Synthesis, Ion-Exchange and Dehydration

Microporous titanosilicate AM-2: Synthesis, ion-exchange and dehydration

Inauguraldissertation der Philosophisch-naturwissenschaftlichen Fakultat¨ der Universitat¨ Bern

vorgelegt von Nicola Dobelin¨ von Basel

Leiter der Arbeit: Prof. Dr. T. Armbruster Laboratorium fur¨ chemische und mineralogische Kristallographie

Von der Philosophisch-naturwissenschaftlichen Fakultat¨ angenommen.

Bern, 12. Januar 2006 Der Dekan Prof. Dr. P. Messerli

Contents

1. Abstract 6

2. Introduction 7 2.1. Zorite, ETS-4 ...... 8 2.2. ETS-10 ...... 9 2.3. Penkvilksite, ETS-14, AM-3 ...... 12 2.4. Umbite, AM-2 ...... 14 2.5. Silico-titanate, CST, TAM-5 ...... 15 2.6. Chemical variations ...... 17 References ...... 18

3. Synthesis 24 3.1. Experimental procedures ...... 24 3.1.1. Reproduction of previously published syntheses (TiS02) ...... 24

3.1.2. Doping with CsCl, NH4Cl, and RbCl (TiS03–TiS05) ...... 25

3.1.3. Lower SiO2 and TiCl3 concentrations (TiS08, TiS09) ...... 26 3.1.4. Agitation and different cooling rates (TiS11) ...... 26 3.1.5. Using anatase as Ti source and KF as F- source (TiS15) ...... 28 3.1.6. Seeds (TiS16) ...... 29

3.1.7. Adding TiCl3 solution prior to KOH (TiS20) ...... 29 3.1.8. Different fill levels of the autoclaves (TiS21–TiS23) ...... 30 3.1.9. Temperatures between 210 and 250 ◦C (TiS25–TiS28) ...... 30 3.1.10. Agitation and high temperature (TiS30–TiS31) ...... 31 3.1.11. Synthesis of UND-1 (TiS32–TiS33) ...... 31 3.1.12. Using HF as fluorine source (TiS34–TiS35) ...... 32 3.1.13. Adjusting the pH (TiS36–TiS42) ...... 32 3.1.14. Using higher concentrations of HCl (TiS43–TiS46) ...... 34 3.1.15. Ti deficiency (TiS48) ...... 34 3.2. Conclusion ...... 37 3.3. SEM images ...... 41 References ...... 59

3 Contents

4. Ion exchange and dehydration of Rb-exchanged AM-2 60 4.1. Abstract ...... 60 4.2. Experimental procedure ...... 60 4.3. Results ...... 64 4.4. Discussion ...... 69 4.5. Tables ...... 75 References ...... 84

5. Structural characterisation of ion-exchanged AM-2 85 5.1. Abstract ...... 85 5.2. Experimental procedure ...... 85 5.3. Results ...... 87 5.3.1. Ion exchange ...... 87 5.3.2. Thermo-gravimetric analysis ...... 89 5.3.3. Structure reﬁnement ...... 89 5.3.4. Dehydration experiments ...... 95 5.4. Discussion ...... 99 5.5. Figures ...... 102 5.5.1. X-ray diffraction patterns ...... 102 5.5.2. Structures ...... 110 5.6. Tables ...... 113 5.6.1. Cell parameters ...... 113

5.6.2. Atomic coordinates, site occupancies, and Beq values ...... 115 5.6.3. Bond angles and distances ...... 118 5.6.4. Bond valence calculations ...... 125 5.6.5. Peak lists ...... 128 References ...... 130

6. Schreyerite, V2Ti3O9: New Occurrence and Crystal Structure 131 6.1. Abstract ...... 131 6.2. Introduction ...... 131 6.3. Geological setting and occurrence of schreyerite ...... 133 6.4. Experimental procedure ...... 134 6.4.1. Chemical analysis ...... 134 6.4.2. Structure determination ...... 134 6.5. Results ...... 135 6.5.1. Chemical composition ...... 135 6.5.2. Crystal structure ...... 136 6.6. Discussion ...... 138 6.7. Acknowledgements ...... 144

4 Contents

References ...... 144

7. Heulandite-Ba, a new zeolite species from Norway 147 7.8. Abstract ...... 147 7.9. Introduction ...... 147 7.10. Heulandite-Ba occurrences ...... 148 7.11. Morphology, physical and optical properties ...... 148 7.12. Chemical composition ...... 150 7.13. X-ray crystallography and crystal structure determination ...... 151 7.13.1. Experimental procedures and results ...... 151 7.13.2. The tetrahedral framework ...... 153 7.13.3. The extra-framework sites ...... 154 7.14. Acknowledgements ...... 156 7.15. References ...... 156

8. The crystal structure of painite CaZrB[Al9O18] revisited 158 8.16. Abstract ...... 158 8.17. Introduction ...... 158 8.18. Experimental methods ...... 159 8.19. Discussion ...... 159 8.20. Acknowledgments ...... 161 8.21. References cited ...... 161

A. Appendix 162 A.1. A sample Fullprof input file ...... 162 A.1.1. Input PCR file with comments ...... 162 A.1.2. The input PCR file ...... 166

B. Acknowledgements 170

C. Curriculum Vitae 171 C.1. Personal Details ...... 171 C.2. Education ...... 171

5 1. Abstract

Microporous titanosilicates have attracted great attention among mineralogists and material scientists during the last decades. Their variable chemical compositions and structural features suggest high potential for technical applications in the classical fields of zeolites such as ion-exchange, catalysis, adsorption, and gas separation, but also in new areas like optoelectronics, batteries, non-linear optics, magnetic materials and sensors. Some materials have been investigated thoroughly and are still subject of many studies (e. g. ETS-10, CST), whereas others have merely been synthesised without further investigation of the structure. The microporous structure AM-2 is an analogue of the natural zirconosilicate umbite, and has been synthesised in various chemical compositions (K2MSi3O9 · H2O, with M = Ti, Zr, Sn, Pb and others). Although several synthesis instructions have been published to date, none of them produces crystals large and pure enough for single-crystal X-ray diffraction (XRD). This technique, however, is preferred for structural analyses due to its higher resolution compared to powder XRD. In the first part our study thus focuses on the optimisation of AM-2 synthesis, striving for larger crystals suitable for single-crystal XRD. Synthesis instructions using colloidal silica and dissolved TiCl3 for hydrothermal reactions turned out to be most suitable for a starting point. By changing various parameters we found a de- pendence between crystal intergrowth and pH of the reaction gel, and to a limited extent between HCl concentration and crystal size. Maximum crystal sizes up to 30 µm were obtained, which is, however, too small for single-crystal XRD. On the other hand, some synthesis products contained aggregates with a high specific surface and crystallite sizes in the range of 1–20 µm, which guaranteed fast ion-exchange reactions and high-quality powder XRD patterns with sharp reflections and low risk for orientation and absorption effects. The ion-exchange properties of AM-2 of initial composition K2TiSi3O9 · H2O were analysed by exchanging K+ for various mono- and divalent cations, namely Rb+, Cs+, Sr2+, Na+, Mn2+, Ca2+, and Cu2+. The exchanged structures were analysed for their dehydration behaviour with the goal to provide a comprehensive characterisation of the Ti-AM-2 structure in terms of selectivity and exchange capacity for various cations, as well as thermal stability in general and as a function of the incorporated cation species. The ion exchange reactions were monitored with ICP-OES and Rietveld refinement of powder XRD data. Dehydration was analysed with powder XRD and thermo-gravimetric analyses. In conclusion, the Ti-AM-2 structure has almost 100% exchange capacity for all analysed cations. The kinetics of the exchange reaction depend on the ionic radius and are faster for small cations than for large ones. In fully hydrated state the H2O concentration varies between 1 and 3 molecules per formula unit. H2O may easily be expelled by heating to 400–500 ◦C. For K- and Rb-bearing AM-2 the dehydration is completely reversible without compromising the structural integrity, whereas all other varieties gradually break down and reach complete amorphisation between 250 and 500 ◦C, depending on the cation species in the cavities. At higher temperatures (700–750 ◦C) new phases crystallise, with structures determined by the exchangeable cations. This phase change is irreversible.

6 2. Introduction

Microporous titanosilicates (TS) constitute a novel family of zeotype materials built from

TiO6-octahedra and SiO4-tetrahedra. Some species show promising properties as catalysts, molecular and ionic sieving materials, and ion exchangers. Some of the various synthetic structures known to date are analogues of natural minerals (tab. 2.1). Due to their large pore aperture of 8 A,˚ ETS-10 (Engelhard titanosilicate number 10) and its derivatives (e. g. the Al-substituted ETAS-10) belong to the most interesting structures for technical applications. Since the first patents on the syntheses of microporous titanosilicates were issued to Kuznicki (1989, 1990) and Kuznicki and Thrush (1990), many reports on the syntheses from different titanium sources and with different organic additives have been published. Fur- thermore, several reports on applications like removal of harmful cations from wastewater (Behrens and Clearfield, 1997; Kuznicki and Thrush, 1991), and the usage of titanosilicates as catalyst in chemical reactions were published (Liepold et al., 1997). Although most of the early reports focused on titanosilicate structures, isostructural materials of other chemical compositions have become a popular field of research recently. Besides titanosilicates more than 20 natural and synthetic zirconium silicates are known to date (Rocha and Anderson, 2000), and for about one-third of them the crystal structures have been solved. The substitution of titanium for zirconium is rather common in natural minerals, hence zirconium silicates often contain a considerable amount of Ti and are structurally closely related to the group of titanosilicates. It is therefore reasonable to expect similar physico-chemical properties for both groups of materials. Interest in the synthesis and structural properties of zirconosilicates has been aroused only in recent years, and often the ion-exchange behaviour and crystal structures have not been studied in detail yet. Lin et al. (1999a) reported the synthesis and structural characterisation of analogues to several natural microporous zirconosilicates like petarasite (AV-3), gaidonnayite (AV-4), and umbite (AM-2). Furthermore Ferreira et al. (2001) synthesised and characterised the synthetic analogues to kostylevite (AV-8), petarasite (AV-3), umbite (AM-2), wadeite, and parakeldyshite (tab. 2.1). The following sections give an overview over some of the most common titano- and zirconosilicate structures that were synthesised or found in nature, and discuss the synthesis, structural topology, and the thereof resulting physical and chemical properties. A more comprehensive list is shown in Table 2.1 or in Ilyushin and Blatov (2002). In addition to this chapter, a review published by Rocha and Anderson (2000) is recommended as introduction to the research on titanosilicates and related structures.

7 2. Introduction

Table 2.1.: Overview of some synthetic microporous silicates and their natural analogues. A more comprehensive list including non-porous materials can be found in Ilyushin and Blatov (2002).

Name Composition Isostr. Analogue Reference Symmetry

AV-3 Na5Zr2Si6O18(Cl,OH) · 2H2O Petarasite Lin et al. (1999a) monoclinic AV-4 Na2ZrSi3O9 · 2H2O Gaidonnayite Lin et al. (1999a) orthorhombic AV-5 Na4K2Ce2Si16O38 · 10H2O Montregianite Ananias et al. (2001) monoclinic AV-6 K2SnSi3O9 · H2O Umbite Lin et al. (1999b) orthorhombic AV-7 K2SnSi3O9 · H2O Kostylevite Lin and Rocha (2001) monoclinic AV-8 Na0.2K1.8ZrSi3O9 · H2O Kostylevite Ferreira et al. (2001) monoclinic AV-9 Na4K2Eu2Si16O8 · 10H2O Montregianite Ananias et al. (2001) monoclinic AV-11 (K2SnSi3O9 · H2O) unknown Lin and Rocha (2001) unknown

AM-1 Na4Ti2Si8O22 · 4H2O unknown Lin et al. (1997) tetragonal AM-2 K2(Ti,Zr)Si3O9 · H2O Umbite Lin et al. (1997) orthorhombic AM-3 K2(Ti,Zr)Si3O9 · H2O Penkvilksite 2O Lin et al. (1997) orthorhombic AM-4 Na3(Na,H)Ti2O2[Si2O6]2 · 2H2O unknown Lin et al. (1997) unknown AM-6 (ETS-10) Brandao˜ et al. (2002) AM-11 unknown unknown Rocha et al. (1998a) unknown AM-13 HNaCa2Si10VO47 unique Brandao˜ et al. (2002) triclinic AM-14 Na2Si4VO11 unique Brandao˜ et al. (2002) triclinic

ETS-4 M6Ti3Si18O25 · nH2O (M = Na, K) Zorite Xu et al. (2001) orthorhombic ETS-10 A M2TiSi5O13 · nH2O (M = Na, K) unique Anderson et al. (1994) tetragonal ETS-10 B M2TiSi5O13 · nH2O (M = Na, K) unique Anderson et al. (1994) monoclinic UND-1 Na2.7K5.3Ti4Si12O36 · 4H2O unique Liu et al. (1997) monoclinic JDF-L1 Na4Ti2Si8O22 · 4H2O unknown Lin et al. (1997) tetragonal TAM-5 (Na,H)2Ti2O3(SiO4) · 2H2O Sitinakite Anthony et al. (1994) tetragonal

2.1. Zorite, ETS-4

Zorite is an orthorhombic titanosilicate mineral that was discovered in 1973 in the Lovozero massif, Kola Peninsula, Russia. The structure was solved in 1979 by Sandomirskii and Belov (1979) and was found to be related to the structure of nenadkevichite (Perrault et al., 1973). Zorite is characterised by a two-dimensional system of channels: A set of 12-membered channels runs in direction [001] and is intersected by a set of 8-membered channels running in direction [010]. The strong diffuse reﬂections usually found in X-ray diffraction patterns were interpreted as a proof for a high degree of disorder. These stacking faults interrupt the large 12-membered channels and thus reduce the adsorption characteristics of disordered zorite to a level far inferior to that expected from ordered material (Philippou and Anderson, 1996). A synthetic counterpart to zorite, ETS-4, was discovered by Kuznicki (1990). However, several authors found slight deviations between the structures of ETS-4 and zorite (Philip- pou and Anderson, 1996; Valtchev et al., 1996; Cruciani et al., 1998). Differences were mainly visible in 29Si MAS n.m.r results, whereas XRD patterns of zorite and ETS-4 are very similar. Possible explanations for these deviations are either differences in chemical compositions such as trace elements and isomorphous substitutions (Valtchev et al., 1996), or differences in the crystal structures. Philippou and Anderson (1996) described the structure of ETS-4 as

8 2.2. ETS-10 an intergrowth of a zorite-type and a nenadkevichite-type structure. However, they were not able to solve the structure of ETS-4 completely. Cruciani et al. (1998), who ﬁnally solved the structure in 1998, found the previously proposed defects in the 12-membered ring system or nenadkevichite-like intergrowths not to be signiﬁcant in their sample. Instead they suggest an interference effect related to 29Si–1H dipolar relaxation processes to be responsible for the lack of reliability and reproducibility in the 29Si MAS n.m.r. measurements. He also found hydrogen bonds to play an important role in the zorite and ETS-4 structure, which is an explanation for the low thermal stability (up to 200–250 ◦C) reported in several studies (Kuznicki, 1989; Kuznicki and Thrush, 1990; Valtchev et al., 1996).

2.2. ETS-10

The framework of ETS-10 consists of corner sharing TiO6-octahedra and SiO4-tetrahedra, which build the negatively charged lattice. As in most microporous structures the charge imbalance is compensated by the integration of cations, mainly Na+ and K+, into the voids (Liepold et al., 1997). Although the first report of ETS-10 syntheses was published in 1989 by Kuznicki (1989), the first successful structure solution was reported by Anderson et al. (1994), who used HR electron microscopy, electron and powder XRD, solid-state NMR, molecular modelling, and chemical analysis to characterise the material. The ETS-10 structure can be decomposed into sheets parallel to the (001) plane, which contain two sets of orthogonal Ti–O–Ti chains. Adjacent sheets can be displaced 0.25 in directions [100] and [010], leading to eight possible locations. The disorder in a random stacking sequence and the small crystal size of the synthetic material (≈ 5 µm, Anderson et al. (1994)) only allow powder XRD analysis and are the reasons for the long period between the discovery and the first successful structure determination. The two ordered phases polymorph A (stacking sequence 1, 3, 1, 3, . . . ) and polymorph B (stacking sequence 1, 3, 6, 8; figure 2.1 left shows all possible displacements) are of special interest due to their regular arrangement of 12-membered rings leading to open channels with a diameter of 7.6 × 4.9 A˚ (fig. 2.1).

The synthesis of ETS-10 is usually performed in the Na2O–K2O–TiO2–SiO2–H2O system. Kuznicki (1989) added ETS-4 crystals as seeds to stimulate crystal growth. Many subsequent studies reported positive effects of organic additives and different precursor materials in terms of improved crystallisation time and higher degree of crystallinity. The inﬂuence of parameters such as temperature, concentration of the ingredients and the amount of seed was investigated by Das et al. (1995). Since seeds of ETS-10 did not show any improvements in reduction of the crystallisation time or in the degree of crystallinity of the product they used ETS-4 seeds instead, which were also more easily obtained. A standard synthesis (Das et al., 1995) starts by preparing two solutions A and B. Solu- tion A contains 63 g of sodium silicate (28.6% SiO2, 8.82% Na2O, 62.58% H2O) and 20 g of

9 2. Introduction

a a b

b 0 1/4 1/2 3/4 1 c 1 2

1/4 3 4

1/2 5 6

3/4 7 8

Figure 2.1.: Left: A projection along [110] shows the large 12-membered channels in the regular stacked polymorph B of ETS-10. Atomic coordinates were published by Anderson et al. (1994). Right: The black dots represent all possible displacements in a stacking sequence in ETS-10. distilled water and is stirred vigorously. Solution B (8.4 g NaOH pellets dissolved in 58.8 g of distilled water) is added slowly under stirring. The gel must be stirred for 15–20 minutes and 54.4 g of TiCl3 (15% solution in HCl) are added dropwise. After another 30 minutes stirring 9.4 g KF · 2H2O must be added and the mixture is stirred for one hour. Finally 1.3 g seeds of ETS-4 are added very slowly and stirred vigorously for one hour until the paste becomes homogeneous (pH 10.8–11.0). Crystallization is carried out in a tightly-capped stainless-steel autoclave at 443 K for ten days. The solid material is then ﬁltered off from the solution, washed with deionized water and dried.

Syntheses by the above procedure with input ratios SiO2/TiO2 = 3, 4, and 7.5 and ETS-4 seeds resulted in either a mixture of ETS-10 and ETS-4 (SiO2/TiO2 = 3, 4), or less crystalline

ETS-10 (SiO2/TiO2 = 7.5) as compared to the ideal input ratio SiO2/TiO2 = 5.7. ETS-10 crystallises as sharp, twinned cuboids with sizes in the range of 1–4 µm. On decreasing seed concentration, the cuboids tend to form larger clusters, but the crystallinity is not inﬂuenced by variations in the quantity of seeds used. Das et al. (1995) also performed syntheses of ETS-10 without seeds. The product, however, was not pure ETS-10 but a mixture of ETS-10 and ETS-4. Valtchev and Mintova (1994) studied the inﬂuence of the organic template tetramety- lammonium (TMA) chloride on the crystallisation of ETS-10. The hydrothermal synthesis was carried out using gels of the molar composition 40 R : 52 Na2O : 42 K2O : 20 TiO2 :

100 SiO2 : 7030 H2O, with R being TMACl. The inorganic reactants used were Na2SiO3 · nH2O (Na2O 18%; SiO2 63%), TiCl4, NaOH, KOH, and distilled water. The experiments were carried out in Teﬂon-lined stainless-steel autoclaves at temperatures of 160, 180, and

10 2.2. ETS-10

200 ◦C. Analyses of the crystals with various techniques displayed a susbtantial difference in the concentration of sodium and water compared to ETS-10 without organic additives.

The chemical compositions were TMA8Na20K6Ti20Si100O252 · 32H2O for TMA-ETS-10, and

Na26K6Ti20Si100O252 · 66 H2O for ETS-10, respectively. Keeping in mind that the size of a hydrated Na ion is close to that of TMA, Valtchev and Mintova suppose that the TMA ions replace the hydrated sodium ions that do not balance framework cations in the ETS- 10 structure. Their study shows that the crystallisation time for ETS-10 is shorter and the product reaches a higher degree of crystallinity if a TMA template is present.

Not only TMACl, but also other organic templates like choline chloride (OHCH2CH2 + − − + − (CH3)3N Cl ) and the bromide salt of hexaethyl diquat-5 (Br (C2H5)3N (CH2)5N (C2H5)3 Br−) show an improving effect on the crystallisation time and the degree of crystallinity in ETS-10 syntheses (Das et al., 1996a). Titanosilicates crystallised in the presence of choline chloride show agglomerates of cuboidal crystals (≤ 0.5 µm) with cuboidal growing clusters (2–4 µm), whereas the crystals produced in presence of hexaethyl diquat-5 form sheaf- shaped polycrystallites (2–4 µm). The use of hexaethyl diquat-5 also leads to purer material than cholin chloride. The inﬂuence of temperature, water content, template concentration, and titanium content, as well as the results of IR studies, TGA/DTA, MAS n.m.r, sorption studies, and catalytic activities are discussed by Das et al. (1996a) and Das et al. (1996c). Besides the presence of organic templates the titanium source plays an important role on the quality (purity, degree of crystallinity, crystal size) of synthetic titanosilicates. The

ﬁrst report on the synthesis of ETS-10 (Kuznicki, 1989) describes a reaction using TiCl3 as titanium source. In order to optimise batch processes and to improve the results of the syntheses, several authors analysed the inﬂuence of different titanium sources on the synthesis products. Rocha et al. (1998b) used TiCl3 and TiO2 in the form of anatase, and found both titanium precursors to be suitable for pure and highly crystalline ETS-10 preparation.

However, TiO2 leads to smaller crystals (≈ 0.5 µm) than TiCl3 (≈ 25 µm). The largest crystals were achieved from a gel with the composition 3.4 Na2O : 1.5 K2O : TiO2 : 5.5 SiO2 :

125 H2O without seeds during 60 hours synthesis time. The use of P25, a mixture composed of 76 wt-% anatase and 24 wt-% rutile, for synthesis of pure ETS-10 was published by Liu and Thomas (1996).

A synthesis using TiCl4, which is less expensive than TiCl3, was ﬁrst published by Das et al. (1996b). Using Ti(SO4)2 as titanium source, Kim et al. (2000) produced almost 100% pure ETS-10 within 18 hours. However, not the Ti compound but a high water content at low alkalinity was found to be responsible for these improvements.

Another possible replacement for the titanium chlorides is TiF4, which is stable in water and in alkaline solutions. Yang et al. (2001) compared the products of syntheses with TiF4 (with and without organic additive TMACl) and P25, and found small amounts of X-ray amorphous TiO2 in the product of the P25 reaction (crystals up to 1 µm in size), whereas pure ETS-10 crystals (1 µm in the presence of TMACl, ≈ 4 µm without TMACl) were ob-

11 2. Introduction

29 tained from TiF4. The resolution of Si MAS n.m.r spectra was better for ETS-10 samples from organic-free syntheses.

2.3. Penkvilksite, ETS-14, AM-3

Penkvilksite is a natural titanosilicate mineral with the ideal chemical formula Na4Ti2Si8O22

· 4H2O that was found in Lovozero, Kola Peninsula, Russia in 1974. Two polytypes were found in Nature. Penkvilksite-2O, discovered and described by Bussen et al. (1974), is orthorhombic, whereas penkvilksite-1M, described by Merlino et al. (1994), is the monoclinic variety. Both the 1M and the 2O polytype have minor substitutions of Ca, Zr, Fe and Al for Na, Ti, and Si, respectively (Liu et al., 1999). The structure of penkvilksite is shown in ﬁgure 2.2 and basically consists of two types of tetrahedra, Si1 and Si2, and one octahedron. Si1 shares two corners with other tetrahedra and two corners with octahedra, whereas Si2 shares three corners with tetrahedra and one corner with an octahedron. Spirals of corner-sharing tetrahedra with a periodicity of six develop along [010]. The stacking of spirals along [001] forms layers of tetrahedra parallel to (100). Neighbouring layers are connected by TiO6 octahedra. Oxygen atoms of the octahedra are all linked to one Ti, one Si, and one Na atom, leading to very regular octahedral coordination around Ti. Additional cations are placed on both sides of the tetrahedral layers in sevenfold coordination (ﬁve O atoms and two H2O molecules). O atoms of H2O molecules are linked to Na and form H bonds to two tetrahedral vertices. The polytypes penkvilksite-1M and penkvilksite-2O represent two maximum degree of order (MDO) polytypes within a family of OD structures built up from two kinds of layers. Four polytypes result from these stacking variations:

• MDO1: 1¯ at 00 and 21 parallel to b (P21/c)

¯ 1 1 • MDO2: 1 at 4 4 and 2 parallel to b (I2/c)

• MDO3: n perpendicular to a and 2 parallel to c (Pnca)

• MDO4: m perpendicular to a and 21 parallel to c (Pmcn)

The monoclinic penkvilksite-1M corresponds to the polytype MDO1, the orthorhombic penkvilksite-2O corresponds to MDO3. Merlino et al. (1994) suggest to call MDO3 penkvilksite-2O1, and the two other MDO polytypes penkvilksite-2M (MDO2) and penkvilksite-2O2

(MDO4). However, the occurrence of penkvilksite-2O2 is unlikely because of crystal chemical constraints. Noticeable differences in the structural arrangements in MDO2 and MDO4 with respect to MDO1 and MDO3 were found. Instead of a 21 screw axis along [010] or a (100) glide plane in MDO1 and MDO3, MDO2 and MDO4 contain twofold axes or a mirror plane, reducing the SiO4 spirals to closed six-membered rings.

12 2.3. Penkvilksite, ETS-14, AM-3

Penkvilksite−2O

a a b Penkvilksite−1M

b c c

Figure 2.2.: The two polytypes penkvilksite-2O (left and centre) and penkvilksite-1M (right) represent the maximum degree of order within a family of OD structures (Merlino et al., 1994).

The synthesis of AM-3, a structural analogue of natural penkvilksite-2O, was published by Lin et al. (1997). An alkaline solution was made by mixing 12.30 g of sodium silicate solution, 2.45 g of NaOH, 0.96 g of NaCl, 1.0 g of KCl, and 5.52 g of H2O. TiCl3 (10.60 g;

15 mol-% solution of TiCl3 in 10 mol-% HCl) was added to the solution and stirred thoroughly. The gel with a composition 5.3 Na2O : 0.7 K2O : 5.3 SiO2 : 1.0 TiO2 : 116 H2O was autoclaved under autogeneous pressure up to 17 days. The samples obtained after 4– 7 days represent a mixture of AM-1, AM-3, ETS-10, and ETS-4. When AM-3 seeds were added to the starting gel the crystallisation time decreased. After 4 days AM-3 became the main phase, contaminated with only small amounts of ETS-10. The resulting crystals were developed as thin plates of maximum 100 µm in diameter. An analogue of penkvilksite-1M was synthesised by Liu et al. (1999). Tetrabutyltitanate (98%), fumed silica (99%) and sodium hydroxide (96%) were used to produce a gel with the initial composition 3.7 SiO2 : 5.9 TiO2 : 1.0 Na2O : 100 H2O : 0.6 (TMA)2O. Sodium hydroxide (0.27 g) was dissolved in 12 ml distilled water. Under vigorous stirring, 5.1 g

Ti(OC4H9)4, 2.5 ml of aqueous TMAOH, and 0.6 g fumed silica were slowly added to the solution. The mixture was stirred until it became homogeneous, then transferred into a teﬂon-lined stainless-steel autoclave, and heated at 473 K for about 20–30 days. The crystals were isolated by ﬁltration, washed with water and dried at 353 K.

The TiO2/SiO2 ratio was found to be important in order to prevent formation of amorphous phases or quartz impurities. Liu et al. (1999) gained the best results with TiO2/SiO2

= 1.53. The resulting crystals (after 35 days crystallisation time) showed a TiO2/SiO2 ratio − of 0.181 and contained no quartz or amorphous impurities. The OH /SiO2 ratio affects

13 2. Introduction

− the crystallisation of penkvilksite-1M, too. Pure crystals were obtained with OH /SiO2 = 0.2–0.6. Higher ratios lead to formation of zorite and amorphous impurities. Alkali metal cations have an important effect on the crystallisation of titanosilicates. In + + + + SiO2–TiO2–M2O–H2O–(TMA)2O (M = Na ,K , or Na + K ) systems the alkali ions inﬂu- ence the titanosilicate species that crystallises: The presence of K+ leads to the formation of ETS-10, whereas at low Na+ penkvilksite-1M, and at high Na+ ETS-4 is obtained. The cation exchange capacity (CEC) of penkvilksite for K+ is relatively low (0.72 meq/g) compared to other titanosilicates and zeolites (2.2–5.3 meq/g, Pabalan and Bertetti (2001)).

2.4. Umbite, AM-2

Umbite is a rare natural zirconosilicate with the ideal formula K2ZrSi3O9 · H2O that was found in the Khibiny alkaline massif on Kola peninsula, Russia. Ilyushin (1993), who described the structure in 1993, found the unit cell to be monoclinic but close to orthorhombic (a = 10.207(2), b = 13.241(4), c = 7.174(1) A,˚ α = 90.01(2)◦, β = 90.01(2)◦, γ = 90.07(2)◦). How- ever, since reﬁnements in the orthorhombic system yielded lower R values, subsequent structure reﬁnements were usually performed in the orthorhombic space group P212121. Natural samples usually show variations in chemical compositions, and different compositions have been proposed for natural umbite:

• KZr(Si2O6)(OH,F) · nH2O, determined by chemical analysis (Ilyushin et al., 1982b)

• K2ZrSi3O9 · H2O, initially reﬁned composition (Ilyushin et al., 1982a)

• K2(Zr0.8Ti0.2)Si3O9 · H2O for umbite, and (K,H)2ZrSi3O9 · nH2O for paraumbite, based on chemical analysis (Khomyakov, 1990)

• K2(Ti0.31Zr0.69)Si3O9 · H2O, revised chemical composition from X-ray structure reﬁne- ment (Ilyushin, 1993)

A monoclinic polymorph with space group P21/c named kostylevite was also found in Na- ture (Ilyushin et al., 1982a). Although substitution of Ti for Zr in potassium zirconosilicates is common, natural occurrences of a Ti endmember have not been found yet, but successful synthesis of the complete solid solution from K2ZrSi3O9 · H2O to K2TiSi3O9 · H2O has been reported (Clearfield et al., 1998; Poojary et al., 1997). A detailed description of an umbite-like framework structure is given in chapter 4 in this work. The ion exchange properties of umbite have been described by Jale et al. (1999). Refluxing a sample in 0.5 M NaCl solution changed the cation composition from 29% Na and 71% K in the original sample to 81% Na and 21% K in the exchanged sample. The Si/Zr ratio of the sample was 3.6, and substitution of Al for Si led to a Si/Al ratio of 76.6. The crystal lost about 13% of its weight upon ignition. The results confirm that at least 80% of the extra-framework cations are exchangeable and Zr and Al are part of the framework.

14 2.5. Silico-titanate, CST, TAM-5

Table 2.2.: Physical and chemical properties of the inorganic sorbents chabazite zeolite and crystalline silicotitanate CST (Bostick, 1999).

Chabazite zeolite CST (IONSIV IE-911) Source GSA Resources, Inc. UOP Molecular Sieves Form Naturally occurring, inorganic Inorganic, engineered pellets zeolite of powder; framework of alu- minosilicate Exchangeable cation Sodium Sodium and hydrogen Cost $ 102/ft3 $ 7200/ft3 Bulk density, g/cm3 0.7 1.0 Particle density, g/cm3 1.73 2.0 Average particle size, µm 480 ± 220 410 ± 110 Moisture content, % 7.71 5.85 (6.07 % for IONSIV IE-910) Order of selectivity in process Na < Mg < Ca < Sr < Cs Mg < Ca < Na < K < Sr < Cs waste simulant Column operating characteris- Material is friable; tends to Material appears to be struc- tics break down prior to complete turally stable; in column test loading, causing column plug- over 10 months, no noticeable ging plugging or ﬁnes produced Pretreatment for near-neutral- Sieve; wash with 2M NaCl; Sieve, wash with H2O; wash pH waste treatment wash with H2O; air dry with 0.1M HCl Ion-exchange capacity, meq/g 2.2 2.5

The synthesis of an umbite-type material named AM-2 was described by Lin et al. (1997). The structure was reﬁned from single-crystal X-ray diffraction data and was published by Zou and Dadachov (2000). A detailed analysis on the synthesis and ion exchange properties of synthetic AM-2 follows in this work.

2.5. Silico-titanate, CST, TAM-5

The term ’silico-titanate’ is a synonym for ’silicon titanate’, ’titanium silicate’, ’silicotitanate’, ’TiSi’ and ’CST’ (crystalline silico-titanate), but it is primarily conﬁned to the description of the group of TAM (Texas A&M University) microporous titanosilicates. These materials became generally known as the most promising materials for the selective sorption of radioactive caesium, strontium, and plutonium over a wide pH range (Anthony et al., 1994). A decrease of the selectivity for caesium in highly alkaline solutions with high sodium concentrations can be compensated by doping with small amounts of niobium, which is incorporated into the framework structure. Many authors use the term ’crystalline silicotitanate’ (CST) as a synonym of TAM-5, the only single-phase material in the group of TAM microporous silicotitanates. This material shows the highest selectivity for caesium and outstanding stability against radiation. The selectivity for caesium is preserved even after exposure to 109 rads (Anthony et al., 2000). Due to these exceptional properties Nb-doped TAM-5 is commercially manufactured by Union Oil Products UOP and available as powder (IONSIV

15 2. Introduction

Na2 Na1

y y z x Figure 2.3.: The framework structure of the microporous crystalline sodium silicotitanate + + TAM-5 projected along [100] and [001]. The channels contain Na /H cations and H2O molecules, and are highly selective for Cs+ (Poojary et al., 1994).

IE-910) and pellets (IONSIV IE-911) (Nyman et al., 2001). The usability of different sorbent materials for the removal of 90Sr and 137Cs from radioactive wastewater has been well investigated by the Oak Ridge National Lab, Tennessee (Bostick and DePaoli, 1999; Bostick, 2001)). Bostick (1999) summarises the advantages of CST compared to the baseline sorbent, chabazite zeolite, as follows:

• high selectivity for strontium—the more difﬁcult ion to capture

• very high selectivity for caesium

• ability to function in high-salt media

• greater sorption capacity for strontium, resulting in a substantial (3–30 times) reduction in ﬁnal waste volume (and disposal cost)

• superior mechanical stability, with no column plugging or ﬁnes generation

A comparison of the physical and chemical properties of chabazite and CST is also shown in Table 2.2. The structure of TAM-5 was solved by Poojary et al. (1994) from X-ray powder data by ab initio methods and is shown in ﬁgure 2.3. The ideal composition is Na2Ti2O3SiO4 · 2H2O, however, due to space limitations some of the Na+ ions are in fact replaced by H+, leading to an approximate composition Na1.64H0.36Ti2O3SiO4 · 2H2O (Moller,¨ 2002; Poojary et al., 1994).

The tetragonal structure of TAM-5 with space group P42/mcm comprises Ti–O octahedra located on a diagonal mirror plane close to the 42 axis. This arrangement generates clusters of four face-sharing TiO6 octahedra. The clusters are connected by SiO4 tetrahedra in directions [100] and [010], and by Ti–O–Ti linkages along [001]. This arrangement of silicate

16 2.6. Chemical variations tetrahedra and titanium octahedra creates channels parallel to the c axis. There are two different sodium sites in the structure. The fully occupied Na1 site bonds to four framework oxygens and two H2O molecules. It is therefore rigidly held in the lattice and is not exchanged for caesium ions. Na2, near the centre of the channel, is only occupied to 64%, the deﬁciency being compensated by protons. Two H2O molecules can be distinguished, one of which is part of the bonding environment of Na1 and is therefore present in both the initial sodium phase and the cesium-exchanged phase. The replacement of Na+ by Cs+ slightly increases the diameter of the channels (Na-CST: a = 7.8082(2) A,˚ c = 11.9735(4) A,˚ Cs-CST: a = 7.8258(2) A,˚ c = 11.9815(4) A˚ (Poojary et al., 1994)). Cs+ replaces Na2 and is located exactly in the centre of the tunnel, while Na1 retains its position. From the centre of the channel the caesium ion bonds to eight framework oxygens with almost ideal Cs–O bonding distances, which explains the high afﬁnity for Cs+ and the outstanding resistance against radiation, heat, and caesium leeching. A second caesium site is also located in the centre of the channel and bonds to four framework oxygens and two H2O molecules.

2.6. Chemical variations

The group of microporous zeotype framework-structures comprehends an enormous variety of chemical compositions, as almost all components of the structures are not limited 4+ 2− to a specific element. The framework structure is built from two components, an M O6 4+ 2− octahedron and a T O4 tetrahedron. In both units the central cation is variable and often solid solutions between several elements are observed. Common elements on the octahedral position are Ti, Zr, Sn, Nb, Y, Pb, and V (Rocha et al., 1998a, 1997; Pertierra et al., 2001). While natural occurrences are usually limited to a solid solution of Ti and Zr, other variations were produced in synthetic materials (Lin et al., 1999b; Pertierra et al., 2002) as pure endmembers, solid solutions, or endmembers doped with small amounts of a substituent. Similar variations are observed on the tetrahedral position, on which the most common element is Si. In Nature substitution of Si for Al was observed (Jale et al., 1999), which is a common feature already known from zeolites. In synthetic structures Si can be partially or completely replaced by Ge (Clearfield, 2001; Sun et al., 2000). The chemical variations mentioned above are largely determined by the composition of the precursor and cannot be altered once the material has formed, whereas the weakly bound channel occupants can in many cases easily be exchanged for other species. Mi- + croporous structures usually incorporate mono- or divalent cations, H2O, NH4 , other small organic or inorganic molecules, or are protonated on the inner surface. In natural materials the channel composition often reflects the chemical composition of the last hydrothermal environment the material was exposed to. Since Na, K, Ca, and Mg are relatively abundant in fluids and compatible with many microporous structures, they are usually found

17 2. Introduction as major constituents in the pore system of naturally occurring microporous materials. In synthetic materials it is a common technique to use the same cations as in Nature as initial channel occupants during the synthesis, and to exchange them later in a hydrothermal ion exchange for the species of interest. Another chemical variation indirectly related to the crystal chemistry is caused by surface effects of charged microporous materials. Surface effects describe the accumulation of dipolar molecules on the surface of zeotype materials, held electrostatically by the negatively charged framework. This feature is used to effectively invert the surface charge of particles, which opens a new field of application in the purification of water from harmful microorganisms (Bowman, 2003). Although phases on the surface of microporous particles are usually not observed by X-ray diffraction, they often serve as an explanation for irregularities between structure refinements and other analytical techniques. For example hydrous phases on the crystal surface may lead to a higher weight loss upon dehydration, which cannot be explained by structural data. They may also tamper the composition of ion-exchange solutions leading to erroneous exchange capacities. The manifoldness of microporous materials makes them interesting candidates for various technical applications, reaching from the treatment of wastewaters (Nyman et al., 2001), to agricultural applications, and to opto-electronic high-tech components (Rocha and An- derson, 2000), just to name a few. With the proceeding exploration of new structures, new variations of known phases and new applications, the demand for researchers systematically investigating these materials rises. Although chemical variations sometimes only have marginal influence on the structure topology, they may change the selectivity for certain cation species, the preference for one element over others, or the stability under certain pT conditions drastically. Systematic investigation of the physico-chemical properties is therefore crucial for the optimisation of existing techniques, as well as for the development of new fields of application.

References

Ananias, D., Ferreira, P., Ferreira, A., Rocha, J., Rainho, J. P., Morais, C. M., and Carlos, L. D. Synthesis and characterisation of novel microporous framework cerium and europium silicates. In Galarneau, A., Di Renzo, F., Fajula, F., and Vedrine, J., editors, Zeolites and mesoporous materials at the dawn of the 21st century, volume studies in surface science and catalysis of 135, page 246. Elsevier, 2001. 05-P-12.

Anderson, M. W., Terasaki, O., Ohsuna, T., Philippou, A., MacKay, S. P., Ferreira, A., Rocha, J., and Lidin, S:. Structure of the microporous titanosilicate ETS-10. Nature, 367:347–351, 1994.

Anthony, R. G., Dosch, R. G., Gu, D., and Philip, C. V. Use of silicotitanates for removing

18 2.6. Chemical variations

cesium and strontium from defense waste. Industrial & engineering chemistry research, 33: 2702–2705, 1994.

Anthony, R. G., Dosch, R. G., and Philip, C. V. Method of using novel silico-titanates. US Patent 6,110,378, 2000.

+ Behrens, E. A. and Clearﬁeld, A. Titanium silicates, M3HTi4O4(SiO4)3 · 4H2O (M = Na , K+), with three-dimensional tunnel structures for the selective removal of strontium and cesium from wastewater solutions. Microporous Materials, 11:65–75, 1997.

Bostick, D. T. Fission product separations testing. Technical Report TTP Number OR16C312, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6221, 1999.

Bostick, D. T. Innovative technology summary report: Fission products separations testing. Technical Report DOE/EM-0575, OST Reference #205, U. S. Department of Energy, 2001.

Bostick, D. T. and DePaoli, S. M. Evaluation of improved techniques for the removal of ﬁssion products from process wastewater and groundwater: FY 1998 and 1999 status. Technical Report ORNL, Oak Ridge National Laboratory, Lockheed Martin, 1999.

Bowman, R. S. Applications of surfactant-modiﬁed zeolites to environmental remediation. Microporous and Mesoporous Materials, 61(1-3):43–56, 2003.

Brandao,˜ P., Philoppou, A., Hanif, N., Ribeiro-Claro, P., Ferreira, A., Anderson, M. W., and Rocha, J. Synthesis and characterization of two novel large-pore crystalline vanadosili- cates. Chemistry of Materials, 14(3):1053–1057, 2002.

Bussen, I. V., Men’shikov, Y. P., Mer’kov, A. N., Nedorezova, A. P., Uspenskaya, Y. I., and Khomyakov, A. P. Penkvilksite, a new titanium-sodium hydrosilicate. Doklady Earth Science Section, 217:126–129, 1974.

Clearﬁeld, A. Structure and ion exchange properties of tunnel type titanium silicates. Solid State Sciences, 3:103–112, 2001.

Clearﬁeld, A., Bortun, A. I., Bortun, L. N., Poojary, D. M., and Khainakov, S. A. On the selec-

tivity regulation of K2ZrSi3O9 · H2O-type ion exchangers. Journal of Molecular Structure, 470:207–213, 1998.

Cruciani, G., De Luca, P., Nastro, A., and Pattison, P. Rietveld reﬁnement of the zorite structure of ETS-4 molecular sieves. Microporous and Mesoporous Materials, 21:143–153, 1998.

Das, T. K., Chandwadkar, A. J., Budhkar, A. P., Belhekar, A. A., and Sivasanker, S. Studies of the synthesis of ETS-10 I. Inﬂuence of synthesis parameters and seed content. Microporous Materials, 4:195–203, 1995.

