

ZFC Zero field cooling

ZF……………………Field cooling

122

Publications related to this thesis

Pseudo-tetragonal tungsten bronze superstructure: a combined solution of the crystal

structure of K6.4(Nb,Ta)36.3O94 with advanced transmission electron microscopy and

neutron diffraction

Robert Paria Sena, Artem A. Babaryk, Sergiy Khainakov, Santiago Garcia-Granda, Nikolay S.

Slobodyanik, Gustaaf Van Tendeloo, Artem M. Abakumova and Joke Hadermann.

Dalton Transactions, 2016, 45, 973-979.

Structural chemistry and magnetic properties of the perovskite Sr3Fe2TeO9

YaweiTang, EmilyC.Hunter, PeterD.Battle, Robert Paria Sena, Joke Hadermann, Maxim Avdeev,

J. M. Cadogan.

Journal of Solid State Chemistry, 2016, 242(1), 86-95.

Structural chemistry and magnetic properties of the perovskite SrLa2Ni2TeO9

Robert Paria Sena, Joke Hadermann, Chun-Mann Chin, Emily C. Hunter, Peter D. Battle

Journal of Solid State Chemistry, 2016, 243, 304-311.

Ba3(Cr0.97(1)Te0.03(1))2TeO9: in Search of Jahn-Teller distorted Cr(II) Oxide

Man-Rong Li, Zheng Deng, Saul H. Lapidus, Peter W. Stephens, Carlo U. Segre, Mark Croft,

Robert Paria Sena, Sun Woo Kim, Joke Hadermann, David Walker, Martha Greenblatt.

Inorg. Chem., 2016, 55(20),10135-10142.

Publications of which the work was not included in this thesis

Hole Doping and Structural Transformation in CsTl1-xHgxCl3

Maria Retuerto, Zhiping Yin, Thomas J. Emge, Peter W. Stephens, Man-Rong Li, Tapati Sarkar,

Mark C. Croft, Alexander Ignatov, Z. Yuan, S. J. Zhang, Changqing Jin, Robert Paria Sena, Joke

Hadermann, Gabriel Kotliar, and Martha Greenblatt.

Inorg. Chem., 2015, 54 (3),1066-1075.

Study of hydrogen peroxide reactions on manganese oxides as a tool to decode the oxygen

reduction reaction mechanism

Anna S. Ryabova, Antoine Bonnefont, Pavel Zagrebin, Tiphaine Poux, Robert Paria Sena, Joke

Hadermann, Artem M. Abakumov,Gwénaëlle Kéranguéven, Sergey Y. Istomin, Evgeny V.

Antipov, Galina A. Tsirlina, Elena R. Savinova.

ChemElectroChem, 2016, 3,1667-1677.

123

Submmited

Crystal Growth and Structure Analysis of Ce18W10O57: A Complex Oxide Containing

Tungsten in an Unusual Trigonal Prismatic Coordination Environment

Abeysinghe, Dileka; Smith, Mark; Yeon, Jeongho; Tran, T. Thao; Paria Sena, Robert;

Hadermann, Joke; Halasyamani, P. Shiv; zur Loye, Hans-Conrad.

Inorganic Chemistry

Ferrimagnetism as a consequence of cation ordering in the perovskite LaSr2Cr2SbO9

Emily Hunter; Robert Paria Sena; Joke Hadermann, Peter D. Battle.

Journal of Solid State Chemistry

Oral presentations at conferences

Solución de la estructura cristalina de K6.4(Nb,Ta)36.3O94 superestructura tugnsteno bronce

pseudo-tetragonal, usando microscopía electrónica avanzada y difracción de neutrones.

Robert Paria Sena, Artem A. Babaryk, Sergiy Khainakov, Santiago Garcia-Granda, Nikolay S.

Slobodyanik, Gustaaf Van Tendeloo, Artem M. Abakumov, Joke Hadermann

ECI2015i-Lima-Perú, from 30 July till 2 August 2015.

Solución de la estructura cristalina del compuesto K6.4(Nb,Ta)36.3O94, usando TEM

avanzado

Robert Paria Sena, Artem A. Babaryk, Sergiy Khainakov, Santiago Garcia-Granda, Nikolay S.

Slobodyanik, Gustaaf Van Tendeloo, Artem M. Abakumov, Joke Hadermann.

Sinapsis 2016-Paris-France, from 11 July till 13 July, 2016.

Poster presentations at conferences

TEM investigation of multiferroic compounds: LaAlO3//BaTiO3/CoFe2O4 and

LaAlO3//BiFeO3/CoFe2O4

R. Paria, N. Pavlovic, A. Hardy, M.K. Van Bael, J. Hadermann, G. Van Tendeloo

Electron Crystallography-Introduction to Electron Diffraction Tomography, 07 -11 April, 2014.

Darmstadt/Mainz-Germany.

Solution of the crystal structure of the K6.4Nb28.2Ta8.1O94 ("K2Nb8O21") pseudo-tetragonal

tungsten bronze superstructure, using advanced transmission electron microscopy.

Robert Paria Sena, Artem A. Babaryk, Sergiy Khainakov, Santiago Garcia-Granda,Nikolay S.

Slobodyanik, Gustaaf Van Tendeloo, Artem M. Abakumov, Joke Hadermann.

IAMNANO 2015-Hamburg-Germany, from 8 July till 10 July 2015.

124

Advanced TEM for crystal structure solution of K6.4Nb28.2Ta8.1O94

Robert Paria Sena, Artem A. Babaryk, Sergiy Khainakov, Santiago Garcia-Granda, Nikolay S.