19 2. Introduction

Das, T. K., Chandwadkar, A. J., Budhkar, A. P., and Sivasanker, S. Studies on the synthesis of ETS-10 II. Use of organic templates. Microporous Materials, 5:401–410, 1996a.

Das, T. K., Chandwadkar, A. J., and Sivasanker, S. A rapid method of synthesizing the titanium silicate ETS-10. Chemical Communications, 10:1105, 1996b.

Das, T. K., Chandwadkar, A. J., and Sivasanker, S. Studies on the synthesis, characterization and catalytic properties of the large pore titanosilicate, ETS-10. Journal of Molecular Catalysis A: Chemical, 107:199–205, 1996c.

Ferreira, P., Ferreira, A., Rocha, J., and Soares, M. R. Synthesis and structural characterization of zirconium silicates. Chemistry of Materials, 13:355–363, 2001.

Ilyushin, G. D. New data on crystal-structure of umbite K2ZrSi3O9 · H2O. Inorganic Mate- rials, 29(7):1128–1133, 1993.

Ilyushin, G. D. and Blatov, V. A. Crystal chemistry of zirconosilicates and their analogs: topological classiﬁcation of MT frameworks and suprapolyhedral invariants. Acta Crys- tallographica, B58:198–218, 2002.

Ilyushin, G. D., Khomyakov, A. P., Shumyatskaya, N. G., Voronkov, A. A., Nevskii, N. N., Ilyukhin, V. V., and Belov, N. V. Crystal structure of a new natural modiﬁcation of

K2ZrSi3O9 · H2O. Doklady Akademii Nauk SSSR, 256(4):860–864, 1982a. (in Russian).

Ilyushin, G. D., Pudovkina, Z. V., and Voronkov, A. A. Crystal structure of a new natural

zirconosilicate K4Zr2Si6O18 · H2O. Doklady Akademii Nauk SSSR, 257(3):608–611, 1982b.

Jale, S. R., Ojo, A., and Fitch, F. R. Synthesis of microporous zirconosilicates containing ZrO6

octahedra and SiO4 tetrahedra. Chemical Communications, 5:411–412, 1999.

Khomyakov, A. P. Mineralogy of ultraagpaitic alkaline rocks (in russian). Nauka, Moscow, 1990.

Kim, W. J., Lee, M. C., Yoo, J. C., and Hayhurst, D. T. Study on the rapid crystallization of ETS-4 and ETS-10. Microporous and Mesoporous Materials, 41:79–88, 2000.

Kuznicki, S. M. Large-pored crystalline titanium molecular sieve zeolites. US Patent 4,853,202, 1989.

Kuznicki, S. M. Preparation of small-pored crystalline titanium molecular sieve zeolites. US Patent 4,938,939, 1990.

Kuznicki, S. M. and Thrush, A. K. European Patent 0405978A1, 1990.

Kuznicki, S. M. and Thrush, K. A. Removal of heavy metals, especially lead, from aqueous systems containing competing ions utilizing wide-pored molecular sieves of the ETS-10 type. US Patent 4,994,191, 1991.

20 2.6. Chemical variations

Liepold, A., Roos, K., Reschetilowski, W., Lin, Z., Rocha, J., Philippou, A., and Anderson, M. W. Characterisation studies on the new microporous aluminium-containing ETS-10 molecular sieve used for processing larger molecules. Microporous Materials, 10:211–224, 1997.

Lin, Z. and Rocha, J. Novel microporous framework stannosilicates. In Galarneau, A., Di Renzo, F., Fajula, F., and Vedrine, J., editors, Zeolites and mesoporous materials at the dawn of the 21st century, volume 135 of studies in surface science and catalysis, page 246. Elsevier, 2001. 05-P-13.

Lin, Z., Rocha, J., Brandao,˜ P., Ferreira, A., Esculcas, A. P., de Jesus, J. D. P., Philippou, A., and Anderson, M. W. Synthesis and structural characterization of microporous umbite, penkvilksite, and other titanosilicates. Journal of Physical Chemistry B, 101:7114–7120, 1997.

Lin, Z., Rocha, J., Ferreira, P., Thursﬁeld, A., Agger, A. J., and Anderson, M. W. Synthesis and structural characterization of microporous framework zirconium silicates. Journal of Physical Chemistry B, 103:957–963, 1999a.

Lin, Z., Rocha, J., and Valente, A. Synthesis and characterisation of a framework microporous stannosilicate. Chemical Communications, 24:2489–2490, 1999b.

Liu, X., Shang, M., and Thomas, J. K. Synthesis and structure of a novel microporous ti-

tanosilicate (UND-1) with a chemical composition of Na2.7K5.3Ti4Si12O36 · 4H2O. Micro- porous Materials, 10:273–281, 1997.

Liu, X. and Thomas, J. K. Synthesis of microporous titanosilicates ETS-10 and ETS-4 using

solid TiO2 as the source of titanium. Chemical Communications, 12:1435, 1996.

Liu, Y., Du, H., Xu, Y., Ding, H., Pang, W., and Yue, Y. Synthesis and characterization of a novel microporous titanosilicate with a structure of penkvilksite-1M. Microporous and Mesoporous Materials, 28:511–517, 1999.

Merlino, S., Pasero, M., Artioli, G., and Khomyakov, A. P. Penkvilksite, a new kind of silicate structure – OD character, X-ray single-crystal (1M), and powder rietveld (2O) reﬁnements of 2 MDO polytypes. American Mineralogist, 79(11-12):1185–1193, 1994.

Moller,¨ T. Selective crystalline inorganic materials as ion exchangers in the treatment of nuclear waste solutions. Academic dissertation, University of Helsinki, Department of Chemistry, 2002.

Nyman, M., Nenoff, T. M., and Headley, T. J. Characterization of UOP IONSIV IE-911. San- dia Report SAND2001-0999, Sandia National Laboratories, Albuquerque, New Mexico 87185 and Livermore, California 94550, 2001.

21 2. Introduction

Pabalan, R. T. and Bertetti, F. P. Natural Zeolites: Occurrence, Properties, Applications, volume 45 of Reviews in Mineralogy & Geochemistry, chapter Cation-Exchange Properties of Natural Zeolites, pages 453–518. Mineralogical Society of America, 2001.

Perrault, P. G., Boucher, C., Vicat, J., Cannillo, E., and Rossi, G. Structure cristalline du

nenadkevichite (Na,K)2−x(Nb,Ti)(O,OH)Si2O6 · 2H2O. Acta Cristallographica, pages 1432– 1438, 1973.

Pertierra, P., Salvado,´ M. A., Garcia-Granda, S., Bortun, A. I., Khainakov, S. A., and Garcia, J. R. Hydrothermal synthesis and structural characterization of framework microporous mixed tin-zirconium silicates with the structure of umbite. Inorganic chemistry communications, 5:824–828, 2002.

Pertierra, P., Salvado,´ M. A., Garcia-Granda, S., Garcia, J. R., Bortun, A. I., Bortun, L. N., and

Clearﬁeld, A. Synthesis and structural study of K2PbSi3O9 · H2O with the structure of kostylevite. Materials Research Bulletin, 36:717–725, 2001.

Philippou, A. and Anderson, M. W. Structural investigation of ETS-4. Zeolites, 16:98–107, 1996.

Poojary, D. M., Bortun, A. I., Bortun, L. N., and Clearﬁeld, A. Synthesis and X-ray powder

structures of K2(ZrSi3O9) · H2O and its ion-exchanged phases with Na and Cs. Inorganic Chemistry, 36:3072–3079, 1997.

Poojary, D. M., Cahill, R. A., and Clearﬁeld, A. Synthesis, crystal structures, and ion- exchange properties of a novel porous titanosilicate. Chemistry of Materials, 6(12):2364– 2368, 1994.

Rocha, J. and Anderson, M. W. Microporous titanosilicates and other novel mixed octahedral-tetrahedral framework oxides. European Journal of Inorganic Chemistry, 2000 (5):801–818, 2000.

Rocha, J., Brandao,˜ P., Phillippou, A., and Anderson, M. W. Synthesis and characterisation of a novel microporous niobium silicate catalyst. Chemical Communications, 24:2687–2688, 1998a.

Rocha, J., Ferreira, A., Lin, Z., and Anderson, M. W. Synthesis of microporous titanosil-

icate ETS-10 from TiCl3 and TiO2: a comprehensive study. Microporous and Mesoporous Materials, 23:253–263, 1998b.

Rocha, J., Ferreira, P., Lin, Z., Brandao,˜ P., Ferreira, A., and de Jesus, J. D. P. The ﬁrst synthetic microporous yttrium silicate containing framework sodium ions. Chemical Commu- nications, 21:2103–2104, 1997.

Sandomirskii, P. A. and Belov, N. V. Sov. Phys. Crystallogr., 24:686–693, 1979.

22 2.6. Chemical variations

Sun, K., Dadachov, M. S., and Zou, X. D. A three-dimensional open-framework germanate containing four-, ﬁve- and six-coordinated germanium. Acta Crystallographica Section C, 56:1092–1094, 2000.

Valtchev, V. and Mintova, S. Synthesis of titanium silicate ETS-10: The effect of tetramety- lammonium on the crystallization kinetics. Zeolites, 14:697–700, 1994.

Valtchev, V., Mintova, S., Mihailova, B., and Konstantinov, L. Comparison of physicochem- ical properties of zorite and ETS-4. Materials Research Bulletin, 31(2):163–169, 1996.

Xu, H., Zhang, Y., and Navrotsky, A. Enthalpies of formation of microporous titanosilicates ETS-4 and ETS-10. Microporous and Mesoporous Materials, 47:285–291, 2001.

Yang, X., Paillaud, J.-L., van Breukelen, H. F. W. J., Kessler, H., and Duprey, E. Synthe-

sis of microporous titanosilicate ETS-10 with TiF4 or TiO2. Microporous and Mesoporous Materials, 46(1):1–11, 2001.

Zou, X. and Dadachov, M. S. K2TiSi3O9 · H2O. Acta Crystallographica Section C, 56:738–739, 2000.

23 3. Synthesis

Several recipes for the hydrothermal synthesis of AM-2 have been published to date (Lin et al., 1997; Poojary et al., 1997; Clearfield et al., 1998; Zou and Dadachov, 2000). They all follow the same approach by mixing a Ti/Zr source and a Si source in a highly alkaline solution, carrying out the synthesis at 180–240 ◦C under autogeneous pressure. Although pure AM-2 is easily obtained by these techniques, the product is usually not suitable for single-crystal X-ray diffraction. The crystallites are too small (1–30 µm) and often strongly intergrown. From the above mentioned publications only Zou and Dadachov (2000) have published single-crystal structure data of AM-2. A reproducible recipe for the synthesis of larger and more idiomorphic single crystals would allow us to analyse the structure and its ion-exchanged and dehydrated variations in more detail. Thus in our study we tried to find correlations between size, morphology and various synthesis parameters. We know from experience that in order to obtain a good diffraction pattern on a conventional diffractometer, crystals of at least 100 µm in diameter are required. The synthesis published by Zou and Dadachov (2000), as well as optimisations kindly provided by Christensen and Zou (2004), served as a starting model for our experiments, since good results have already been reported by these authors. The following section lists syntheses done in the course of this work in chronological order, together with a preliminary analysis of the results. Chemicals and vendors are listed in Table 3.1, and the equipment we used is shown in figure 3.1. SEM images of the syntheses products are given in section 3.3 at the end of this chapter.

3.1. Experimental procedures

3.1.1. Reproduction of previously published syntheses (TiS02)

In batch TiS02 we tried to reproduce the synthesis of AM-2 published by Zou and Dadachov

(2000) using dissolved TiCl3 as Ti source. Quantities were changed according to the latest optimisations by these authors. 9.89 g KOH, 4.00 g colloidal SiO2 (Ludox), and 2.78 g KCl were added to 29.4 g H2O and stirred until all solid material was dissolved. 17.96 g TiCl3 solution were added dropwise. The blackish gel was stirred thoroughly until it was homogeneous and ﬁlled into Teﬂon-lined stainless-steel autoclaves. The synthesis was carried out at 200 ◦C during 168 hours (7 days). Afterwards the autoclaves were cooled to room temperature during several hours before opening.

24 3.1. Experimental procedures

Figure 3.1.: The autoclaves used for hydrothermal AM-2 syntheses consist of a Teﬂon liner and a stainless-steel mantle. They allow syntheses up to 250 ◦C at autogeneous pressure. The dismantled bomb on the right has a custom-built screw fastening to mount it on the rotation axis in the furnace.

The resulting white powder was washed with deionised water and acetone, and dried at 50 ◦C for several minutes. Powder XRD analysis showed pure Ti-AM-2 within the detection limit. Scanning electron microscope images show aggregates of less than 20 µm in diameter composed of small crystals with diameters less than 10 µm in the longest direction. The synthesis product was thus not suitable for single-crystal XRD experiments.

3.1.2. Doping with CsCl, NH4Cl, and RbCl (TiS03–TiS05)

In batches TiS03, TiS04, and TiS05 small amounts of monovalent cations other than K+ were introduced to the synthesis. At the same time pT conditions and reaction time were changed. A base gel was prepared from 15.45 g H2O, 5.02 g KOH, 2.00 g SiO2 solution and

9.01 g TiCl3 solution. The blackish gel with pH 13.6 was stirred thoroughly until all solid material was dissolved. All batches were heated at 180 ◦C for 402.5 hours (17 days).

0.71 g CsCl (TiS03), 2.00 g NH4Cl (TiS04), and 0.89 g RbCl (TiS05) were added to the base gel as a replacement for KCl used in TiS02. The prolongation of the reaction time and the lower temperature had no positive effect on the crystal growth compared to TiS02. As shown in the SEM image the material of TiS03 and TiS05 also forms aggregates of less than 20 µm in diameter, composed of even smaller single crystals than TiS02. The products were analysed with EDS to check whether the additional cation was incorporated into the structure, but no Cs and Rb was found within the detection

25 3. Synthesis

Table 3.1.: List of chemicals used in this work.

Product Vendor Article Description

H2O deionised KOH Hanseler¨ AG 6-4712-1 Kalii hydroxidum in rotul 500 g

SiO2 (Ludox) Aldrich 420816-1L Ludox HS-40 colloidal silica 1 l 40 wt-% suspension in water KF Sigma-Aldrich 402931-100g Potassium ﬂuoride 100 g 99+%, A.C.S. reagent

TiCl3-solution Riedel-deHaen¨ 14010 Titanium(III) cloride solution 15% in hycdrochloric acid 250 ml Assay of TiCl3: min 15% Free acid (as HCl): 5–10%

TiO2 (anatase) Fluka Chemika 89490 Titanium Dioxide purum; >99% 250 g HCl Merck 1.00316.1000 Hydrochloric Acid 25% GR pro analysi 1 l HCl Hanseler¨ AG 20-1625-2 Acid hydrochloricum 25% 1 kg limit of the EDS device (fig. 3.2). The synthesis product from batch TiS04 was much more flaky than TiS03 and TiS05. X- ray powder diffraction showed that it was X-ray amorphous (fig. 3.2) and, according to EDS analysis, composed of almost pure SiO2.

3.1.3. Lower SiO2 and TiCl3 concentrations (TiS08, TiS09)

TiS08 and TiS09 were prepared with only 10% and 20% of the Ti and Si concentration used in TiS02. The synthesis was carried out at 230 ◦C during 119 hours (5 days). KOH and KCl were added in the same amounts as in TiS02. The resulting powder was ﬂuffy, and electron micrographs show no recognisable single crystals (see section 3.3). Due to the undesirable results the material was not further analysed.

3.1.4. Agitation and different cooling rates (TiS11)

TiS11 was prepared with concentrations equal to TiS02. The initial gel was stirred vigorously in the autoclaves, which were then mounted on a rotating device in the oven (ﬁg.

26 3.1. Experimental procedures

TiS05

TiS04 Intensity [Counts]

TiS03

TiS02

10 20 30 40 50 60 Angle [°2θ]

C ONa Si S K K Ti Ti

TiS05

TiS04 Intensity [Counts]

TiS03

TiS02

0 1 2 3 4 5 6 Energy [keV]

Figure 3.2.: XRD (top) and EDS (bottom) pattern of syntheses TiS02–TiS05. C peaks in the EDS pattern come from the carbon coating of the sample.

27 3. Synthesis

(a) (b)

Figure 3.3.: The furnace used for syntheses was equipped with a rotating device which could hold up to six autoclaves at a time. It was driven by an electric motor whose the speed could be adjusted from approximately 10 minutes to 10 seconds per revolution. Depending on the inclination of the bombs relative to the rotation axis, the amount of turbulence in the solution could be inﬂuenced. Turning the bomb axis perpendicular to the rotation axis caused the most turbulence (a), whereas turning it parallel to the rotation axis caused almost no turbulence (b).

3.3). During the synthesis the bombs were rotated at approximately 3 rounds per minute in order to cause turbulence in the solution. The autoclave axes were approximately 10◦ inclined to the rotation axis. After 5 days at 200 ◦C the three bombs were cooled at different cooling rates. TiS11A was quenched and cooled to room temperature within a few minutes, TiS11B was cooled at ambient temperature and required approximately 1–2 hours to reach room temperature, and TiS11C was left in the switched-off and closed oven, which took more than 5 hours to cool to room temperature. During that time it remained mounted on the rotating device and was agitated. SEM images show platy crystallites of approximately 5 × 1.5 × 0.1 µm3 in diameter, forming aggregates of sizes up to 10 µm. The aggregates are either radial (TiS10A), or smaller but more aligned (TiS10B/C). Powder XRD analysis revealed pure AM-2 at a high degree of crystallinity.

3.1.5. Using anatase as Ti source and KF as F- source (TiS15)

Previous AM-2 syntheses suffered from high nucleation rate and strong intergrowth. Using a less soluble Ti source may have a positive effect on the crystal size and shape, thus TiO2 in the form of anatase was used in TiS15 rather than dissolved TiCl3. Three autoclaves were

28 3.1. Experimental procedures

ﬁlled with a gel of composition 15.11 g H2O, 5.03 g KOH, 2.00 g SiO2 (Ludox), and 1.00 g

TiO2 (anatase). Furthermore 2.00 g KF were added to autoclaves TiS15B and C, and 1.41 g KCl to autoclave TiS15C. The reaction was carried out at 180 ◦C during ﬁve days. According to XRD analysis, AM-2 with minor residual anatase was produced (ﬁg. 3.4). No formation of aggregates and only minor intergrowths were observed. The crystals were small (3–10 µm) but very idiomorphic. Additional KF and KCl lead to no visible difference in the product.

3.1.6. Seeds (TiS16)

In TiS16 we prepared solutions with compositions similar to TiS02, but with several additives. The base gel was prepared from 60.07 g H2O, 8.00 g SiO2, 20.16 g KOH, and 8.06 g

KF, and was distributed equally onto four autoclaves. 9.05 g TiCl3 solution were added to each of TiS16D, E, and F. TiS16A contained a mixture of 4.60 g TiCl3 solution and 1.04 g TiO2 in the form of anatase powder, with a total Ti/Si ratio of 1/3. TiS16D, E, and F contained different amounts of seeds taken from batch TiS15A (TiS16D: 0.1 g, TiS16E: 0.5 g, TiS16F: 0.027 g). All syntheses were carried out at 200 ◦C during 10 days.

The presence of TiO2 in TiS16A lead again to the formation of pure idiomorphic AM- 2 crystals with only minor intergrowths. The average length of crystals is approximately 3 µm. The anatase-free syntheses D, E, and F showed strongly aligned aggregates rather than single crystals, but no systematic difference could be found among the three samples.

3.1.7. Adding TiCl3 solution prior to KOH (TiS20)

0.102 g KOH and 2.60 g Ludox HS-40 were added to 59.89 g H2O under continuous stirring.

The pH of the solution was 12.8. 1.00 g KF and 5.765 g TiCl3-solution were added and the pH dropped to approximately 1.0. The solution became dark brown, but remained transparent. Adding 4.26 g KOH in small portions caused the pH to rise to 14 and the colour of the solution changed from transparent brown to opaque violet. When the solution became homogeneous again, it was transferred into three autoclaves and placed in the cold oven, which was then switched on and reached 215 ◦C at 5 ◦C per minute. After 100 hours the autoclaves were removed from the oven and cooled at room temperature. The material was washed with water and acetone and dried at 50 ◦C. SEM images show idiomorphic crystals of approximately 10 µm size (section 3.3). Most of the material forms crystals of more or less the same size. This was the most successful synthesis so far in terms of crystal size and idiomorphism. The main difference to previous syntheses was that the pH dropped to acidic when TiCl3 solution was added, and that most of the KOH was added afterwards. That way the transition from acidic to alkaline was slower, and the associated precipitation of Ti phases was less violent.

29 3. Synthesis

TiS16F

TiS16E

TiS16D

TiS16A Intensity [Counts] TiS15C

TiS15B

* TiS15A

10 20 30 40 50 60 Angle [°2θ]

Figure 3.4.: Powder X-ray diffraction patterns of synthesis batches TiS15 and TiS16. In all cases pure AM-2 was formed. Reﬂections marked with asterisks belong to residual anatase.

3.1.8. Different ﬁll levels of the autoclaves (TiS21–TiS23)

Repetition of experiment TiS20, but with different free volumina in the Teflon liners. In TiS21 all autoclaves were filled to approximately 25%, whereas in all previous experiments the fill level was approximately 50%. In TiS22 each autoclave was filled to a different level. Free volumina were 15.7 cm3 (TiS22A), 13.74 cm3 (TiS22B), and 11.8 cm3 (TiS22C). In TiS23 10 ml were filled in each of the 45 ml-autoclave and the bombs were agitated once per day. The syntheses were carried out at 230 ◦C during 96 hours. No systematic difference between different fill levels was observed. All batches yielded AM-2 with maximum crystal sizes up to 30 µm. Although individual crystals were larger than in TiS20, much more amorphous material was produced. As the amorphous fraction does not seem to depend on the fill levels as well, we assume that the higher temperature has a negative influence on the crystallisation.

3.1.9. Temperatures between 210 and 250 ◦C (TiS25–TiS28)

In TiS25 and TiS26 80.06 g H2O, 13.77 g KOH, 5.07 g Ludox HS-40 (pH 14.1), 1.28 g KF and 7.61 g TiCl3-solution (pH 14.0) were mixed under continuous stirring. The solution was distributed into four autoclaves. TiS25 was heated for 96 hours at 240 ◦C, TiS26 for 96

30 3.1. Experimental procedures hours at 220 ◦C. TiS26 was left at room temperature for approximately 3 hours before the synthesis was started because the oven was not ready. The same amounts were used for TiS27 and TiS28, which were heated for 96 hours at 250 ◦C (TiS27) and 210 ◦C (TiS28). After the synthesis TiS27 and TiS28 were cleaned in water with an ultrasound device for a few minutes. TiS25 contained two generations of crystals, smaller ones with diameters around 10 µm, and some bigger ones with diameters up to 50–80 µm. TiS26 contained large amounts of amorphous material. Some few crystals >30 µm were found, but most of the material formed small intergrown particles or aggregates up to 100 µm in diameter. Although the crystals in TiS27 and TiS28 were not as big as in TiS25, the ultrasound treatment lead to very clean surfaces. Most particles were smaller than 10 µm, with only a few exceptions up to 30 µm in the longest direction. TiS25 yielded the best results in terms of crystal size and shape so far.

3.1.10. Agitation and high temperature (TiS30–TiS31)

80.06 g H2O, 13.77 g KOH, 5.07 g Ludox HS-40, 1.28 g KF and 7.61 g TiCl3-solution were mixed under continuous stirring. The solution was distributed into four autoclaves. Two of them were attached to the rotating device and agitated with approximately 3 rounds per minute (TiS30), while the other two autoclaves were standing still (TiS31). The synthesis was carried out at 240 ◦C for 240 hours. Lots of amorphous material was produced. The ﬁne grained fraction observed in the previous batches was not present, instead an amorphous gel-like phase mainly composed of K and Cl formed. The biggest crystals were still approximately 30–40 µm long, but embedded in the amorphous matrix. Washing the synthesis product in order to separate the amorphous material from the crystals turned out to be impossible. No systematic difference was observed between TiS30 and TiS31.

3.1.11. Synthesis of UND-1 (TiS32–TiS33)

A slightly different synthesis method described by Liu et al. (1997) for the production of UND-1 was reproduced. However, instead of P25 we used anatase, and Ludox HS-40 was used as silica source. 30.01 g H2O, 22.48 g Ludox HS-40, 2.41 g anatase, 5.42 g NaOH, 5.15 g KF were mixed and stirred for 30 minutes. The gel was transferred into 3 Teﬂon-lined autoclaves and heated for 4 days at 220 ◦C. The same amounts were used for TiS33, which was heated to 220 ◦C for 41 (TiS33A), 65 (TiS33B), and 89 (TiS33C) hours. In all cases pure AM-2 was produced instead of UND-1. Idiomorphic crystals of sizes up to 30 µm formed, but started to decompose again after four days. Due to the shorter reaction times in TiS33, no decomposition was observed.

31 3. Synthesis

3.1.12. Using HF as ﬂuorine source (TiS34–TiS35)

Good results have been reported by Christensen and Zou (2004) for AM-2 syntheses using HF as ﬂuorine source.

TiS34: 60.02 g H2O, 1.27 g KOH, and 2.52 g Ludox HS-40 were mixed together in a 250 ml polypropylene beaker under continuous stirring. 0.828 ml HF (40 wt-%, the same amount of mol F as in 0.968 g KF) and 4.8 ml TiCl3-solution were added. The solution became brown/yellow and transparent. KOH was added (1.72, 1.10, and 6.01 g), and after each step the solution was stirred until the reaction had terminated. For the synthesis the gel was transferred into three autoclaves and heated at 215 ◦C for 100 hours.

TiS35: 60.05 g H2O, 10.29 g KOH and 3.75 g Ludox HS-40 were mixed together in a 250 ml polypropylene beaker. 0.828 ml HF and 4.75 ml TiCl3-solution were added under continuous stirring. The solution was transferred into three autoclaves and placed in a preheated oven. The reaction took place during 96 hours at 230 ◦C. Both batches yielded pure AM-2 crystals of two generations; the large crystals were up to 30 µm long, and the small ones reach approximately 5 µm. The difference between the two groups is more pronounced in TiS34. In TiS35 the crystal surfaces were covered with small particles and more intergrowths were observed, whereas in TiS34 the crystals were cleaner and more idiomorphic. Due to the precautions required for working with HF and the absence of substantial improvements in the synthesis product, subsequent experiments were done with KF.

3.1.13. Adjusting the pH (TiS36–TiS42)

In previous experiments we observed a correlation between the pH of the initial gel and the shape of AM-2 crystals. Thus in batches TiS36 and TiS37 a pH range from 12.4–14.0 was analysed in relatively coarse steps, in order to get an overview of the inﬂuence on the crystals. Since the correlation was conﬁrmed, the pH range from 13.1–13.9 was covered in small steps with batches TiS38–TiS42. An initial gel was prepared from 120.21 g H2O,

3.042 g KOH, 5.238 g Ludox, 1.82 g KF, and 11.199 g TiCl3-sol, which were added one by one under stirring. The pH was then adjusted by adding KOH until the desired starting value was reached. From this solution 20 ml were transferred into an autoclave, and the pH of the remaining solution was adjusted to the next value. This was repeated until all six autoclaves were filled. KOH amounts and pH values are listed in Table 3.2. SEM images show that the optimum pH lies in the range of 13.4–13.6. From pH 13.0 to 14.0 the morphology evolves from fibrous to strongly intergrown aggregates, to single crystals of homogeneous size, to single crystals of two size generations, and finally to a mixture of small single crystals and amorphous components (fig. 3.8).

32 3.1. Experimental procedures

Table 3.2.: Additional KOH and corresponding pH values of batches TiS36–TiS42.

∗ ∗∗ ∗∗∗ Name Vol. [ml] Initial pH +KOH [g] KOHtotal [g] pH pH TiS36/1 136.5 7.6 2.437 2.437 13.0 12.25 TiS36/2 116.5 7.6 0.87 3.456 13.2 13.1 TiS36/3 96.5 7.6 1.488 5.561 13.4 13.4 TiS36/4 76.5 7.6 1.55 8.327 13.6 13.6 TiS36/5 56.5 7.6 1.85 12.796 13.8 13.6 TiS36/6 36.5 7.6 2.154 20.852 14.0 13.5 TiS37/1 136.5 7.8 0.98 0.980 12.4 7.5 TiS37/2 116.5 7.8 0.29 1.320 12.7 9.5 TiS37/3 96.5 7.8 0.365 1.836 13.0 11.3 TiS37/4 76.5 7.8 0.304 2.379 13.2 12.2 TiS37/5 56.5 7.8 0.289 3.077 13.3 12.7 TiS37/6 36.5 7.8 0.298 4.191 14.5 13.1 TiS38/1 136.5 7.8 1.819 1.819 13.1 11.2 TiS38/2 116.5 7.8 0.099 1.935 13.1 11.5 TiS38/3 96.5 7.8 0.09 2.062 13.1 11.8 TiS38/4 76.5 7.8 0.083 2.21 13.2 11.9 TiS38/5 56.5 7.8 0.09 2.428 13.2 12.15 TiS38/6 36.5 7.8 0.1 2.802 13.3 12.7 TiS39/1 136.5 7.3 2.83 2.83 13.2 12.7 TiS39/2 116.5 7.3 0.1 2.947 13.3 12.8 TiS39/3 96.5 7.3 0.09 3.074 13.3 12.85 TiS39/4 76.5 7.3 0.089 3.233 13.3 12.7 TiS39/5 56.5 7.3 0.097 3.468 13.4 12.6 TiS39/6 36.5 7.3 0.104 3.857 13.4 13.2 TiS40/1 136.5 7.6 3.62 3.62 13.5 TiS40/2 116.5 7.6 0.104 3.742 13.5 TiS40/3 96.5 7.6 0.098 3.88 13.5 TiS40/4 76.5 7.6 0.091 4.043 13.6 TiS40/5 56.5 7.6 0.1 4.284 13.6 TiS40/6 36.5 7.6 0.091 4.625 13.7 TiS41/1 137.5 7.6 4.51 4.51 13.6 13.35 TiS41/2 117.5 7.6 0.092 4.618 13.6 13.4 TiS41/3 97.5 7.6 0.096 4.753 13.7 13.4 TiS41/4 77.5 7.6 0.086 4.906 13.7 13.45 TiS41/5 57.5 7.6 0.093 5.128 13.7 13.45 TiS41/6 37.5 7.6 0.088 5.451 13.7 13.5 TiS42/1 136.5 7.6 5.33 5.33 13.7 13.4 TiS42/2 116.5 7.6 0.103 5.451 13.7 13.4 TiS42/3 96.5 7.6 0.1 5.592 13.8 13.4 TiS42/4 76.5 7.6 0.093 5.758 13.8 13.5 TiS42/5 56.5 7.6 0.099 5.997 13.8 13.5 TiS42/6 36.5 7.6 0.087 6.323 13.9 13.5 ∗ total amount of KOH added to the initial gel, normalised to 136.5 ml (137.5 ml for TiS41 ∗∗ before the synthesis ∗∗∗ after the synthesis

33 3. Synthesis

H2O +TiCl3 in HCl +KOH +KOH 14

12 5-10% HCl 20-30% HCl 10 Figure 3.5.: The pH values during AM- 2 syntheses with TiCl dissolved in 8 3 5–10% HCl and 20–30% HCl. The pH 6 main difference is that in the latter case

4 the pH drops to highly acidic values when TiCl3 solution is added (step 3). 2 This prevents sudden precipitation of 0 1 2 3 4 Ti phases. See ﬁgure 3.6 for further ex- Step planation.

3.1.14. Using higher concentrations of HCl (TiS43–TiS46)

The TiCl3 solution used in our experiments was different from the one used by Zou and Dadachov (2000) and Christensen and Zou (2004) in that the concentration of free HCl was lower. In our case TiCl3 was dissolved in 5–10% HCl, whereas Zou and co-workers used

TiCl3 dissolved in 20–30% HCl. Although the pH of the initial gel was adjusted to the same value in both cases, it drops to strongly acidic values during the preparation if higher HCl concentrations are used. A similar pH development was already observed in TiS20 and lead to good results. In order to introduce the same amount of HCl, we enriched the TiCl3 solution with additional HCl (25 mol-%) prior to mixing the initial gel. KOH was used to adjust the pH to 13.0–14.0. In TiS43 the pH range from 12.9 to 13.6 was covered in coarse steps, whereas in TiS44–TiS46 the pH from 13.2 to 13.9 was covered in small steps. TiS43/6 and TiS44/2 conﬁrm our assumption that the pH range from 13.4 to 13.6 is ideal for the formation of single crystals. Crystals of up to 30 µm in the longest direction were produced. Largest crystals were produced with HCl concentrations corresponding 20–30% in the TiCl3 solution.

3.1.15. Ti deﬁciency (TiS48)

A synthesis with lower Ti-concentration was performed. Deﬁciency of Ti may reduce nucleation rate and produce larger crystals. All other components (except KOH) were present in the same concentration as usual. KOH amounts were varied to adjust the pH in all autoclaves to 13.6.

3.11 g KOH were added to 111.37 g H2O under stirring. The pH measured was 13.7. 5.082 g Ludox and 1.82 g KF were added under stirring. 11.4 g HCl (25%) caused the pH to drop to 0.9. Small amounts of TiCl3-solution were added, and for each step the pH was adjusted to 13.6 by adding KOH. 20 ml of the gel were transferred into an autoclave, and more TiCl3 and KOH were added to the remaining solution. Six samples with different TiCl3 and KOH concentrations, but with pH 13.6 were prepared (TiCl3 solution [g]/KOH [g]): I:

34 3.1. Experimental procedures ∗∗∗ pH 144.5 ml ) 4 ∗∗ ( 13.5 13.4 13.4 13.3 13.3 13.9 13.8 13.9 13.65 13.8 13.6 13.7 13.5 13.7 13.45 13.113.213.413.513.6 11.7 12.3 12.9 13.4 13.0 13.4 13.3 13.5 13.6 13.7 144.0 ml, ) [g] pH 3 ( ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) 4 4 4 4 4 3 3 3 3 3 1 1 1 1 1 2 2 2 2 2 ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ∗ total 148.0 ml, ) 2 ( 141.25 ml, ) KOH [g] KOH 1 ( + 0.2 0.16 14.388 0.2 0.20 13.868 0.2 0.28 13.420 0.2 0.30 12.941 0.2 0.29 12.850 0.2 12.19 12.190 13.2 0.1 0.20 13.061 0.1 0.26 12.406 0.1 0.27 11.821 0.1 0.29 11.359 0.1 0.29 10.960 0.1 10.623 10.623 13.6 13.3 − − − − − − − − − − − − O [g] HCl 25% [g] Vol. [ml] Min. pH 2 Additional KOH and corresponding pH values of batches TiS43–TiS46. Min. pH is the minimum pH the solution reached after the synthesis before the synthesis total amount of KOH added to the initial gel, normalised to the initial volume of solution: during preparation. ∗ ∗∗ ∗∗∗ TiS46/6 105.00 20.01 44.5 TiS46/5 105.00 20.01 64.5 TiS46/4 105.00 20.01 84.5 TiS46/3 105.00 20.01 104.5 TiS46/2 105.00 20.01 124.5 TiS46/1 105.00 20.01 144.5 TiS45/6 108.01 16.00 44.0 TiS45/5 108.01 16.00 64.0 TiS45/4 108.01 16.00 84.0 TiS45/3 108.01 16.00 104.0 TiS45/2 108.01 16.00 124.0 NameTiS43/1TiS43/2TiS43/3 H 111.32TiS43/4 111.32TiS43/5 111.32TiS43/6 111.32 11.60TiS44/1 111.32 11.60TiS44/2 111.32 11.60TiS44/3 100.00 11.60TiS44/4 141.25 100.00 11.60TiS44/5 121.25 100.00 11.60TiS44/6 101.25 100.00 26.66TiS45/1 100.00 81.25 26.66 100.00 0.3 61.25 26.66 0.3 41.25 108.01 26.66 0.3 148.0 26.66 128.0 0.3 26.66 108.0 0.3 7.00 16.00 0.3 0.29 88.0 0.38 68.0 0.0 48.0 0.0 0.30 7.000 144.0 0.0 0.28 7.338 0.20 0.0 7.686 15.98 0.0 8.389 12.9 0.0 0.29 9.035 0.288 9.720 0.29 15.980 0.27 16.315 10.9 0.19 16.710 13.4 17.198 17.785 18.371 Table 3.3.:

35 3. Synthesis

Figure 3.6.: Two slightly different approaches were used for the AM-2 synthesis. Top row: A highly alkaline solution was prepared from H2O and KOH (a). Additives like KCl and KF were added (b), and ﬁnally TiCl3 dissolved in HCl was added (c). The solution immediately turned to black and opaque and precipitation of Ti phases occurred. After some stirring the solution became homogeneous and bluish (d), and turned to white after several minutes (e). Bottom row: Starting with a less alkaline solution with additives KCl and KF (a), the pH dropped to acidic when TiCl3 solution was added. Ti remained dissolved (b) and the mixture became brown and transparent. Note that TiCl3 solution is violet and becomes brown as soon as it comes in contact with KF. Adding KOH also causes moderate precipitation as long as the solution is acidic (c), but the ﬂakes dissolve quickly. At the transition from acidic to alkaline the gel becomes black and viscous and starts foaming (d). When the reaction has settled the solution becomes blue and opaque (e). After some time it also turns to white, but it takes much longer than in the upper example. Using higher HCl concentrations leads to the same development as seen in the bottom row.

36 3.2. Conclusion

+2.00/+4.26; II: +1.668/+1.44; III: +1.333/+1.02; IV: +1.00/+0.68; V: +0.661/+0.475; VI: +0.335/+0.276. Note that these amounts are cumulative and that the amount of solution decreased by 20 ml with every step. The evolution of crystal morphology with increasing Ti concentration is similar to the pH dependency. At low concentrations the material is nanocrystalline or amorphous, with increasing Ti concentration spherical aggregates with crystallites growing radially from the nucleus appear. These crystallites become more aligned and form idiomorphic single crystals at a Ti/Si ratio of 1:3. Higher Ti concentrations lead to the formation of amorphous material and less homogeneous size distribution of single crystals. The maximum crystal size is 20–30 µm and hence not superior to other results of our study.

3.2. Conclusion

The adjustable parameters of the hydrothermal AM-2 synthesis are manifold. Physical parameters such as reaction time, temperature, heating and cooling rates, agitation, and fill levels, as well as chemical parameters such as Ti/Si source material, pH, mixing order, and various additives influence the species and the crystal shape and size of the synthesis product. The combination of all these parameters leads to a huge multidimensional array which cannot be investigated systematically with reasonable effort. By changing several parameters at a time in a more or less random manner we were able to develop a feeling for the stability field of AM-2, and we could quickly exclude the most unpromising strategies. The synthesis works in a relatively wide range of temperature and reaction time and is not sen- sitive to fill levels of the autoclaves. AM-2 is produced at least between 180 and 240 ◦C, best results were achieved at 215–220 ◦C. The reaction normally takes place in three days. Af- ter 72 hours no residual compounds were found, and longer reaction time did not improve the results. We neither found a dependency to heating and cooling rates, but one must keep in mind that fast temperature changes were not possible in our experiments due to the high heat capacity of the stainless-steel autoclave mantles. As far as agitation is concerned the formation of aggregates is slightly inhibited, leading to smaller particles. However, the size of the crystallites is not affected, and in cases where single crystals were obtained no improvement was observed. Variations in chemical composition are much more complex since there are virtually no limits as to what additives can be used. Hence we confined ourselves to only marginal variations from the initial composition by Christensen and Zou (2004). We changed the pH, the Ti/Si ratio, and Ti and Si sources. We also introduced variable amounts of Cl− and F− from different sources and monovalent cations other than K+. The bottom line is that at a Ti/Si ratio of 1/3 the largest single crystals are obtained, and that the crucial parameter determining the crystal morphology is the pH. Idiomorphic single crystals are obtained at pH = 13.4–13.6; lower values lead to formation of aggregates. The shape of these aggregates

37 3. Synthesis

Figure 3.7.: SEM images of AM-2 crystals synthesised by Christensen and Zou (2004). The size of the crystals is sufficient for single-crystal X-ray diffraction, but neither we nor the above authors were able to reproduce these outstanding results. depends on the amount of F− added to the solution. Fluorine-free syntheses lead to radial crystal growth where crystallites grow from the nucleus in all directions. The aggregates become discoidal, spherical, or bow-tie-shaped. In the presence of F− the crystallites are aligned, but still strongly intergrown. The F− source seems to be largely irrelevant, as no systematic difference between HF and KF syntheses was observed. The presence of other + + + monovalent cations, e. g. Cs , Rb , and NH4 , did not influence the crystal morphology + in a positive way. NH4 even inhibited the formation of AM-2 and only an amorphous substance was produced. It is also remarkable that no substantial amounts of Rb+ and Cs+ were incorporated into the structure, as both elements are known to be compatible with the AM-2 channel system. A possible explanation for this observation is that the selectivity for K+ is higher than for Cs+ and Rb+. Additional HCl had to a limited extent a positive effect on the crystal size. Although the pH of the final solution was compensated by additional KOH, more HCl had two consequences on the synthesis: i) A higher Cl− concentration and ii) a different development of the pH during preparation. Higher Cl− concentration can also be obtained by adding more KCl. However, this did not have the same positive effect on the crystal size. We therefore assume that the lower pH during preparation is mainly responsible for the difference in crystal size. Adding TiCl3 dissolved in HCl to an alkaline solution causes precipitation of Ti phases and the solution becomes dark violet and opaque. If the pH drops to the acidic milieu, however, Ti remains dissolved and the solution becomes olive green and transparent (fig. 3.5). Adding solid KOH bit by bit to this acidic solution causes precipitation as soon as

38 3.2. Conclusion the pH goes to alkaline, but the process is much slower and the reaction is less violent. The solution becomes lukewarm and foams. Despite these improvements we were only able to increase the crystal size by approximately 20%. Thus we assume that the potential for increasing crystal sizes in this transition is very limited. The crystals shown in figure 3.7 prove that by the herein described synthesis procedure it is possible to produce crystals suitable for single-crystal X-ray diffraction. Although it may still be difficult to obtain a diffraction pattern on conventional diffractometers due to the small size, it is at least possible to mount them on needles and they produce excellent diffraction patterns on synchrotron radiation devices. The crux is that neither Christensen and Zou (2004), who produced the crystals shown in figure 3.7, nor we were able to reproduce the excellent results, although we tried hard to reconstruct the experimental conditions. One must keep in mind that these crystals are the result of the very first attempt to synthesise AM-2 in autoclaves which have never been in contact with AM-2 before. It is thus possible that impurities from the previous usage or surface conditions in the Teflon liner had a positive influence on the crystal growth. Although we tried to take into account the latter by boring the inner surface or by using virgin Teflon liners, we had no information about the experiments done prior to the AM-2 synthesis in the autoclaves of Christensen and Zou.

39 3. Synthesis

(a) pH = 13.1 (b) pH = 13.3 (c) pH = 13.6

(d) pH = 13.8 (e) pH = 13.9

Figure 3.8.: pH dependency of the AM-2 synthesis. The most idiomorphic crystals with the least intergrowths are formed at pH = 13.6. The width of each image is approximately 64 µm.

40 3.3. SEM images

3.3. SEM images

41 3. Synthesis

42 3.3. SEM images

43 3. Synthesis

44 3.3. SEM images

45 3. Synthesis

46 3.3. SEM images

47 3. Synthesis

48 3.3. SEM images

49 3. Synthesis

50 3.3. SEM images

51 3. Synthesis

52 3.3. SEM images

53 3. Synthesis

54 3.3. SEM images

55 3. Synthesis

56 3.3. SEM images

57 3. Synthesis

58 3.3. SEM images

References

Christensen, K. E. and Zou, X. personal communication, 2004.

Clearﬁeld, A., Bortun, A. I., Bortun, L. N., Poojary, D. M., and Khainakov, S. A. On the selec-

tivity regulation of K2ZrSi3O9 · H2O-type ion exchangers. Journal of Molecular Structure, 470:207–213, 1998.

Lin, Z., Rocha, J., Brandao,˜ P., Ferreira, A., Esculcas, A. P., de Jesus, J. D. Pedrosa, Philip- pou, A., and Anderson, M. W. Synthesis and structural characterization of microporous umbite, penkvilksite, and other titanosilicates. Journal of Physical Chemistry B, 101:7114– 7120, 1997.

Liu, X., Shang, M., and Thomas, J. K. Synthesis and structure of a novel microporous ti-

tanosilicate (UND-1) with a chemical composition of Na2.7K5.3Ti4Si12O36 · 4H2O. Micro- porous Materials, 10:273–281, 1997.

Poojary, D. M., Bortun, A. I., Bortun, L. N., and Clearﬁeld, A. Synthesis and X-ray powder

structures of K2(ZrSi3O9) · H2O and its ion-exchanged phases with Na and Cs. Inorganic Chemistry, 36:3072–3079, 1997.

Zou, X. and Dadachov, M. S. K2TiSi3O9 · H2O. Acta Crystallographica Section C, 56:738–739, 2000.

59 4. Ion exchange and dehydration of Rb-exchanged AM-2

4.1. Abstract

Rb-exchange and thermal stability of the microporous titanosilicate AM-2 were analysed by powder X-ray diffraction, thermo-gravimetric analysis, and ICP-OES. The dehydration and thermal stability of the exchanged structure were monitored with powder XRD. Crystal structures were reﬁned with Rietveld methods at 25 and 400 ◦C. The AM-2 structure was found to easily incorporate Rb+ by replacing K+. After four exchange cycles and 166 hours ◦ reaction time at 90 C, the chemical composition was reﬁned to K0.18Rb1.82TiSi3O9 · H2O. Extrapolation suggests that higher exchange ratios may be obtained after further cycles.

The H2O molecule was expelled by heating, leading to a completely dehydrated structure at 360 ◦C. Dehydration came along with a change of symmetry from the orthorhombic space group P212121 to the monoclinic space group P21, which is completely reversible after rehydration. The AM-2 structure breaks down above 600 ◦C and at 750 ◦C a wadeite-type phase crystallises. This transformation is irreversible and leads to immobilisation of Rb+.

4.2. Experimental procedure

Rb ion-exchange was done on AM-2 powder from synthesis batch TiS11. This material was chosen on the basis of its visual appearance on electron micrographs. It has a small crystallite size in the region of about 1 µm3 or less, a high speciﬁc surface, and due to the formation of aggregates it can easily be ﬁltered from the exchange solution. Due to the same properties the material also seems to be ideal for powder X-ray diffraction experiments. From the poorly oriented aggregation preferred orientation effects are supposed to be moderate and negligible. A 1 M exchange solution was prepared by mixing 24.184 g RbCl and 200 ml deionised

H2O. 0.4024 g AM-2 of composition K2TiSi3O9 · H2O were mixed with 20 ml exchange solution and heated to 90 ◦C at ambient pressure in a covered Teﬂon beaker. In order to renew the solution the AM-2 powder was ﬁltered off, washed with water and acetone and dried for 15 minutes at 50 ◦C. The used solution was kept for further analysis, and the remaining powder was again mixed with 20 ml of fresh solution. The progress of the exchange process was monitored by quantitative analysis of the K+ concentration in the exchange solution, experimental parameters and ICP-OES results are listed in Table 4.1. Evaporation was taken into account by measuring the amount of solution remaining after each cycle.

60 4.2. Experimental procedure

100

Figure 4.1.: K+ concentration in the 40 mother liquid, measured with ICP- Exchange ratio [%] OES. The experimental values were ﬁtted with an exponential function 20 f (x) = −100 · exp(−x/51.44) + 100. According to this function, 99.9% ion exchange is reached after approxi- 0 0 50 100 150 200 250 300 mately 355 hours. The error bars show Time [h] a range of ±5%.

As known from other microporous silicate structures, dehydration can either be induced by heating (Dobelin¨ and Armbruster, 2003), or by exposing the material to vacuum. Since the first method is technically simpler and presumably of more practical value, the dehydration process and total weight loss between 25 and 400 ◦C were measured by thermo- ◦ −1 gravimetry (TGA) in N2 atmosphere with a heating rate of 5 C min . The influence of the dehydration process on the crystal structure was monitored by powder X-ray diffraction with a Philips X’Pert Pro powder diffractometer in Bragg-Brentano theta-2theta configuration. The device was equipped with an Anton Paar HTK-1200 heating chamber, which allows in-situ data acquisition from room temperature up to 1200 ◦C, at ambient pressure and in natural or controlled atmosphere. The Ni-filtered CuKα X-radiation from a fine-focus tube lead to a maximum resolution of 1.006 A˚ at 100◦ 2θ. Data was collected at 25 ◦C, 50 ◦C and further in steps of 50 ◦C up to 400 ◦C. The sample was heated at a rate of 5 ◦C min−1 and during data collection the temperature was held constant. After the heating sequence the sample cooled down to 25 ◦C and a final data set was collected. In a second heating cycle data was collected at 25, 400, 500, 600, 700, and 750 ◦C, and again after cooling down to room temperature. The experiment was conducted in room atmosphere, moreover the connectors for the atmosphere control system were kept open during the entire procedure in order to allow air circulation. The 25, 400, and 750 ◦C data sets were recorded from 9 to 100◦ 2θ and merged from 10 successively recorded measurements in order to obtain maximum resolution and a better signal-to-noise ratio. Prior to merging the individual patterns were checked for zero-point shift, but no displacement within the detection limit was found. All other data sets were recorded from 9 to 60◦ 2θ and used to monitor phase transitions on the basis of unit cell variations. Data sets collected above

61 4. Ion exchange and dehydration of Rb-exchanged AM-2

Rb1.92K0.08TiSi3O9 × H2O

11.1

11.0

10.9 Step: -4.64 wt% = 1.2 H2O pfu

10.8 Weight [mg]

10.7

10.6

10.5 100 200 300 400 Temperature [°C]

Figure 4.2.: Dehydration monitored by thermo-gravimetric analysis. The fully hydrated AM-2 structure loses 4.64% of its initial mass when heated from 25 to 400 ◦C. This corresponds to 1.2 H2O molecules per formula unit. room temperature were corrected for thermal expansion of the sample stage by the formula ∆2θ = (2s · cos θ)/R, with s being the displacement from the focus circle in millimetres, R being the goniometer radius in millimetres, and θ, 2θ given in radians. Values for s were provided by the manufacturer. The crystal structures were refined with the software FULLPROF (Rodr´ıguez-Carvajal, 2005). Refinement parameters of the hydrated and dehydrated phase are listed in table 4.2. As a starting model for the 25 ◦C-structure the AM-2 structure published by Zou and Dadachov (2000) was used. In the first stage of the refinement, only profile parameters such as zero-point shift, provisional background points and cell parameters were refined in LeBail profile fitting mode. The Thomson-Cox-Hastings pseudo-Voigt function (no. 7 in FULLPROF) was used to model the reflections. Parameters U, V, W, X, Y of the instrument resolution function were determined with a corundum standard sample prior to data collection, and held constant during the entire refinement. After the cell parameters were determined, the model structure was introduced and refined in Rietveld mode with fully occupied Rb on the co-ordinates of the K sites. The first refined parameters were the scale factor, an overall displacement parameter for the framework atoms and the profile parameters mentioned above. Subsequently the co-ordinates of the heavy atoms Rb, Ti, Si, and

finally O including the H2O site were released for refinement. For the channel occupants a common isotropic displacement parameter was refined, while Boverall was used for the framework atoms. The refined framework showed distorted SiO4 and TiO6 polyhedra, thus soft distance constraints were applied to all Si–O (1.626(5) A)˚ and Ti–O bonds (1.962(5) A).˚

62 4.2. Experimental procedure

Table 4.1.: Experimental parameters and K+ concentrations of the Rb ion exchange of AM-2.

Cycle No. AM-2 [g] Dur. [h] Sol. after [ml] K+ conc. [%] 1 0.4024 5 19.9 18.98 2 0.3828 18 19.7 23.87 3 0.3700 72 19.0 36.78 4 0.3570 71 19.0 16.24 Total: 95.86

Distances for soft constraints correspond to average Ti–O and Si–O distances in the single- crystal structure published by Zou and Dadachov (2000). In the final step occupancies of Rb sites were refined. After the refinement difference Fourier plots showed remaining electron densities not higher than ±0.4 e−. A parameter IG describing isotropic peak broadening of the Gaussian component as a function of the crystallite size was refined for all phases during early stages of the refinement. For the Thomson-Cox-Hastings pseudo-Voigt function, Gaussian and Lorentzian components of the peak shape are computed from the shape parameters U, V, W, X, Y, and IG as follows:

IG FWHM = U · tan2 θ + V · tan θ + W + gau cos2 θ

Y FWHM = X · tan θ + lor cos θ Anisotropic contributions of micro-strain were not refined and are not shown in the formulae above. The average isotropic crystallite size can be calculated from IG with the Scherrer- formula (size and λ in A):˚ 180 · λ Size = √ π · IG For crystallite sizes greater than approximately 1 µm IG converges to zero. In that case it is recommended not to refine this parameter, but rather fix it to zero in order to reduce the number of refined parameters. The refined hydrated structure served as starting model for the dehydrated high-temperature structure. The transition from the low-temperature to the high-temperature phase lead to splitting of [h0l] peaks in the powder diffraction pattern. The new pattern was indexed in space group P21, a subgroup of P212121, with unique b axis. Due to the lower symmetry all atoms had to be duplicated by the symmetry operation (x + 1/2, −y + 1/2, −z) and shifted by z + 0.25 in order to obtain a standard compliant unit cell set-up. The refinement sequence was similar to the hydrated structure. Soft distance constraints for framework atoms were necessary in order to achieve a stable refinement.

63 4. Ion exchange and dehydration of Rb-exchanged AM-2

Table 4.2.: Experimental and reﬁnement parameters for Rb-exchanged AM-2.

Temperature [◦C] 25 400

Space group P212121 P21 a [A]˚ 10.0851(2) 10.0755(2) b [A]˚ 13.0157(2) 12.8625(2) c [A]˚ 7.2255(1) 7.1587(1) β [◦] 90 91.754(1) Cell volume [A˚ 3] 948.45(3) 927.30(3) RBragg [%] 3.48 5.27 Rp [%] 8.21 11.5 Rwp [%] 7.64 10.6 χ2 0.39 0.623 No. of variables 62 105 No. of indep. reﬂections 581 801 U 0.102452 0.102452 V −0.064479 −0.064479 W 0.013693 0.013693 X 0.108989 0.108989 Y 0.046079 0.046079

4.3. Results

As measured by ICP-OES analysis of the exchange solution, a total of 96(5)% of the K+ ions were exchanged for Rb+ (tab. 4.1). The pH of fresh exchange solution was neutral (7.18), and alkaline after the first exchange cycle (10.25). Alkalinity was less distinct after the following cycles (cycles 2–4: 8.85, 8.08, 7.28), and the stability of Ti-AM-2 was not compromised at any time (Bortun et al., 2000). Since there is no reason for a change of pH caused by the ion exchange process, we assume that minor by-products of the synthesis formed on the surface of AM-2 particles, which could not be removed by washing the material after the synthesis, but were dissolved under hydrothermal conditions during the ion exchange. This could also explain the discrepancy of the exchange ratio obtained by structure refinements and ICP-OES analysis of the mother liquid. The total weight loss measured by TGA amounts to 4.64% of the initial mass. Assuming an ion exchange ratio of 96% this corresponds to 1.2 H2O molecules per formula unit. As shown in figure 4.2, the release of H2O is steady with a maximum loss rate between 200 and 250 ◦C, and subsides above 320 ◦C. Exposed to ambient temperature and humidity (approximately 20 ◦C and 50% relative humidity in our case) the structure completely re- hydrates within a few hours (Lin et al., 1999). The hydrated structure shows the same orthorhombic microporous framework as the AM-2 structure published by Zou and Dadachov (2000), but it has a 3% larger cell volume.

The H2O site Ow1 is located near the centre of a triangle of Rb/K sites in bonding distance to the corners (Ow1–Rb2: 2.63(2), Ow1–Rb1: 2.93(2), Ow1–Rb2: 3.22(2) A).˚ Two distinct cation positions were found. Since on both sites electron densities were not sufﬁcient for

64 4.3. Results complete Rb occupation, they were reﬁned with scattering factors of both Rb and K, assuming incomplete ion exchange. The chemical composition was reﬁned to K0.18Rb1.82TiSi3O9

· H2O. Although these cation sites are not solely occupied by Rb, they are herein after referred to as Rb1 and Rb2. For the description of bonding environments a maximum Rb–O bond length of 3.37 A˚ was assumed, which corresponds to a bond valence of 5% (Brese and O’Keeffe, 1991). Weak reflections of a second phase with cell parameters slightly different from the main phase were also observed. This phase was refined in LeBail profile fitting mode in order to optimise the pattern fit, but intensities were too weak to extract any useful structure information. The cell parameters were refined to a = 10.008(1), b = 13.012(1), and c = 7.2116(6) A˚ in the orthorhombic space group P212121 and lie thus between the initial K-bearing cell published by Zou and Dadachov (2000) and our Rb-exchanged cell. The concentration of this phase was well below 5%. Peak broadening was observed in the diffraction pattern of the minor phase, which was refined as a crystallite size effect leading to an average size of 30 nm. X-ray diffraction patterns recorded during heating show a decrease of the unit cell dimensions between 50 and 350 ◦C (fig. 4.3). Between 150 and 200 ◦C a distinct phase change from orthorhombic to monoclinic symmetry occurs, which coincides with the highest loss ◦ rate of H2O. Complete dehydration is reached at approximately 320 C. The high-temperature phase remains stable up to 600 ◦C and completely reverts to the higher symmetric low-temperature phase after cooling down to room temperature (tab. 4.3). The monoclinic unit cell is slightly smaller than the orthorhombic cell (tab. 4.2 and fig. 4.3); the cell volume decreased by 2.22%, a remained almost constant, and b and c shortened by approximately 1%. Due to high temperature data collection, high displacement parameters, particularly for weakly bound channel occupants, were to be expected. Thus one common displacement factor was refined isotropically for all channel occupants, and Boverall was used for all framework atoms. To our knowledge the results presented in Table 4.6 are the first structure refinement of this monoclinic high-temperature phase. The topology of the framework is identical to the orthorhombic hydrated phase. Changes are mainly observed in P–O–P angles (P: central atom of polyhedra, O: vertex oxygen, tab. 4.9), but no rearrangement of polyhedra occurred. Due to the absence of H2O molecules in the voids, the bonding situations of Rb cations changed slightly. In 0.96(5) A˚ distance to Rb2b, a new partially occupied site Rb3a was refined. Occupancy factors of these two sites were coupled to a sum of 1.0 and refined to 30% on Rb2b and 70% on Rb3a. The resolution of the experimental data was not sufficient to refine overlapping K and Rb on channel sites, therefore only scattering factors of Rb were used, but occupancies <1.0 were allowed on Rb1a, Rb1b, and Rb2a. The sum of electrons on channel cation sites was refined to 140.86(2), which agrees within two standard deviations with the 139.4(4) e− found in the orthorhombic structure. The refined composition is thus

65 4. Ion exchange and dehydration of Rb-exchanged AM-2

10.20 Orthorhombic phase 13.05 Monoclinic phase 10.15 13.00

10.10 12.95 a [Å] b [Å]

10.05 12.90

10.00 12.85 0 100 200 300 400 0 100 200 300 400

7.30 950

] 945 7.25 3

940 7.20 c [Å] 935

7.15 Cell volume [Å 930

7.10 925 0 100 200 300 400 0 100 200 300 400 Temp [°C] Temp [°C]

Figure 4.3.: Changes of cell parameters upon dehydration from 25–400 ◦C. a remains almost constant, but b, c, and the cell volume decrease mainly between 100 and 250 ◦C. In this temperature range most of the H2O is released. Errors for cell dimensions are smaller than the size of the symbols. Cell parameters are also listed in Table 4.3.

Rb1.80K0.20TiSi3O9. The splitting of Rb2 into Rb2b and Rb3a may be an artefact and may also be interpreted as one anisotropically disordered site. This is supported by the high Beq values for channel occupants shown in Table 4.6. Our X-ray powder data do not allow distinction between one strongly disordered site and two partially occupied sites ca. 1 A˚ apart. Atomic co-ordinates, occupancy factors and displacement parameters for the hydrated and dehydrated phase are listed in tables 4.4 and 4.6. Measured and refined profiles for both structures are shown in figure 4.4. The dehydrated structure persists up to 600 ◦C. No increasing amorphous fraction was observed in X-ray diffraction patterns. Between 600 and 700 ◦C the crystal structure completely breaks down and at 700 ◦C the material becomes amorphous. Heating to 750 ◦C leads to crystallisation of several phases (fig. 4.7), one of which being a wadeite-type structure of composition KxRb2−xTiSi3O9, other phases could not be identified from powder X- ray diffraction patterns. This phase transformation is irreversible and was already observed by Bortun et al. (2000) for K2TiSi3O9. Peak positions of the diffraction pattern collected at room temperature are given in Table 4.13.

66 4.3. Results

Rb-AM-2 (25 °C) Iobs Icalc Intensity [Counts]

10 20 30 40 50 60 70 80 90 100 Angle [°2θ]

Rb-AM-2 (400 °C) Iobs Icalc Intensity [Counts]

10 20 30 40 50 60 70 80 90 100 Angle [°2θ]

Figure 4.4.: Observed and calculated proﬁles for the Rietveld reﬁnement of Rb-exchanged ◦ ◦ AM-2 at 25 C (top, orthorhombic, space group P212121) and 400 C (bottom, monoclinic, space group P21). The difference curve on the bottom is drawn in the same intensity scale.

67 4. Ion exchange and dehydration of Rb-exchanged AM-2

25 °C

750 °C

700 °C Intensity [Counts] 600 °C

500 °C

400 °C

10 20 30 40 50 60 Angle [°2θ]

* * *

* * * * 025 °C

400 °C

350 °C

300 °C

250 °C

Intensity [Counts] 200 °C

150 °C

100 °C

050 °C

025 °C

10 20 30 40 50 60 Angle [°2θ]

Figure 4.5.: Bottom: Dehydration of Rb-exchanged AM-2 monitored by powder X-ray diffraction shows a distinct phase change between 150 and 200 ◦C which is completely reversible immediately after room temperature has been reached again. The monoclinic phase is marked with a black bar on the right, whereas the white bar marks the orthorhombic phase. Regions with obvious differences between the two phases are marked with asterisks. Top: The AM-2 structure breaks down between 600 and 700 ◦C, and a wadeite-type phase crystallises between 700 and 750 ◦C, which remains stable at room temperature.

68 4.4. Discussion

Table 4.3.: Overview of unit cells described in this chapter.

Composition T [◦C] a [A]˚ b [A]˚ c [A]˚ β Vol. [A˚ 3] ∗ K2TiSi3O9 · H2O 25 9.941(1) 12.946(1) 7.1543(7) 90 920.7(2) K0.18Rb1.82TiSi3O9 · H2O 25 10.0851(2) 13.0157(2) 7.2255(1) 90 948.45(3) KxRb2−xTiSi3O9 · H2O 25 10.008(1) 13.012(1) 7.2116(6) 90 939.1(2) K0.18Rb1.82TiSi3O9 · H2O 50 10.0880(2) 13.0190(2) 7.2253(1) 90 948.93(3) K0.18Rb1.82TiSi3O9 · H2O 100 10.0926(2) 13.0103(2) 7.2220(1) 90 948.31(3) K0.18Rb1.82TiSi3O9 · H2O 150 10.0903(2) 12.9810(2) 7.2113(1) 90 944.55(3) K0.18Rb1.82TiSi3O9 200 10.0871(3) 12.8898(3) 7.1762(2) 91.709(2) 932.64(4) K0.18Rb1.82TiSi3O9 250 10.0828(3) 12.8716(4) 7.1627(2) 91.981(2) 929.03(5) K0.18Rb1.82TiSi3O9 300 10.0823(3) 12.8679(3) 7.1602(2) 91.968(1) 928.40(4) K0.18Rb1.82TiSi3O9 350 10.0737(2) 12.8552(3) 7.1545(2) 91.873(1) 926.01(4) K0.18Rb1.82TiSi3O9 400 10.0755(2) 12.8625(2) 7.1587(1) 91.754(1) 927.30(3) ∗∗ K0.18Rb1.82TiSi3O9 · H2O 25 10.0623(2) 13.0085(2) 7.2131(1) 90 944.16(3) ∗∗∗ KxRb2−xTiSi3O9 750 6.6599(5) 10.1397(7) 389.49(5) ∗∗∗ ∗∗ KxRb2−xTiSi3O9 25 6.5850(3) 10.0792(5) 378.50(3) ∗ Zou and Dadachov (2000) ∗∗ after heating cycle ∗∗∗ Wadeite (P63/m)

4.4. Discussion

The structure of AM-2 is made up of a framework of TiO6 octahedra and SiO4 tetrahedra (fig. 4.10). There are no direct Ti–O–Ti connections, but alternating . . . –O–Ti–O–Si–. . . chains along a, b, [101] and [101¯]. A one-dimensional open channel system is running along c consisting of eight-membered rings of approximately 6 × 8 A˚ in diameter. These continuous channels are interconnected by a set of non-continuous tubes composed of seven- and eight- membered rings in the (001) plane (fig. 4.6). The polyhedral framework has a negative charge of 2e− per formula unit (pfu), which is compensated by the incorporation of cations into the channel system. Under ambient ◦ conditions one H2O molecule is present. Although the weight loss from 25 to 400 C monitored by TG analysis corresponds to 1.2 H2O molecules pfu, only one ordered H2O site was found in the structure refinement. This leads to the assumption that either additional

H2O is disordered and the resolution of the powder X-ray diffraction data is not sufficient to locate the remaining sites, or that additional H2O is adsorbed on the crystal surface. Rb-exchange and dehydration are both processes with a high degree of completion. The exchange process plotted in figure 4.1 shows an exponential character and indicates that >99% ion exchange could have been reached with only one or two more exchange cycles. Clearfield et al. (1998) reported even higher selectivity for Rb+ from AM-2 structures with Zr/Ti ratios >50%, whereas for ratios <50% a higher preference for K+ was observed. Since we reached almost complete exchange with the pure Ti endmember, we assume that the same process with a Zr/Ti ratio >50% leads to an equal exchange ratio but in a shorter reaction time under the same conditions.

69 4. Ion exchange and dehydration of Rb-exchanged AM-2

Figure 4.6.: The channel system of AM-2 shows continuous channels running along c. These main channels are interconnected by seven- and eight-membered rings allowing relatively free migration of channel occupants among the tubes.

The cation site Rb1 is located almost exactly at the centre of an 8-membered ring made from six tetrahedra and two octahedra (fig. 4.8). It bonds to eight framework oxygens with bonding distances between 2.85(1) and 3.31(2) A,˚ seven of which belong to ring-building polyhedra. Bond eight is almost perpendicular to the ring plane fixing Rb1 in three dimensions. Towards the main channel Rb1 bonds to the H2O site Ow1 with a bond length of 2.93(2) A.˚ The second channel occupant Rb2 is located at the intersection of the main channel with the interconnecting channel, the latter joining the main channel in a 7-membered ring. At this location the framework leaves relatively large void areas, allowing Rb2 to reside above the ring plane in almost ideal bonding distance to the ring-forming polyhedra (fig. 4.8).

Towards the centre of the main channel it bonds to two H2O sites, on the opposing side it bonds to ﬁve framework oxygens of the 7-membered ring with bond lengths between 2.94(1) and 3.13(1) A.˚ Bond valence calculations for all framework atoms show low values on the vertices of the TiO6 octahedron (tab. 4.11), which are thus the preferred bonding counterparts for exchangeable cations. The AM-2 structure allows cations to take positions in bonding distance to the framework regularly distributed around TiO6 octahedra in a rectangular arrangement. This conﬁguration minimises repulsive forces between channel occupants, and since they are located off-centre in the main channels, the pore system remains open and allows cations and H2O molecules to diffuse in the crystal. The second phase observed in the hydrated diffraction pattern gradually disappeared with increasing temperature and was no longer observed in the dehydrated and re-hydrated

70 4.4. Discussion phases. We assume that this minor phase represents cores with incomplete ion exchange. Variable exchange ratios may be responsible for the observed peak broadening and ele- vated temperature may have induced enough diffusion to even out inhomogeneities in the channel content. Despite the change to monoclinic symmetry the framework of the dehydrated structure shows only minor changes compared to the orthorhombic variant. The loss of symmetry operators leads to slightly less regular channel cross sections, but the framework topology remains unchanged. The sites Rb1a and Rb1b located in the centre of the 8-membered rings remain at their position in bonding distance to two octahedra. They bond to five (Rb1a) and four (Rb1b) vertices of ring-polyhedra, and to three vertices lying outside of the ring plane (fig. 4.9). As in the hydrated structure the sites are strongly fixed in three dimensions. Rb2a also remains at its initial position at the intersection of the main channel and the intercon- nections. It bonds to seven framework oxygens, six of which being vertices of octahedra. On the splitted position, Rb3a maintains its position at the centre of the 7-membered ring, slightly shifted out of the ring plane towards the centre of the main channel. It bonds to five octahedral vertices and to O2a. The second position Rb2b is more off-centre from the 7- membered ring and bonds four times to octahedra, and also to O2a. Framework oxygen O2a is pulled towards the centre of the 7-membered ring (fig. 4.9). This causes the O2a-sharing tetrahedra to rotate and to distort the connected octahedron Tia, which becomes slightly trapezoid in cross section. The angle of the shifted vertices O1(a)–Ti(a)–O3(a) changes from 93.7(7)◦ in the orthorhombic phase, to 83(1)◦ in the monoclinic phase. In addition, units composed of one octahedron and two tetrahedra rotate slightly around [010] and cause the monoclinic angle to increase above 90◦. Distances of Rb sites to framework oxygens are listed in Table 4.8. As for the hydrated structure, bond valence calculations show a lower valence sum for oxygens which are vertices of TiO6 octahedra than for oxygens bonding only to Si. These sites are preferred bonding partners for Rb cations also in the dehydrated structure (tab. 4.12). The remarkably low bond valence sum of O6(a/b) can be explained by the un- favourable positions to bond to Rb sites, yet the central atoms Ti(a/b) remain at the centre of the octahedra and bonds to octahedral vertices remain almost regular. A shortening of the bonds to O6(a/b), as expected for a second order Jahn-Teller effect (Kunz and Brown, 1995), was neither observed in our structure refinements, nor in the single-crystal structure by Zou and Dadachov (2000) (tab. 4.10). The wadeite structure that crystallises from the amorphous phase above 700 ◦C is a hexagonal ring-silicate which completely immobilises Rb ions by incorporating them into the silicate framework. In addition one or more other phases form in quantities comparable to the wadeite phase. We were not able to identify these materials by X-ray diffraction. The bottom line of this observation is that some of the exchangeable Rb can be immobilised by heating the AM-2 material to 750 ◦C, but considerable amounts of Rb will form phases of

71 4. Ion exchange and dehydration of Rb-exchanged AM-2

* *

Intensity [Counts] * *

* * * * * * * * * 750 °C * *

10 20 30 40 50 60 70 80 Angle [°2θ]

Figure 4.7.: The wadeite-type structure which crystallises from Rb-exchanged AM-2 above 700 ◦C. Reflections belonging to this phase are marked with asterisks, unlabelled peaks belong to unknown phases. See Table 4.13 for a list of reflections. yet unknown stability. A transformation from orthorhombic to monoclinic symmetry similar to the one we observed between 150 and 200 ◦C was described by Pertierra et al. (2004), who performed dehydration experiments on AM-2 with Sn in octahedral co-ordination (K2SnSi3O9 · H2O). Although the dehydration of their sample was complete at 200 ◦C, the transformation occurs between 150 and 200 ◦C. These authors suggest that the flexibility of the AM-2 structure is not only decisive for its cation exchange capacity, but also prevents the crystal structure from breaking down upon loss of H2O. A more rigid kostylevite-type polymorph of composition K2PbSi3O9 · H2O, which is structurally closely related to AM-2, was not able to withstand the dehydration and gradually lost its crystallinity upon dehydration. From detailed analysis of the Rb-exchanged structure we conclude that the framework topology of hydrated AM-2 provides an excellent bonding environment for two Rb cations. Both sites can maintain positions in bonding distance to the negatively charged octahedra with repulsive forces between neighbouring cations being irrelevant to a large extent. The dehydrated structure provides a similar environment with minor distortions but without rearrangement of the framework topology. All exchangeable positions are located off-centre in the main channel, leaving enough space for migration of cations and H2O molecules through the channel system. In contrast to our clearly orthorhombic hydrated phase, the structure of umbite, a natural mineral with AM-2 structure and the composition K2(Zr0.86

72 4.4. Discussion

Ow1 1

Rb1 Rb2 1 Rb1 Rb2 a a c 2

b b c

Figure 4.8.: In the fully hydrated phase the cation position Rb1 is located in the centre of a 8-membered ring (1), whereas Rb2 resides at the intersection of the main channel and an interconnecting channel, the latter joining the main channel in a 7-membered ring (2) in the side wall. The projected planes of the two rings are shown as bold lines in the left ﬁgure, main channels are indicated by broken lines. The centre and right ﬁgure show the two rings projected roughly perpendicular to the ring planes.

Rb1a Rb1b

c c

b b

Tib

Tia

O2a Rb2a Tia Tib Rb3a Rb2b a a

c c

Figure 4.9.: The bonding situations of Rb cations in dehydrated AM-2. Top: Rb1a/b maintain their positions in the centre of the 8-membered ring. The ring becomes slightly irregular, and both cation positions lose one bond to the framework. Bottom: Rb2a remains almost unchanged, but the stronger bonds of Rb2b and Rb3a to the framework oxygen O2a lead to a rotation of the two connected tetrahedra. This also induces a distortion of octahedron Tia, which becomes trapezoid in cross section.

73 4. Ion exchange and dehydration of Rb-exchanged AM-2

(a) 25 ◦C

(b) 400 ◦C

Figure 4.10.: The two phases of Rb-exchanged AM-2 at 25 and 400 ◦C. Small spheres are cations (Rb, K), large spheres are H2O molecules reﬁned as oxygen. Notice the 21 screw axis in the centre of the large channels of the hydrated phase, which is lost upon dehydration.

74 4.5. Tables

Ti0.14)Si3O9 · H2O, was described by Ilyushin (1993) as monoclinic, but close to orthorhombic with a unique c axis. Although we cannot confirm this symmetry for pure Ti-AM-2 with different monovalent channel occupants, we also observed a flexibility of the framework which allows the structure to transform between orthorhombic and monoclinic symmetry without losing its integrity. This flexibility seems to be preserved over a wide range of chemical compositions of the framework structure.

4.5. Tables

Table 4.4.: Atomic coordinates, site occupancies and displacement parameters for fully hy- ◦ drated Rb-exchanged AM-2 at 25 C (orthorhombic, P212121).

Atom x/a y/b z/c Occupancy Beq Ti 0.0545(4) 0.2848(3) 0.7442(6) 1 0.28(7) Rb1 0.4355(3) 0.5827(2) 0.7694(6) 0.99(1) 2.9(2) K1 0.4355(3) 0.5827(2) 0.7694(6) 0.01(1) 2.9(2) Rb2 0.2034(4) 0.1308(3) 0.3105(5) 0.83(1) 2.9(2) K2 0.2034(4) 0.1308(3) 0.3105(5) 0.17(1) 2.9(2) Si1 0.0416(5) 0.5485(4) 0.7538(9) 1 0.28(7) Si2 0.3331(5) 0.3231(4) 0.9707(7) 1 0.28(7) Si3 0.3525(6) 0.3251(5) 0.5656(7) 1 0.28(7) O1 0.1765(8) 0.297(1) 0.956(1) 1 0.28(7) O2 0.405(1) 0.344(1) 0.7741(8) 1 0.28(7) O3 0.1939(7) 0.309(2) 0.561(1) 1 0.28(7) O4 0.154(1) 0.5736(9) 0.599(2) 1 0.28(7) O5 0.948(2) 0.264(1) 0.519(2) 1 0.28(7) O6 0.9256(9) 0.6336(5) 0.741(3) 1 0.28(7) O7 0.115(1) 0.5702(8) 0.950(1) 1 0.28(7) O8 0.907(1) 0.2704(9) 0.921(2) 1 0.28(7) O9 1.004(1) 0.4294(4) 0.719(3) 1 0.28(7) Ow1 0.197(2) 0.936(2) 0.405(2) 1 2.9(2)

75 4. Ion exchange and dehydration of Rb-exchanged AM-2

Table 4.5.: Bond angles and distances in framework polyhedra of Rb-exchanged AM-2 at ◦ 25 C (orthorhombic, P212121).

Atom 1–Atom 2–Atom 3 Angle [◦] d12 [A]˚ d23 [A]˚ d13 [A]˚ O1–Ti–O3 93.7(7) 1.97(1) 1.96(1) 2.86(1) O1–Ti–O6 88.5(9) 1.97(1) 1.981(8) 2.76(2) O1–Ti–O8 88.7(8) 1.97(1) 1.97(1) 2.75(2) O1–Ti–O9 99.2(9) 1.97(1) 1.960(7) 2.99(2) O3–Ti–O5 81.6(8) 1.96(1) 1.97(1) 2.57(2) O3–Ti–O6 97(1) 1.96(1) 1.981(8) 2.96(2) O3–Ti–O9 88.2(8) 1.96(1) 1.960(7) 2.73(2) O5–Ti–O6 87.8(9) 1.97(1) 1.981(8) 2.74(2) O5–Ti–O8 96.3(9) 1.97(1) 1.97(1) 2.94(2) O5–Ti–O9 85.0(8) 1.97(1) 1.960(7) 2.65(2) O6–Ti–O8 87.0(8) 1.981(8) 1.97(1) 2.72(2) O8–Ti–O9 87.3(8) 1.97(1) 1.960(7) 2.71(2) O4–Si1–O6 109(1) 1.62(1) 1.61(1) 2.64(2) O4–Si1–O7 104(1) 1.62(1) 1.63(1) 2.57(2) O4–Si1–O9 105(1) 1.62(1) 1.617(8) 2.56(2) O6–Si1–O7 105(1) 1.61(1) 1.63(1) 2.58(2) O6–Si1–O9 118.5(7) 1.61(1) 1.617(8) 2.776(9) O7–Si1–O9 114(1) 1.63(1) 1.617(8) 2.73(2) O1–Si2–O2 114.3(9) 1.62(1) 1.617(9) 2.72(1) O1–Si2–O4 107(1) 1.62(1) 1.64(1) 2.62(2) O1–Si2–O8 109(1) 1.62(1) 1.63(1) 2.64(2) O2–Si2–O4 109(1) 1.617(9) 1.64(1) 2.65(2) O2–Si2–O8 110(1) 1.617(9) 1.63(1) 2.66(2) O4–Si2–O8 108(1) 1.64(1) 1.63(1) 2.64(2) O2–Si3–O3 111.3(9) 1.614(8) 1.61(1) 2.67(1) O2–Si3–O5 105(1) 1.614(8) 1.63(2) 2.58(2) O2–Si3–O7 106.6(9) 1.614(8) 1.63(1) 2.60(1) O3–Si3–O5 119(1) 1.61(1) 1.63(2) 2.79(2) O3–Si3–O7 107(1) 1.61(1) 1.63(1) 2.61(2) O5–Si3–O7 107(1) 1.63(2) 1.63(1) 2.61(2)

76 4.5. Tables

Table 4.6.: Atomic coordinates, site occupancies and displacement parameters for dehy- ◦ drated Rb-exchanged AM-2 at 400 C (monoclinic, P21).

Atom x/a y/b z/c Occupancy Beq Tia 0.0673(4) 0.3110(3) −0.0111(6) 1 1.02(8) Tib 0.5455(4) 0.2415(3) 0.5058(6) 1 1.02(8) Rb1a 0.435(1) 0.625(1) −0.005(2) 1.00(2) 4.2(2) Rb1b 0.951(1) 0.938(2) 0.498(2) 0.88(2) 4.2(2) Rb2a 0.202(1) 0.163(1) 0.443(2) 0.924(9) 4.2(2) Rb2b 0.732(3) 0.430(3) 0.868(7) 0.30(3) 4.2(2) Rb3a 0.710(2) 0.416(1) −0.005(3) 0.70(3) 4.2(2) Si1a 0.0415(6) 0.5773(4) 0.0035(7) 1 1.02(8) Si1b 0.5499(5) 0.9794(4) 0.5003(8) 1 1.02(8) Si2a 0.3751(6) 0.3659(4) 0.1806(7) 1 1.02(8) Si2b 0.8441(5) 0.2184(4) 0.2848(7) 1 1.02(8) Si3a 0.3506(5) 0.3587(4) 0.7838(7) 1 1.02(8) Si3b 0.8418(5) 0.2047(4) 0.6990(8) 1 1.02(8) O1a 0.223(2) 0.327(4) 0.159(2) 1 1.02(8) O1b 0.690(2) 0.242(4) 0.329(3) 1 1.02(8) O2a 0.436(2) 0.382(3) −0.0243(7) 1 1.02(8) O2b 0.896(2) 0.183(3) 0.4920(8) 1 1.02(8) O3a 0.202(2) 0.310(3) 0.798(3) 1 1.02(8) O3b 0.685(1) 0.231(3) 0.703(2) 1 1.02(8) O4a 0.138(2) 0.613(1) 0.838(3) 1 1.02(8) O4b 0.630(3) 0.974(1) 0.702(2) 1 1.02(8) O5a 0.946(2) 0.293(1) 0.772(2) 1 1.02(8) O5b 0.412(4) 0.2513(7) 0.699(5) 1 1.02(8) O6a 0.9165(9) 0.6584(5) −0.015(6) 1 1.02(8) O6b 0.430(1) 0.8941(4) 0.495(7) 1 1.02(8) O7a 0.108(4) 0.5949(8) 0.212(2) 1 1.02(8) O7b 0.677(1) 0.943(2) 0.383(3) 1 1.02(8) O8a 0.940(4) 0.305(1) 0.191(5) 1 1.02(8) O8b 0.411(2) 0.263(1) 0.305(3) 1 1.02(8) O9a 0.003(1) 0.4549(4) −0.005(5) 1 1.02(8) O9b 0.498(2) 0.0960(4) 0.454(4) 1 1.02(8)

77 4. Ion exchange and dehydration of Rb-exchanged AM-2

Table 4.7.: Bond angles and distances in framework polyhedra of Rb-exchanged AM-2 at ◦ 400 C (monoclinic, P21).

Atom 1–Atom 2–Atom 3 Angle [◦] d12 [A]˚ d23 [A]˚ d13 [A]˚ O1a–Tia–O3a 83(1) 1.97(2) 1.96(2) 2.60(3) O1a–Tia–O6a 89(2) 1.97(2) 1.978(8) 2.77(4) O1a–Tia–O8a 94(2) 1.97(2) 1.96(4) 2.88(4) O1a–Tia–O9a 98(2) 1.97(2) 1.963(8) 2.97(4) O3a–Tia–O5a 83(1) 1.96(2) 1.96(2) 2.59(3) O3a–Tia–O6a 90(2) 1.96(2) 1.978(8) 2.78(4) O3a–Tia–O9a 105(2) 1.96(2) 1.963(8) 3.11(3) O5a–Tia–O6a 90(1) 1.96(2) 1.978(8) 2.79(3) O5a–Tia–O8a 100(2) 1.96(2) 1.96(4) 3.01(4) O5a–Tia–O9a 86(1) 1.96(2) 1.963(8) 2.68(3) O6a–Tia–O8a 87(2) 1.978(8) 1.96(4) 2.71(4) O8a–Tia–O9a 78(2) 1.96(4) 1.963(8) 2.47(3) O1b–Tib–O3b 86(1) 1.96(2) 1.96(1) 2.69(2) O1b–Tib–O6b 84(2) 1.96(2) 1.978(7) 2.64(5) O1b–Tib–O8b 92(1) 1.96(2) 1.96(2) 2.82(3) O1b–Tib–O9b 94(2) 1.96(2) 1.964(9) 2.86(4) O3b–Tib–O5b 89(2) 1.96(1) 1.96(4) 2.76(4) O3b–Tib–O6b 89(2) 1.96(1) 1.978(7) 2.77(4) O3b–Tib–O9b 104(2) 1.96(1) 1.964(9) 3.09(3) O5b–Tib–O6b 91(2) 1.96(4) 1.978(7) 2.82(4) O5b–Tib–O8b 92(2) 1.96(4) 1.96(2) 2.82(4) O5b–Tib–O9b 92(2) 1.96(4) 1.964(9) 2.81(3) O6b–Tib–O8b 87(1) 1.978(7) 1.96(2) 2.70(3) O8b–Tib–O9b 81(1) 1.96(2) 1.964(9) 2.54(2) O4a–Si1a–O6a 104(2) 1.62(2) 1.64(1) 2.56(3) O4a–Si1a–O7a 113(2) 1.62(2) 1.63(2) 2.71(3) O4a–Si1a–O9a 114(2) 1.62(2) 1.623(8) 2.71(2) O6a–Si1a–O7a 106(2) 1.64(1) 1.63(2) 2.62(4) O6a–Si1a–O9a 115.5(7) 1.64(1) 1.623(8) 2.758(9) O7a–Si1a–O9a 105(2) 1.63(2) 1.623(8) 2.58(3) O4b–Si1b–O6b 110(2) 1.63(2) 1.63(1) 2.67(4) O4b–Si1b–O7b 94(2) 1.63(2) 1.62(2) 2.38(3) O4b–Si1b–O9b 112(2) 1.63(2) 1.62(1) 2.69(3) O6b–Si1b–O7b 113(2) 1.63(1) 1.62(2) 2.71(2) O6b–Si1b–O9b 112.6(8) 1.63(1) 1.62(1) 2.70(1) O7b–Si1b–O9b 114(2) 1.62(2) 1.62(1) 2.73(2) O1a–Si2a–O2a 110(2) 1.62(2) 1.62(1) 2.65(3) O1a–Si2a–O4b 106(3) 1.62(2) 1.62(2) 2.58(4) O1a–Si2a–O8b 90(2) 1.62(2) 1.63(2) 2.29(3) O2a–Si2a–O4b 112(2) 1.62(1) 1.62(2) 2.69(2) O2a–Si2a–O8b 121(2) 1.62(1) 1.63(2) 2.83(3) O4b–Si2a–O8b 115(1) 1.62(2) 1.63(2) 2.74(2)

78 4.5. Tables

Table 4.7 continued Atom 1–Atom 2–Atom 3 Angle [◦] d12 [A]˚ d23 [A]˚ d13 [A]˚ O1b–Si2b–O2b 99(2) 1.62(2) 1.62(1) 2.48(3) O1b–Si2b–O4a 112(2) 1.62(2) 1.63(2) 2.70(4) O1b–Si2b–O8a 122(3) 1.62(2) 1.64(3) 2.85(4) O2b–Si2b–O4a 103(2) 1.62(1) 1.63(2) 2.54(3) O2b–Si2b–O8a 113(3) 1.62(1) 1.64(3) 2.72(4) O4a–Si2b–O8a 106(2) 1.63(2) 1.64(3) 2.60(2) O2a–Si3a–O3a 119(2) 1.63(1) 1.63(2) 2.80(3) O2a–Si3a–O5b 106(2) 1.63(1) 1.64(2) 2.60(4) O2a–Si3a–O7b 125(2) 1.63(1) 1.63(2) 2.89(2) O3a–Si3a–O5b 93(3) 1.63(2) 1.64(2) 2.38(4) O3a–Si3a–O7b 99(2) 1.63(2) 1.63(2) 2.49(4) O5b–Si3a–O7b 110(2) 1.64(2) 1.63(2) 2.69(3) O2b–Si3b–O3b 114(2) 1.62(1) 1.62(1) 2.72(2) O2b–Si3b–O5a 100(2) 1.62(1) 1.62(2) 2.49(3) O2b–Si3b–O7a 96(1) 1.62(1) 1.62(2) 2.40(2) O3b–Si3b–O5a 118(2) 1.62(1) 1.62(2) 2.78(2) O3b–Si3b–O7a 118(3) 1.62(1) 1.62(2) 2.78(4) O5a–Si3b–O7a 107(2) 1.62(2) 1.62(2) 2.61(2)

◦ Table 4.8.: Rb–O distances ≤ 3.37 A˚ in the hydrated orthorhombic (25 C, P212121) and the ◦ dehydrated monoclinic (400 C, P21) Rb-exchanged AM-2 structure.