Slobodyanik, Gustaaf Van Tendeloo, Artem M. Abakumova, Joke Hadermann.

Summer school on Fundamental crystallography which has been organized at the University of

Antwerp, from 27 June till 2 July, 2016.

Crystal structure solution of K6.4(Nb,Ta)36.3O94 compound, by using advanced TEM

Robert Paria Sena, Artem A. Babaryk, Sergiy Khainakov, Santiago Garcia-Granda, Nikolay S.

Slobodyanik, Gustaaf Van Tendeloo, Artem M. Abakumov, Joke Hadermann.

EMC 2016-Lyon-France, which has been held at Lyon Convention Center, from 28 August-2

September, 2016.

125

Acknowledgement

First of all, I would like to thank my promotors, who gave me this incredible opportunity to do a

PhD in Physics. I am immensely grateful to Prof. Dr. Joke Hadermann, who taught me so much

about crystallography and how to work with the electron microscope. Without her

encouragement and guidance, I would not have been able to finish my thesis. Joke, it was a

pleasure to work together with such a nice person like you. Thank you for your invaluable

scientific contributions to each of our discussions, meetings and experiments, as well as your

patience with my many questions. I would also like to sincerely thank my co-promotor Prof. Dr.

Gustaaf Van Tendeloo for his many important suggestions and advice concerning my research.

Thank you both for everything.

Furthermore, I would also like to thank to the members of my PhD committee: Prof. Dr. Paul

Scheunders (chairperson), Prof. Dr. Michiel Wouters, Prof. Dr. Artem Abakumov and Prof. Dr.

Peter Battle. I am thankful for their constructive comments and evaluation of my thesis, which

were both very important for improving the final version.

Thanks also goes to my collaborators from different universities: Prof. Dr. Petter Battle of the

University of Oxford, Prof. Dr. Martha Greenbladt of the University of Rutgerds, Dr. Artem

Babaryk of the University of Kyiv, Prof. Dr. Artem Abakumov of the Skoltech Institute, Prof. Dr.

Elena Savinova of the University of Straussburg,. Their work on the synthesis and XRPD, NDP

and magnetometry characterization of my samples was instrumental to my research, and for their

invaluable scientific contributions to the peer-reviewed submitted and published papers. Besides,

my thanks goes for the other collaborators Prof. Dr. Marlies K. Van Bael and Dr. Nikolina of the

University of Hasselt, Dr. Vikas Shabadi of the University of Darmstadt, Dr. Monica Burriel of

the University of Liverpool.

Next, I would like to thank to all EMAT members for their contributions and support to my PhD

work, Prof. Dr. Dirk Van Dyck, Prof. Dr. Nick Schryvers, Prof. Dr. Sara Bals, Prof. Dr. Jo

Verbeeck, Prof. Dr. Sandra Van Aert, Prof. Dr. Dirk Lamoen. I would also like to say thanks to

my Peruvian friends for their friendship and the good moments that we shared in Antwerp:

Ricardo, who thought me a great deal about working with the electron microscope, Ivan, who

help me with the simulations and theoretical aspects, to my roommate Marcos, and finally to

Fiorella. Besides, I would like to thank to Prof. Dr. Walter Estrada from National University of

Engineering, who guided me to do a PhD. There are many friends(colleagues) from EMAT that I

would like to mention for their invaluable contributions my training in the use of microscopes

and valuable discussions about my works. A special thanks goes to Thomas and Maria, for their

help in understanding the various microscopes, and for the wonderful times we have had

together, Maria and Dmitry, for helping me with my experiments at the beginning of my PhD,

Armand, for training me in the use of the microscopes and Benham, for helping me with the

strain analysis of my data. During the last months of my PhD, I also learned a great deal from

my friends (colleagues) of crystallography group: Maria, Olesia, Caroline, and Mylène. Finally,

126

I also thank Caroline and Marnik for their help with the translations, and all other present and

former EMAT PhD and postdoc members: Alex, Andrea, Annelies, Annick, Antonios, Bart,

Carolien, Daniele, Dmitry, Dimitry, Elena, Elsa, Eva B., Eva G., Hamed, Hans, Hosni, Ge,

Giulio, Gerardo, Gunnar, Jarmo, Julie, Kadir, Karel, Kirsten, Knut, Kristof, Laura, Lucian,

Marnik, Martin, Masha, Mert, Mylène, Naomi, Nathalie, Olesia, Nicolas, Paromita, Roeland,

Ruben, Saeid, Shirley, Shyam, Stuart, Svetlana, Thais, Thomas, Tyché, Tom, Vahid, Vladimir,

Yang, Xiayang, Zhi-Yi. Thank you all for the past four years. I would not have made it without

your help and friendship and forever I will keep this nice experience.

My gratitude also goes to all administrative and technical staff for their help with various aspects

of my PhD career, such as: Liesbet, for helping me with the cover and printing of my thesis, Niek

for training and helping me in the microscopes, Stijn, Tine, Frederic, Andre, Ludo, for the

samples preparation, Lydia, Hilde, Marianne and Sabine for the help with the administrative

staffs, and finally to Koen for his support with technical things.

Last but not least, I would like to thank my beloved family: my parents Adalberto and Beatriz, my

brother Carlos, my sisters Yessica and Dennise, my brother-in-law Edward, my nephew Bryam,

and all my other close relatives, as well as my others friends. Your support, encouragement and

love during my PhD meant a great deal to me. Muchas gracias a cada uno de ustedes!

Robert