25 ◦C 400 ◦C Atom 1–Atom 2 Distance [A]˚ Atom 1–Atom 2 Distance [A]˚ Rb1–O1 2.98(1) Rb1a–O1b 3.00(3) Rb1–O2 3.13(1) Rb1a–O2a 3.13(4) Rb1–O3 2.85(1) Rb1a–O3b 2.86(3) Rb1–O4 3.10(1) Rb1a–O4a 3.16(2) Rb1–O4 3.26(1) Rb1a–O4b 3.00(2) Rb1–O5 2.94(1) Rb1a–O5b 3.10(3) Rb1–O7 3.09(1) Rb1a–O8b 3.22(2) Rb1–O9 3.31(2) Rb1a–O9b 3.32(3) Rb1–Ow1 2.93(2) Rb1b–O1a 3.37(3) Rb1b–O2b 3.21(4) Rb1b–O3a 3.06(3) Rb1b–O5a 2.90(2) Rb1b–O7a 2.97(2) Rb1b–O7b 2.85(2) Rb1b–O8a 2.98(4) Rb2–O1 3.37(2) Rb2a–O1a 2.94(4) Rb2–O2 3.09(1) Rb2a–O2b 3.13(2) Rb2–O3 2.95(2) Rb2a–O3a 3.17(3) Rb2–O5 3.08(2) Rb2a–O5b 2.98(4)

79 4. Ion exchange and dehydration of Rb-exchanged AM-2

Table 4.8 continued 25 ◦C 400 ◦C Atom 1–Atom 2 Distance [A]˚ Atom 1–Atom 2 Distance [A]˚ Rb2–O8 2.94(1) Rb2a–O6a 3.26(4) Rb2–O9 3.13(1) Rb2a–O8b 2.68(2) Rb2–Ow1 2.63(2) Rb2a–O9b 3.10(2) Rb2–Ow1 3.22(2) Rb2b–Rb3a 0.96(5) Rb2b–O2a 3.16(4) Rb2b–O3b 2.85(5) Rb2b–O5a 2.89(4) Rb2b–O6b 3.06(6) Rb2b–O9a 2.87(4) Rb3a–O1b 3.28(4) Rb3a–O2a 2.80(2) Rb3a–O3b 3.18(4) Rb3a–O5a 3.31(3) Rb3a–O8a 3.03(4) Rb3a–O9a 2.99(2)

Table 4.9.: Angles between central atoms of polyhedra in the orthorhombic hydrated and monoclinic dehydrated structure.

25 ◦C 400 ◦C Atom 1–Atom 2–Atom 3 Angle [◦] Atom 1–Atom 2–Atom 3 Angle [◦] Ti–O1–Si2 132.8(6) Tia–O1a–Si2a 146(1) Si2–O2–Si3 130.4(6) Tib–O1b–Si2b 149(1) Ti–O3–Si3 135.8(6) Si2a–O2a–Si3a 122.4(7) Si1–O4–Si2 127.8(8) Si2b–O2b–Si3b 132.8(8) Ti–O5–Si3 137.2(8) Tia–O3a–Si3a 134(1) Ti–O6–Si1 127.3(6) Tib–O3b–Si3b 132.8(7) Si1–O7–Si3 133.0(8) Si1a–O4a–Si2b 135(1) Ti–O8–Si2 136.7(7) Si1b–O4b–Si2a 118(1) Ti–O9–Si1 147.6(5) Tia–O5a–Si3b 136.1(9) Tib–O5b–Si3a 126(1) Tia–O6a–Si1a 124.3(6) Tib–O6b–Si1b 125.3(5) Si1a–O7a–Si3b 126(1) Si1b–O7b–Si3a 116(1) Tia–O8a–Si2b 137(2) Tib–O8b–Si2a 130.7(9) Tia–O9a–Si1a 146.7(5) Tib–O9b–Si1b 140.2(6)

80 4.5. Tables

Table 4.10.: Bond valence calculations for the AM-2 published by Zou and Dadachov (2000). K–O bond lengths ≤ 3.6 A˚ were considered.

Ti K1 K2 Si1 Si2 Si3 Sum lattice Sum total O1 0.753 0.130 1.087 1.840 1.970 O2 0.145 0.082 0.989 0.989 1.978 2.206 O3 0.687 0.254 0.245 1.093 1.780 2.279 O4 0.155 0.976 1.022 1.998 2.235 0.082 O5 0.633 0.193 0.195 1.099 1.733 2.178 0.057 O6 0.607 0.082 1.050 1.656 1.738 O7 0.030 0.042 0.989 0.981 1.971 2.224 0.182 O8 0.674 0.237 1.079 1.753 1.990 O9 0.689 0.077 0.149 1.126 1.815 2.041 Ow1 0.249 0.389 0 0.638 Sum 4.042 1.498 1.478 4.141 4.177 4.163

Table 4.11.: Bond valence calculations for orthorhombic Rb-exchanged AM-2 at 25 ◦C. Rb–O bond lengths ≤ 3.6 A˚ were considered.

Ti Rb1 Rb2 Si1 Si2 Si3 Sum lattice Sum total O1 0.658 0.144 0.050 1.056 1.713 1.908 O2 0.096 0.107 1.064 1.073 2.137 2.340 O3 0.676 0.205 0.156 1.084 1.760 2.121 O4 0.104 1.056 1.000 2.056 2.160 O5 0.658 0.068 0.110 1.027 0.040 1.685 1.903 O6 0.638 0.036 1.084 1.723 1.759 O7 0.107 1.027 1.027 0.036 2.055 2.198 O8 0.658 0.032 0.160 1.027 1.685 1.877 O9 0.676 0.059 0.096 1.064 1.740 1.895 Ow1 0.165 0.371 0.075 0 0.611 Sum 3.963 1.015 1.203 4.232 4.147 4.212

81 4. Ion exchange and dehydration of Rb-exchanged AM-2 9 .6 .5 .0 .5 .2 1.886 2.102 2.288 2.080 2.231 1.724 1.787 1.984 1.703 2.083 1.717 1.741 2.083 1.924 1.671 2.219 1.056 1.676 1.027 2.153 1.644 2.218 1.676 2.083 1.905 2.219 2.284 1.731 2.083 1.056 1.000 1.000 1.027 1.703 1.932 1.056 1.056 1.731 1.924 2.111 1.056 1.056 4.222 1.731 1.027 1.027 1.027 4.082 1.713 1.027 1.056 4.138 1.047 1.056 1.027 4.194 0.140 4.166 0.194 1.027 0.126 4.130 1.000 0.038 0.104 0.707 1.056 0.116 0.855 1.056 0.324 0.028 0.982 1.056 0.205 0.994 0.030 0.144 0.057 0.148 1.056 0.059 0.924 0.068 0.669 0.184 4.015 0.075 3.999 0.676 0.084 0.144 Sum 0.205 0.670 0.234 O9b 0.179 O9a 0.064 0.089 0.676 0.644 O8b 0.046 0.096 0.104 0.136 O8a 0.086 0.089 0.077 0.676 O7b 0.116 O7a 0.644 O6b 0.199 O6a 0.676 0.676 O5b 0.160 0.096 O5a 0.050 O4b O4a 0.136 0.676 O3b 0.676 O3a O2b O2a 0.658 O1b O1a al 4.12.: Table i i baR1 baR2 baS1 ibS2 ibS3 ibSmltieSmtotal Sum lattice Sum Si3b Si3a Si2b Si2a Si1b Si1a Rb3a Rb2b Rb2a Rb1b Rb1a Tib Tia odvlnecluain o oolncR-xhne M2a 400 at AM-2 Rb-exchanged monoclinic for calculations valence Bond 0.030 ◦ .R– odlengths bond Rb–O C. .8 2.532 2.083 ≤ 3.6 eeconsidered. were A ˚

82 4.5. Tables

Table 4.13.: Peak positions of the diffraction pattern collected at room temperature after heating to 750 ◦C. hkl indices of some strong reﬂections belonging to the wadeite-type structure are given in the ﬁfth column. Other phases are unknown.

2θ d [A]˚ Abs. intens. [cts] Rel. intens.[%] hklwadeite 15.348 5.768 1538 28 (010) 17.404 5.091 3064 72 (002) 20.996 4.228 938 12 23.360 3.805 1031 15 (012) 24.236 3.669 588 3 24.829 3.583 3906 96 25.539 3.485 1242 21 27.501 3.241 3290 79 29.381 3.037 3790 94 30.560 2.923 1950 42 (013) 31.186 2.866 4002 100 (020) 35.496 2.527 2634 62 (004) 36.021 2.491 2171 49 (022) 37.459 2.399 516 3 38.936 2.311 1021 17 (014) 39.888 2.258 1077 19 40.372 2.232 863 14 41.677 2.165 494 3 42.997 2.102 515 4 44.339 2.041 1509 32 45.590 1.988 760 11 46.844 1.938 566 6 47.639 1.907 815 13 48.049 1.892 621 7 50.397 1.809 768 12 50.471 1.807 709 10 50.546 1.804 709 10 51.209 1.782 601 7 51.581 1.770 759 11 52.826 1.732 568 6 53.748 1.704 647 8 53.870 1.700 604 7 54.507 1.682 724 10 54.933 1.670 520 5 55.952 1.642 636 8 56.051 1.639 668 9 57.065 1.613 536 5 57.732 1.596 740 11 58.105 1.586 987 18

83 4. Ion exchange and dehydration of Rb-exchanged AM-2

References

Bortun, A. I., Bortun, L. N., Poojary, D. M., Xiang, O., and Clearﬁeld, A. Synthesis, characterization, and ion exchange behaviour of a framework potassium titanium trisilicate

K2TiSi3O9 · H2O and its protonated phases. Chemistry of Materials, 12:294–305, 2000.

Brese, N. E. and O’Keeffe, M. Bond-valence parameters for solids. Acta Crystallographica, B47:192–197, 1991.

Clearﬁeld, A., Bortun, A. I., Bortun, L. N., Poojary, D. M., and Khainakov, S. A. On the selec-

tivity regulation of K2ZrSi3O9 · H2O-type ion exchangers. Journal of Molecular Structure, 470:207–213, 1998.

Dobelin,¨ N. and Armbruster, T. Stepwise dehydration and change of framework topology in Cd-exchanged heulandite. Microporous and Mesoporous Materials, 61:85–103, 2003.

Ilyushin, G. D. New data on crystal-structure of umbite K2ZrSi3O9 · H2O. Inorganic Mate- rials, 29(7):1128–1133, 1993.

Kunz, M. and Brown, I. D. Out-of-center distortions around octahedrally coordinated d0 transition metals. Journal of Solid State Chemistry, 115:395–406, 1995.

Pertierra, P., Salvado,´ M. A., Garc´ıa-Granda, S., Khainakov, S. A., and Garc´ıa, J. R. Thermal

behavior of K2MSi3O9 · H2O with the structure of umbite (M = Sn) and kostylevite (M = Pb) minerals. Thermochimica Acta, 423:113–119, 2004.

Rodr´ıguez-Carvajal, J. FullProf 2000. Computer program, 2005. Version 3.20.

Zou, X. and Dadachov, M. S. K2TiSi3O9 · H2O. Acta Crystallographica Section C, 56:738–739, 2000.

84 5. Structural characterisation of ion-exchanged AM-2

5.1. Abstract

The microporous titanosilicate AM-2 was used for ion-exchange experiments with various mono- and bivalent cations (Cs+, Sr2+, Na+, Mn2+, Ca2+, and Cu2+). The exchange process was monitored with ICP-OES, and the exchanged structures were reﬁned from powder X-ray diffraction data using the Rietveld method. Dehydration was studied by thermo- gravimetric analysis and powder XRD in order to determine the thermal stabilities. Cu- exchange could not be realised due to a strongly acidic exchange solution, leading to dissolution of AM-2. All other cations considered in this study could be exchanged against K. The kinetics of the exchange reaction at 90 ◦C depended on the ionic radius and ranged from ca. 200 hours for Mn, to >1700 hours for Cs to reach an exchange ratio of 99%. The cation and

H2O arrangement in the cavities follows a trend: Monovalent cation positions can clearly be distinguished from H2O sites, whereas bivalent cations share their positions with H2O molecules, leading to a high degree of disorder among the channel occupants. The Na- exchanged structure was the only one to show lowering of symmetry from orthorhombic

(space group P212121) to monoclinic (space group P21/c) induced by ion-exchange Na → K. The unexchanged AM-2 structure was able to withstand dehydration by lowering its ◦ symmetry to monoclinic (space group P21) above 250 C, and back to orthorhombic upon rehydration and cooling to room temperature. All other structures showed an irreversible gradual degradation of the crystal structure to X-ray amorphous. The maximum thermal stability depended on size and valence of the exchanged cation. The X-ray amorphous products recrystallised above 700–750 ◦C, leading to new phases which are structurally not related to AM-2.

5.2. Experimental procedure

Ion exchange was done on AM-2 powder of composition K2TiSi3O9 · H2O from synthesis batches TiS11 and TiS16. These series were chosen since no impurities were found in SEM micrographs and powder XRD analyses, and the particles are well crystallised and form intergrowths with large specific surfaces. The formation of aggregates allows the powder to be filtered from the solution without plugging the filter, and reduces orientation effects in powder XRD samples. 1 M exchange solutions were prepared by dissolving CsCl, NaCl,

SrCl2, MnCl2, CuCl2, and CaCl2 in H2O. 0.3–0.6 g of the powder were mixed with 20 ml

85 5. Structural characterisation of ion-exchanged AM-2

Table 5.1.: Developing of pH values during the ion-exchange process. 1 M solutions of x+ mono- and divalent cations were prepared by dissolving M Clx in H2O, with M being Cs+, Na+, Sr2+, Mn2+, Cu2+, and Ca2+. pH values from the Na-exchange are not available. Durations (hours) of exchange cycles are given in parentheses.

Initial I (5) II (18) III (96) IV (71) V (71) VI (71) VII (95) CsCl 7.31 10.68 10.00 9.40 9.97 10.01 9.92 9.68 NaCl − − − − − − − − SrCl2 6.32 6.68 6.19 5.44 5.74 5.55 5.55 5.57 Initial I (5) II (18) III (72) IV (71)

MnCl2 5.11 6.17 5.88 5.45 5.44 CuCl2 2.58 2.70 2.62 2.57 2.56 CaCl2 6.21 8.33 6.55 5.94 3.70 of the exchange solution and treated at 90 ◦C in covered Teflon beakers for several hours at ambient pressure. After each cycle the powder was filtered off, washed with deionised water and acetone, and dried for several minutes at 50 ◦C. The used exchange solution was kept for further analysis, and the powder was again mixed with 20 ml of fresh solution and treated at 90 ◦C. Four to seven cycles were performed, depending on the exchange ratio obtained after the fourth cycle. Reaction times and the pH values of the exchange solutions after each pass are given in Table 5.1. About 0.01–0.15 g powder were lost per cycle, either by dissolution or in the filter. K+ concentrations in the exchange solution were measured by ICP-OES analysis in order to determine the exchange ratio.

The exchangeable H2O content was determined by thermo-gravimetric analysis (TGA). Prior to the TGA experiments the powders were kept in open phials for several days to ◦ allow the material to regain H2O possibly lost by washing with acetone and drying at 50 C. ◦ ◦ −1 TGA data was collected from 25–400 C at a heating rate of 5 C min in N2 atmosphere. Structural changes upon dehydration were monitored by powder X-ray diffraction on a Philips X’Pert Pro multipurpose diffractometer in theta-2theta configuration, operating at 40 mA/40 kV. The device was equipped with an Anton Paar HTK-1200 heating chamber allowing in-situ data acquisition from room temperature to 1200 ◦C in natural or controlled atmosphere. Data sets used for Rietveld structure analysis were recorded from 9–100◦ 2θ and merged from 10 successively recorded data sets. Those used for unit cell determination were recorded from 9–60◦ 2θ. Ni-filtered CuKα X-radiation from a fine-focus tube was used, leading to a maximum resolution of 1.006 A˚ at 100◦ 2θ. The programmable divergence and anti-scatter slits were fixed to 1/8◦ during data collection, and a fixed anti-scatter slit of ◦ 1/4 aperture was inserted in the incident beam. The samples were prepared in an Al2O3 ceramics sample holder of 10 mm in diameter remaining in horizontal position during the entire data collection. High-temperature data was collected in two passes: In the first sequence the sample was heated from 25 to 400 ◦C at a heating rate of 5 ◦C min−1. Data sets were collected at 25, 50, 100, 150, 200, 250, 300, 350, and 400 ◦C. During data collection the

86 5.3. Results temperature was held constant. The 25 and 400 ◦C data sets were used for Rietveld structure analysis, whereas the intermediate sets were used to monitor changes of the unit cell. After the 400 ◦C data set, another 25 ◦C pattern was collected to monitor the reversibility of the dehydration reaction. In a second heating sequence the sample was heated to 800 ◦C. Data sets were recorded at 400, 500, 600, 700, 750, and 800 ◦C, and again at 25 ◦C after heating. Data sets recorded above room temperature were corrected for displacement caused by thermal expansion of the sample holder. Following formula was used for correction: ∆2θ = (2s · cos θ)/R; s being the displacement from the focus circle in millimetres, R being the goniometer radius in millimetres, and θ, 2θ given in radians. Displacement distances s were provided by the manufacturer of the heating chamber. Crystal structures were reﬁned with the Rietveld programme FULLPROF (Rodr´ıguez-

Carvajal, 2005). The AM-2 structure of composition K2TiSi3O9 · H2O refined from single- crystal X-ray data and published by Zou and Dadachov (2000) served as starting model for the TiSi3O9 framework. Soft distance constraints were applied to Si–O and Ti–O bonds. Ideal Si–O distances were set to 1.6259 A,˚ Ti–O distances to 1.962 A,˚ which corresponds to the average Si–O and Ti–O distances in the structure of Zou and Dadachov (2000). Standard deviations for the constraints were set to 0.02 A˚ for high-quality, and to 0.01 A˚ for the lower quality data sets. One common isotropic displacement parameter Boverall was refined for all framework atoms, and depending on the quality of the data set, isotropic displacement parameters for channel occupants were either refined with the framework in Boverall, as a common Bchannel factor for all channel occupants independent of the framework, or as independent groups of B factors for ordered and disordered channel occupants. Occupancies of framework atoms were fixed to 1.0, whereas those of channel occupants were refined with arbitrary scattering factors (K, O, or the exchanged cation) in the first stage of the refinement. Later we assigned elements to these extraframework sites on the basis of electron density, distances to framework atoms, distances to other channel occupants, charge balance, and exchange ratios determined by ICP analysis. If no distinction between cations and H2O sites could be made, all channel sites were refined with scattering factors of the exchanged cation, without restrictions for occupancies.

5.3. Results

5.3.1. Ion exchange

Due to the ion-exchange the Mn-exchanged powder became dark brown to almost black. The Cu-exchanged powder showed a very light turquoise tint, whereas all other powders remained white. K+ concentrations in the exchange solutions determined by ICP-OES analysis are shown in figure 5.1. Since we expect the extraction of K+ to follow the exponential function f (x) = 100 − 100 · e(x · k) which converges at 100%, we fitted the experimental data for the parameter k. According to the fitted functions, 99% K+ would be extracted

87 5. Structural characterisation of ion-exchanged AM-2

Cs-AM-2 Sr-AM-2 100

40 Exchange ratio [%]

20 k = -0.00262 k = -0.00554

0 0 100 200 300 400 500 0 100 200 300 400 500

Mn-AM-2 Cu-AM-2 Ca-AM-2 100

50 Exchange ratio [%]

k = -0.023 k = -0.08 k = -0.0168 0 0 100 200 0 100 200 0 100 200 Duration [h]

Figure 5.1.: The ion exchange processes were monitored by ICP analysis. Exchange ratios were ﬁtted for k with the function f (x) = 100 − 100 · e(x · k). Error bars indicate an error range of ±5%.

88 5.3. Results after 1758 h (Cs), 831 h (Sr), 200 h (Mn), and 274 h (Ca). K+ extractions obtained in our experiments were 71(5)% (Cs), 89(5)% (Sr), 95(5)% (Mn), and 95(5)% (Ca). ICP data for Na exchange is not available. The pH of fresh exchange solutions is close to neutral, with the exception of the Cu-solution, which was strongly acidic (tab. 5.1). After the first ion-exchange pass, the Cs-solution was strongly alkaline and remained at pH 10.0 ± 1 for the following cycles. The Sr- and Mn-solutions were close to neutral after the first cycles, and became slightly acidic (pH 5.55) with proceeding ion-exchange. The Ca-solution became alkaline after the first cycle, and steadily droped to the acidic milieu with proceeding reactions. It is to be expected that longer reaction time or additional Ca-exchange cycles would have dissolved the AM-2 structure. The Cu-solution was strongly acidic and remained at pH 2.5–2.7 during the entire experiment. ICP analysis showed that 90% of the initial K+ was extracted after only two cycles. Since it has been shown that AM-2 is not stable in acidic milieus (Bor- tun et al., 2000), we assume that the material was dissolved and the release of K+ seen in ICP results originates from dissolution rather than from the ion-exchange process.

5.3.2. Thermo-gravimetric analysis

In figures 5.2–5.3 the Cs-, Sr-, and Na-exchanged structures show a well defined dehydration behaviour. The release of H2O starts at room temperature and subsides around 400– 500 ◦C. The unexchanged AM-2 structure shows a steadily increasing weight loss with a maximum release rate around 330 ◦C, which rapidly diminishes between 330 and 400 ◦C. In contrast the weight-loss curve of the Cs-exchanged type shows three bumps between ◦ 25 and 60, 60 and 280, and 280 and 370 C. The total loss of weight amounts to 1.76 H2O molecules pfu. A similar characteristic was observed for the Sr-exchanged structure. A first ◦ bump between 25 and 120 C releasing 1.0 H2O molecules pfu is followed by a steady slope ◦ ◦ ◦ up to 350 C, with a weak step at 200 C. From 350 to 500 C 0.37 H2O are released, leading to a total weight loss of 2.94 H2O molecules pfu. In contrast to the previous structures, the Na-exchanged type loses weight not until 40 ◦C. Two well-defined bumps between 40 and 160, and 160 and 360 ◦C are observed, the first corresponding to a release of 0.54, the second ◦ to 1.03 H2O molecules pfu. The weight loss completely subsides around 400 C. The Mn-exchanged phase loses almost 20 wt-% between 25 and 100 ◦C. Since such high

H2O concentrations in the channel system are very unlikely for the AM-2 structure, we assume that hydrous phases exist on the surface of the AM-2 particles or as impurities in the powder. Impurities in the XRD pattern conﬁrmed the presence of additional phases.

5.3.3. Structure reﬁnement

Rietveld structure reﬁnements were done for all room-temperature phases except the Cu- exchanged one, as well as for the Cs-exchanged phase at 400 ◦C. Reﬁnement parameters 2− are listed in Table 5.3. The topologies of the (TiSi3O9) frameworks are identical in all re-

89 5. Structural characterisation of ion-exchanged AM-2

Weight [mg] Weight [mg] 9.4 8.6 8.8 9.0 9.2 9.6 9.8 14.0 14.2 14.4 14.6 14.8 15.0 0 0 0 0 500 400 300 200 100 = 0.99 H Step: -3.66 wt% 0 0 0 400 300 200 100 2 O pfu Step: -2.36 wt% = 0.64 H = 0.94 H Step: -3.49 wt% 2 O pfu 2 Step: -5.35 wt% = 1.1 H O pfu Temperature [°C] Temperature [°C] 2 O pfu = 2.94 H Step: -10.85 wt% 2 O pfu SrTiSi K 2 TiSi 3 O 3 O 9 × 2H 9 × H 2 2 O O

Weight [mg] 13.0 13.2 13.4 13.6 13.8 14.0 iue5.2.: Figure tprgt,adS-xhne bto-et M2uo heat- 400 upon to AM-2 ing (bottom-left) Sr-exchanged and (top-right), ie ntetprgtcorner. top-right the in given = 0.30 H Step: -1.09 wt% = 1.29 H Step: -4.623 wt% 2 egtls fKbaig(o-et,Cs-exchanged (top-left), K-bearing of loss Weight O pfu 2 O pfu ◦ 0 0 0 400 300 200 100 ,maue yTA h endcmoiinis composition reﬁned The TGA. by measured C, Temperature [°C] Step: -6.283 wt% = 1.76 H Cs 1.323 2 O pfu K 0.718 TiSi 3 O 9 × 1.29H 2 O

90 5.3. Results O 2 × 3.04H 9 O 3 MnTiSi O pfu 2 = 7.83 H Step: -26.663 wt% Temperature [°C] O pfu 2 O pfu 2 Step: -6.993 wt% = 2.05 H 100 200 300 400 Weight loss of Na- (top-left), Mn-exchanged (top- Step: -19.67 wt% = 5.78 H C, measured by TGA. The reﬁned composition is given ◦ right), and Cu-exchanged (bottom-left) AM-2 upon400 heating to in the top-right corner.

Figure 5.3.:

8.5 7.5 6.5 9.0 8.0 7.0 Weight [mg] Weight O 2 × 1.9H 9 O 3 TiSi Step: -1.64 wt% 0.16 K O pfu 2 1.82 Step: -3.48 wt% Na Step: -7.905 wt% = 1.57 H Temperature [°C] Temperature [°C] Step: -7.98 wt% O pfu 2 Step: -5.195 wt% = 1.03 H 100 200 300 400 100 200 300 400 O pfu 2 Step: -1.94 wt% Step: -0.92 wt% Step: -2.71 wt% = 0.54 H

9.8 9.6 9.4 9.2

10.0 9.8 9.6 9.4 9.2 9.0

Weight [mg] Weight Weight [mg] Weight

91 5. Structural characterisation of ion-exchanged AM-2

fined structures. Corner-sharing chains of alternating TiO6 octahedra and SiO4 tetrahedra run along [110], [010], and [101], and are connected among each other to form a three- dimensional framework. Void areas concatenated along [001] form an open channel system containing mono- or bivalent cations or protons, compensating the negative charge of the framework, as well as a varying number of H2O molecules. The main channels are connected with their neighbouring ones by crosslinks meeting the main channels in rings made of six tetrahedra and two octahedra on one end, hitherto referred to as the 8-membered ring, and of four tetrahedra and three octahedra on the other end, referred to as the 7-membered ring (fig. 5.4). Results of the refinements, including atomic coordinates, site occupancy factors, cell parameters, bond distances and angles, and bond valence calculations, are given in Tables 5.4–5.33 in section 5.6. Bond valences were calculated with the programme BOND VALENCE WIZARD (Orlov et al., 1998).

Cs-exchanged AM-2 Rietveld refinement of the Cs-exchanged structure showed that a subordinate second phase with a somewhat smaller cell than the main phase was present. The cell dimensions of this minor phase are similar to the unexchanged structure, thus we assume that cores with incomplete ion-exchange remained. Three sites were identified in the main channel. An electron density of 57.3 e− and bonding distances to framework oxygens between 2.85(2) and 3.46(2) A˚ on Cs1 indicate that this site is fully occupied by Cs. It is located exactly at the centre of the 8-membered ring. 29.2 e− were found on the second cation site Cs2/K2, which corresponds to a Cs/K ratio of 1/3 on a fully occupied site. Bonding distances to framework oxygens lie between 3.00(3) and 3.42(3) A.˚ Cs2/K2 is located at the intersection of the main channel with the interconnection, bonding to ligands of the 7-membered ring polyhedra. The H2O site Ow1 is located in the main channel, in the middle of Cs1 and two ◦ Cs2/K2 sites. The total H2O content released between 70 and 280 C was 1.29 H2O pfu according to thermo-gravimetric analysis. This indicates that the H2O released between room temperature and 70 ◦C is either highly disordered structural water, hydrated phases on the particle surfaces, or released from X-ray amorphous hydrous phases. Remaining electron densities in difference Fourier plots were highly disordered, and the difference curve of the powder XRD pattern shows a good fit of the experimental data (fig. 5.7). The diffraction pattern of the 400 ◦C data set showed noticeably less intensity and broader reflections, and was thus only refined to an RBragg value of 8.38%. Polyhedra, mainly octahedra, became strongly distorted, but the AM-2 framework topology remained intact. The two Cs sites retained their positions in the 8-membered ring and at the junction of the main channel with the crosslink, but no H2O was found anymore. While we could clearly distinguish between a fully (Cs1) and partially exchanged site (Cs2/K2) in the room-temperature structure, the electron densities of Cs1 and Cs2 in the high-temperature structure show that Cs/K ratios on both sites have been levelled out (tab. 5.11).

92 5.3. Results

A A

B 1 1 1 2 2 1 2 a

Figure 5.4.: The AM-2 framework structure has open channels running along c (A), which are interconnected by rings made of four tetrahedra and three octahedra (1), and six tetrahedra and two octahedra (2). This builds a second type of channels running along b (B). However, the aperture of the seven-membered ring (1) is approximately 4.5 A˚ in diameter, which is too narrow for some cation species to pass. Thus migration mainly takes place in the A-channel.

Sr-exchanged AM-2

The Sr-exchanged phase was also refined in the orthorhombic space group P212121. The unit cell volume is almost identical to the unexchanged phase, but smaller than the Cs- exchanged cell (tab. 5.3). Four channel sites were found and refined with scattering factors of Sr. The actual cation and H2O distribution is presumably rather complex and could not be resolved by powder XRD. The electron densities on Sr1–Sr3 correspond to 0.322(4)–0.345(8) Sr ions, and to 0.248(7) Sr ions on Sr4. Sr1 resides at the centre of the 8-membered ring with bonding distances between 3.06 and 3.53 A.˚ Although we have seen cations on that position in previous structures, the bonding distances are too long for Sr–O bonds (tab. 5.2). In fact this site is highly disordered and an atomic displacement parameter (ADP) could not be refined. ADPs for all channel occupants were refined together with the framework, which explains the high B values of framework atoms shown in Table 5.12. The long bonding distances may also indicate that the site is preferred by the remaining potassium ions. Similar to the K- and Cs-structures, the second cation Sr2 sits at the intersection of the main channel and the crosslink. Due to the shorter Sr–O bonding distance, it moved from the centre of the main channel towards the centre of the 7-membered ring and bonds to ligands of the

93 5. Structural characterisation of ion-exchanged AM-2 ring-building polyhedra (ﬁg. 5.20). The distance between Sr1 and Sr2 is 2.56(1) A,˚ which excludes simultaneous occupation of both sites with cations. In the centre of the main channel, two sites Sr3 and Sr4 were found. As they are only 1.42 A˚ apart, the sum of occupancies cannot exceed 1.0. Furthermore, distances <3.0 A˚ to Sr1 and Sr2 denote that these sites also interact, which leads to a complex set of rules for possible combinations of cations, H2O molecules, and vacancies on all exchangeable sites. Hence the values for occupancies given in Table 5.12 should be regarded as electron densities originating from various elements.

Na-exchanged AM-2 Due to Na-exchange the symmetry changed from orthorhombic to monoclinic. The same change was also observed by Bortun et al. (2000) for protonated AM-2 of composition

K1.26H0.74TiSi3O9 · 1.8H2O. Their structure served as a starting model for our refinement in space group P21/c. Note that the axis labels differ from the orthorhombic setup used in this study in order to obtain a standard setup with a unique b axis. The change in symmetry is caused rotation of polyhedra, but the topology of the framework remained intact and the microporous nature and channel configuration characteristic for AM-2 structures were preserved. An exchangeable site Na1 was found between the 7- and 8-membered rings. Electron density and bonding distances indicate that the site is fully occupied with Na. In contrast to above described structures of exchanged varieties, Na1 moved out of the 8-membered ring plane towards the 7-membered ring and bonds to framework oxygens of both rings. A second site Na2/K2 was refined to a composition of 0.84 Na and 0.16 K. Bond- ing distances are clearly longer than those of Na1. The site is located in the main channel and bonds to the 7-membered ring. It is, however, shifted towards one side of the ring and bonds to Ow1 (fig. 5.21). Two H2O sites Ow1 and Ow2 were refined, but due to strong disorder atomic displacement parameters had to be fixed to a reasonable value, which in turn pre- vented accurate refinement of occupancy factors. Despite the similar chemical composition and space group, Na-AM-2 is structurally different from kostylevite, another monoclinic polymorph of AM-2 (Ilyushin et al., 1982).

Mn-exchanged AM-2 The powder XRD pattern of Mn-exchanged AM-2 contained impurities which made it impossible to reach a fit of the same quality as obtained in previous examples (fig. 5.8). Since the structure of the contaminant was unknown, we could not consider it in profile matching mode. Instead we excluded regions with obvious overlap from the refinement. Four exchangeable sites were refined with scattering factors of Mn and O, coupled to a total occupancy of 1.0 per site (Mn1/Ow1, Mn2/Ow2, Mn3/Ow3), and as an H2O site with un- restricted occupancy (Ow4). As shown in Table 5.14, this lead to a sum of 1.04 Mn pfu, which is in agreement with the stoichiometric requirements. Mn1/Ow1 moved to the ring plane in the centre of the 7-membered ring—a position which cannot be occupied by larger cations due to the narrow aperture. Mn2/Ow2 is located in the main channel. It is displaced

94 5.3. Results

Table 5.2.: Cation–oxygen distances in 8-fold coordination used for structure examinations, taken from Shannon (1976). Distances corresponding to 5% bondvalence were considered as the longest acceptable bond lengths between channel occupants and framework oxygens.

Atom 1 Ionic radius [A]˚ Atom 2 Ionic radius [A]˚ Sum [A]˚ 5% bondvalence [A]˚ Mn2+ 1.10 O2− 1.28 2.38 2.90 Ca2+ 1.26 O2− 1.28 2.54 3.08 Na+ 1.32 O2− 1.28 2.60 2.91 Sr2+ 1.40 O2− 1.28 2.68 3.23 K+ 1.65 O2− 1.28 2.93 3.24 Cs+ 1.88 O2− 1.28 3.16 3.53 from the void cage of the intersection of the main channel with the crosslink channel, to the narrowest place in the main channel: A ring membered by four octahedra and four tetrahedra. High displacement parameters indicate a strong disorder on that position. Mn3/Ow3 sits close to Mn1/Ow1 in the middle of the 8-membered ring. Due to the high displacement parameters, we cannot clearly say whether it is an individual site or part of a strongly anisotropically blurred site including both Mn1/Ow1 and Mn3/Ow3. In the latter case, Ow4, which is located in the main channel at the intersection with the crosslink, may also be part of a sausage-like smeared site running across the crosslink channel (ﬁg. 5.22).

Cu-exchange Due to the acidic conditions during the Cu exchange, the AM-2 structure was completely dissolved. Instead two new phases were found: Clinoatacamite, a monoclinic copper chloride hydroxide of composition ClCu2(OH)3, and poorly crystalline anatase (TiO2). Cell parameters for both phases are given in Table 5.9.

5.3.4. Dehydration experiments

Cs-exchanged AM-2 The powder XRD patterns of the heating sequences of Cs-exchanged AM-2 are shown in figure 5.9. No changes were observed between room temperature and 150 ◦C. From 150 to 250 ◦C a change in peak intensity and width occurred, which coincides with the highest release of H2O according to TGA results (fig. 5.2). The pattern remained unchanged up to 400 ◦C, and reverted to initial relative intensities, but showing less sharp reflections after cooling down to room temperature. The cell dimensions remained almost constant during the entire procedure (fig. 5.5). The sudden change from 350 to 400 ◦C shown in figure 5.5 may be related to the fact that in all refinements up to 350 ◦C a second phase was refined separately, but could no longer be resolved in the 400 ◦C pattern (tab. 5.5). The broadening of reflections may be related to disintegration of crystallites caused by the discharge of ◦ H2O. The second heating sequence starting at 400 C shows that the dehydrated structure

95 5. Structural characterisation of ion-exchanged AM-2 is stable up to 600 ◦C, and completely breaks down to X-ray amorphous between 600 and 700 ◦C. Between 750 and 800 ◦C new phases crystallise and remain stable after cooling to room temperature (ﬁg. 5.15). The main phase is a caesium titanium silicate of composition

Cs2(TiSi6O15) that was described by Nyman et al. (2000). The second phase of composition

Cs(TiSi2O6.5) was published by Balmer et al. (1997) and occurred in smaller amounts.

Sr-exchanged AM-2 Major changes are evident in the diffraction pattern of Sr-AM-2 between 50 and 100 ◦C (ﬁg.

5.10). According to the TGA results one H2O molecule was expelled in that range, which obviously leads to major rearrangements among the channel occupants. At the same time the cell volume decreased by more than 5% (tab. 5.7). Further heating lead to a steady loss of crystallinity, and the material remained in its poorly crystalline state after cooling down to room temperature. It became completely amorphous at 600 ◦C, and between 700 and 750 ◦C new phases recrystallised, which could not be identiﬁed. A list of reﬂections of these new phases measured at room temperature is given in Table 5.32.

Na-exchanged AM-2 The monoclinic Na-exchanged structure showed the most pronounced decrease of unit cell dimensions. A steady increase of the monoclinic angle from 91.946(2)◦ at 25 ◦C, to 94.164(4)◦ at 400 ◦C, as well as a constant decrease of the monoclinic axis are observed. Be- tween 25 and 400 ◦C the cell volume decreases by almost 9%. Whereas the crystallinity is reduced upon heating to 500 ◦C, the structure completely breaks down between 500 and 600 ◦C and a new phase crystallises between 750 and 800 ◦C (fig. 5.11). A pattern with similar peak positions was published by Khainakov et al. (1999). It belongs to a material of composition Na2Ti8O13(SiO4)2 with unknown structure. We were able to index the reflections in the orthorhombic space group P222 with cell constants a = 7.7408(2), b = 7.6414(1), and c = 12.0275(3) A,˚ but we did not manage to identify the structure. Reflection positions are listed in Table 5.33 and the pattern is shown in figure 5.16.

Mn-exchanged AM-2 Dehydration of the Mn-exchanged phase lead to a constant decrease of the cell volume, mainly caused by a decrease of a. Between 150 and 200 ◦C the structure started to break down and became completely amorphous above 300 ◦C (ﬁg. 5.12). Hence cell parameters were only reﬁned up to 200 ◦C (tab. 5.6).

Cu phases Anatase was not affected by high temperature, but clinoatacamite changed its monoclinic angle from 99.6209(9)◦ at room temperature to 90.686(8)◦ at 250 ◦C, and became almost orthorhombic. Between 250 and 300 ◦C it recrystallised as tenorite (CuO) and remained stable up to 400 ◦C and after cooling down to room temperature (ﬁg. 5.13).

96 5.3. Results Cs-AM-2 Sr-AM-2 Mn-AM-2 Ca-AM-2 Na-AM-2 K-AM-2 Temp [°C] 0 100 200 300 400 0 100 200 300 400

980 960 940 920 900 880 860 840 820 800

1000

13.50 13.40 13.30 13.20 13.10 13.00 12.90 12.80 12.70 12.60 12.50 12.40

] [Å volume Cell b [Å] b 3 Temp [°C] 0 100 200 300 400 0 100 200 300 400 Cell parameters of all ion-exchanged AM-2 structures as a function of temperature. Note that for the Ca-exchanged phase

9.90 9.80 9.70 9.60 9.50 9.40 9.30 9.20 7.50 7.40 7.30 7.20 7.10 7.00 6.90 6.80 6.70 6.60 6.50 6.40

10.30 10.20 10.10 10.00 c [Å] c a [Å] a only room-temperature data is available. Figure 5.5.:

97 5. Structural characterisation of ion-exchanged AM-2 Y W U endvrals6 56 06 62 348 64 542 60 909 578 67 597 55 605 62 reflections Indep. variables Refined R [ Temperature R R [ volume Cell [ c [ b [ a group Space endcm.Cs comp. Refined V χ X β 2 wp p Bragg [ ]727()722()715()1.132 .742 7.1395(1) 7.0734(2) 12.8193(2) 7.1857(2) 7.2424(6) 7.2870(2) A] ]1.173 023()1.033 .731 001()10.0365(2) 10.0619(3) 7.0763(1) 10.0033(3) 10.2939(8) 10.2147(3) A] ]1.893 2821 288()984()1.454 12.8204(3) 12.8455(4) 9.8244(2) 12.8286(4) 12.872(1) 13.0819(3) A] ◦ ˚ ˚ ˚ % 001. 401. 0815.7 20.8 13.4 14.0 18.6 10.0 [%] 09 09.4()9 90 90 91.946(2) 90 90 90 ] % .81. 311. 8115.8 18.1 15.1 13.1 17.6 9.68 [%] % .183 .366 238.79 12.3 6.67 4.43 8.38 4.01 [%] A ˚ ◦ ]2 0 52 525 25 25 25 400 25 C] 3 7.44 5.()921()806()942()918.65(3) 914.23(5) 890.69(3) 922.13(5) 959.6(1) 973.74(4) ] sA- sA- rA- aA- nA- Ca-AM-2 Mn-AM-2 Na-AM-2 Sr-AM-2 Cs-AM-2 Cs-AM-2 − .4090067 .4090067 .4090.046079 0.108989 0.013693 0.046079 0.108989 0.013693 0.046079 0.108989 0.013693 0.046079 0.108989 0.013693 0.046079 0.108989 0.013693 0.046079 0.108989 0.013693 0.064479 .6 .9 .8 .90782.192 0.102452 0.798 0.102452 1.39 0.102452 0.102452 0.584 0.102452 0.496 0.102452 0.363 P 2 · 1.323 1 1.29H 2 1 2 K 1 0.718 2 al 5.3.: Table O TiSi 3 O 9 xeietladrfieetprmtr o o-xhne AM-2. ion-exchanged for parameters refinement and Experimental − 0.064479 Cs P 2 1.16 1 2 1 K 2 1 0.84 TiSi 3 O 9 − 0.064479 Sr P 2 · 0.976 1 2.272H 2 1 2 TiSi 1 3 2 O O 9 − 0.064479 Na P 2 · 1 1.824 1.9H / P c K 2 0.16 O TiSi 3 O 9 − 0.064479 Mn 2 · 1 3.03H 1.04 2 1 2 TiSi 1 2 O 3 O 9 − 0.064479 Ca P 2 · 1 1.207 2.25H 2 1 2 TiSi 1 2 O 3 O 9

98 5.4. Discussion

Figure 5.6.: The polyhedral models used to draw the crystal structures in this study are somewhat misleading as to the void areas in the channels. For the sake of clearness we reduced the ionic radii of channel occupants (left). The same section drawn with correct radii (Shannon, 1976) shows that the channel occupants ﬁll large parts of the voids and leave only little space for migration of cations and H2O molecules.

5.4. Discussion

Unexchanged AM-2

The chemical composition of our AM-2 powder prior to the ion exchange was K2TiSi3O9 ·

H2O. An AM-2 structure of the same composition was refined from single-crystal data and published by Zou and Dadachov (2000) (fig. 5.17). Two K sites K1 and K2 were found. K1 is located in the centre of the 8-membered ring, bonding to eight framework oxygens with distances between 2.77 and 3.21 A.˚ The second site K2 sits at the intersection of the main channel with the crosslink and bonds to framework oxygens of the 7-membered ring with distances between 2.78 and 3.19 A.˚ The H2O site O10 is located in the centre of an irregular tetrahedron made up by one K1 and three K2 sites. Distances between O10 and the nearest K sites lie between 2.61 and 2.78 A.˚ The structure adapts to dehydration by changing its symmetry to monoclinic (fig. 5.14) without losing crystallinity, and completely reverts to orthorhombic upon rehydration.

Cs-exchanged AM-2 Cs is the largest cation analysed in this study. Its ionic radius has certain consequences on the migration behaviour: i) The ion exchange process is slow because incorporated Cs atoms plug the channel system, ii) the migration capability is limited because the Cs ion is too vo- luminous to pass through the crosslink channels, and iii) fast heating rates may destroy the integrity of the crystal due to the large ions reducing the mobility of H2O molecules (ﬁg. 5.6). Despite these drawbacks we suppose that >95% Cs exchange can be reached, albeit slowly compared to other elements. We have no explanation for the high pH of the exchange solution at the time. Although it was neutral prior to the ion-exchange, the pH became 10 ± 1

99 5. Structural characterisation of ion-exchanged AM-2 after the reaction throughout the experiment. We suppose that unknown phases formed on the surface of AM-2 particles, or that X-ray amorphous phases were present in the synthesis product which could not be detected directly by our analysis methods. Although bond valence calculations of the 25 ◦C structure indicate underbonding of O3, O5, and massively of O6 (tab. 5.26), it is obvious that the presence of channel occupants leads to a much more balanced structure. Underbonding of O6 was also observed in other AM-2 structures, including the one derived from single-crystal X-ray data by Zou and Dadachov (2000). It is related to the spatial position of O6, allowing no direct bonds to channel occupants. It is assumed that the effect is further emphasised by the soft distance constraints we applied to Ti/Si–O bonds in polyhedra, which prevent decrease of the distance Ti–O6. At 400 ◦C (tab. 5.27), bond valences show similar deviations from ideal values. In both cases we cannot assess whether this is related to the distance constraints or to structural features. The peak broadening and loss of intensity observed in the diffraction patterns between 25 and 400 ◦C, interpreted as reduction of crystallite size, may either be related to a breakdown caused by excessive release of H2O, or by associated strain caused by rearrangements of the Cs ions. We conclude that the Cs selectivity of AM-2 is high and the exchanged structure is stable up to 150 ◦C. However, the exchange kinetics is relatively slow. Furthermore, dehydration above 150 ◦C may lead to a ﬁner grained or amorphous fraction.

Sr-exchanged AM-2 The Sr exchange is noticeably faster than the Cs exchange, and following the trend seen in figure 5.1, an exchange ratio >95% will be reached after about 500 hours. We were particularly interested in how a divalent cation would fit into the AM-2 channels. In structure refinements with monovalent extraframework cations we could clearly distinguish between cation and H2O sites. In the Sr-exchanged structure, on the other hand, cations and H2O molecules are distributed evenly onto several sites, leading to variable H2O/Sr ratio on all extraframework positions. Upon heating the structure gradually became X-ray amorphous, as there is obviously no way to withstand the stronger bonds of divalent cations to the framework without the buffering effect of H2O molecules. Bond valence calculations shown in Table 5.28 were calculated without taking into account the actual occupancy of the Sr sites and are thus too high, but nevertheless they show how the deficits of bond valences on ligands of the TiO6 octahedron are being compensated by the Sr sites. In conclusion the selectivity of AM-2 for Sr is high, with exchange rates superior to Cs, but with a thermal stability inferior to the Cs-exchanged structure, as Sr-exchanged AM-2 starts to collapse at 300 ◦C.

Na-, Mn, and Ca-exchanged AM-2 The Na, Mn, and Ca exchanged phases show that the AM-2 structure is also able to take up cations smaller than K. In all cases the exchange process was faster and more advanced than in the Cs and Sr experiments. Exchange ratios >95% were easily reached in all three

100 5.4. Discussion cases. A substancial difference was observed between Na and divalent cations, which is in agreement with the above mentioned observation: Na and H2O molecules take different positions in the cavities, whereas the divalent cations Mn and Ca share their positions with H2O molecules. Even though Mn and Ca are smaller than Na, only the latter changes the symmetry of the unit cell to monoclinic. The Na-exchanged phase starts to gradually collapse above 200 ◦C. While in the case of Cs we suggested that the breakdown is related to plugging of the channels and the therewith associated pressure of H2O at high temperatures, we assume that in the Na-exchanged structure it is caused by the stronger bonding of the Na sites to the framework. The Mn-exchanged phase, on the other hand, completely breaks down between 200 and 300 ◦C and hence supports our assumption that AM-2 with divalent channel occupants is only stable if a certain amount of H2O is present in the cavities. The AM-2 structure is capable of taking up cations of various sizes. We observed exchange ratios of 70 to almost 100% for ionic radii ranging from 1.10 A˚ (Mn) to 1.88 A˚ (Cs). The reaction kinetics of the exchange process was proportional to the ionic radius of the exchanging cation, in our experiments Mn < Ca < Sr < Cs. Furthermore we found several trends related to the valence of the cations. In structures loaded with monovalent cations such as K+, Na+, Cs+, and Rb+ (see chapter 4), we were mostly able to distinguish between H2O sites and cation sites in structure refinements. Certain sites were preferred by cations, whereas other sites were occupied by H2O molecules. Divalent cations such as Ca2+, Sr2+, and Mn2+, on the other hand, were distributed onto several sites shared with ◦ H2O molecules. K-bearing and Rb-exchanged AM-2 proved to be stable up to 400 C and easily revert to the initial structure after cooling down to room temperature, whereas both larger (Cs) and smaller (Na) monovalent cations caused a gradual decay of the structure, albeit for different reasons. The thermal stability of phases containing divalent cations seems to depend on the ionic potential (charge/radius) of the channel occupants. Mn-exchanged AM-2 loses its crystallinity around 250 ◦C, whereas the Sr-bearing variety shows weak reflection up to 400 ◦C in powder XRD patterns. With these simple rules of thumb and the easily determined parameters (charge and radius of the cation) we can roughly predict the exchange behaviour and thermal stability of AM-2 to a limited extent. For future research we suggest focusing on chemical variations of the framework, in particular substitutions for Ti, which have already been reported to influence the selectivity for certain cation species (Clearfield et al., 1998).

101 5. Structural characterisation of ion-exchanged AM-2

5.5. Figures

5.5.1. X-ray diffraction patterns

Cs-AM-2 (25 °C) Iobs Icalc Intensity [Counts]

10 20 30 40 50 60 70 80 90 100 Angle [°2θ]

Cs-AM-2 (400 °C) Iobs Icalc Intensity [Counts]

10 20 30 40 50 60 70 80 90 100 Angle [°2θ]

Figure 5.7.: Observed and calculated XRD patterns of Cs-exchanged AM-2 at room temperature (top) and 400 ◦C (bottom). The difference curves are drawn in the same intensity scale.

102 5.5. Figures obs calc obs calc I I I I ] ] 60 70 80 90 100 60 70 80 90 100 θ θ Angle [°2 Angle [°2 50 50 Ca-AM-2 (25 °C) Na-AM-2 (25 °C)

10 20 30 40 10 20 30 40

Intensity [Counts] Intensity [Counts] Intensity obs calc obs calc I I I I ] ] 60 70 80 90 100 60 70 80 90 100 θ θ Angle [°2 Angle [°2 50 50 Sr-AM-2 (25 °C) Mn-AM-2 (25 °C) Observed and calculated diffraction patterns of Sr-, Na-, Mn- and Ca-exchanged AM-2. Difference curves are drawn in the

10 20 30 40 10 20 30 40

Intensity [Counts] Intensity [Counts] Intensity same intensity scale. Figure 5.8.:

103 5. Structural characterisation of ion-exchanged AM-2

25 °C

800 °C

750 °C

Intensity [counts] 700 °C

600 °C

500 °C

400 °C

10 20 30 40 50 60 Angle [°2θ]

25 °C

400 °C

350 °C

300 °C

250 °C

200 °C Intensity [counts] 150 °C

100 °C

50 °C

25 °C

10 20 30 40 50 60 Angle [°2θ]

Figure 5.9.: XRPD patterns of the heating sequences from 25 to 400 ◦C (bottom), and from 400 to 800 ◦C (top) of Cs-exchanged AM-2. Stability fields are marked with bars on the right hand side. In the white field the AM-2 structure is stable, in the cross-hatched field it becomes X-ray amorphous, and in the black range newly crystallised phases are stable.

104 5.5. Figures

25 °C

800 °C

750 °C

700 °C Intensity [counts] 600 °C

500 °C

400 °C

10 20 30 40 50 60 Angle [°2θ]

25 °C

400 °C

350 °C

300 °C

250 °C

200 °C Intensity [counts] 150 °C

100 °C

50 °C

25 °C

10 20 30 40 50 60 Angle [°2θ]

Figure 5.10.: XRPD patterns of the heating sequences from 25 to 400 ◦C (bottom), and from 400 to 800 ◦C (top) of Sr-exchanged AM-2. Crystallinity gradually decreases (white bar on the right hand side), and becomes amorphous (cross-hatched bar) around 350–400 ◦C. Above 700 ◦C new phases crystallise (black bar) and remain stable at room temperature.

105 5. Structural characterisation of ion-exchanged AM-2

I · 0.1 25 °C

I · 0.5 800 °C

750 °C

700 °C Intensity [counts] 600 °C

500 °C

400 °C

10 20 30 40 50 60 Angle [°2θ]

25 °C

400 °C

350 °C

300 °C

250 °C

200 °C Intensity [counts] 150 °C

100 °C

50 °C

25 °C

10 20 30 40 50 60 Angle [°2θ]

Figure 5.11.: Na-exchanged AM-2 is stable up to 500 ◦C, but loses crystallinity gradually (white bar on the right). Between 600 and 750 ◦C it becomes amorphous (cross-hatched range), and forms new phases above 750 ◦C (black bar), which remain stable at room temperature.

106 5.5. Figures

25 °C

400 °C

350 °C

300 °C

250 °C

200 °C Intensity [counts] 150 °C

100 °C

50 °C

25 °C

10 20 30 40 50 60 Angle [°2θ]

Figure 5.12.: Powder XRD pattern of Mn-exchanged AM-2 between 20 and 400 ◦C. The material is stable up to 200 ◦C, and becomes amorphous above 250 ◦C (cross-hatched bar).

25 °C

400 °C

350 °C

300 °C

250 °C

200 °C Intensity [counts] 150 °C

100 °C

50 °C

25 °C

10 20 30 40 50 60 Angle [°2θ]

Figure 5.13.: During the Cu-exchange, the AM-2 structure is dissolved and clinoatacamite and anatase crystallise. Between 250 and 300 ◦C clinoatacamite disappears and tenorite is formed (black bar).

107 5. Structural characterisation of ion-exchanged AM-2

25 °C

400 °C

350 °C

300 °C

250 °C

200 °C Intensity [counts] 150 °C

100 °C

50 °C

25 °C

10 20 30 40 50 60 Angle [°2θ]

Figure 5.14.: The initial K-bearing AM-2 structure is orthorhombic from room temperature up to 250 ◦C. Between 250 and 300 ◦C the symmetry changes to monoclinic and the structure remains stable up to 400 ◦C (black bar on the right). Upon cooling to room temperature the symmetry completely reverts to orthorhombic.

108 5.5. Figures 004 Intensity [Counts]

* 002 221 400 -221 200 321 111 * *

022 * * * *

10 20 30 40 50 60 70 Angle [°2θ]

Figure 5.15.: Above 750 ◦C two new phases crystallise from Cs-exchanged AM-2. The major phase, with peaks labelled in roman font, is a caesium titanium silicate of composition Cs2(TiSi6O15) described by Nyman et al. (2000). The minor phase of composition Cs(TiSi2O6.5) is labelled in bold italic font and was described by Balmer et al. (1997). Peaks marked with asterisks belong to the corundum sample holder and are strongly shifted due to displacement from the focus circle. 030 020 ] 1/2 [Counts 040 401 1/2 113 022 Intensity 003 010 100 060 110 050 006 232 070 225 326 008

10 20 30 40 50 60 70 80 90 100 Angle [°2θ]

Figure 5.16.: A new phase recrystallises above 750 ◦C from Na-exchanged AM-2. The peaks were indexed in the orthorhombic space group P222 and correspond to√ a pattern published by Khainakov et al. (1999). Note that the unit of the y-axis is counts, which reveals more details of low intensity.

109 5. Structural characterisation of ion-exchanged AM-2

5.5.2. Structures

K2 H1 a H2 O10

Figure 5.17.: The initial K-bearing AM-2 structure published by Zou and Dadachov (2000) contains two distinct K sites and one H2O molecule per formula unit. In contrast to + this structure, our powder X-ray data was not sufﬁcient to locate H sites from H2O molecules. Note that the axis labels were changed to the setup used in this study.

Cs1 c Ow1

b Cs1

Cs2/K2 Cs2/K2 a a Ow1 Ow1

b c

Figure 5.18.: The structure of Cs-exchanged AM-2 at room temperature in different pro- jections, showing the Cs sites located at the centre of the 8-membered (middle) and 7- membered (right) rings.

110 5.5. Figures

a Cs2/K2

Cs1/K1 b

Figure 5.19.: The structure of Cs-exchanged AM-2 at 400 ◦C, projected along c.

Sr1

Sr2 Sr4 Sr3

Figure 5.20.: The structure of Sr-exchanged AM-2 at room temperature. Sr and H2O molecules are distributed on the sites Sr1–Sr4.

111 5. Structural characterisation of ion-exchanged AM-2

Na2 Ow1 Ow1

b Na2/K2 b Ow2

Na1 c a

Figure 5.21.: The monoclinic crystal structure of Na-exchanged AM-2 projected along a (left) and c (right). The right projection shows how the Na2 site moved out of the centre of the 7-membered ring closer to the framework. Note that the axis labels were changed to a standard setup for the monoclinic cell.

Mn3/Ow3

3 Ow4 a 1 4 a Mn2/Ow2 Mn1/Ow1 c b b

Figure 5.22.: The crystal structure of Mn-exchanged AM-2 at room temperature, projected along c (left). The right part shows a section containing the 7- and 8-membered rings and the extra-framework sites Mn1/Ow1, Mn3/Ow3, and Ow4, labelled only with the corresponding numbers. Mn1/Ow1 is located at the centre of the 7-membered ring and is small enough to pass through. High disorder indicates that the sites are poorly deﬁned along the line 3–1–4, leading to a sausage-like diffuse electron density.

112 5.6. Tables

5.6. Tables

5.6.1. Cell parameters

Table 5.4.: Unit cell parameters of unexchanged K-bearing AM-2.

Temp. [◦C] a [A]˚ b [A]˚ c [A]˚ α [◦] β [◦] γ [◦] Volume [A˚ 3]

25 9.9111(1) 12.9460(1) 7.13870(7) 90 90 90 915.96(2) 050 9.9134(2) 12.9440(2) 7.1385(1) 90 90 90 916.01(3) 100 9.9235(2) 12.9480(2) 7.1410(1) 90 90 90 917.54(3) 150 9.9343(2) 12.9527(2) 7.1445(1) 90 90 90 919.33(3) 200 9.9435(2) 12.9494(2) 7.1458(1) 90 90 90 920.11(3) 250 9.9193(2) 12.9223(2) 7.1322(1) 90 90 90 914.20(3) 300 9.9283(2) 12.8080(2) 7.0781(1) 90 93.029(1) 90 898.80(3) 350 9.9241(2) 12.8102(2) 7.0809(1) 90 92.995(1) 90 898.97(3) 400 9.9273(2) 12.8164(2) 7.08612(8) 90 92.927(1) 90 900.40(2) 25 9.9075(2) 12.9382(2) 7.1350(1) 90 90 90 914.61(3)

Table 5.5.: Unit cell dimensions of Cs-exchanged AM-2. Due to incomplete ion exchange a minor phase with a smaller unit cell was also found and could be reﬁned up to 350 ◦C.

Main phase Minor phase

Temp. [◦C] a [A]˚ b [A]˚ c [A]˚ a [A]˚ b [A]˚ c [A]˚

25 10.2147(3) 13.0819(3) 7.2870(2) 9.9422(6) 12.9639(6) 7.1421(4) 50 10.1870(3) 13.0802(3) 7.2713(2) 9.9366(5) 12.9521(3) 7.1402(3) 100 10.1908(3) 13.0741(3) 7.2617(2) 9.9438(5) 12.9566(4) 7.1393(3) 150 10.1945(3) 13.0479(3) 7.2529(2) 9.9467(5) 12.9580(5) 7.1520(3) 200 10.1944(5) 13.0133(7) 7.2302(4) 9.9594(6) 12.9526(7) 7.1588(4) 250 10.1964(9) 13.020(1) 7.2188(6) 9.948(1) 12.906(1) 7.1475(8) 300 10.200(1) 13.043(1) 7.2247(7) 9.940(1) 12.900(1) 7.1585(9) 350 10.198(1) 13.048(1) 7.2221(7) 9.932(1) 12.891(1) 7.1626(9) 400 10.2942(8) 12.874(1) 7.2436(6) – – –

Table 5.6.: Unit cell dimensions of Mn-exchanged AM-2.

Temp. [◦C] a [A]˚ b [A]˚ c [A]˚ 25 10.0619(3) 12.8455(4) 7.0734(2) 50 10.0105(3) 12.8290(4) 7.0446(2) 100 9.9480(3) 12.8173(4) 7.0146(2) 150 9.9037(4) 12.8292(4) 6.9959(2) 200 9.7657(7) 12.8490(9) 6.9530(4)

113 5. Structural characterisation of ion-exchanged AM-2

Table 5.7.: Unit cell dimensions of Sr-exchanged AM-2.

Main phase Minor phase Temp. [◦C] a [A]˚ b [A]˚ c [A]˚ a [A]˚ b [A]˚ c [A]˚ 25 10.0033(3) 12.8286(4) 7.1857(2) 10.1090(6) 13.7142(1) 7.1761(4) 50 10.0034(3) 12.8222(4) 7.1790(2) 10.167(3) 13.685(4) 7.226(2) 100 9.9132(4) 12.7062(3) 6.9216(2) 10.031(1) 13.773(2) 7.2049(7) 150 9.8511(6) 12.6835(6) 6.9173(3) 10.013(1) 13.733(1) 7.1820(6) 200 9.8507(9) 12.7000(8) 6.9135(4) – – – 250 9.865(1) 12.7097(9) 6.9367(5) – – – 300 9.910(1) 12.744(1) 6.9672(6) – – – 350 9.923(2) 12.744(2) 6.9733(8) – – – 400 9.915(2) 12.717(3) 7.006(1) – – –

Table 5.8.: Unit cell dimensions of Na-exchanged AM-2.

Temp. [◦C] a [A]˚ b [A]˚ c [A]˚ α [◦] β [◦] γ [◦] Volume [A˚ 3] 25 7.0763(1) 9.8244(2) 12.8193(2) 90 91.946(2) 90 890.69(3) 50 7.0696(2) 9.8195(3) 12.8150(2) 90 91.960(2) 90 889.09(4) 100 7.0593(2) 9.7300(3) 12.8010(3) 90 92.290(2) 90 878.56(4) 150 7.0494(2) 9.6045(3) 12.8078(3) 90 92.785(3) 90 866.15(4) 200 7.0458(2) 9.5776(3) 12.8135(3) 90 92.766(2) 90 863.68(4) 250 7.0290(2) 9.4655(4) 12.8024(4) 90 92.906(3) 90 850.68(5) 300 7.0062(3) 9.2513(5) 12.7437(5) 90 93.696(3) 90 824.28(6) 350 7.0015(3) 9.1914(4) 12.7294(4) 90 93.788(3) 90 817.38(5) 400 7.0113(3) 9.1816(4) 12.7467(5) 90 94.164(4) 90 818.40(6)

Table 5.9.: Phases and unit cell dimensions of the clinoatacamite, anatase, and tenorite phases that formed during the Cu-exchange of AM-2 and the thermal treatment. The last data set was measured at 25 ◦C after the heating sequence.

Temp. [◦C] Phase a [A]˚ b [A]˚ c [A]˚ β [◦] 25 Clinoatacamite 6.16924(7) 6.82003(8) 9.1209(1) 99.6209(9) Anatase 3.7976(2) 3.7976(2) 9.505(1) 90 250 Clinoatacamite 5.4770(5) 6.8481(2) 9.1145(4) 90.686(8) Anatase 3.7934(4) 3.7934(4) 9.470(2) 90 400 Tenorite 4.7064(1) 3.44189(9) 5.1541(1) 99.0440(9) Anatase 3.8042(2) 3.8042(2) 9.497(1) 90 25 Tenorite 4.6903(2) 3.4278(2) 5.1340(3) 99.436(1) Anatase 3.7957(4) 3.7957(4) 9.461(2) 90

114 5.6. Tables

5.6.2. Atomic coordinates, site occupancies, and Beq values

Table 5.10.: Atomic coordinates, site occupancies and displacement parameters for fully hydrated Cs-exchanged AM-2 at 25 ◦C.

Atom x/a y/b z/c Occupancy Beq Ti 0.0601(7) 0.2887(4) 0.745(1) 1 2.02(3) Si1 0.042(1) 0.5460(6) 0.759(2) 1 2.02(3) Si2 0.338(1) 0.3229(9) 0.963(2) 1 2.02(3) Si3 0.336(1) 0.3254(8) 0.543(2) 1 2.02(3) O1 0.183(1) 0.314(2) 0.949(3) 1 2.02(3) O2 0.388(2) 0.346(1) 0.754(2) 1 2.02(3) O3 0.208(2) 0.254(2) 0.581(3) 1 2.02(3) O4 0.117(3) 0.573(2) 0.565(3) 1 2.02(3) O5 0.935(3) 0.261(2) 0.539(4) 1 2.02(3) O6 0.923(2) 0.6334(9) 0.803(4) 1 2.02(3) O7 0.161(2) 0.572(2) 0.909(3) 1 2.02(3) O8 0.923(3) 0.263(2) 0.933(4) 1 2.02(3) O9 0.999(2) 0.4293(8) 0.755(9) 1 2.02(3) Cs1 0.0507(3) 0.4086(2) 0.249(1) 1.041(6) 3 Cs2 0.2037(6) 0.1090(5) 0.216(1) 0.282(5) 3 K2 0.2037(6) 0.1090(5) 0.216(1) 0.718(5) 3 Ow1 0.299(2) 0.048(2) 0.915(3) 1.29(3) 3

Table 5.11.: Atomic coordinates, site occupancies and displacement parameters for fully hydrated Cs-exchanged AM-2 at 400 ◦C.

Atom x/a y/b z/c Occupancy Beq Ti 0.0595(7) 0.2795(6) 0.744(1) 1 3.2(2) Si1 0.454(1) 0.4578(8) 0.264(2) 1 3.2(2) Si2 0.367(1) 0.3141(9) 0.968(1) 1 3.2(2) Si3 0.335(1) 0.3277(9) 0.562(2) 1 3.2(2) O1 0.222(2) 0.297(4) 0.887(4) 1 3.2(2) O2 0.413(2) 0.344(2) 0.757(2) 1 3.2(2) O3 0.193(2) 0.275(4) 0.553(4) 1 3.2(2) O4 0.147(3) 0.586(2) 0.610(4) 1 3.2(2) O5 0.947(4) 0.245(3) 0.534(4) 1 3.2(2) O6 0.928(2) 0.625(1) 0.790(7) 1 3.2(2) O7 0.142(4) 0.563(2) 0.943(5) 1 3.2(2) O8 0.898(2) 0.279(2) 0.889(4) 1 3.2(2) O9 0.990(2) 0.4229(8) 0.79(1) 1 3.2(2) Cs1 0.0634(6) 0.4018(6) 0.250(3) 0.47(1) 1.6(3) K1 0.0634(6) 0.4018(6) 0.250(3) 0.53(1) 1.6(3) Cs2 0.2187(6) 0.1011(5) 0.225(1) 0.69(1) 1.6(3) K2 0.2187(6) 0.1011(5) 0.225(1) 0.31(1) 1.6(3)

115 5. Structural characterisation of ion-exchanged AM-2

Table 5.12.: Atomic coordinates, site occupancies and displacement parameters for fully hydrated Sr-exchanged AM-2 at 25 ◦C.

Atom x/a y/b z/c Occupancy Beq Ti 0.0345(4) 0.2909(3) 0.7386(7) 1 3.1(1) Si1 0.0515(5) 0.5553(4) 0.7519(8) 1 3.1(1) Si2 0.3386(5) 0.3141(5) 0.9376(6) 1 3.1(1) Si3 0.3310(6) 0.3492(4) 0.5482(8) 1 3.1(1) O1 0.1788(8) 0.289(2) 0.923(2) 1 3.1(1) O2 0.4135(9) 0.339(1) 0.7420(8) 1 3.1(1) O3 0.196(1) 0.282(1) 0.585(3) 1 3.1(1) O4 0.140(2) 0.5832(9) 0.569(2) 1 3.1(1) O5 0.905(2) 0.2502(9) 0.547(3) 1 3.1(1) O6 0.9318(8) 0.6405(4) 0.749(5) 1 3.1(1) O7 0.131(2) 0.5412(6) 0.948(2) 1 3.1(1) O8 0.873(1) 0.259(2) 0.883(2) 1 3.1(1) O9 0.994(2) 0.4382(4) 0.777(4) 1 3.1(1) Sr1 0.453(1) 0.5976(8) 0.773(2) 0.322(4) 3.1(1) Sr2 0.1324(9) 0.2155(7) 0.237(2) 0.336(4) 3.1(1) Sr3 0.186(1) 0.936(1) 0.473(2) 0.345(8) 3.1(1) Sr4 0.243(2) 0.032(2) 0.417(3) 0.248(7) 3.1(1)

Table 5.13.: Atomic coordinates, site occupancies and displacement parameters for fully hydrated Na-exchanged AM-2 at 25 ◦C.

Atom x/a y/b z/c Occupancy Beq Ti1 0.2617(5) 0.7077(4) 0.2140(3) 1 0.94(2) Si1 −0.0185(8) 0.4409(6) 0.1729(4) 1 0.94(2) Si2 0.7104(8) 0.2933(6) 0.0479(4) 1 0.94(2) Si3 0.4006(8) 0.3872(5) 0.1711(4) 1 0.94(2) O1 0.241(2) 0.674(1) 0.3693(7) 1 0.94(2) O2 0.040(1) 0.595(1) 0.193(1) 1 0.94(2) O3 0.279(2) 0.755(1) 0.0728(7) 1 0.94(2) O4 0.526(1) 0.797(1) 0.2381(8) 1 0.94(2) O5 0.126(1) 0.872(1) 0.2296(7) 1 0.94(2) O6 0.415(2) 0.541(1) 0.1950(8) 1 0.94(2) O7 0.181(1) 0.353(1) 0.1516(9) 1 0.94(2) O8 0.484(1) 0.344(1) 0.0603(8) 1 0.94(2) O9 0.840(1) 0.425(1) 0.0762(8) 1 0.94(2) Na1 0.695(1) 0.6755(8) 0.1135(6) 0.984(8) 2 Na2 0.407(1) 0.4803(7) 0.5969(4) 0.84(1) 2 K2 0.407(1) 0.4803(7) 0.5969(4) 0.16(1) 2 Ow1 0.106(1) 0.0909(7) 0.1143(4) 0.84(1) 2 Ow2 0.133(2) 0.379(1) 0.4402(8) 1.06(1) 2

116 5.6. Tables

Table 5.14.: Atomic coordinates, site occupancies and displacement parameters for fully hydrated Mn-exchanged AM-2 at 25 ◦C.

Atom x/a y/b z/c Occupancy Beq Ti 0.0443(7) 0.2884(5) 0.751(2) 1 2.80(7) Si1 0.045(1) 0.5505(7) 0.767(2) 1 2.80(7) Si2 0.339(1) 0.3239(9) 0.973(2) 1 2.80(7) Si3 0.318(1) 0.332(1) 0.545(2) 1 2.80(7) O1 0.177(2) 0.309(2) 0.956(3) 1 2.80(7) O2 0.355(2) 0.364(1) 0.756(2) 1 2.80(7) O3 0.169(2) 0.315(2) 0.543(3) 1 2.80(7) O4 0.116(3) 0.560(2) 0.560(3) 1 2.80(7) O5 0.891(2) 0.240(2) 0.609(4) 1 2.80(7) O6 0.919(2) 0.625(1) 0.777(6) 1 2.80(7) O7 0.100(2) 0.566(2) 0.991(3) 1 2.80(7) O8 0.894(2) 0.289(2) 0.920(3) 1 2.80(7) O9 0.971(2) 0.433(1) 0.745(6) 1 2.80(7) Mn1 0.069(1) 0.274(1) 0.256(5) 0.25(1) 3.7(6) Ow1 0.069(1) 0.274(1) 0.256(5) 0.75(1) 3.7(6) Mn2 0.214(2) 0.921(1) 0.435(2) 0.48(2) 7(1) Ow2 0.214(2) 0.921(1) 0.435(2) 0.52(2) 7(1) Mn3 0.521(2) 0.079(1) 0.745(4) 0.31(1) 3.7(6) Ow3 0.521(2) 0.079(1) 0.745(4) 0.69(1) 3.7(6) Ow4 0.168(2) 0.132(1) 0.295(4) 1.07(3) 2.80(7)

Table 5.15.: Atomic coordinates, site occupancies and displacement parameters for fully hydrated Ca-exchanged AM-2 at 25 ◦C.

Atom x/a y/b z/c Occupancy Beq Ti 0.0400(4) 0.2850(3) 0.7443(8) 1 2.68(5) Si1 0.0406(6) 0.5488(4) 0.757(1) 1 2.68(5) Si2 0.3276(6) 0.3212(5) 0.9725(9) 1 2.68(5) Si3 0.3340(6) 0.3273(5) 0.5484(9) 1 2.68(5) O1 0.1727(9) 0.305(2) 0.950(2) 1 2.68(5) O2 0.370(1) 0.355(1) 0.764(1) 1 2.68(5) O3 0.1793(8) 0.307(2) 0.560(2) 1 2.68(5) O4 0.097(1) 0.5701(9) 0.541(1) 1 2.68(5) O5 0.927(1) 0.2682(9) 0.513(2) 1 2.68(5) O6 0.9202(9) 0.6309(6) 0.749(4) 1 2.68(5) O7 0.150(1) 0.5657(9) 0.927(2) 1 2.68(5) O8 0.883(1) 0.2774(9) 0.905(2) 1 2.68(5) O9 0.982(1) 0.4310(5) 0.749(4) 1 2.68(5) Ca1 0.0301(8) 0.4190(6) 0.247(3) 0.653(4) 2.68(5) Ca2 0.721(1) 0.5735(8) 0.551(1) 0.554(5) 2.68(5) Ow1 0.075(1) 0.2515(8) 0.255(3) 1.16(1) 2.68(5) Ow2 0.184(1) 0.101(1) 0.320(2) 1.09(2) 2.68(5)

117 5. Structural characterisation of ion-exchanged AM-2

5.6.3. Bond angles and distances

Table 5.16.: Bond angles and distances in framework polyhedra of Cs-exchanged AM-2 at 25 ◦C.

Atom 1–Atom 2–Atom 3 Angle [◦] d12 [A]˚ d23 [A]˚ d13 [A]˚ O1–Ti–O3 90(1) 1.97(2) 1.98(2) 2.81(3) O1–Ti–O6 104(2) 1.97(2) 2.07(1) 3.18(3) O1–Ti–O8 88(2) 1.97(2) 1.99(3) 2.74(3) O1–Ti–O9 91(2) 1.97(2) 1.94(1) 2.79(4) O3–Ti–O5 90(2) 1.98(2) 2.00(3) 2.81(4) O3–Ti–O6 67(1) 1.98(2) 2.07(1) 2.24(3) O3–Ti–O9 119(2) 1.98(2) 1.94(1) 3.38(3) O5–Ti–O6 75(1) 2.00(3) 2.07(1) 2.49(3) O5–Ti–O8 92(2) 2.00(3) 1.99(3) 2.87(4) O5–Ti–O9 90(2) 2.00(3) 1.94(1) 2.79(5) O6–Ti–O8 91(2) 2.07(1) 1.99(3) 2.88(4) O8–Ti–O9 85(2) 1.99(3) 1.94(1) 2.65(4) O4–Si1–O6 111(2) 1.64(3) 1.70(2) 2.75(4) O4–Si1–O7 100(2) 1.64(3) 1.67(2) 2.55(3) O4–Si1–O9 109(3) 1.64(3) 1.59(1) 2.63(4) O6–Si1–O7 105(2) 1.70(2) 1.67(2) 2.67(3) O6–Si1–O9 117(1) 1.70(2) 1.59(1) 2.80(2) O7–Si1–O9 114(2) 1.67(2) 1.59(1) 2.73(4) O1–Si2–O2 105(2) 1.59(2) 1.63(2) 2.57(2) O1–Si2–O4 112(2) 1.59(2) 1.61(2) 2.66(3) O1–Si2–O8 121(2) 1.59(2) 1.61(3) 2.79(3) O2–Si2–O4 101(2) 1.63(2) 1.61(2) 2.50(3) O2–Si2–O8 114(2) 1.63(2) 1.61(3) 2.71(3) O4–Si2–O8 103(2) 1.61(2) 1.61(3) 2.52(3) O2–Si3–O3 101(2) 1.65(2) 1.63(2) 2.53(3) O2–Si3–O5 105(2) 1.65(2) 1.63(3) 2.59(3) O2–Si3–O7 114(2) 1.65(2) 1.66(3) 2.78(3) O3–Si3–O5 99(2) 1.63(2) 1.63(3) 2.48(4) O3–Si3–O7 125(2) 1.63(2) 1.66(3) 2.92(3) O5–Si3–O7 110(2) 1.63(3) 1.66(3) 2.69(3)

118 5.6. Tables

Table 5.17.: Bond angles and distances in framework polyhedra of Sr-exchanged AM-2 at 25 ◦C.

Atom 1–Atom 2–Atom 3 Angle [◦] d12 [A]˚ d23 [A]˚ d13 [A]˚ O1–Ti–O3 77(1) 1.96(1) 1.96(2) 2.44(3) O1–Ti–O5 164(1) 1.96(1) 1.96(2) 3.88(2) O1–Ti–O6 80(1) 1.96(1) 1.961(7) 2.53(3) O1–Ti–O8 104.2(8) 1.96(1) 1.96(1) 3.09(1) O1–Ti–O9 94(1) 1.96(1) 1.951(8) 2.85(2) O3–Ti–O5 98(1) 1.96(2) 1.96(2) 2.95(2) O3–Ti–O6 80(1) 1.96(2) 1.961(7) 2.52(2) O3–Ti–O8 165(1) 1.96(2) 1.96(1) 3.88(2) O3–Ti–O9 108(1) 1.96(2) 1.951(8) 3.16(2) O5–Ti–O6 83(1) 1.96(2) 1.961(7) 2.61(3) O5–Ti–O8 77(1) 1.96(2) 1.96(1) 2.44(3) O5–Ti–O9 103(1) 1.96(2) 1.951(8) 3.06(2) O6–Ti–O8 85(1) 1.961(7) 1.96(1) 2.65(2) O6–Ti–O9 169.1(6) 1.961(7) 1.951(8) 3.894(8) O8–Ti–O9 88(1) 1.96(1) 1.951(8) 2.71(2) O4–Si1–O6 104(2) 1.62(2) 1.621(9) 2.56(3) O4–Si1–O7 117(1) 1.62(2) 1.63(2) 2.78(2) O4–Si1–O9 119(2) 1.62(2) 1.617(9) 2.79(2) O6–Si1–O7 116(2) 1.621(9) 1.63(2) 2.76(3) O6–Si1–O9 111.6(7) 1.621(9) 1.617(9) 2.678(9) O7–Si1–O9 88(1) 1.63(2) 1.617(9) 2.26(2) O1–Si2–O2 116(1) 1.63(1) 1.625(9) 2.76(1) O1–Si2–O4 109(1) 1.63(1) 1.64(1) 2.66(2) O1–Si2–O8 98(1) 1.63(1) 1.63(2) 2.47(2) O2–Si2–O4 106(1) 1.625(9) 1.64(1) 2.61(2) O2–Si2–O8 134(1) 1.625(9) 1.63(2) 3.00(2) O4–Si2–O8 89(1) 1.64(1) 1.63(2) 2.28(2) O2–Si3–O3 104(1) 1.624(9) 1.62(2) 2.56(2) O2–Si3–O5 94(1) 1.624(9) 1.62(2) 2.37(2) O2–Si3–O7 109(1) 1.624(9) 1.62(1) 2.65(2) O3–Si3–O5 92(1) 1.62(2) 1.62(2) 2.33(2) O3–Si3–O7 136(1) 1.62(2) 1.62(1) 3.01(2) O5–Si3–O7 113(1) 1.62(2) 1.62(1) 2.71(1)

119 5. Structural characterisation of ion-exchanged AM-2

Table 5.18.: Bond angles and distances in framework polyhedra of Na-exchanged AM-2 at 25 ◦C.

Atom 1–Atom 2–Atom 3 Angle [◦] d12 [A]˚ d23 [A]˚ d13 [A]˚ O1–Ti1–O2 87.8(7) 2.03(1) 1.93(1) 2.75(1) O1–Ti1–O4 90.8(7) 2.03(1) 2.08(1) 2.92(1) O1–Ti1–O5 88.8(7) 2.03(1) 1.89(1) 2.74(1) O1–Ti1–O6 92.6(7) 2.03(1) 1.98(1) 2.90(1) O2–Ti1–O3 94.8(8) 1.93(1) 1.88(1) 2.80(1) O2–Ti1–O5 95.1(8) 1.93(1) 1.89(1) 2.82(1) O2–Ti1–O6 87.6(7) 1.93(1) 1.98(1) 2.71(1) O3–Ti1–O4 87.4(7) 1.88(1) 2.08(1) 2.73(1) O3–Ti1–O5 86.7(7) 1.88(1) 1.89(1) 2.58(1) O3–Ti1–O6 91.8(8) 1.88(1) 1.98(1) 2.77(1) O4–Ti1–O5 94.8(8) 2.08(1) 1.89(1) 2.92(1) O4–Ti1–O6 82.6(7) 2.08(1) 1.98(1) 2.68(1) O2–Si1–O5 114(1) 1.59(1) 1.63(1) 2.70(1) O2–Si1–O7 107(1) 1.59(1) 1.68(1) 2.64(1) O2–Si1–O9 112(1) 1.59(1) 1.57(1) 2.62(1) O5–Si1–O7 109(1) 1.63(1) 1.68(1) 2.70(1) O5–Si1–O9 105(1) 1.63(1) 1.57(1) 2.55(1) O7–Si1–O9 110(1) 1.68(1) 1.57(1) 2.66(1) O1–Si2–O3 113.8(9) 1.61(1) 1.62(1) 2.70(1) O1–Si2–O8 110(1) 1.61(1) 1.69(1) 2.70(1) O1–Si2–O9 109(1) 1.61(1) 1.62(1) 2.63(1) O3–Si2–O8 105(1) 1.62(1) 1.69(1) 2.62(1) O3–Si2–O9 114(1) 1.62(1) 1.62(1) 2.72(1) O8–Si2–O9 105.8(9) 1.69(1) 1.62(1) 2.65(1) O4–Si3–O6 113(1) 1.54(1) 1.55(1) 2.58(1) O4–Si3–O7 108(1) 1.54(1) 1.60(1) 2.53(1) O4–Si3–O8 113(1) 1.54(1) 1.61(1) 2.63(1) O6–Si3–O7 107(1) 1.55(1) 1.60(1) 2.53(1) O6–Si3–O8 114(1) 1.55(1) 1.61(1) 2.65(1) O7–Si3–O8 101(1) 1.60(1) 1.61(1) 2.48(1)

120 5.6. Tables

Table 5.19.: Bond angles and distances in framework polyhedra of Mn-exchanged AM-2 at 25 ◦C.

Atom 1–Atom 2–Atom 3 Angle [◦] d12 [A]˚ d23 [A]˚ d13 [A]˚ O1–Ti–O3 95(2) 1.99(2) 1.97(2) 2.92(3) O1–Ti–O6 95(2) 1.99(2) 2.13(2) 3.03(4) O1–Ti–O8 94(1) 1.99(2) 1.93(2) 2.87(3) O1–Ti–O9 98(2) 1.99(2) 2.00(2) 3.01(3) O3–Ti–O5 100(2) 1.97(2) 1.94(2) 2.99(3) O3–Ti–O6 89(1) 1.97(2) 2.13(2) 2.89(3) O3–Ti–O9 93(2) 1.97(2) 2.00(2) 2.88(3) O5–Ti–O6 77(1) 1.94(2) 2.13(2) 2.55(3) O5–Ti–O8 72(2) 1.94(2) 1.93(2) 2.29(3) O5–Ti–O9 90(1) 1.94(2) 2.00(2) 2.77(3) O6–Ti–O8 102(2) 2.13(2) 1.93(2) 3.15(3) O8–Ti–O9 74(1) 1.93(2) 2.00(2) 2.35(3) O4–Si1–O6 110(3) 1.63(3) 1.60(2) 2.65(4) O4–Si1–O7 134(2) 1.63(3) 1.68(2) 3.05(3) O4–Si1–O9 100(2) 1.63(3) 1.69(2) 2.55(4) O6–Si1–O7 98(2) 1.60(2) 1.68(2) 2.48(4) O6–Si1–O9 101(1) 1.60(2) 1.69(2) 2.54(2) O7–Si1–O9 110(2) 1.68(2) 1.69(2) 2.76(4) O1–Si2–O2 94(2) 1.64(2) 1.63(2) 2.39(3) O1–Si2–O4 114(2) 1.64(2) 1.68(2) 2.78(3) O1–Si2–O8 105(2) 1.64(2) 1.73(2) 2.67(3) O2–Si2–O4 92(2) 1.63(2) 1.68(2) 2.38(3) O2–Si2–O8 130(2) 1.63(2) 1.73(2) 3.04(3) O4–Si2–O8 120(2) 1.68(2) 1.73(2) 2.95(3) O2–Si3–O3 106(2) 1.59(2) 1.51(2) 2.48(3) O2–Si3–O5 133(2) 1.59(2) 1.61(3) 2.93(3) O2–Si3–O7 84(2) 1.59(2) 1.60(2) 2.13(3) O3–Si3–O5 111(2) 1.51(2) 1.61(3) 2.58(3) O3–Si3–O7 129(2) 1.51(2) 1.60(2) 2.81(3) O5–Si3–O7 94(2) 1.61(3) 1.60(2) 2.35(3)

121 5. Structural characterisation of ion-exchanged AM-2

Table 5.20.: Bond angles and distances in framework polyhedra of Mn-exchanged AM-2 at 25 ◦C.

Atom 1–Atom 2–Atom 3 Angle [◦] d12 [A]˚ d23 [A]˚ d13 [A]˚ O1–Ti–O3 89.9(9) 2.00(1) 1.94(1) 2.79(2) O1–Ti–O6 89(1) 2.00(1) 2.016(9) 2.81(2) O1–Ti–O8 96.4(8) 2.00(1) 1.96(1) 2.95(1) O1–Ti–O9 94(1) 2.00(1) 1.960(8) 2.89(2) O3–Ti–O5 82.3(8) 1.94(1) 2.02(1) 2.60(1) O3–Ti–O6 91(1) 1.94(1) 2.016(9) 2.82(2) O3–Ti–O9 95(1) 1.94(1) 1.960(8) 2.87(2) O5–Ti–O6 92(1) 2.02(1) 2.016(9) 2.89(2) O5–Ti–O8 91.2(9) 2.02(1) 1.96(1) 2.84(2) O5–Ti–O9 87(1) 2.02(1) 1.960(8) 2.74(2) O6–Ti–O8 95.6(9) 2.016(9) 1.96(1) 2.94(2) O8–Ti–O9 78.5(8) 1.96(1) 1.960(8) 2.48(2) O4–Si1–O6 97(1) 1.66(1) 1.60(1) 2.44(2) O4–Si1–O7 116(1) 1.66(1) 1.65(1) 2.81(2) O4–Si1–O9 104(1) 1.66(1) 1.621(9) 2.58(2) O6–Si1–O7 116(1) 1.60(1) 1.65(1) 2.76(2) O6–Si1–O9 109.8(8) 1.60(1) 1.621(9) 2.64(1) O7–Si1–O9 113(1) 1.65(1) 1.621(9) 2.73(2) O1–Si2–O2 102(1) 1.58(1) 1.61(1) 2.47(1) O1–Si2–O4 126(1) 1.58(1) 1.66(1) 2.89(2) O1–Si2–O8 107(1) 1.58(1) 1.63(1) 2.58(2) O2–Si2–O4 85.8(9) 1.61(1) 1.66(1) 2.23(1) O2–Si2–O8 128(1) 1.61(1) 1.63(1) 2.91(2) O4–Si2–O8 110(1) 1.66(1) 1.63(1) 2.69(2) O2–Si3–O3 102(1) 1.62(1) 1.58(1) 2.48(1) O2–Si3–O5 107(1) 1.62(1) 1.60(1) 2.59(1) O2–Si3–O7 107(1) 1.62(1) 1.63(1) 2.62(1) O3–Si3–O5 118(1) 1.58(1) 1.60(1) 2.71(2) O3–Si3–O7 106(1) 1.58(1) 1.63(1) 2.55(2) O5–Si3–O7 116(1) 1.60(1) 1.63(1) 2.74(2)

122 5.6. Tables

Table 5.21.: Cs–O distances ≤ 3.53 A˚ in the hydrated orthorhombic Cs-exchanged AM-2 structure.

Atom 1–Atom 2 Distance [A]˚ Atom 1–Atom 2 Distance [A]˚ Cs1–O1 2.85(2) Cs2/K2–O1 3.32(3) Cs1–O2 3.27(2) Cs2/K2–O2 3.29(2) Cs1–O4 3.22(2) Cs2/K2–O3 3.27(2) Cs1–O5 3.10(3) Cs2/K2–O5 3.42(3) Cs1–O7 3.46(2) Cs2/K2–O8 3.00(3) Cs1–O7 3.18(2) Cs2/K2–O9 3.07(2) Cs1–O8 3.26(3) Cs2/K2–Ow1 2.53(2) Cs1–Ow1 2.89(2) Cs2/K2–Ow1 2.52(2) Ow1–O6 3.39(3) Ow1–O6 2.98(3) Ow1–O8 2.99(3) Ow1–O9 3.17(5) Ow1–Cs1 2.89(2) Ow1–Cs2/K2 2.53(2) Ow1–Cs2/K2 2.52(2)

Table 5.22.: Sr–O distances ≤ 3.60 A˚ in the hydrated Sr-exchanged AM-2 structure.

Atom 1–Atom 2 Distance [A]˚ Atom 1–Atom 2 Distance [A]˚ Sr1–O1 3.19(2) Sr2–O1 2.49(2) Sr1–O2 3.34(2) Sr2–O2 2.31(1) Sr1–O2 3.38(2) Sr2–O3 2.72(2) Sr1–O3 3.10(2) Sr2–O5 3.21(2) Sr1–O4 3.46(2) Sr2–O5 3.17(2) Sr1–O4 3.28(2) Sr2–O8 2.58(2) Sr1–O5 3.12(2) Sr1–O7 3.53(2) Sr1–O7 3.06(2) Sr3–O1 3.22(3) Sr4–O3 3.46(3) Sr3–O6 3.10(2) Sr4–O5 3.24(3) Sr3–O6 3.50(3) Sr4–O6 3.15(3) Sr3–O7 3.49(2) Sr4–O6 3.28(3) Sr3–O8 2.57(3) Sr4–O9 2.90(3) Sr3–O9 2.54(3) Sr4–O9 3.45(3)

123 5. Structural characterisation of ion-exchanged AM-2

Table 5.23.: Bonding distances around channel occupants in the hydrated Na-exchanged AM-2 structure.

Atom 1–Atom 2 Distance [A]˚ Atom 1–Atom 2 Distance [A]˚ Na1–O2 2.73(1) Na2/K2–O1 2.93(1) Na1–O3 3.07(1) Na2/K2–O3 2.77(1) Na1–O4 2.35(1) Na2/K2–O4 2.94(1) Na1–O6 2.62(1) Na2/K2–O5 3.03(1) Na1–O8 2.53(1) Na2/K2–O6 2.92(1) Na1–O9 2.71(1) Na2/K2–Ow1 2.26(1) Na1–Ow2 2.45(1) Na2/K2–Ow2 2.92(1) Ow1–O1 2.61(1) Ow2–O1 3.15(2) Ow1–O2 2.71(1) Ow2–O2 3.46(2) Ow1–O3 3.57(1) Ow2–O3 3.16(2) Ow1–O5 2.61(1) Ow2–O4 3.47(2) Ow1–O7 2.67(1) Ow2–O5 2.80(1) Ow1–Ow2 2.76(1) Ow2–O7 3.55(2) Ow1–Ow2 2.27(1) Ow2–Ow2 3.44(2)

Table 5.24.: Mn–O distances ≤ 2.90 A˚ in the hydrated orthorhombic Mn-exchanged AM-2 structure.

Atom 1–Atom 2 Distance [A]˚ Atom 1–Atom 2 Distance [A]˚ Mn1–O1 2.43(4) Mn2–O6 2.62(3) Mn1–O2 2.79(2) Mn2–O8 2.26(3) Mn1–O3 2.33(4) Mn2–Ow3 2.72(3) Mn1–Ow3 1.95(2) Mn2–Ow4 2.89(3) Mn1–Ow4 2.10(2) Mn3–O1 3.00(3) Ow4–O1 3.30(3) Mn3–O2 3.03(2) Ow4–O2 3.17(2) Mn3–O3 2.87(3) Ow4–O3 2.93(3) Mn3–O4 2.95(3) Ow4–O5 2.86(3) Mn3–O7 2.75(3) Ow4–O6 3.15(5) Mn3–O8 3.19(3) Ow4–O7 3.20(3) Mn3–O9 3.51(5) Ow4–O8 2.91(3) Mn3–Ow1 1.95(2) Ow4–O9 3.17(3) Mn3–Ow2 2.72(3) Mn3–Ow4 3.33(2)

Table 5.25.: Ca–O distances ≤ 3.08 A˚ in the hydrated orthorhombic Ca-exchanged AM-2 structure.

Atom 1–Atom 2 Distance [A]˚ Atom 1–Atom 2 Distance [A]˚ Ca1–O1 2.94(2) Ca2–O6 2.56(2) Ca1–O2 3.07(1) Ca2–O8 2.41(2) Ca1–O3 3.05(2) Ca2–O9 2.97(2) Ca1–O4 2.94(2) Ca2–Ow2 2.45(2) Ca1–O5 2.90(2) Ca2–Ow2 2.84(2) Ca1–Ow1 2.19(1)

124 5.6. Tables

5.6.4. Bond valence calculations

Table 5.26.: Bond valence calculations for Cs-exchanged AM-2 at 25 ◦C. Cs–O bond lengths ≤ 3.53 A˚ were considered. According to the reﬁned occupancies, bond valences for Cs2 and K2 were multiplied by 0.28 and 0.72.

Ti Cs1 Cs2 K2 Si1 Si2 Si3 Sum lattice Sum total O1 0.658 0.310 0.024 0.029 1.114 1.772 2.136 O2 0.100 0.026 0.031 1.027 0.973 2.001 2.158 O3 0.640 0.048 0.028 0.033 1.027 1.668 1.777 O4 0.114 1.000 1.084 2.084 2.199 O5 0.607 0.158 0.019 0.022 1.027 1.634 1.833 O6 0.502 0.850 1.352 1.352 O7 0.060 0.922 0.947 0.127 1.869 2.056 O8 0.623 0.102 0.058 0.069 1.084 1.708 1.937 O9 0.713 0.048 0.057 1.145 1.858 1.963 Ow1 0.278 0.212 0.252 0.743 Sum 3.743 1.298 0.415 0.494 3.917 4.310 3.976

Table 5.27.: Bond valence calculations for Cs-exchanged AM-2 at 400 ◦C. Cs–O bond lengths ≤ 3.53 A˚ were considered.

Ti Cs1 Cs2 Si1 Si2 Si3 Sum lattice Sum total O1 0.640 0.074 0.052 1.084 1.725 1.851 O2 0.051 0.111 0.981 1.014 1.995 2.157 O3 0.694 0.186 0.097 1.084 1.779 2.061 O4 0.138 1.056 0.973 2.029 2.167 O5 0.676 0.150 1.000 1.676 1.825 O6 0.592 1.027 1.619 1.619 O7 0.142 0.973 0.922 1.895 2.113 0.076 1.895 2.113 O8 0.658 0.054 0.717 1.056 1.713 2.484 O9 0.607 0.061 0.346 0.971 1.577 1.984 Sum 3.867 0.931 1.323 4.027 4.095 4.020

125 5. Structural characterisation of ion-exchanged AM-2

Table 5.28.: Bond valence calculations for Sr-exchanged AM-2 at 25 ◦C. Sr–O bond lengths ≤ 3.60 A˚ were considered. Total sums are generally too high because Sr sites were treated like fully rather than partialy occupied.

Ti Sr1 Sr2 Sr3 Sr4 Si1 Si2 Si3 Sum lattice Sum total O1 0.676 0.055 0.366 0.051 1.027 1.703 2.175 O2 0.037 0.595 1.041 1.044 0.033 2.086 2.751 O3 0.676 0.070 0.197 0.027 1.056 1.731 2.025 O4 0.027 1.056 1.000 0.043 2.056 2.125 O5 0.676 0.067 0.052 0.048 1.056 0.058 1.731 1.957 O6 0.674 0.070 0.061 1.053 0.024 0.043 1.727 1.926 O7 0.022 0.025 1.027 1.056 0.078 2.083 2.208 O8 0.676 0.287 0.295 1.027 1.703 2.285 O9 0.692 0.320 0.121 1.064 0.027 1.757 2.224 Ow1 0.303 0.103 0.406 Ow2 0.303 0.044 0.121 0.468 Ow3 0.103 0.044 0.042 0.189 Sum 4.069 0.838 1.902 0.931 0.491 4.200 4.096 4.211

Table 5.29.: Bond valence calculations for Na-exchanged AM-2 at 25 ◦C. Na/K–O bond lengths ≤ 3.60 A˚ were considered. Bonds between H2O molecules and framework oxygens are not considered in the calculation, hence the low total sum on O1, which is in bonding distance to both Ow1 and Ow2.

Ti Na1 Na2/K2 Si1 Si2 Si3 Sum lattice Sum total O1 0.559 0.046 1.084 1.644 1.690 O2 0.733 0.082 1.145 1.878 1.959 O3 0.839 0.033 0.073 1.056 1.894 2.000 O4 0.489 0.228 0.011 1.310 0.046 1.799 2.084 O5 0.817 0.036 1.027 1.844 1.880 O6 0.640 0.110 0.049 1.275 1.916 2.074 O7 0.009 0.898 1.114 0.009 2.012 2.029 O8 0.140 0.019 0.874 1.084 1.958 2.117 O9 0.086 1.208 1.056 2.264 2.350 Ow1 0.291 0.291 Ow2 0.174 0.049 0.008 0.231 Sum 4.076 0.870 0.628 4.278 4.069 4.784

126 5.6. Tables

Table 5.30.: Bond valence calculations for Mn-exchanged AM-2 at 25 ◦C. Mn–O bond lengths ≤ 3.60 A˚ were considered.

Ti Mn1 Mn2 Mn3 Si1 Si2 Si3 Sum lattice Sum total O1 0.623 0.177 0.025 0.038 1.000 1.623 1.863 O2 0.067 0.036 1.027 1.145 2.172 2.275 O3 0.658 0.232 0.054 1.421 2.079 2.365 O4 0.043 1.027 0.898 1.925 1.968 O5 0.713 0.029 1.084 0.014 1.798 1.840 O6 0.427 0.106 1.114 0.008 1.541 1.655 O7 0.075 0.898 1.114 2.012 2.086 O8 0.733 0.042 0.281 0.023 0.784 0.008 1.517 1.870 O9 0.607 0.012 0.010 0.874 0.046 1.480 1.547 Ow1 0.649 0.649 Ow2 0.125 0.081 0.206 Ow3 0.042 0.081 0.123 Ow4 0.047 0.051 0.098 Sum 3.760 0.736 0.656 1.008 3.913 3.709 4.764

Table 5.31.: Bond valence calculations for Ca-exchanged AM-2 at 25 ◦C. Ca–O bond lengths ≤ 3.60 A˚ were considered.

Ti Ca1 Ca2 Si1 Si2 Si3 Sum lattice Sum total O1 0.607 0.072 0.040 1.176 1.783 1.894 O2 0.051 1.084 1.056 2.140 2.191 O3 0.713 0.054 1.176 1.889 1.943 O4 0.072 0.947 0.947 1.895 1.967 O5 0.575 0.080 1.114 1.689 1.769 O6 0.581 0.201 1.114 1.695 1.896 O7 0.037 0.973 1.027 0.017 2.001 2.055 O8 0.676 0.017 0.302 1.027 1.703 2.022 O9 0.676 0.016 1.053 0.066 1.728 1.811 Ow1 0.547 0.547 Ow2 0.035 0.271 0.094 0.400 Sum 3.827 0.982 0.991 4.088 4.235 4.373

127 5. Structural characterisation of ion-exchanged AM-2

5.6.5. Peak lists

Table 5.32.: Positions of the XRD reﬂections of a new phase that crystallised from Sr- exchanged AM-2 above 700 ◦C. The data was collected at room temperature using CuKα X-radiation.

2θ d [A]˚ Iabs [cts] Irel [%] 2θ d [A]˚ Iabs [cts] Irel [%] 21.610 4.109 61 5 56.432 1.629 28 2 27.648 3.224 362 27 57.158 1.610 324 24 29.684 3.007 970 71 57.320 1.606 202 15 30.145 2.962 139 10 58.514 1.576 132 10 33.893 2.643 361 27 59.343 1.556 241 18 34.746 2.580 43 3 59.730 1.547 82 6 35.491 2.527 1360 100 60.414 1.531 319 23 35.584 2.521 282 21 60.839 1.521 48 4 35.620 2.518 342 25 61.872 1.498 42 3 38.503 2.336 97 7 62.789 1.479 100 7 41.801 2.159 105 8 66.190 1.411 123 9 42.992 2.102 114 8 66.385 1.407 61 4 43.254 2.090 450 33 67.416 1.388 67 5 43.921 2.060 44 3 67.894 1.379 187 14 44.636 2.028 99 7 68.091 1.376 78 6 44.760 2.023 209 15 68.414 1.370 38 3 46.172 1.964 99 7 73.266 1.291 92 7 47.319 1.919 115 8 74.395 1.274 66 5 47.982 1.894 134 10 75.448 1.259 80 6 48.476 1.876 262 19 76.560 1.243 79 6 48.921 1.860 73 5 76.860 1.239 35 3 49.617 1.836 135 10 78.654 1.215 49 4 49.989 1.823 403 30 80.433 1.193 59 4 52.206 1.751 132 10 82.392 1.170 39 3 52.349 1.746 57 4 85.439 1.135 103 8 53.740 1.704 31 2 94.987 1.045 87 6 54.581 1.680 97 7 95.314 1.042 59 4

128 5.6. Tables

Table 5.33.: Positions of the XRD reﬂections of a new phase that crystallised from Na- exchanged AM-2 above 750 ◦C. The reﬂection pattern is similar to a compound of composition Na2Ti8O13(SiO4)2 with unknown structure described by Khainakov et al. Khainakov et al. (1999). The data was collected at room temperature with CuKα X- radiation.

2θ d [A]˚ Iabs [cts] Irel [%] 2θ d [A]˚ Iabs [cts] Irel [%] 11.608 7.617 3352 5 49.110 1.854 79 < 1 13.784 6.419 171 < 1 50.563 1.804 563 1 16.359 5.414 1815 3 50.705 1.799 321 < 1 22.188 4.003 6989 11 54.558 1.681 177 < 1 23.309 3.813 37757 58 56.657 1.623 106 < 1 25.862 3.442 388 1 56.818 1.619 56 < 1 26.004 3.424 449 1 57.237 1.608 286 < 1 27.111 3.286 574 1 57.412 1.604 177 < 1 27.477 3.243 1322 2 58.393 1.579 157 < 1 27.733 3.214 7302 11 58.605 1.574 226 < 1 31.752 2.816 239 < 1 58.789 1.569 118 < 1 32.356 2.765 149 < 1 59.939 1.542 82 < 1 33.070 2.707 201 < 1 60.560 1.528 2274 4 34.243 2.616 538 1 60.734 1.524 1330 2 34.328 2.610 587 1 61.850 1.499 518 1 34.496 2.598 857 1 62.019 1.495 273 < 1 35.264 2.543 64671 100 63.543 1.463 617 1 37.210 2.414 231 < 1 63.727 1.459 333 1 38.425 2.341 226 < 1 70.283 1.338 167 < 1 38.517 2.335 195 < 1 70.477 1.335 143 < 1 40.176 2.243 48 < 1 73.548 1.287 74 < 1 41.256 2.186 93 < 1 74.458 1.273 3325 5 42.114 2.144 160 < 1 74.682 1.270 1897 3 42.240 2.138 80 < 1 75.580 1.257 101 < 1 42.617 2.120 104 < 1 75.960 1.252 102 < 1 42.759 2.113 89 < 1 87.801 1.111 130 < 1 45.226 2.003 1280 2 88.073 1.108 58 < 1 45.350 1.998 975 2 89.784 1.091 1093 2 47.584 1.909 15964 25 90.077 1.089 626 1 47.728 1.904 9213 14 93.243 1.060 99 < 1

129 5. Structural characterisation of ion-exchanged AM-2

References

Balmer, M. L., Huang, Q., Wong-Ng, W., Roth, R. S., and Santoro, A. Neutron and X-ray

diffraction study of the crystal structure of CsTiSi2O6.5. Journal of Solid State Chemistry, 130(1):97–102, 1997.

Bortun, A. I., Bortun, L. N., Poojary, D. M., Xiang, O., and Clearﬁeld, A. Synthesis, characterization, and ion exchange behaviour of a framework potassium titanium trisilicate

K2TiSi3O9 · H2O and its protonated phases. Chemistry of Materials, 12:294–305, 2000.

Clearﬁeld, A., Bortun, A. I., Bortun, L. N., Poojary, D. M., and Khainakov, S. A. On the selec-

tivity regulation of K2ZrSi3O9 · H2O-type ion exchangers. Journal of Molecular Structure, 470:207–213, 1998.

Ilyushin, G. D., Khomyakov, A. P., Shumyatskaya, N. G., Voronkov, A. A., Nevskii, N. N., Ilyukhin, V. V., and Belov, N. V. Crystal structure of a new natural modiﬁcation of

K2ZrSi3O9 · H2O. Doklady Akademii Nauk SSSR, 256(4):860–864, 1982. (in Russian).

Khainakov, S. A., Bortun, A. I., Bortun, L. N., Clearﬁeld, A., Trobajo, C., and Garc´ıa, J. R. Hydrothermal synthesis and characterization of alkali metal titanium silicates. Journal of Materials Chemistry, 9:269–272, 1999.

Nyman, M., Bonhomme, F., Teter, D. M., Maxwell, R. S., Gu, B. X., Wang, L. M., Ewing, R. C., and Nenoff, T. M. Integrated experimental and computational methods for structure determination and characterization of a new, highly stable cesium silicotitanate phase,

Cs2TiSi6O15 (SNL-A). Chemistry of Materials, 12(11):3449–3458, 2000.

Orlov, I. P., Popov, K. A., and Urusov, V. S. Program for predicting interatomic distances in crystals by the bond valence method. Journal of Structural Chemistry, 39(4):575–579, 1998.

Rodr´ıguez-Carvajal, J. FullProf 2000. Computer program, 2005. Version 3.20.

Shannon, R. D. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Crystallographica, A(32):751–767, 1976.

Zou, X. and Dadachov, M. S. K2TiSi3O9 · H2O. Acta Crystallographica Section C, 56:738–739, 2000.

130 6. Schreyerite, V2Ti3O9: New Occurrence and Crystal Structure

Nicola Dobelin¨ 1, Leonid Z. Reznitsky2, Evgeny V. Sklyarov2, Thomas Armbruster1, Olaf Medenbach3

1Laboratorium fürchemische und mineralogische Kristallographie, UniversitätBern, Freiestrasse 3, CH-3012 Bern, Switzerland 2Institute of the Earth’s Crust, Siberian Branch of the Russian Academy of Science, Irkutsk 664033, Russia 3Institut fürGeologie, Mineralogie und Geophysik, Ruhr-Universität,D-44780 Bochum, Germany

6.1. Abstract

A new occurrence of schreyerite, V2Ti3O9, has recently been found in metamorphic rocks of the Sludyanka complex at the southern shore of Lake Baikal, Russia. In contrast to previously known schreyerite lamellae, which are intergrown with rutile, crystals from the Sludyanka complex occur as isolated single crystals associated with titanite, allowing single-crystal X-ray diffraction experiments. The chemical composition was determined with an electron microprobe giving the composition (V1.785Cr0.157Fe0.036)(Ti2.536V0.468)O9.A peculiarity of this schreyerite is the partial substitution of V4+ for Ti4+. The crystal structure was determined by single-crystal X-ray diffraction and reﬁned in the monoclinic space ◦ group C2/c (a = 17.102(2), b = 5.0253(5), c = 7.0579(8) A,˚ β = 106.636(10) ) to R1 = 2.84%. The structure is in agreement with a qualitative model of Grey et al. (1973) determined on the basis of electron diffraction and X-ray powder data for synthetic (Fe,Cr)2Ti3O9. Reinves- tigation of schreyerite from the type locality indicates that this sample has cell dimensions and symmetry corresponding to the Sludyanka sample. The structure of schreyerite may be considered as a 1 : 1 polysome composed of slabs of berdesinskiite, V2TiO5, and Ti2O4

(high-pressure phase of TiO2 with the α-PbO2 structure).

6.2. Introduction

Schreyerite, V2Ti3O9, was ﬁrst discovered in metamorphic rocks of the Kwale district, Kenya (Medenbach and Schmetzer, 1978; Bernhardt et al., 1983). It also occurs at a pyrite deposit in Sartra, Sweden (Zakrzewski et al., 1982), at the Pb-Zn ore deposit Rampura Agucha, In- dia (Holler¨ and Stumpﬂ, 1995), and in metamorphic rocks of the Ol’khon complex on the

131 6. Schreyerite, V2Ti3O9: New Occurrence and Crystal Structure western shore of Lake Baikal, Russia (Boronikhin et al., 1990). Schreyerite usually occurs as polysynthetically twinned exsolution lamellae in rutile, and until now, single-crystal X-ray structure solution has not been possible. Recently, schreyerite was found in metamorphic rocks of the Sludyanka complex at the southern shore of Lake Baikal, Russia. Instead of the usual intergrowths with rutile, single crystals of schreyerite occur, associated with titanite. ◦ Phase equilibria in the system V2O3–TiO2 at 1200 C (Brach et al., 1977) yielded the intermediate phases V2TiO5 (berdesinskiite), V2Ti2O7 and V2Ti4O11.V2Ti3O9 was also found, but it was interpreted at this temperature as a metastable phase with a free energy slightly larger than those of the adjacent phases V2Ti2O7, and V2Ti4O11. Notice that both vana- 3+ 4+ 3+ 4+ dium and titanium end-members V2 V3 O9 and Ti2 Ti3 O9 can be prepared easily (Brach et al., 1977), but these homologues have a triclinic structure of different topology (LePage et al., 1991; Marezio et al., 1977) compared to monoclinic V2Ti3O9(schreyerite). It is not clear which structure type is represented by metastable V2Ti3O9(Brach et al., 1977).

Attempts to synthesise and to derive a structural model of M5O9 compound in the sys- ◦ tem (Cr,Fe)2Tin−2O2n−1 began with Grey and Reid (1972). Oxidation of ilmenite at 770 C ◦ yielded the reaction: 6 FeTiO3 + 3O → 2 Fe2Ti3O9 + Fe2O3. Above 800 C, Fe2Ti3O9 decomposed to pseudobrookite and rutile. Between 1050 and 1250 ◦C, predominantly single phase (Fe,Cr)2Ti3O9 was obtained, and electron diffraction conﬁrmed the unit-cell dimensions previously derived from X-ray powder data for Fe2Ti3O9.

The structures of (Cr,Fe)2Tin−2O2n−1, with n = 6, 7, 8, 9 are closely related to that of rutile and may be derived from it by crystallographic shear (Magneli-Andersson phases). Grey and Reid (1972) recognised that the members with n = 3, 4, and 5 are closely related to the high-pressure modiﬁcation of TiO2 with the structure of α-PbO2, and may also be derived from it by crystallographic shear. Subsequently, they depicted (their Figure 5) an inferred ◦ structure for (Fe,Cr)2Ti3O9 (monoclinic with a = 7.1 , b = 4.9, c = 18.6 A,˚ β = 119.7 ). How- ever, this model proved to be incorrect and was revised by Grey et al. (1973). As will be shown in this paper, the new improved model is the correct structure of schreyerite. When schreyerite V2Ti3O9was described as a new mineral, Medenbach and Schmetzer (1978) no- ticed the similarity of its powder-diffraction pattern and that of (Fe,Cr)2Ti3O9, and they indexed their powder data both with the setting of Grey and Reid (1972) and with the setting of Grey et al. (1973). The aim of the present study is to provide structural details of schreyerite from a new locality at the Baikal region (Russia) and to show that this schreyerite with slightly modiﬁed chemical composition has the same structure as schreyerite from the type locality described by Medenbach and Schmetzer (1978) and Bernhardt et al. (1983). In addition, the excellent study by Grey et al. (1973) provides the basis for the description of the schreyerite structure as a 1:1 member of a polysomatic series composed of alternating slabs of berdesinskiite

V2TiO5 (Bernhardt et al., 1983) and Ti2O4 (high-pressure phase of TiO2 with the α-PbO2 structure).

132 6.3. Geological setting and occurrence of schreyerite

6.3. Geological setting and occurrence of schreyerite

The Sludyanka complex belongs to one of the metamorphic terranes of the Central Asian foldbelt. The terrain is situated to the south of Lake Baikal near the boundary with the Siberian craton. The Sludyanka complex includes folded supracrustal series of intercalated maﬁc granulite, gneiss, marble and carbonate-silicate rocks metamorphosed in granulite facies. Metamorphism of the Sludyanka group corresponds to granulite facies of moderate pressure (climax P,T-conditions: T = 800–830 ◦C, P = 6–8 kbar). The age of metamorphism and syn- and late metamorphic magmatic and metasomatic processes is Early Paleozoic (460–490 Ma), the age of the protolith is unknown and is thought to be Early Paleozoic or Proterozoic. One of the most typical types of metasedimentary rocks in the Sludyanka complex consists of diopside, quartz and calcite, indicating siliceous-dolomite sediments as the protolith. These rocks are usually referred to as quartz-diopside rocks and include several assemblages according to the content of the principal minerals: diopsidite, diopside-quartz, diopside-calcite rocks, calciphire, diopside-bearing quartzite (up to 80–90% of Qtz). Apatite can be relatively abundant (1–3 up to 50 wt-%), particularly in metaphosphorites. Thin layers and lenses, characterized by relatively high contents of Cr and V (0.1–0.7 wt-% Cr2O3 +

V2O3, with Cr/V ratio from 15:1 to 1:6) and various Cr-V minerals, occur in this type of rock (Reznitsky and Sklyarov, 1996). The most diverse assemblages of chromium and vanadium minerals (as well as Cr-V-bearing minerals, mostly in trace amounts) occur usually in high- silica (80–95 wt-%) lenses and thin layers in diopside-calcite rocks and marbles. Studying the new mineral vanadiumdravite of the tourmaline group from such high-silica lenses with high V/Cr ratio, we have found two unknown Ti-V oxides (Reznitsky et al., 2001) and later, after careful investigations of the heavy fraction, other ore minerals including schreyerite. Cr-V mineral associations in investigated samples are similar, differing only in detail. They include in association with quartz (Cr-V)-bearing diopside, tremolite, di- and trioc- tahedral micas, Cr-bearing tourmalines of the dravite-vanadiumdravite series, Cr-bearing goldmanite, eskolaite-karelianite, V-bearing chromite and titanite, and barite. Sporadi- cally occurring ilmenite, magnetite and anatase also contain vanadium. Together with schreyerite and berdesinskiite, unknown Ti-V oxides and two Ba-Ti-V oxides (one of the mannardite-ankangite type and another unknown oxide belonging to the derbylite group) are present. In all V-Ti and Ba-V-Ti oxides of this association, combination of microprobe and structure data indicates the presence of V4+ and V3+, and isomorphic substitution V4+ → Ti4+. 3+ 4+ Therefore, an extensive solid-solution of the form V3O5(V2 V O5)–V2TiO5 (berdesinskiite) is possible. Schreyerite from Sludyanka forms anhedral, rarely subhedral grains (ﬁg. 6.1), sometimes as inclusions in quartz, but usually intergrown with silicates and opaque Cr-V minerals. The maximum size of grains is 100 × 150 µm.

133 6. Schreyerite, V2Ti3O9: New Occurrence and Crystal Structure

Figure 6.1.: Scanning electron-microscopic image of schreyerite crystal studied by single-crystal X-ray diffraction. Width of picture is 0.14 mm.

6.4. Experimental procedure

6.4.1. Chemical analysis

After crushing the rock sample to grain sizes of 0.2 × 0.3 mm, opaque minerals were extracted by electromagnetic separation using pure Clerichi solution. These grains were pressed into plastic, polished, and analysed with a JCXA-733 electron microprobe at the Institute of Geochemistry, Irkutsk, as well as with a modiﬁed MAR-3 microprobe at the Ge- ological Institute of Ulan-Ude. Both devices were operated at 15 kV acceleration voltage and 20 nA beam current using a ﬁne-focused beam (2 and 10 µm). Raw data were reduced and converted into oxide weight percentages using the programs ZAF and PAP (Pouchou and Pichoir, 1991). The following standards were used: Synthetic V2O5 (V), chromite UV- 126, 79/12 and Cr-spinel (Cr, Fe, Al), rutile and ilmenite (Ti), blue diopside (Si, Mg), and spessartine (Mn). For additional control, some grains were analysed on an ABT-55 electron microscope equipped with a LINK AN-1000/S85 device, using pure metals (Ti, V, Cr, Mn,

Fe) and oxides (SiO2, MgO, Al2O3) as standards. The results of the analyses using different instruments gave almost the same concentrations of elements. Grains for X-ray analysis were selected after semi-quantitative analysis on a LEO electron microscope to check for uniform composition of the grains.

6.4.2. Structure determination

The structure of a schreyerite crystal (ca. 0.10 × 0.07 × 0.05 mm) was studied by single- crystal X-ray diffraction on an Enraf-Nonius CAD4 diffractometer using graphite-monochromated MoKα X-radiation. Cell dimensions were obtained from high-angle data (θ > 20◦), giving a C-centred monoclinic lattice with a = 17.102(2), b = 5.0253(5), c = 7.0579(8) A,˚ β = 106.636(10)◦, V = 581.20(12) A˚ 3. Data reduction, including background, Lorentz and polarisation corrections and an empirical absorption correction based on ψ scans, was done using the SDP program library (Enraf Nonius, 1983). As indicated by systematic absences,

134 6.5. Results

Table 6.1.: Representative chemical analyses and average composition of schreyerite from the Sludyanka complex, Russia.

1 2 3 4 5 6 7

SiO2 0.00 0.00 0.00 0.26 0.00 0.22 0.12 (0.00–0.34) TiO2 51.47 51.93 51.78 53.46 53.3 52.8 51.91 (49.89–53.46) Al2O3 0.00 0.00 0.00 0.14 0.00 0.00 0.04 (0.00–0.21) Cr2O3 4.11 4.15 5.20 2.86 3.93 3.63 3.06 (0.79–5.20) V2O3* 42.57 41.72 41.57 42.74 41.14 41.35 43.26 (41.72–45.11) Fe2O3 0.97 0.95 0.60 0.19 0.82 0.75 0.75 (0.00–1.13) MnO 0.00 0.00 0.00 0.00 0.00 0.32 0.11 (0.00–0.32) MgO 0.00 0.06 0.00 0.00 0.00 – 0.04 (0.00–0.24) total 99.12 98.81 99.09 99.65 99.19 99.07 99.30 (96.58–100.55)

VO2** 10.19 9.66 9.84 8.17 8.27 8.86 9.94 (8.17–11.45) V2O3** 33.36 32.99 32.62 35.36 33.67 33.35 34.28 (32.13–35.76) total 100.1 99.74 100.04 100.44 99.99 99.93 100.26 (97.57–101.54) 3+ 4+ * all vanadium as V2O3; ** V and V calculated according to the ideal formula. 1–6: analyses of individual grains; 7: average composition of schreyerite (17 analyses), maximal and minimal values in parentheses the structure was solved in space group C2/c by direct methods and refined using the program SHELX-97 (Sheldrick, 1997), converging at R1 = 2.84%. A total of 1432 reflections was measured, of which 1277 were unique. Refinements were done with anisotropic displacement parameters for all sites. The refined chemical composition was constrained to 4+ 3+ the ideal composition of schreyerite (V2Ti3O9) because the scattering factors of Ti ,V , V4+ and Cr3+ are very similar. In addition, single-crystal X-ray data were collected for the same (twinned) schreyerite crystal intergrown with rutile used by Bernhardt et al. (1983) for the improved X-ray powder data of type-locality schreyerite from Kenia. Our recollected data showed that the diffraction pattern can be indexed in space group C2/c with a = 17.023(9), b = 5.012(2), c = 7.058(5) A,˚ β = 106.20(4)◦, V = 578.3(6) A˚ 3. This confirms our suggestion that type-locality schreyerite and the crystal from the Baikal region have the same structure.

6.5. Results

6.5.1. Chemical composition

According to electron-microprobe analysis, the main components of the mineral are Ti and V, with minor amounts of Cr and Fe (tab. 6.1). The ratio Ti : R3+ (Cr, V, Fe, Al) is nearly 1 : 1, which is not consistent with the reﬁned ratio R4+ :R3+ = 3 : 2. This can be explained by the fact that Ti4+ is partly replaced by V4+. The V4+ :V3+ ratio was calculated to maintain charge balance and to be consistent with the reﬁned composition (tab. 6.2). Usual substitu- 4+ 4+ 3+ 3+ tions are Ti ← V and V ← Cr (up to 5.2 wt-% Cr2O3, or up to 0.27 a.f.u.), and minor

135 6. Schreyerite, V2Ti3O9: New Occurrence and Crystal Structure

Table 6.2.: Unit formulae (9 O atoms and 5 cations) for schreyerite based on results in Table 6.1.

1 2 3 4 5 6 7 Si – – – 0.017 – 0.014 0.006 (0.000–0.022) Ti 2.519 2.549 2.536 2.600 2.610 2.586 2.536 (2.475–2.610) V4+ 0.481 0.457 0.464 0.383 0.390 0.418 0.468 (0.390–0.536) Al – – – 0.011 – – 0.003 (0.000–0.016) V3+ 1.741 1.727 1.703 1.834 1.758 1.741 1.785 (1.665–1.880) Cr3+ 0.212 0.214 0.268 0.146 0.202 0.187 0.157 (0.040–0.268) Fe3+ 0.047 0.047 0.029 0.009 0.040 0.037 0.036 (0.009–0.054) Mn – – – – – 0.018 0.005 (0.000–0.018) Mg – 0.006 – – – – 0.004 (0.000–0.023)

3+ Fe (up to 1.13 wt-% Fe2O3, or 0.05 a.f.u.). Contents of other analysed elements (Si, Al, Mn, Mg) are insigniﬁcant or not detected. The grain selected for X-ray investigation is relatively uniform in composition (three determinations, mean composition TiO2: 51.83, V2O3: 44.41,

Cr2O3: 3.04 wt-%) and corresponds to the average composition of schreyerite as determined by electron-microprobe analysis.

6.5.2. Crystal structure

Details of the diffraction experiment and structure reﬁnement are given in Table 6.3. Atom coordinate sites, occupancies, and anisotropic displacement parameters of the reﬁned structure are given in Tables 6.4 and 6.5. The three cation sites Ti1, Ti2 and V3 are each coordinated by six oxygen atoms within rather similar bond distances: Ti1–O: 1.987–2.025, Ti2–O: 1.913–2.070, V3–O: 1.851–2.191 (tab. 6.6). The oxygen atoms bonding to Ti1 form an almost perfect octahedron with bond angles close to 180◦ and 90◦, respectively, whereas the Ti2 and V3 octahedra are slightly distorted. These octahedra are connected to two different types of chains of edge-sharing octahedra usually referred to as the berdesinskiite-type chain and α-

PbO2-type chain. As shown in ﬁgure 6.2, these chains are connected to staircase-like sheets, which alternate to form a framework. There are two distinct tetravalent sites and one trivalent site, leading to an ionic ratio R4+ :R3+ = 3 : 2. As electron-microprobe analysis shows that all three sites are occupied by elements with similar scattering factors (Ti, V, Cr), one cannot clearly assign each element to a particular position. In addition, EMP analysis indicates that vanadium occurs in both trivalent and tetravalent oxidation state, and partly replaces Ti4+. However, as the structure contains one face-sharing octahedron V3, one can assume that this site is occupied by the trivalent cation in order to minimise repulsion between central cations. Connections of octahedra Ti1 and Ti2 are by edge- and corner-sharing, which leads to larger separation between central cations and less repulsion in the case of high positive charge.

136 6.5. Results

Table 6.3.: X-ray data collection and reﬁnement parameters for schreyerite.

Sample IV8-13 Crystal size (mm) 0.10 × 0.07 × 0.05 Chemical composition (V1.785Cr0.157Fe0.036)(Ti2.536V0.468)O9 Diffractometer Enraf Nonius CAD4 X-ray radiation fine focus sealed tube, MoKα X-ray power 50 kV, 40 mA Temperature 293 K Space group C2/c Z 4 Cell dimensions a, b, c (A)˚ 17.102(2), 5.0253(5), 7.0579(8) β (◦) 106.636(10) Cell volume (A˚ 3) 581.20(12) Absorption correction Empirical ψ scans Maximum 2θ 69.83◦ Measured reflection 1432 Index range 0 ≤ h ≤ 27; 0 ≤ k ≤ 8; −11 ≤ l ≤ 10 Unique reflections 1277 Reflections > 4σ I 727 Rint 0.0195 R 0.0572 Number of l.s. parameters 67 GooF 0.999 R1,Fo > 4σ (Fo) 0.0284 R1, all data 0.0775 2 wR2 (on Fo) 0.0923

Table 6.4.: Atom positional parameters, and occupancies for schreyerite from the Sludyanka complex, Russia.

2 Atom Occupancies x/a y/b z/c Beq (A˚ ) Ti1 1 0 0 0 0.233(7) Ti2 1 −0.21608(2) 0.0067(1) −0.07227(6) 0.560(6) V3 1 −0.07555(3) −0.5010(1) 0.10036(6) 0.612(6) O1 1 −0.1085(1) 0.1585(4) −0.0203(3) 0.47(3) O2 1 −0.1742(1) −0.3316(4) 0.0731(3) 0.50(2) O3 1 −0.0474(1) −0.3463(4) −0.1300(3) 0.40(2) O4 1 −0.2262(1) −0.1679(4) −0.3190(3) 0.52(2) O5 1 0 0.1965(5) −0.25 0.37(3)

137 6. Schreyerite, V2Ti3O9: New Occurrence and Crystal Structure

Table 6.5.: Anisotropic displacement parameters for schreyerite from the Sludyanka complex, Russia.

Atom U11 U22 U33 U12 U13 U23 Ti1 0.0035(2) 0.0021(2) 0.0032(2) 0.0003(3) 0.0009(2) 0.0000(3) Ti2 0.0072(2) 0.0058(2) 0.0074(2) 0.0007(2) 0.0007(1) −0.0011(3) V3 0.0101(2) 0.0068(2) 0.0053(2) −0.0011(2) 0.0005(1) 0.0003(2) O1 0.0040(7) 0.0049(8) 0.0089(7) −0.0005(6) 0.0017(6) −0.0008(6) O2 0.0055(7) 0.0051(7) 0.0091(7) 0.0005(7) 0.0034(6) 0.0004(6) O3 0.0061(8) 0.0041(7) 0.0051(7) −0.0004(6) 0.0016(6) −0.0006(6) O4 0.0081(7) 0.0063(8) 0.0058(7) −0.0016(7) 0.0025(6) −0.0005(6) O5 0.007(1) 0.003(1) 0.0039(9) 0 0.0018(8) 0

6.6. Discussion

In the system Cr2Tin−2O2n−1, the mixed oxides with n = 6, 7, 8 and 9 are isomorphous with the homologues series TinO2n−1 and VnO2n−1 (Andersson et al., 1959). First attempts to 3+ describe the crystal structure of Me2 Tin−2O2n−1-like phases with n < 6 from X-ray powder patterns and electron diffraction were published by Grey and Reid (1972). According to the nomenclature by Grey et al. (1973), the structure-building sheets are referred to as P slabs (V3O5, berdesinkiite-type) and Q slabs (α-PbO2-type, Ti2O4), respectively. Grey and Reid (1972) described the structure of a high temperature phase, n = 5, of composition

(Cr,Fe)2Ti3O9 as a P2Q2 polysome with stacking sequence PPQQ. . . Later, that structure was revised by the same authors (Grey et al., 1973) and described as a PQPQ. . . sequence, for the reason that the new structural model was most compatible with the observed electron- diffraction pattern. This is in perfect agreement with the fact that polysomes with maximum alternation of P and Q slabs are the most abundant in Nature. Periodic intergrowths of the corresponding structural slabs can be expressed as pM3O5 · qTi2O4, or PpQq. This approach also allows derivation of phases with non-integer values of n. The unit cell Grey et al. (1973) used to describe the revised structure is illustrative in that the monoclinic angle describes the ,,slope” of the P and Q slabs (ﬁg. 6.3). However, their cell (a = 7.06, b = 5.01, c = 25.06 A,˚ β = 139◦) has a rather obtuse monoclinic angle and is A-centred, and can be transformed (2 0 1; 0 −1 0; −1 0 0) to a reduced C-centred setting, as used in the present study. The original setting by Grey and Reid (1972), a = 7.06, b = 5.01, c = 18.74 A,˚ β = 119◦, is also A-centred and can be transformed (2 0 1; 0 1 0; −1 0 0) to the standard setting for space group C2/c. The various settings are summarised in Table 6.7 and ﬁgure 6.3.

The end-member phase berdesinskiite V2TiO5 (P slab) of the polysomatic series was described by Bernhardt et al. (1983) as a mineral. However, the structure has not been published for a crystal of composition V2TiO5, but has been assumed to be analogous to Fe2TiO5 (Drofenik et al., 1981), based on corresponding cell-dimensions and symmetry (fig. 6.4). Our own structure refinements of berdesinskiite (from the same Baikal locality as schreyerite) confirm this assumption.

138 6.6. Discussion

Figure 6.2.: A polyhedral representation of the schreyerite structure projected approximately along [010]. Two types of chains can be distinguished: Dark polyhedra form berdesinskiite-type (V3O5-type) chains of edge-shearing octahedra, whereas the bright polyhedra form α-PbO2-type chains of edge-sharing octahedra. The chains are interconnected to staircase-like layers (P and Q slabs). Different stacking sequences of P and Q slabs are possible. Schreyerite shows maximum alternation, also expressed as P1Q1.

To our knowledge, the P1Q1 schreyerite structure is the only polysome of this series in the system V2O3–TiO2 that has been found in Nature to date. The synthetic phases V2Ti2O7, ◦ V2Ti4O11 and V2Ti8O15 (Brach et al., 1977) in the system V2Tin−2O2n−1 at 1200 C, corresponding to n = 4, 6 and 8, are isomorphous with members of the homologues series

TinO2n−1 and VnO2n−1, and are thus not built by P and Q slabs. At high temperature ◦ (T > 1200 C) in the more complex system Cr2O3–Fe2O3–TiO2–ZrO2, several phases occur which are assembled of P and Q slabs (Grey et al., 1973): For n = 4 with the sequence QPP. . . , for n = 6 with QPQQP. . . , and for n = 8 with PQQPQPQQ. . . . These n values are the same as those found in V2Tin−2O2n−1, indicating the possibility of polymorphism. If additional V2Tin−2O2n−1 phases with a structure composed of berdesinskiite (V2TiO5) and Ti2O4

(α-PbO2 structure) slabs should be found in Nature, they should be regarded as schreyerite polysomes, adopting the rules for a polysomatic series.

The only polysome of the general pM3O5 · qM2O4 series (except schreyerite) for which a crystal-structure reﬁnement is available is the n = 4 member CrFeTi2O7 (Grey and Mumme,

1972), which has the stacking sequence P2Q1 (ﬁg. 6.5). The structure of most other (P, Q) polysomes have not been solved from single-crystal X-ray data but are derived as ,,models” from systematic features in electron-diffraction patterns (Grey et al., 1973). Peculiar to schreyerite from the Baikal region is the partial V4+ → Ti4+ substitution in- dicative of highly oxidising conditions. Similar substitutions have also been observed for

139 6. Schreyerite, V2Ti3O9: New Occurrence and Crystal Structure

Figure 6.3.: The schreyerite structure shown as closest oxygen-packing projected normal to the (010) (top) and (201) (bottom) planes. The large light spheres represent oxygen atoms, small dark spheres represent octahedrally coordinated cations. Three different monoclinic unit-cells are shown: (1) the unit cell presented in this study, (2) the unit cell reported by Grey et al. (1973), and (3) the unit cell reported by Grey and Reid (1972). Axis labels and indices apply to unit cell 1.

140 6.6. Discussion

Figure 6.4.: End-members of the polysomatic series PpQq are the structures of berdesinskiite (left, V2TiO5) (Drofenik et al., 1981; Bernhardt et al., 1983) and α-PbO2 (right).

Table 6.6.: Interatomic distances (A)˚ and angles (◦) of the reﬁned schreyerite structure.

Ti1– Distance Angles O1 1.989(2) O1 1.989(2) 180.0(1) O5 2.024(1) 91.45(7) 88.55(7) O5 2.024(1) 88.55(7) 91.45(7) 180 O3 2.027(2) 93.87(8) 86.13(8) 82.87(8) 97.13(8) O3 2.027(2) 86.13(8) 93.87(8) 97.13(8) 82.87(8) 180.0(2) Ti1– O1 O1 O5 O5 O3 Ti2– Distance Angles O4 1.915(2) O1 1.929(2) 100.78(9) O4 1.953(2) 97.05(5) 95.23(9) O2 2.011(2) 90.18(9) 93.27(8) 167.62(9) O4 2.017(2) 169.7(1) 89.52(8) 81.36(8) 89.73(8) O2 2.073(2) 87.55(8) 169.57(8) 89.93(8) 80.32(8) 82.28(9) Ti2– O4 O1 O4 O2 O4 V3– Distance Angles O2 1.852(2) O1 1.924(2) 102.72(8) O3 1.981(2) 103.99(8) 92.35(9) O3 1.984(2) 99.39(9) 95.43(8) 153.1(1) O5 2.086(2) 97.45(8) 159.80(7) 81.52(7) 82.35(6) O3 2.193(2) 173.07(8) 83.25(8) 79.04(9) 76.33(8) 76.71(6) V3– O2 O1 O3 O3 O5

141 6. Schreyerite, V2Ti3O9: New Occurrence and Crystal Structure

Table 6.7.: An overview of the different unit cells mentioned in this study. A: The cell published by Grey and Reid (1972). B: The revised cell published by Grey et al. (1973). C: Schreyerite from Kenia, reinvestigated in this study. Original data were published by Bernhardt et al. (1983). D: Schreyerite from Sludyanka complex described in the present study.

ABCD a (A)˚ 18.74 25.06 17.023(9) 17.102(2) b (A)˚ 5.01 5.01 5.012(2) 5.0253(5) c (A)˚ 7.06 7.06 7.058(5) 7.0579(8) β (◦) 119.4 139.35 106.20(4) 106.636(10) V(A˚ 3) 577.48 577.43 578.3(6) 581.20(12) berdesinskiite from the same locality (to be published). The occurrence of both V4+ and V3+ are rare in minerals. We are only aware of one other example, the new mineral zoltaiite 4+ 3+ 2+ Ba(V , Ti)2(V , Ti, Fe , Cr, Mg)12(Si, Al)2O27 (Bartholomew et al., 2005). According to a recent personal communication with W. Schreyer, Bochum, experiments were done at Bochum some 15 years ago in order to synthesise schreyerite (and perhaps ◦ berdesinskiite) under hydrothermal conditions. At 3 kbar, 700 C, a mixture of V2O3 and

TiO2 in the stoichiometric ratio of 1 : 3 had reacted to single-phase schreyerite after 3 days at an oxygen fugacity of the wustite-magnetite buffer and in the presence of water. However, addition of as much as 30 wt-% of excess V2O3 relative to that in V2Ti3O9 did not produce any additional berdesinskiite, solely schreyerite. This indicates solid solution in schreyerite, with V4+ replacing Ti4+ as well. In addition, Kosuge and Kachi (1975) present a phase diagram of the system TiO2–VO2–Ti2O3–V2O3 and emphasise complete solid-solution in the pseudo-binary system VnO2n−1–V2Tin−2O2n−1 (3 ≤ n ≤ 7). Unfortunately, their reference for this part of the phase diagram (Okinaka, H., Makino, T., Kosuge, K., Kachi, S., to be published) could not be located. Their complete solid-solution between V5O9 and V2Ti3O9 suggests that their V2Ti3O9phase is not isotypic with schreyerite, but related to triclinic

V5O9 (LePage et al., 1991).

There are still several open questions concerning the binary system V2O3?TiO2, particularly under geologically relevant P,T conditions. Smyslova et al. (1982) reported a monoclinic dimorph of schreyerite, named kyzylkumite, with a = 33.80, b = 4.578, c = 19.99 A,˚ β = 93.4◦ (determined by Laue and oscillation photographs). This mineral was found in ﬁne veins cutting siliceous schists, associated with chlorite, pyrite and rutile. The crystal structure of this dimorph could not be investigated due to complex twinning and low crystal quality. Nevertheless, the cell dimensions and X-ray powder data are signiﬁcantly different from those of triclinic V5O9 and monoclinic schreyerite V2Ti3O9. The periodicity of 4.578 A˚ along b emphasises that kyzylkumite represents a structure composed of two octahedral layers where each layer is ca. 2.3 A˚ thick. Kyzylkumite is thus structurally not related to the polysomatic series derived from V2TiO5 and Ti2O4 (with α-PbO2 structure). Most probably,

142 6.6. Discussion

Figure 6.5.: The structure of (CrFe)Ti2O7 (Grey and Mumme, 1972) represents another polysome of the pM3O5 · qM2O4 series with the stacking sequence P2Q1.

143 6. Schreyerite, V2Ti3O9: New Occurrence and Crystal Structure

kyzylkumite has some structural relation to tivanite TiVO3(OH) (Grey and Nickel, 1981) and carmichaelite (Ti,Cr,Fe)2O3(OH) (Wang et al., 2000), both of which have in common an octahedral two-layer structure and a periodicity of 4.56 A˚ along b.

Olkhonskite (Cr,V)2Ti3O9 (Koneva et al., 1996) is isotypic with schreyerite. An important step toward understanding the stability of phases in the system Cr2O3–Fe2O3–TiO2 was taken by Pownceby et al. (2001) who explored the ternary phase diagram between 1000 and ◦ ◦ 1300 C. At 1200 C, the ternary phases (Cr,Fe)2TiO5 (berdesinskiite type), (Cr,Fe)2Ti2O7

(P2Q polysome, Grey and Mumme (1972)), (Cr,Fe)6Ti7O23 (PQPPQPQPPQ polysome, Grey et al. (1973)), (Cr,Fe)4Ti5O16 (QPQPPQP polysome, Grey et al. (1973)) and (Cr,Fe)2Ti3O9 (PQPQ polysome, olkhonskite) were analysed. At 1100 and 1000 ◦C, no ternary phase between (Cr,Fe)2TiO5 (berdesinskiite type) and (Cr,Fe)2Ti3O9 (PQPQ polysome, olkhonskite) was found. This also agrees with the predominant occurrence of olkhonskite (Cr,V)2Ti3O9 as lamellae in rutile and (Cr,Fe)2TiO5 (berdesinskiite type) as rims around olkhonskite, described by Koneva (2002) for the Cr-V-Ti oxide minerals of the Olkhon series of the lake Baikal area. Surprisingly, Koneva (2002) also analysed (by electron microprobe) a few tiny inclusions of (Cr,V)2Ti4O11 and (Cr,V)2Ti2O7 in rutile. In particular, (Cr,V)2Ti2O7 should not be found in rutile if olkhonshite is a stable phase in this assemblage. However, as shown in this study, a (Cr,V,Fe)/Ti ratio of 1 : 1 does not necessarily indicate a phase of (Cr,V)2Ti2O7 composition. If vanadium occurs as V4+ and V3+, the latter inclusions could also be interpreted as schreyerite-olkhonshite with partial substitution of Ti4+ by V4+.

6.7. Acknowledgements

The study of mineral composition was partly supported by Russian Foundation of Basic Re- searches (project 05-05-64200). We are grateful to Nick Karmanov, Lyudmila Suvorova and Eugene Galuskin for the help in microprobe and electron-microscope investigations. N.D. and T.A. acknowledge support from the Swiss Science Foundation. The authors acknowledge comments and corrections by W. Schreyer and H.-J. Bernhardt.

References

Andersson, S., Sundholm, A., and Magneli,´ A. A homologous series of mixed titanium

chromium oxides Tin−2Cr2O2n−1 isomorphous with the series TinO2n−1 and VnO2n−1. Acta Chemica Scandinavica, 13:989–997, 1959.

Bartholomew, P. R., Mancini, F, Cahill, C., Harlow, G. E., and Bernhardt, H.-J. Zoltaiite, a new barium-vanadium nesosubsilicate mineral from British Columbia: Description and crystal structure. American Mineralogist, 90(10):1655–1660, 2005.

Bernhardt, H.-J., Schmetzer, K., and Medenbach, O. Berdesinskiite, V2TiO5, a new mineral

144 6.7. Acknowledgements

from Kenya and additional data for schreyerite, V2Ti3O9. Neues Jahrbuch f¨urMineralogie Monatshefte, 3:110–118, 1983.

Boronikhin, V. A., Dmitrieva, M. T., Koneva, A. A., and Korikovskii, S. P. Chro- mian schreyerite and chromium-vanadium-manganese-iron-zinc spinel in sillimanite- cordierite quartzitic schists in of the Olkhon region, southwestern Baikalia. Doklady Akademii Nauk SSSR, 314(2):447–451, 1990.

Brach, B., Grey, I. E., and Li, C. Phase equilibriums in the vanadium(3+) oxide-titanium(3+) oxide-titanium(4+) oxide system. Journal of Solid State Chemistry, 20:29–41, 1977.

Drofenik, M., Golic,ˇ L., Hanzel,˘ D., Krasevec,ˇ V., Prodan, A., Bakker, M., and Kolar,

D. A new monoclinic phase in the Fe2O3–TiO2 system. I. structure determination and Mossbauer¨ spectroscopy. Journal of Solid State Chemistry, 40(1):47–51, 1981.

Enraf Nonius, . Structure determination package (SDP). (computer program), 1983.

Grey, I. E. and Mumme, W. G. The structure of CrFeTi2O7. Journal of Solid State Chemistry, 5: 168–173, 1972.

Grey, I. E. and Nickel, E. H. Tivanite, a new oxyhydroxide mineral from western australia, and its structural relationship to rutile and diaspore. American Mineralogist, 66:866–871, 1981.

Grey, I. E. and Reid, A. F. Shear structure compounds (Cr,Fe)2Tin−2O2n−1 derived from the

α-PbO2 structural type. Journal of Solid State Chemistry, 4:186–194, 1972.

Grey, I. E., Reid, A. F., and Allpress, J. G. Compounds in the system Cr2O3–Fe2O3–TiO2–

ZrO2 based on intergrowth of the α-PbO2 and V3O5 structural types. Journal of Solid State Chemistry, 8:86–99, 1973.

Holler,¨ W. and Stumpﬂ, E. F. Cr-v oxides from the Rampura-Agucha Pb-Zn-(Ag) deposit, Rajasthan, India. Canadian Mineralogist, 33:745–752, 1995.

Koneva, A. A. Cr-V oxides in metamorphic rocks, Lake Baikal, Russia. Neues Jahrbuch f¨ur Mineralogie Monatshefte, 12:541–550, 2002.

Koneva, A. A., Piskunova, L. F., Ushchapovskaya, Z. F., and Konev, A. A. Olkhonskite

(Cr,V)2Ti3O9 – a new mineral from the Priolkhon’ye (abstract). American Mineralogist, 81: 251–252, 1996.

Kosuge, K. and Kachi, S. Electron-diffraction and electron-microscopic observation of the

pseudo binary TiO2–V2O3 system. Chimica Scripta, 8:70–83, 1975.

LePage, Y., Bordet, P., and Marezio, M. Valence ordering in V5O9 below 120 K. Journal of Solid State Chemistry, 92(2):380–385, 1991.

145 6. Schreyerite, V2Ti3O9: New Occurrence and Crystal Structure

Marezio, M., Tranqui, D., Lakkis, S., and Schlenker, C. Phase transitions in Ti5O9 single crystals: Electrical conductivity, magnetic susceptibility, speciﬁc heat, electron paramagnetic resonance, and structural aspects. Physical Review B, 16:2811–2821, 1977.

Medenbach, O. and Schmetzer, K. Schreyerite, V2Ti3O9, a new mineral. American Mineralo- gist, 63:1182–1186, 1978.

Pouchou, J. L. and Pichoir, F. Electron probe quantitation, chapter Quantitative analysis of homogeneous or stratiﬁed microvolumes applying the model ,,PAP”, pages 31–75. Plenum Press, New York, 1991.

Pownceby, M. I., Fisher-White, M. J., and Swamy, V. Phase relations in the system Fe2O3 + ◦ Cr2O3 + TiO2 between 1000 and 1300 C and the stability of (Cr,Fe)2Tin−2O2n−1 crystallographic shear structure compounds. Journal of Solid State Chemistry, 161(1):45–56, 2001.

Reznitsky, L. Z. and Sklyarov, E. V. Unique Cr-V mineral association in metacarbonate rocks of the Sludyanka, Russia. In Proceedings of the 30th International Geological Congress, Beijin, China, page 446, 1996.

Reznitsky, L. Z., Sklyarov, E. V., Ushchapovskaya, Z. F., Nartova, A. A., N. V. and Kashaev, Karmanov, N. S., Kanakin, S. V., Smolin, A. S., and Nekrasova, E. A. Vanadiumdravite

NaMg3V6[Si6O18][BO3]3(OH)4, a new mineral of the tourmaline group. In Proceedings Russian Mineralogical Society, pages 59–72, 2001.

Sheldrick, G. M. SHELX-97. University of Gottingen,¨ Germany, 1997. computer program.

Smyslova, I. G., Komkov, A. I., Pavshukov, V. V., and Kuznetsova, N. V. Kyzylkumite,

V2Ti3O9, a new mineral of vanadium and titanium of a group of complex oxides (abstract). American Mineralogist, 67:855–856, 1982.

Wang, L., Rouse, R. C., Essene, E. J., Peacor, D. R., and Zhang, Y. Carmichaelite, a new hydroxyl-bearing titanate from Garnet Ridge, Arizona. American Mineralogist, 85:792–800, 2000.

Zakrzewski, M. A., Burke, E. A. J., and Lustenhouwer, W. J. Vuorelainenite, a new spinel, and associated minerals from the Satra (Doverstorp) pyrite deposit, central Sweden. Cana- dian Mineralogist, 20:281–290, 1982.

146 Heulandite-Ba, a new zeolite species from Norway

ALF OLAV LARSEN1,FRED STEINAR NORDRUM2,NICOLA DÖBELIN3,THOMAS ARMBRUSTER3, OLE V. PETERSEN4 and MURIEL ERAMBERT5

1Norsk Hydro ASA, Research Centre Porsgrunn, P. O. Box 2560, N-3907 Porsgrunn, Norway Corresponding author, e-mail: [email protected] 2Norwegian Mining Museum, P. O. Box 18, N-3602 Kongsberg, Norway 3Laboratorium für chemische und mineralogische Kristallographie, University of Bern, Freiestr. 3, CH-3012 Bern, Switzerland 4Geological Museum, University of Copenhagen, Øster Voldgade 5–7, DK-1350 Copenhagen, Denmark 5Institute for Geology, University of Oslo, P. O. Box 1047 Blindern, N-0316 Oslo, Norway

Abstract: Heulandite-Ba, ideally (Ba,Ca,Sr,K,Na)5Al9Si27O72·22H2O, is a new zeolite species in the heulandite series, occurring as an accessory mineral in hydrothermal veins of the Kongsberg silver deposit type at the Northern Ravnås prospect, southern Vinoren, 14 km NNW of Kongsberg town, Kongsberg ore district, Flesberg community, Buskerud county, Norway. The mineral has also been found at the Bratteskjerpet mine, Saggrenda near Kongsberg, and in hydrothermal veins in quartzite at Sjoa in Sel community, Oppland county. Heulandite-Ba occurs as well developed, thick tabular, trapezoidal crystals up to 4 mm across, showing the forms {100}, {010}, {001}, 111} and {201}. The mineral is colourless to white, rarely very pale yellowish white or pale beige, with a white streak; transparent to translucent, with a vitreous lustre, pearly on {010}. The mineral has a perfect {010} cleavage; subconchoidal to uneven fracture. It is non-fluorescent in long- or short-wave ultraviolet light. The Mohs’ hardness is 3½; Dmeas = 2.35(1) and Dcalc 3 2 = 2.350 g/cm . Heulandite-Ba is biaxial positive with n [ = 1.5056(5), n q = 1.5064(5) and n * = 1.5150(5); = 0.0094, n(mean) = 1.5090. 2V * (calc) = 34.1°, 2V * (meas) = 38(1)°; distinct dispersion, r > v; [ , c varying from ≅ 39° to ≅ 51° in obtuse angle q , * =b.An average of 14 electron microprobe analyses on heulandite-Ba from the Northern Ravnås prospect, Kongsberg, gave SiO2 54.26, Al2O3 15.27, MgO <0.1, CaO 2.65, SrO 1.03, BaO 12.76, Na2O 0.34, K2O 0.58, H2O 13.1 (from TGA), total 99.99, corresponding to (Ba2.49Ca1.41Sr0.30K0.37Na0.33) 7 4.90Al8.96Si27.00O72.00·21.75H2O on the basis of 72 framework oxygen atoms. Chemical zoning is frequent, with transitions to heulandite-Ca and heulandite-Sr. Heulandite-Ba is monoclinic, C2/m,witha = 17.738(3), b = 17.856(2), c = 7.419(1) Å, q = 116.55(2)°, V = 2102.0(7) Å3, Z =1. The strongest five X-ray diffraction lines of the powder pattern [d in Å(I)(hkl)] are: 2.973(100)(151), 3.978(97)(131), 7.941(66)(200), 4.650(66)(-131), 2.807(65)(–621). The crystal structure refinements (R = 3.5 %) of heulandite-Ba were done in space groups C2/m, Cm, C2, and C1, but refinements in space groups with lower symmetry than C2/m did not improve the structural model. Key-words: heulandite-Ba, new mineral, analysis, crystal structure, Norway.

Introduction landite and clinoptilolite on the basis of the Si/Al ratio. Al- rich crystals with Si/Al < 4.0 are heulandites, the less alu- Heulandite was first described by Brooke (1822), and has minous ones (Si/Al & 4.0) are regarded as clinoptilolites. generally been considered as a calcium dominant zeolite. Significant amount of barium in heulandite has been re- Over the years, however, chemical analyses reveal that other ported from a few localities. Lovisato (1897) used the term extra-framework cations may be dominant. As a conse- barium heulandite (heulandite baritica) for a barium-bear- quence, with the establishment of series among the zeolites, ing heulandite from Sardinia, Italy. The reported chemical International Mineralogical Association, Commission on analysis showed a content of 2.55 wt.% BaO, and the che- New Mineral and Mineral Names, introduced four species mical formula based on his analysis is (Na3.02Ca1.84 among the heulandite series (Coombs et al., 1997): heulan- Ba0.48) 7 5.34Al9.46Si27.00O72·27.08H2O. The mineral is a calci- dite-Ca, heulandite-Sr, heulandite-Na and heulandite-K. an heulandite-Na, and barium is only a minor component. Clinoptilolite is an isostructural zeolite mineral that dif- The name barium heulandite has also been used by Mnatsa- fers from heulandite for a lower concentration of Al. kanyan et al. (1970) for a barium bearing heulandite-Ca Coombs et al. (1997) suggested to distinguish between heu- with the chemical formula (Ca3.65Ba0.18Sr0.14Na0.99K0.72) 7 5.68

147 7. Heulandite-Ba, a new zeolite species from Norway

Al9.90Si26.37O72·23.85H2O, although this mineral has even less lents at depths between 3 and 4 km, with a hydrostatic pres- Ba than that of Lovisato (1897). Ogawa (1967) reported a heu- sure (from a fluid column) of about 350 atmospheres. Im- landite with 4.8 wt.% BaO from Japan. Ceˇ rny´ & Povondra portant constituents in the silver bearing calcite veins are (1969) reported a strontian heulandite-Ca with 2.44 wt.% quartz, barite, fluorite, native silver, argentite, pyrite, sphal- BaO from Czechoslovakia, while Miller & Ghent (1973) de- erite, chalcopyrite and coal blende (bitumen), but many oth- scribed a barian-strontian heulandite-Ca from Alberta, Cana- er minerals are present in small amounts (Neumann, 1944; da, with a barium content varying from 4.6 to 6.9 wt.% BaO. Johnsen, 1986 and 1987; Bancroft et al., 2001). Many min- During an investigation on minerals of the brewsterite se- erals occur in more than one generation. ries and their associated minerals in Norway (Larsen et al., The Kongsberg Silver Mines were in operation 1623– 2003), the first barium dominant heulandite, heulandite-Ba, 1958, being historically the largest mining enterprise in was discovered at the Northern Ravnås silver prospect, Norway. About 300 mines and prospects have been in oper- southern Vinoren, Flesberg community, Buskerud county, ation within the Kongsberg ore district (Moen, 1967; Berg, Norway. The occurrence is situated 14 km NNW of Kongs- 1998; Helleberg, 2000). berg town, within the Kongsberg ore district. Further inves- At the Northern Ravnås prospect, heulandite-Ba occurs tigations revealed that heulandite-Ba also occurred at the in calcite-quartz veins of the Kongsberg silver ore type. The Bratteskjerpet silver mine, near Saggrenda 7 km SW of veins range from a few mm to 10 cm in thickness. Barite, py- Kongsberg town, and in a road cut at Sjoa, Sel community, rite, chalcopyrite, sphalerite, galena, silver, acanthite and Oppland county. fluorite are also present. Late stage minerals include zeolites The name heulandite-Ba is in accordance with the no- of the brewsterite series, the heulandite series, and harmo- menclature for new species of previously known zeolite se- tome. A late generation calcite usually filled the last empty ries, using the root name and a suffix including a hyphen and space in the hydrothermal veins, although small cavities are the chemical symbol of the dominant extra-framework cat- still quite frequently observed. ion (Coombs et al., 1997). The new mineral species and its The find of barium dominant zeolites at Vinoren promted name have been approved by the Commission on New Min- a re-investigation of previously found heulandites, as well erals and Mineral Names, International Mineralogical As- as new search on mine dumps (Nordrum et al., 2003). Sub- sociation (CNMMN no. 2003–001). The holotype specimen sequently, heulandite-Ba with transition to heulandite-Ca of heulandite-Ba from the Northern Ravnås prospect is was identified in samples from the dumps of the Bratteskjer- housed in the collection of Geological Museum, University pet mine near Saggrenda, in the southern part of the Kongs- of Oslo (Catalogue No. 33929). berg ore district. Also at this locality, the minerals of the heulandite series are associated with minerals of the brewsterite series, harmotome, calcite and minor pyrite. Heulandite-Ba occurrences Strongly zoned crystals composed of heulandite Ba, heulandite-Sr and heulandite-Ca together with quartz, hematite, The heulandite-Ba type locality, the Northern Ravnås pro- rutile, anatase, chlorite, albite and minerals of the brewster- spect, is situated within the Kongsberg ore district, which ite series, were discovered in a sample from a roadcut south comprises an area of about 30 km in length (in a north – of Åmot farm, Sjoa, Sel community, Oppland county, Nor- south direction) and 15 km in width. The district is dominat- way. The mineralization occurs along NE-SW striking, ed by strongly metamorphosed and tectonized rocks, mainly nearly vertical fissures in a metasandstone of late-Precam- quartz-plagioclase-biotite gneisses, mica and chlorite brian age, metamorphosed during the Caledonian orogeny schists, amphibolites (metagabbros and metadolerites), and (Strand, 1951 and 1967; Siedlecka et al., 1987). The deposit granite gneisses (Bugge, 1917; Starmer, 1985). Roughly represents a low-temperature, alpine vein mineralization, parallel to the north – south strike of the rocks are zones and the hydrothermal solutions were probably active during (called fahlbands) with disseminated sulphides, predomi- the Caledonian orogeny. nantly pyrite and pyrrhotite. Age determinations on gneisses have indicated two episodes of metamorphism, about 1600 Ma and 1100–1200 Ma, respectively (Jacobsen Morphology, physical and optical properties & Heier, 1978) The Precambrian rocks are cut by Permian dolerite Heulandite-Ba from the Northern Ravnås prospect occurs as dykes, hydrothermal quartz veins and calcite veins deposit- well developed, thick tabular, trapezoidal crystals up to 4 ed along fissures and faults caused by the development of mm across, dominated by the pinacoids {100}, {010} and the Oslo Rift (Ihlen, 1986). The silver occurrences are gen- {001}, modified by {111} and {201}. Aggregates of inter- erally found at the intersections between the calcite veins grown crystals up to a few cm across are quite common. and the fahlbands. The calcite veins normally dip steeply, Heulandite-Ba from the Bratteskjerpet mine and Sjoa has and most of them have an east – west strike. They measure generally a similar morphology, but the size of the individu- from a fraction of a millimetre up to about 50 cm in width, al crystals is smaller, usually less than 1 mm across. The typ- and mostly from a few metres to 100 m in length. Breccia- ical habit of heulandite-Ba crystals is shown in Fig. 1. tion within the veins is common. According to Segalstad Heulandite-Ba is colourless to white, rarely very pale yel- (1985 and 2000) the vein minerals were precipitated from lowish white or pale beige. The streak is white. It is transpar- hydrothermal solutions at temperatures in the range 200– ent to translucent with a vitreous lustre, pearly on {010}. 300 oC and a salinity between 0 and 35 wt.% NaCl-equiva- The mineral has a perfect {010} cleavage. The fracture is

148 7.11. Morphology, physical and optical properties

Table 1. Chemical composition (in oxide weight-%) of minerals of the heulandite series from Northern Ravnås prospect (R1 with compositional ranges in parentheses – R5), Bratteskjerpet mine (B1–B6) and Sjoa (S1–S3), corresponding number of atoms per formula unit based on 72 framework oxygen atoms, and balance error (E %) according to Alietti et al. (1977).

R1 R2 R3 R4 R5 B1 B2 B3 B4 B5 B6 S1 S2 S3

SiO2 54.26 (52.51-55.91) 53.15 54.20 54.41 55.65 54.51 55.36 54.96 54.96 55.57 55.43 56.54 57.21 53.29 Al2O3 15.27 (14.54-15.77) 14.68 14.54 15.35 16.00 16.06 15.46 15.52 15.27 15.64 15.18 17.44 15.65 15.94 MgO <0.1 (<0.1) <0.1 <0.1 0.46 <0.1 0.10 0.13 <0.1 <0.1 <0.1 0.20 <0.1 <0.1 <0.1 CaO 2.65 (2.30-2.87) 2.32 2.59 3.40 4.17 2.25 2.93 3.43 3.54 3.85 3.24 4.48 2.82 1.27 SrO 1.03 (0.64-2.18) 0.98 1.55 1.12 5.26 1.75 1.22 2.45 2.72 3.89 1.49 7.90 6.30 2.58 BaO 12.76 (11.17-13.84) 13.84 11.17 9.70 3.20 12.67 11.81 8.00 7.30 5.85 10.22 1.07 5.78 16.57

Na2O 0.34 (0.23-0.50) 0.41 0.26 0.22 0.76 0.58 0.53 0.67 0.65 0.68 0.59 0.44 0.08 0.39 K2O 0.58 (0.49-0.86) 0.52 0.53 0.71 0.31 0.68 0.62 0.48 0.51 0.53 0.62 1.20 1.44 0.76 H2O 13.1 Total 99.99 85.90 84.84 85.37 85.35 88.60 88.06 85.51 84.95 86.01 86.97 89.07 89.28 90.80 Si 27.008 27.052 27.338 26.943 26.839 26.708 27.000 27.003 27.085 26.950 27.094 26.280 27.095 26.466 Al 8.958 8.806 8.644 8.958 9.094 9.274 8.886 8.987 8.869 8.940 8.745 9.554 8.736 9.330 Mg 0.000 0.340 0.073 0.095 0.146 Ca 1.413 1.265 1.400 1.804 2.155 1.181 1.531 1.806 1.869 2.001 1.697 2.231 1.431 0.676 Sr 0.297 0.289 0.453 0.322 1.471 0.497 0.345 0.698 0.777 1.094 0.422 2.129 1.730 0.743 Ba 2.489 2.760 2.208 1.882 0.605 2.433 2.257 1.540 1.410 1.112 1.958 0.195 1.073 3.225 Na 0.328 0.405 0.254 0.211 0.711 0.551 0.501 0.638 0.621 0.639 0.559 0.397 0.073 0.376 K 0.368 0.338 0.341 0.449 0.191 0.425 0.386 0.301 0.321 0.328 0.387 0.712 0.870 0.482

H2O 21.747 E (%)-1.50 -6.05 -0.84 -4.25 -2.88 -0.75 -4.88 -0.44 -2.04 -4.68 -6.89 -6.51 -7.17 -8.04

tion of the † – T variation method to be: n [ = 1.5056(5), n q 2 = 1.5064(5) and n * = 1.5150(5); = 0.0094, n(mean) = 1.5090. 2V * calculated= 34.1°, 2V * measured directly on the spindle stage = 38(1)°. Heulandite-Ba shows a distinct dispersion, r > v. Both the refractive indices and 2V * are in good agreement with the values given for ordinary heulandite (heulandite-Ca) and in particular with those given for strontian heulandite (Ceˇ rny´ & Povonda, 1969; Lucchetti et al., 1982). According to Palmer & Gunter (2000), as well as previous authors, the mean refractive index of natural divalent-exchanged heulandite series zeolites increases with increasing atomic number, and the refractive index parallel to b, in casu * , increases at a greater rate than the refractive indices in (010). The refractive indices of heulandite-Ba confirm their results. The refractive indices of heulandite-Ba also confirm the results of Boles (1972) that samples with Si/Al ratio e 3.5 corresponds to the highest values of the mean refractive indices of heulandite series minerals. Fig. 1. Drawing of a heulandite-Ba crystal from the Northern Ravnås [ , ≅ ≅ prospect. Heulandite-Ba has c varying from 39° to 51° in obtuse angle q , * = b. The optical orientation for original heulandite (heulandite-Ca) given in most literature, [ , a= subconchoidal to uneven. No fluorescence in long- or short- 0–34°and q , c = 0 – 32° (see for instance Tröger, 1982) wave ultraviolet light was observed. The Mohs’ hardness is refers to an orthorhombic pseudocell with q ≅ 91½°. When 3½. The measured density, determined by the sink/float referred to the monoclinic cell, currently accepted for origi- method using di-iodomethane diluted with acetone, is nal heulandite (heulandite-Ca), this corresponds to [ , c= 2.35(1) g/cm3. The same value, 2.350 g/cm3, is calculated 0 to 34° in obtuse angle q ,and q , a = 26½° in obtuse angle from the empirical formula, refined cell dimensions (based q to 7½ ° in acute angle q . The optical orientation of heu- on XRD powder data), and Z = 1. Finely ground heulandite- landite-Ba thus corresponds well with that of ordinary heu- Ba easily decomposes in warm 6M HCl, leaving silica as a landite (heulandite-Ca), and in particular with the optical powder. The mineral is rather resistant to cold, diluted acid. orientation given by Ceˇ rny´ & Provodra (1969) for the pre- Heulandite-Ba is biaxial positive. The following refrac- dominant part of strontian heulandite crystals studied, disre- tive indices for † = 589 nm were determined by means of the garding some extreme values they obtained in a limited rim microrefractometer spindle-stage, using calcite as refrac- zone of one growth sector. tometer crystal (Medenbach, 1985) and under the applica- Heulandite-Ba shows complicated sector zoning with ti-

149 7. Heulandite-Ba, a new zeolite species from Norway

Fig. 2. Backscatter image of a heulandite crystal from Sjoa. Darker areas are enriched in Ca and Sr, while lighter areas are enriched in Fig. 3. Variation in divalent cation content in heulandite from Sjoa; Ba. Scale bar is 0.1 mm. Ba (squares), Sr (triangles) and Ca (crosses). ny differences in refractive indices between some of the zones; twinning has been observed in a few cases. and patchy structure, and with a discrete outer rim in the AGladstone-Dale calculation gives a compatibility in- crystals. The mineral consists of calcian heulandite-Ba index of 0.011, which is regarded as superior (Mandarino, terwoven with barian strontian heulandite-Ca (Table 1, ana- 1981). lyses B1–B5). The rim is a calcian heulandite-Ba (Table 1, analysis B6). Backscatter images of heulandite from Sjoa show that the Chemical composition crystals are strongly zoned, and often showing oscillatory crystallization (Fig. 2). The inner part of the crystals is Chemical analyses of heulandites were carried out by means strontian heulandite-Ca (Table 1, analysis S1). Discrete of a CAMECA SX-100 electron microprobe, operating in zones outwards are composed of calcian heulandite-Sr of wavelength-dispersive mode. The operating conditions varied composition (represented by analysis S2 in Table 1). were as follows: operating voltage 15 kV, beam current 5 The outer part of the crystals, with relatively sharp boundary nA, and a beam diameter of 20 µm. The following standards against the inner part, is heulandite-Ba (Table 1, analysis [ [ [ were used: wollastonite (SiK ,CaK ), Al2O3(AlK ), MgO S3). This is the most barium rich member of the heulandite [ [ [ (MgK ), Sr-silicate glass (SrL ), BaSO4 (BaL ) albite series ever observed (16.57 wt.% BaO). A compilation of 22 (NaK [ ) and orthoclase (KK [ ). Count times for all elements analysis points, showing the variation in composition were 10 seconds, exept for K (20 seconds). The water con- among the extra-framework divalent cations, is given in Fig. tent is derived from the thermogravimetric analysis. The an- 3. The levels of concentration of Ca and Sr display an in- alytical results are given in Table 1. verse correlation towards Ba, and clearly demonstrate the Backscatter images show that heulandite from the North- complete solid solution of these elements in heulandite. ern Ravnås prospect is compositionally heterogeneous with The reliability of the zeolite compositions is demonstrat- an irregular patchy or mottled structure. The mean chemical ed by the balance errors E which are < 10% for all the ana- composition (14 analysis points) of the dominating part of lysed heulandites (Alietti et al., 1977). the heulandite-Ba crystal individuals corresponds to the for- The thermogravimetric analysis of heulandite-Ba from mula (based on 72 framework oxygen atoms) (Ba2.49Ca1.41 the Northern Ravnås prospect was done using a Mettler Sr0.30K0.37Na0.33) 7 4.90Al8.96Si27.00O72.00·21.75H2O(Table1, TG50 instrument linked to a Mettler M3 microbalance. Ni- analysis R1), with a compositional range (anhydrous) vary- trogen was used as a purge gas, with a flow rate of 100 mL/ ing from (Ba2.76Ca1.26Sr0.29Na0.40K0.34) 7 5.05Al8.81Si27.05O72.00 min. Finely ground sample (7.3650 mg) was heated from to (Ba2.20Ca1.40Sr0.45K0.34Na0.25) 7 4.64Al8.64Si27.34O72.00 (Table 35 °C to 1000 °C at a heating rate of 40 °C per minute. The 1, analyses R2 and R3). Minor parts of the crystals are de- thermogravimetric curve is shown in Fig. 4. Weight loss oc- pleted in barium, with typical chemical composition (Ba1.88 curred in three steps: 1) 35–250 °C, 8.2 wt.% loss, corre- Ca1.80Mg0.34Sr0.32K0.45Na0.21) 7 5.00Al8.96Si26.94O72.00 (Table 1, sponding to 13.6 H2O, 2) 250–520 °C, 3.4 wt.% loss, corre- analysis R4). A relatively small part, which make up the sponding to 5.6 H2O, and 3) 520–815 °C, 1.5 wt.% loss, cor- crystal core, is actually a strontian barian heulandite-Ca respondingto2.5H2O. A total of 13.1 wt.% loss occurred with chemical composition (Ca2.15Sr1.47Ba0.60Na0.71 between 35 °C and 1000 °C, and this value is assumed as the K0.19) 7 5.12Al9.09Si26.84O72.00 (Table 1, analysis R5). total amount of water in heulandite-Ba. Above 900 °C there Heulandite from the Bratteskjerpet mine is chemically was a minor, but gradual loss of weight, probably due to heterogeneous, shown by backscatter images as a mottled evaporation of alkali. The trippel-stage decomposition of

150 7.13. X-ray crystallography and crystal structure determination

Fig. 4. Thermogravimetric curve of heulandite-Ba.

Fig. 5. Infrared spectrum of heulandite-Ba. heulandite-Ba, caused by differences in water bonding, is Northern Ravnås prospect were obtained using a Philips typical to that of many heulandites (Gottardi & Galli, 1985). X’pert diffractometer equipped with automatic divergence [ Heulandite-Ba from the Northern Ravnås prospect was slits and diffracted-beam graphite monochromator (CuK 1 handpicked under a binocular microscope. The material was radiation, † = 1.54056 Å). Data were collected from 3° to treated with diluted hydrochloric acid in order to remove 70° 2 ’ in steps of 0.01° 2 ’ and 5 s counting time per step. traces of calcite. Carefully washed and air dried material Si (NBS 640a) was used for calibration of the diffractome- was ground, pressed into a KBr pellet, and the infrared spec- ter. The X-ray powder diffraction data is shown in Table 2. trum was recorded over the region 400–4000 cm-1 using a Indexing and least squares refinement were done by the Perkin Elmer S-2000 FT-IR spectrometer (Fig. 5). The program CELREF (Laugier & Bochu, 1999). The unit cell spectrum shows broad absorption bands at 3603 cm-1 and dimensions found are a = 17.762(3) Å, b = 17.904(2) Å, 3459 cm-1 (O-H stretching), and a sharp absorption band at c = 7.422(1) Å, q = 116.49(1)°, and V = 2112.5 Å3,which 1631 cm-1 (H-O-H bending). Bands due to absorption by te- is slightly different from the unit cell dimensions found by trahedral bonds appear at (cm-1, w – weak, m – medium. s – the crystal structure refinement: a = 17.738(3) Å, b = strong, b – broad) 1192 w, 1030 sb, 778 w, 718 m, 661 w, 17.856(2) Å, c = 7.419(1) Å, q = 116.55(2)°, and V = 597 m, 519 w and 456 m. 2102.0 Å3. The differences are probably due to variations in chemical composition among the material used for the investigations. Calculation of a powder diffraction pattern X-ray crystallography and crystal structure using the program POWDERCELL (Kraus & Nolze, determination 1996) shows that heulandite-Ba has ' 130 lines with intensity >1 in the region 5–70° 2 ’ , most of them above 35° 2 ’ . Experimental procedures and results Obtaining a high quality X-ray powder pattern by a conventional Bragg-Brentano diffractometer is therefore X-ray powder diffraction data on heulandite-Ba from the problematic because of insufficient resolution. The pow-

151 7. Heulandite-Ba, a new zeolite species from Norway

Table 2. X-ray powder diffraction data for heulandite-Ba. Id(obs.) d(calc.) hkl Id(obs.) d(calc.) hkl 4 2.077 2.076 4 6 1 26 8.946 8.952 0 2 0 6 2.058 2.057 1 1 3 66 7.941 7.948 2 0 0 11 2.024 2.024 -6 6 2 10 6.846 6.853 -1 1 1 6 1.968 1.968 -8 4 1 7 6.640 6.643 0 0 1 9 1.959 1.959 -1 5 3 24 5.942 5.944 2 2 0 8 1.939 1.940 8 2 0 5 5.414 5.417 -2 2 1 8 1.923 1.923 2 0 3 17 5.266 5.267 -3 1 1 5 1.873 1.871 -3 9 1 59 5.116 5.117 1 1 1 4 1.854 1.855 -4 0 4 28 5.085 5.081 3 1 0 2 1.839 1.839 -5 1 4 66 4.650 4.650 -1 3 1 6 1.829 1.830 -7 5 3 10 4.480 4.476 0 4 0 5 1.819 1.819 5 3 2 26 4.378 4.377 -4 0 1 7 1.791 1.790 0 10 0 97 3.978 3.980 1 3 1 9 1.776 1.777 5 7 1 16 3.904 3.900 2 4 0 13 1.770 1.770 -7 7 2 14 3.838 3.839 2 2 1 5 1.729 1.729 6 0 2 18 3.738 3.739 -2 4 1 8 1.700 1.699 -10 0 3 18 3.711 3.712 0 4 1 7 1.672 1.672 -5 9 2 5 3.633 3.632 4 2 0 4 1.654 1.654 -4 8 3 48 3.564 3.565 -3 1 2 7 1.617 1.616 2 6 3 17 3.484 3.482 -5 1 1 6 1.594 1.594 3 5 3 42 3.428 3.427 -2 2 2 5 1.565 1.565 10 2 0 22 3.403 3.402 -4 0 2 4 1.553 1.553 -6 6 4 17 3.323 3.321 0 0 2 5 1.523 1.523 7 3 2 56 3.181 3.180 -4 2 2 7 1.495 1.495 9 3 1 45 3.131 3.130 5 1 0 5 1.486 1.486 -3 11 2 24 3.074 3.075 -1 3 2 5 1.459 1.458 2 8 3 26 3.038 3.040 -5 1 2 4 1.454 1.454 -7 9 3 25 2.994 2.992 3 3 1 5 1.439 1.439 -5 11 2 100 2.973 2.974 1 5 1 6 1.417 1.417 -2 0 5 6 2.890 2.890 4 0 1 3 1.400 1.400 -2 8 4 4 2.856 2.856 -2 4 2 3 1.393 1.393 10 2 1 65 2.807 2.807 -6 2 1 3 1.388 1.387 -12 4 3 29 2.734 2.733 -2 6 1 6 1.371 1.371 -9 7 4 10 2.667 2.667 0 4 2 8 1.358 1.357 -7 5 5 7 2.636 2.633 -6 2 2 4 1.345 1.346 -10 2 5 7 2.557 2.559 2 2 2 5 2.539 2.541 6 2 0 5 2.523 2.525 1 7 0 der diffraction pattern in Table 2 reports only well discern- 6 2.495 2.496 -7 1 1 ible lines. 6 2.488 2.487 3 5 1 The crystal structure of heulandite-Ba from the Northern 10 2.465 2.466 -4 6 1 Ravnås prospect was studied by single-crystal X-ray dif- 7 2.442 2.442 2 6 1 fraction on an Enraf-Nonius CAD4 diffractometer with 18 2.429 2.430 -7 1 2 graphite-monochromated MoK [ radiation at 293 K. Data 5 2.383 2.386 4 6 0 reduction, including background and Lorentz-polarization 6 2.368 2.368 -4 2 3 corrections and an empirical absorption correction based on 10 2.348 2.347 -2 2 3 ^ 4 2.325 2.325 -2 6 2 -scans, was performed using the Enraf Nonius SDP program library (Enraf Nonius, 1983). The structure was re- 6 2.309 2.307 1 7 1 2 6 2.301 2.303 3 7 0 fined on F by least squares method using the program 5 2.293 2.293 2 4 2 SHELXTL (Sheldrick, 1997) in space group C2/m and the 6 2.284 2.284 -3 3 3 subgroups C2, Cm,andC1. Refinements in space groups of 7 2.242 2.244 -4 6 2 lower symmetry, however, did not improve the accuracy of 6 2.236 2.238 0 8 0 the structure model. Refinement in C2/m yielded an R1val- 3 2.210 2.208 3 3 2 ue of 3.5 %. Detailed information on the data collection and 4 2.199 2.199 -6 2 3 structure refinement are summarized in Table 3. In the fol- 6 2.154 2.154 2 8 0 lowing description tetrahedral cation sites (Si, Al) are la- 2 2.137 2.137 -2 4 3 belled with T, framework oxygens with O, extra-framework 8 2.124 2.125 4 0 2 cations with the appropriate chemical symbol (Ba, Ca, Na, 5 2.116 2.115 -7 1 3 K), and oxygen atoms of extra-framework H2O molecules 16 2.093 2.093 6 2 1 with W.

152 7.13. X-ray crystallography and crystal structure determination

Table 3. Experimental details of the crystal structure refinement of Table 4. Final atomic positional parameters and Beq values (standard heulandite-Ba. deviations in parentheses) for heulandite-Ba.

˚ 2 Crystal size (mm) 0.300 × 0.125 × 0.250 Atom sof x/a y/b z/c Beq (A ) Composition refined (Ba2.20Ca1.57K1.61Na0.21) 7 5.59(Al,Si)36O72 T1 1 0.17934(4) 0.16850(4) 0.09567(9) 1.070(9) by X-ray data ·20.75H2O T2 1 0.28857(4) 0.08986(3) 0.50023(9) 1.088(9) Composition by EMP (ave- (Ba2.49Ca1.41Sr0.30K0.37Na0.33) 7 4.90Al8.96 T3 1 0.29214(4) 0.30922(4) 0.28287(9) 1.092(9) rage) Si27.00O72.00·21.75H2O T4 1 0.43568(4) 0.20123(3) 0.58874(9) 1.079(9) Space group C2/m T5 1 0 0.21324(5) 0 1.13(1) ˚ a (A) 17.738(3) O1 1 0.3047(2) 0 0.5494(4) 2.46(5) ˚ b (A) 17.856(2) O2 1 0.2684(1) 0.3798(1) 0.3889(3) 2.39(3) ˚ c (A) 7.419(1) O3 1 0.3171(1) 0.3492(1) 0.1177(3) 2.06(3) q (°) 116.55(2) O4 1 0.2385(1) 0.1047(1) 0.2545(3) 2.11(3) ˚ 3 V (A ) 2102.0(7) O5 1 0.5 0.1733(2) 0.5 2.45(5) ’ max (°) 29.975 O6 1 0.0821(1) 0.1586(1) 0.0636(3) 1.83(3) hkl (min., max.) -24 e h e 23, -25 e k e 25, -1 e l e 10 O7 1 0.3745(2) 0.2660(1) 0.4498(3) 3.09(4) K U Scan type 1.5° + 0.35 · tan O8 1 0.0086(1) 0.2677(1) 0.1850(3) 2.50(3) Measured reflections 7423 O9 1 0.2115(1) 0.2531(1) 0.1753(3) 2.13(3) Unique reflections 3165 O10 1 0.3839(1) 0.1275(1) 0.5997(3) 2.25(3) Rint (%) 1.83 Ba1 0.359(2) 0.35504(5) 0.5 0.3237(1) 2.48(1) c Obs. reflections (I 8 2 (I)) 2731 Ba2 0.190(6) 0.2528(3) 0.5 0.0768(7) 2.67(6) R1 (%) 3.54 Ca 0.392(8) 0.4584(1) 0 0.8014(6) 2.20(4) wR2 (%) 9.76 K 0.41(2) 0.219(1) 0.5 -0.013(3) 5.9(2) GooF 1.195 Na 0.05(1) 0.463(2) 0 0.725(9) 1.9(9)* W1 0.75(2) 0.5774(3) 0.0816(2) 0.9728(7) 3.91(8) W2 0.36(2) 0.405(2) 0.5 0.155(4) 15(2)* W3 1 0.5 0 0.5 7.5(2) W4 0.641(2) 0.425(1) 0.5 0.091(4) 8.4(4) The refined chemical composition (Ba2.20Ca1.57K1.61 Na ) (Al,Si) O ·21.15H O approximately agrees W5 0.5 0.4824(6) 0.4051(4) 0.507(4) 7.6(2) 0.21 7 5.59 36 72 2 W6 0.72(2) 0.4074(6) 0.5 0.719(2) 9.2(3) with the results of the chemical analysis, but in contrast to W7 0.25(2) 0.605(1) 0.091(1) 0.922(3) 6.0(5)* the electron microprobe analysis no Sr was found. Instead W8 0.15(1) 0.385(2) 0.5 0.481(5) 4.74* higher populations of the Ca, and in particular K sites, were refined, which is probably caused by partial, but highly dis- Starred atoms were refined isotropically. Anisotropically refined ordered, presence of Sr on the same sites. The sum of elec- atoms are given in the form of the isotropic equivalent thermal parameter defined as B =8/3 ‘ 2 7 ( 7 (U a *a*a a )). trons of the extra-framework sites (XRD: 187.5 e- pfu, EM- eq i j ij i j i j PA: 189.7 e - pfu), however, is almost identical, and thus XRD and EMP experiments are in good agreement. Extra-framework cations were distinguished from H2O axis, (ii) an eight-membered B channel parallel to the c axis, molecules on the basis of small atomic displacement param- (iii) eight-membered C channels parallel to the a axis and eters, as well as appropriate bonding distances to framework [102]. The structure of heulandite-Ba was refined in the oxygens. From various studies describing the cation distri- monoclinic space group C2/m, and the data set shows an ac- bution in natural heulandite/clinoptilolite (Armbruster & ceptable quality compared to heulandite structure refine- Gunter, 2001), we know that the preferred location of a cat- ments by Yang & Armbruster (1996) and Stolz & Armbrus- ion in the pore system depends on its ionic radius, therefore ter (2000). Atom coordinates, occupancies, and Beq values, the A channel generally contains a Na site (ionic radius 1.16 as well as anisotropic displacement parameters are listed in Å in octahedral coordination) close to the channel wall. This Table 4 and Table 5, respectively. site often also contains Ca (ionic radius 1.14 Å). The Ca site For this study, the comparison of our natural heulandite- in the B channel is usually Na free, and a K site is located Ba with the structure refinement of a Ba-exchanged clinop- close to the intersection of the A and the C channel. This in- tilolite performed by Petrov et al. (1985) is of particular information, as well as the similarity of the ionic radii of K and terest. The Ba content in the exchanged sample is 2.54 Ba at- Ba, led us to the assumption of a preferred accumulation of oms per formula unit (pfu) normalized to 72 O, and 2.49 Ba Ba in the centre of the C ring, and served as an initial model atoms pfu in our natural sample. Although the cell dimen- for the extra-framework cation arrangement. sions of our sample deviate slightly from both the original and exchanged material reported by Petrov et al. (1985), the cell volumes coincide within two standard deviations. The tetrahedral framework To identify T-sites with high Al concentration we use the average bond length in tetrahedra from central cation to li- The tetrahedral framework of heulandite-Ba shows the gands (Si-O = 1.61 Å, Al-O = 1.75 Å) (Kunz & Armbruster, common HEU topology forming a two-dimensional system 1990; Alberti & Gottardi, 1988). As shown in Table 6, T2 of connected cavities, which leads to three types of structur- has the highest Al concentration, while T4 has almost ideal al channels: (i) a ten-membered A channel parallel to the c Si-O distances. Therefore it is to be expected that extra-

153 7. Heulandite-Ba, a new zeolite species from Norway

Table 5. Anisotropic atomic displacement parameters (standard deviations in parentheses) for heulandite-Ba.

Atom U11 U22 U33 U12 U13 U23 T1 0.0112(3) 0.0171(3) 0.0122(3) -0.0007(2) 0.0051(2) 0.0007(2) T2 0.0152(3) 0.0120(3) 0.0143(3) 0.0001(2) 0.0069(2) 0.0001(2) T3 0.0146(3) 0.0152(3) 0.0124(3) 0.0006(2) 0.0066(2) 0.0006(2) T4 0.0120(3) 0.0163(3) 0.0121(3) -0.0006(2) 0.0049(2) 0.0003(2) T5 0.0117(3) 0.0179(4) 0.0124(4) 0 0.0045(3) 0 O1 0.041(2) 0.015(1) 0.027(1) 0 0.006(1) 0 O2 0.032(1) 0.034(1) 0.033(1) -0.0030(8) 0.0212(9) -0.0094(8) O3 0.035(1) 0.030(1) 0.0204(8) -0.0051(8) 0.0182(8) -0.0017(7) O4 0.0275(9) 0.029(1) 0.0191(8) 0.0087(8) 0.0068(7) 0.0044(7) O5 0.035(1) 0.034(2) 0.038(2) 0 0.028(1) 0 O6 0.0153(7) 0.0250(9) 0.0290(9) 0.0004(7) 0.0099(7) 0.0022(7) O7 0.036(1) 0.037(1) 0.035(1) 0.016(1) 0.0078(9) 0.017(1) O8 0.0268(9) 0.037(1) 0.0251(9) 0.0000(8) 0.0065(8) -0.0144(8) O9 0.0227(9) 0.0248(9) 0.035(1) -0.0091(7) 0.0147(8) -0.0090(8) O10 0.0237(9) 0.027(1) 0.033(1) -0.0082(8) 0.0114(8) 0.0004(8) Ba1 0.0362(4) 0.0228(3) 0.0351(4) 0 0.0156(3) 0 Ba2 0.041(2) 0.0143(9) 0.057(2) 0 0.031(1) 0 Ca 0.018(1) 0.031(1) 0.028(2) 0 0.0047(8) 0 K 0.065(5) 0.061(3) 0.108(8) 0 0.047(6) 0 W1 0.049(2) 0.050(2) 0.048(2) 0.000(2) 0.020(2) -0.003(2) W3 0.098(6) 0.046(4) 0.151(9) 0 0.063(6) 0 W4 0.09(1) 0.072(8) 0.15(1) 0 0.044(9) 0 W5 0.060(8) 0.058(4) 0.134(7) -0.007(4) 0.01(1) 0.002(6) W6 0.078(6) 0.15(1) 0.124(9) 0 0.046(6) 0

Table 6. Interatomic bond distances (A˚ ) in Si, Al tetrahedra of heu- (Stolz et al., 2000). However, Yang & Armbruster (1996) landite-Ba. and Stolz et al. (2000) reported that this kind of symmetry lowering can only be detected in homoionic cation ex- T1 – T4 – changed structures. In heulandite-Ba there is no indication O3 1.643(2) O5 1.629(1) of symmetry lowering relative to space group C2/m. O4 1.638(2) O7 1.605(2) Distortions of the framework are commonly described by O6 1.641(2) O8 1.618(2) O9 1.629(2) O10 1.628(2) angles between corner sharing tetrahedra. T-O-T connec- Mean 1.638(2) Mean 1.620(2) tions are stable in the range from approximately 135° to 170° (Gibbs, 1982). At lower angles the repulsive forces be- T2 – T5 – tween T sites become too strong to form a stable framework. O1 1.642(1) O6 1.637(2) In heulandite-Ba all angles are between 136.0° and 161.9° O2 1.654(2) O6 1.637(2) (Table 7). O4 1.654(2) O8 1.633(2) O10 1.656(2) O8 1.633(2) Mean 1.652(2) Mean 1.635(2) The extra-framework sites

T3 – The extra-framework cation arrangement is directly related O2 1.637(2) to the Si, Al distribution on tetrahedral sites. As expected O3 1.641(2) from the high Al concentration on T2, three cation sites, O7 1.627(2) Ba1, Ba2, and K, were found near the centre of the C ring O9 1.633(2) (Fig. 6 and 7). All detected cation sites lie on the mirror Mean 1.635(2) plane. Ba1, with the highest Ba population of 36 %, bonds to four framework oxygen atoms and is coordinated by the four highly populated H2O sites W4, W5 (2×), and W6 (Fig. framework cations preferentially bond to T2 ligands to com- 7, left). At a distance from 1.92 Å to Ba1 a less populated pensate the charge imbalance caused by the substitution of site Ba2 (19 %) is located close to the centre of the C ring as Al3+ for Si4+. Armbruster (2001) reported that asymmetric well (Fig. 7, middle). This site bonds to five tetrahedral distribution of Al is also reflected in asymmetric distribu- framework oxygens and is also coordinated by four less tion of extra-framework cations, and although the Si, Al ar- populated H2OsitesW4,W7orW1(2×),andW8.Bonding rangement is usually not sufficient to influence the symme- distances range from 2.54(2) to 3.08(1) Å for Ba1, and from try of the HEU framework, asymmetric distribution of ex- 2.86(2) to 3.20(2) Å for Ba2, respectively, and thus lie in the tra-framework cations can induce a symmetry lowering acceptable range for Ba-O bond lengths. A third site la-

154 7.13. X-ray crystallography and crystal structure determination

Table 7. T-O-T angles in heulandite-Ba.

T – O – T angle (°) T2 – O1 – T2 155.5(2) T2 – O2 – T3 146.8(1) T1 – O3 – T3 140.9(1) T1 – O4 – T2 139.5(1) T4 – O5 – T4 144.4(2) T1 – O6 – T5 136.0(1) T3 – O7 – T4 161.8(2) T4 – O8 – T5 149.4(1) T1 – O9 – T3 146.6(1) T2 – O10 – T4 142.5(1) Fig. 8. A polyhedral representation of the structure of Ba-exchanged clinoptilolite described by Petrov et al. (1985) is shown in a projection parallel to [001]. Ions drawn in broken lines (Na and Mg) represent sites with occupancies close to the detection limit (symmetric equivalent sites are only labelled once). All Ba is located in a cluster

in the C ring (K + Ba1, Ba2). The fact that less H2O sites were found by Petrov et al. can probably be explained by the lower resolution of powder X-ray structure refinements compared to single-crystal X-ray data.

Table 8. Interatomic bond distances (A˚ ) of the extra-framework cation sites in heulandite-Ba.

Ba1– Ca– Fig. 6. The A (left) and B (right) channel of heulandite-Ba shown in W4 2.54(2) W1 (2x)* 2.407(5) a polyhedral representation of the HEU structure projected along W5 (2x) 3.08(1) W1 (2x) 2.507(7) [001]. Ba1, Ba2, and K are located close to T2, the tetrahedron with W6 2.65(1) W3 2.649(5) the highest Al concentration. Large spheres represent extra-frame- O2 (2x) 2.804(2( W7 (2x)* 2.85(2) work cations, small spheres are H2O molecules. O3 (2x) 3.020(2) O1 2.525(4) O10 (2x) 2.720(3) Ba2 – W1 (2x)* 3.20(2) Na – W4 3.01(2) W1 (2x)* 2.50(3) W7 (2x)* 2.86(2) W3 2.04(7) W8 2.86(3) W7 (2x)* 2.79(3) O2 (2x) 3.077(4) O1 2.52(3) O3 (2x) 2.888(3) O10 (2x) 2.62(2) O4 3.147(3)

K– W1 (2x)* 2.867(9) W2 2.95(4) Fig. 7. The three cation site Ba1, Ba2 and K are located near the cen- W4 3.38(2) tre of the C ring, here shown in a projection parallel [101]. Each site W7 (2x)* 2.48(2) bonds to framework oxygens, preferentially to vertice of the Al-rich O3 (2x) 3.116(9) T2 tetrahedron, and is irregularly coordinated by H2O molecules. O4 (2x) 2.910(8) * Starred bonds do not occur simultaneously. belled K is also part of the cluster at the intersection of the A and C channel (Fig. 7, right). It contains K, Sr, and minor The cation composition in the B channel agrees with other Ba, and bonds to four framework oxygens and to additional structure refinements of natural heulandites/clinoptiloli- H2O molecules. Atomic displacement parameters and their tes (Armbruster & Gunter, 1991): Two unique sites labelled standard deviations of this site are significantly higher com- Ca and Na are occupied by the corresponding elements to pared to all other extra-framework cation sites (see Table 5), 39 % and 5 %, respectively. The distance between the two indicating overlap of different atomic species such as Sr, K, positions amounts to 0.61 Å, thus neighbouring sites cannot and Ba. This fact was not considered in the structure refine- be occupied simultaneously. Although Armbruster & Gunter ment, instead, the K site was treated as if it was partially oc- (2001) reported that Na in the B channel is very uncommon, cupied by K+. Since minor Sr and Ba share the site, the re- small amounts are reasonable due to the very similar atomic finement results in a too high concentration of K. radii of Na and Ca. Na sites in the B channel are also known

155 7. Heulandite-Ba, a new zeolite species from Norway

from Na-exchanged heulandites (Yang & Armbruster, Armbruster, T. (2001): Clinoptilolite-heulandite: applications and 1996). Both Ca and Na form three bonds to framework oxy- basic research. in "Studies in surface science and catalysis, 135. st gens and to the partially occupied H2O sites W1, W7, and Zeolites and mesoporous materials at the dawn of the 21 centu- W3, the latter being situated on the intersection of the mirror ry”, Elsevier Science B. V., 13-27. Armbruster, T. & Gunter, M.E. (1991): Stepwise dehydration of heu- plane and the twofold axis. The distance between the H2O sites W1 and W7, with occupancies 75 % and 25 %, respec- landite-clinoptilolite from Succor Creek, Oregon, U.S.A.: A sin- tively, amounts to 0.74 Å and again inhibits simultaneous gle-crystal X-ray study at 100 K. Am. Mineral., 76, 1872-1883. occupancy of both sites. The two sites, however, can be re- –, – (2001): Crystal structures of natural zeolites. in “Natural zeolites: occurrence, properties, applications”, Reviews in mineralo- garded as a fully occupied H2O site that is split into two parts. A list of interatomic bonding distances of the extra- gy and geochemistry, vol. 45, Mineralogical Society of America, 1-67. framework cation sites to the coordinating oxygen atoms is Bancroft, P., Nordrum, F.S., Lyckberg, P. (2001): Kongsberg revisit- shown in Table 8. The extra-framework sites Na, W2, W7, ed. Miner. Rec., 32, 181-205. and W8 were refined isotropically due to their low popula- Berg, B.I. (1998): Gruveteknikk ved Kongsberg Sølvverk 1623– tions. 1914. Senter for teknologi og samfunn, report 37, 586 p. Prior to this study, heulandites/clinoptilolites with high Boles, J.R. (1972): Composition, optical properties, cell dimensions, Ba concentrations were only known from cation-exchange and thermal stability of some heulandite group zeolites. Am. Min- experiments. The structure of an exchanged clinoptilolite eral., 57, 1463-1493. sample with the formula (Ba2.54Ca0.42Mg0.18K0.34Na0.10) 7 3.58 Brooke, H.J. (1822): On the comptonite of Vesuvius, the brewsterite (Al6.20Si29.62) 7 35.82O72·19.7H2O was refined from powder X- of Scotland, the stilbite and the heulandite. Edinb. Phil. J., l6, ray diffraction data and described by Petrov et al. (1985) and 112-115. shown in Fig. 8. Surprisingly, our natural sample has almost Bugge, C. (1917): Kongsbergfeltets geologi. Nor. Geol. Unders., 82, the same Ba concentration (2.49 Ba pfu). The cation distri- 1-272. bution in the channels is nearly identical: In both samples all Cˇ erny´, P. & Povondra, P. (1969): A polycationic, strontian heulan- Ba is accumulated in the centre of the C ring distributed dite; comments on crystal chemistry and classification of heulan- among three cation sites, one of which is shared by two ele- dite and clinoptilolite. N. Jb. Miner. Mh., 1969, 349-361. ments. Our K site (containing K, Sr, and minor Ba) is occu- Coombs, D.S., Alberti, A., Armbruster, T., Artioli, G., Colella, C., pied by Ba in the exchanged sample and located closer to the Galli, E., Grice, J.D., Liebau, F., Mandarino, J.A., Minato, H., centre of the B channel. Our Ba2 site is shared by Ba and K, Nickel, E.H., Passaglia, E., Peacor, D.R., Quartieri, S., Rinaldi, R., Ross, M., Sheppard, R.A., Tillmanns, E., Vezzalini, G. and our Ba1 site contains only minor amounts of Na in the (1997): Recommended nomenclature for zeolite minerals: Re- structure described by Petrov et al. (1985). The weakly oc- port of the subcommittee on zeolites of the International Mineral- cupied Na site in the B channel is not present in the ex- ogical Association, Commission on New Mineral and Mineral changed sample. Instead a very low-populated Mg site in Names. Can. Mineral., 35, 1571-1606 (also published (1998): the centre of the A channel (on the intersection of the two- Miner. Mag., 62, 533-571). fold axis and the mirror plane) was found. In general, the lo- Enraf Nonius (1983): Structure determination package (SDP). Enraf cation of the cation sites of both structures roughly agrees, Nonius, Delft, The Netherlands. apart from the very low-populated sites close to the detec- Gibbs, G.V. (1982): Molecules as models for binding in silicates. tion limit. Differences are found in the distribution of the Am. Mineral., 67, 421-450. elements on the clustered sites in the C ring. Gottardi, G. & Galli, E. (1985): Natural zeolites. Springer-Verlag, Berlin, Heidelberg, New York, Tokyo, 409 p. Acknowledgements: We are indebted to P. Berget for pro- Helleberg, O.A. (2000): Kongsberg Sølvverk 1623-1958. Forlaget viding samples from the Sjoa locality, and H. Kristiansen Langs Lågen, Kongsberg, 374 p. and B. Jacobsen for providing samples from Vinoren. We Ihlen, P.M. (1986): The metallogeny of the Kongsberg district. Sver. thank I. Liberger, A. Henriksen and T. Bach, all at Norsk Geol. Unders., Ser. Ca, 59, 30-32. Hydro ASA, Research Centre Porsgrunn, for carrying out Jacobsen, S.B. & Heier, K.S. (1978): Rb-Sr isotope systematics in the TG analysis, the IR spectrometric analysis and the SEM metamorphic rocks, Kongsberg sector, south Norway. Lithos, 11, images, respectively. Nicola Döbelin and Thomas Armbru- 257-276. Johnsen, O. (1986): Famous mineral localities: the Kongsberg silver ster acknowledge support from the Swiss National Science mines, Norway. Miner. Rec., 17, 19-36. Foundation (Grant 20-65084.01 to TA: crystal chemistry of – (1987): Silber aus Norwegen. Zur Bergbaugeschichte und über die minerals). We thank A. Alberti, an anonymous referee, as- Mineralienschätze. Emser Hefte, 8,1-48. sociate editor E. Passaglia and chief editor R. Altherr for Kraus, W. & Nolze, G. (1996): POWDERCELL – a program for the their comments. representation and manipulation of crystal structures and calculation of the resulting X-ray powder patterns. J. Appl. Cryst., 29, 301-303. References Kunz, M. & Armbruster, T. (1990): Difference displacement parameters in alkali feldspars – Effects of (Si,Al) order-disorder. Am. Alberti, A. & Gottardi, G. (1988): The determination of the Al-content Mineral., 75, 141-149. in the tetrahedra of framework silicates. Z. Kristallogr., 184, 49-61. Larsen, A.O., Nordrum, F.S., Erambert, M. (2003): Mineraler i Alietti, A., Brigatti, M.F., Poppi, L. (1977): Natural Ca-rich clinopti- brewsterittserien fra norske lokaliteter. Norsk Bergverksmuseum lolites (heulandites of group 3): new data and review. N. Jb. Min- Skrift, 25, 41-42. er. Mh., 1977, 493-501. Laugier, J. & Bochu, B. (1999): CELREF: Cell parameters refine-

156 7.15. References

ment program from powder diffraction diagram. Laboratoire des nes Landssammenslutnings Industrigruppe, Bergforskningen Mater´ iaux et du Geni´ e Physique, Ecole Nationale Super´ ieure de (abstract), 100. Physique de Grenoble (INPG), Grenoble, France. – (2000): Native silver from Kongsberg. in “Highlights, selected at- Lovisato, D. (1897): Notizia sopra una heulandite baritica di Pula tractions, Natural History Museums and Botanical Garden” , E. con acceno alle zeoliti finora trovate in Sardegna. Atti della Reale Roaldset, & S.-E. Sjulsen, eds. University of Oslo, 34-39. Acc. Lincei, Rend., 5. Ser., 6, 260-264. Sheldrick, G.M. (1997): SHELX-97: Program for crystal structure Lucchetti, G., Massa, B., Penco, A.M. (1982): Strontian heulandite refinement. Univ. Göttingen, Germany. from Campegli (Eastern Ligurian ophiolites, Italy). N. Jb. Miner. Siedlecka, A., Nystuen, J.P., Englund, J.O., Hossack, J. (1987): Lil- Mh., 1982, 541-550. lehammer, geological map 1:250000. Norges Geologiske Un- Mandarino, J.A. (1981): The Gladstone-Dale relationship: Part IV. dersøkelse, Trondheim. The compatibility concept and its application. Can. Mineral., 19, Starmer, I.C. (1985): The geology of the Kongsberg district and the 441-450. evolution of the entire Kongsberg sector, South Norway. Nor. Ge- Medenbach, O. (1985): A new microrefractometer spindle-stage and ol. Unders., Bull., 401, 35-58. its application. Fortschr. Miner., 61, 111-133. Stolz, J. & Armbruster, T. (2000): Mg2+,Mn2+,Cd2+,Sr2+,andCu2+ Miller, B.E. & Ghent, E.D. (1973): Laumontite and barian-strontian exchange in heulandite single crystals: X-ray structure refine- heulandite from the Blairmore Group (Cretaceous), Alberta. ments. in “Natural zeolites for the third millennium”, C. Colella Can. Mineral., 12, 188-192. & F.A. Mumpton, eds. De Frede Editore, Napoli, 119-138. Mnatsakanyan, A.Kh., Khurshudyan, E.Kh., Revazova, N.V. Stolz, J., Yang, P., Armbruster, T. (2000): Cd-exchanged heulandite: (1970): Zeolites from the Upper Cretaceous volcanic formations symmetry lowering and site preference. Microporous and Meso- in the northeastern part of the Armenian SSR. Zap. Arm. Otd. porous Materials, 37, 233-242. Vses. Mineral. Obshchest., 4, 141-160 (in Russian). Strand, T. (1951): The Sel and Vågå map areas. Nor. Geol. Unders., Moen, K. (1967): Kongsberg Sølvverk 1623-1957. Universitetsfor- 178, 1-117. laget, Oslo, 482 p. – (1967): Stratigraphy and structure of Eocambrian and younger de- Neumann, H. (1944): Silver deposits at Kongsberg. Nor. Geol. Un- posits in a part of the Gudbrandsdal valley district, South Nor- ders., 162, 1-133. way. Nor. Geol. Unders., 251, 93-106. Nordrum, F.S., Larsen, A.O., Erambert, M. (2003): Minerals of the Tröger, W.E. (1982): Optische Bestimmung der Gesteinsbildenden heulandite series in Norway – a progress report. Norsk Berg- Minerale, Teil 1 Bestimmungstabellen, 5. neubearbeitete Auflage verksmuseum Skrift, 25, 51-62. von H.U. Bambauer, F. Taborszky, H.D. Trochim. E. Schweizer- Ogawa, T. (1967): On the varieties of heulandites. J. Sci. Hiroshima bart’sche Verlagsbuchhandlung (Nägele und Obermiller), Stutt- Univ., Ser. C (Geol. Mineral.), 5, 267-286. gart, 188 p. Palmer, J.L. & Gunter, M.E. (2000): Optical properties of natural Yang, P. & Armbruster, T. (1996): Na, K, Rb, and Cs exchange in and cation exchanged heulandite group zeolites. Am. Mineral., heulandite single- crystals: X-ray structure refinements at 100 K. 85, 225-230. J. Solid State Chem., 123, 140-149. Petrov, O.E., Filizova, L.D., Kirov, G.N. (1985): Cation distribution in the clinoptilolite structure: Ba-exchanged sample. Compt. Rend. l’Acad. Bulg. Sci., 38, 603-606. Received 17 November 2003 Segalstad, T.V. (1985): Sølvdannelsen i Kongsberg sølvforekomst. Modified version received 22 June 2004 Ore geology symposium “Nye malmtyper i Norge”, Bergverke- Accepted 11 November 2004

157 7KHFU\VWDOVWUXFWXUHRISDLQLWH&D=U%>$O2@UHYLVLWHG

7+20$6$50%5867(5 1,&2/$'g%(/,1$'2/)3(5(77,'(7/()*h17+(5(5,&5(866(5 $1'%(51$5'*52%e7<

/DERUDWRULXPIUFKHPLVFKHPLQHUDORJLVFKH.ULVWDOORJUDSKLH8QLYHUVLW\RI%HUQ)UHLHVWUDVVH&+%HUQ6ZLW]HUODQG *56*HPUHVHDUFK6ZLVVODE/WG+LUVFKPDWWVWUDVVH&+/X]HUQ6ZLW]HUODQG /DERUDWRU\RI,QRUJDQLF&KHPLVWU\³(OHPHQWDODQG7UDFH$QDO\VLV(7++|QJJHUEHUJ+&,*&+=ULFK6ZLW]HUODQG ,QVWLWXWHRI0LQHUDORJ\DQG3HWURJUDSK\6RQQHJJVWUDVVH(7+=HQWUXP&+=ULFK6ZLW]HUODQG 'HSDUWPHQWRI*HRVFLHQFHV8QLYHUVLW\RI)ULERXUJ3pUROOHV&+)ULERXUJ6ZLW]HUODQG

$%675$&7 7KHFU\VWDOVWUXFWXUHRIWKHUDUHKH[DJRQDOPLQHUDOSDLQLWH>D F c@IURP

0RJRN 0\DQPDU ZLWKWKHLGHDOFRPSRVLWLRQ&D=U%>$O2@ZDVUHGHWHUPLQHGE\VLQJOHFU\VWDO

;UD\GLIIUDFWLRQ6WUXFWXUHUHÀQHPHQWVZHUHSHUIRUPHGLQVSDFHJURXSV3PDQG37KHFHQ

WURV\PPHWULF3PPRGHO\LHOGHGH[FHOOHQWDJUHHPHQW 5 UHÁHFWLRQV!T,REV SDUDPHWHUV ZLWKWKHREVHUYHGGLIIUDFWLRQGDWDZLWKRXWDQ\XQXVXDODWRPLFGLVSODFHPHQWSDUDPHWHUV

WKXVWKHDFHQWULF3PRGHOZDVUHMHFWHG$SUHYLRXVVWUXFWXUDOVWXG\FODLPHGWKDWSDLQLWHZDVQRQ

FHQWURV\PPHWULFDQGGLIIHUHGIURPWKHUHODWHGVWUXFWXUHVRIMHUHPHMHYLWH%> $O 2+ 2@DQG

ÁXRERULWH%>0J )2+ 2@LQKDYLQJORZHUV\PPHWU\

7KHVWUXFWXUHRISDLQLWHFRPSULVHVDIUDPHZRUNRI$O2RFWDKHGUDWKDWIHDWXUHVWZRW\SHVRIFKDQQHOV SDUDOOHOWRWKHFD[LV2QHFKDQQHOKDVDWULJRQDOFURVVVHFWLRQDQGLVRFFXSLHGE\WKUHHIROGFRRUGLQDWHG %DQG=ULQVL[IROGSULVPDWLFFRRUGLQDWLRQ7KHRWKHUFKDQQHOKDVDKH[DJRQDOFURVVVHFWLRQDQGLV RFFXSLHGE\&D&KHPLFDODQDO\VLVE\ODVHUDEODWLRQLQGXFWLYHO\FRXSOHGSODVPDPDVVVSHFWURPHWU\ LQGLFDWHGWKDWWKHFU\VWDOVWXGLHGKDVVLJQLÀFDQWVXEVWLWXWLRQRI1DIRU&D FD FKDUJHEDODQFHGE\ 7L UHSODFLQJRFWDKHGUDO$OOHDGLQJWRWKHIRUPXOD&D1D$O7L&U9=U+I%2

,1752'8&7,21 PDUEOHVDQGVFKLVWV7KLVLVDOVRWKHRULJLQRIDOOKLWKHUWRUHSRUWHG 3DLQLWH ZDV ÀUVW GHVFULEHG E\ &ODULQJEXOO*RUGRQ HW DO VSHFLPHQV7KHRULJLQDO3URWHUR]RLFVHGLPHQWVZHUHUHJLRQDOO\ DQGLVVWLOORQHRIWKHZRUOG·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ÀHG $UDNL ZKRGHVFULEHLWDVDKH[DJRQDO>D F WKURXJKWHVWLQJLQWKH*56*HPUHVHDUFKODERUDWRU\LQ%DQJNRN c@ULJLGDQGGHQVHRFWDKHGUDOIUDPHZRUNZLWKVSDFH 7KDLODQG DQGLVODEHOHGSDLQLWHQR ZRUOG·VODUJHVWJHP JURXS3GHYLDWLQJVOLJKWO\IURPWKHFHQWURV\PPHWULFVSDFH TXDOLW\IDFHWHGVDPSOH 3DLQLWHQRDDODUJHURXJKIUDJPHQW JURXS3P,QLWLDOO\0RRUHDQG$UDNL DVVXPHGVSDFH RIFWV GLPHQVLRQVttPP ZDVGLVFRYHUHG JURXS3PEXWDFRUUHVSRQGLQJUHÀQHPHQWFRQYHUJHGDW5 GXULQJDQH[SHGLWLRQ 3HUHWWL LQ0D\WR0RJRNFORVH \LHOGLQJXQH[SHFWHGO\ODUJHGLVSODFHPHQWSDUDPHWHUVIRU WRWKHW\SHORFDOLW\WKHSULYDWHJRYHUQPHQWOLFHQVHG2KQ*DLQJ WKH$OVLWHV %HT ²c 7KLVZDVWDNHQDVHYLGHQFHWKDW PLQHDW6DJDLQJ$VPDOOIUDJPHQW FW RIWKLVURXJKSDLQLWH WKHVWUXFWXUHZDVLQIDFWQRQFHQWURV\PPHWULFDQGDVXEVHTXHQW ZDVODEHOHGSDLQLWHQRE7KHGDWDSUHVHQWHGLQWKLVSDSHUZHUH UHÀQHPHQWLQVSDFHJURXS 3FRQYHUJHGDWOHDGLQJWR REWDLQHGIURPSDLQLWHQRV FKHPLVWU\ DQGE FKHPLVWU\DQG UHDVRQDEOHLVRWURSLFGLVSODFHPHQWSDUDPHWHUVIRUWKH$OVLWHV VWUXFWXUH ZKHUHDVWKHODUJHPRWKHUSLHFHUHPDLQHGLQ0RJRN )XUWKHUPRUH0RRUHDQG$UDNL VSHFXODWHGWKDWWKHUHODWHG 7KHW\SHORFDOLW\IRUSDLQLWHLVORFDWHGLQWKH0RJRN%HOWRI VWUXFWXUHRIMHUHPHMHYLWH$O %2 )ZLOODOVRUHYHDOVSDFHJURXS 3LQIXWXUHVWDWHRIWKHDUWVWUXFWXUHUHÀQHPHQWV6LQFH*ROR YDVWLNRYHWDO QHZVWUXFWXUHUHÀQHPHQWVRIMHUHPHMHYLWH (PDLOWKRPDVDUPEUXVWHU#NULVWXQLEHFK KDYHEHHQSXEOLVKHGE\6RNRORYDHWDO DQG5RGHOODVHW

158 8.18. Experimental methods

SURJUDPOLEUDU\ (QUDI1RQLXV 6LQJOHFU\VWDOVWUXFWXUHUHÀQHPHQWVXVLQJWKH DO \LHOGLQJ3PV\PPHWU\ZLWKRXWDQ\LQGLFDWLRQRI GHYLDWLRQIURPFHQWURV\PPHWU\7KXVWKHV\PPHWU\SUREOHPRI SURJUDP6+(/; 6KHOGULFN ZHUHGRQHLQVSDFHJURXSV3DQG3P +RZHYHUWKHH[FHOOHQWUHVXOWDQW5YDOXHRIIRUVSDFHJURXS3PZLWKRXW SDLQLWHQHHGHGWREHUHYLVLWHG,WLVDOVRLPSRUWDQWWRQRWHWKDWWKH DQ\LQGLFDWLRQRIXQXVXDODQLVRWURSLFGLVSODFHPHQWSDUDPHWHUVOHGWRUHMHFWLRQRIWKH

TXHVWLRQRIQRQFHQWURV\PPHWU\RUFHQWURV\PPHWU\KDVLPSRU 3PRGHO6LWHRFFXSDQF\UHÀQHPHQWVFOHDUO\LQGLFDWHGWKDWWKHHOHFWURQGHQVLW\DW WDQWEHDULQJVRQWKHSK\VLFDOSURSHUWLHVRIDPDWHULDO WKH&DVLWHLVVLJQLÀFDQWO\EHORZHcDVH[SHFWHGIRUFRPSOHWH&DRFFXSDQF\ ,QRXUUHÀQHPHQWZHÀ[HGWKHSRSXODWLRQRIERWK=UDQG%WRIXOORFFXSDQF\WR (;3(5,0(17$/0(7+2'6 FRSHZLWKVWURQJFRUUHODWLRQVRIWKHDWRPLFGLVSODFHPHQWSDUDPHWHURI=UZLWKWKH RYHUDOOVFDOHIDFWRUDQGWRUHGXFHWKHQXPEHURIUHÀQHGSDUDPHWHUV3DUDPHWHUV 0DMRUPLQRUDQGWUDFHHOHPHQWFRQFHQWUDWLRQVZHUHPHDVXUHGE\HOHFWURQ UHODWHGWRWKHGDWDFROOHFWLRQDQGWKHVWUXFWXUHUHÀQHPHQWDUHOLVWHGLQ7DEOH PLFURSUREHDQDO\VLV FXWVWRQHQR DQGODVHUDEODWLRQLQGXFWLYHO\FRXSOHGSODVPD $WRPLFFRRUGLQDWHVDQGLVRWURSLFGLVSODFHPHQWSDUDPHWHUV %HT DUHVKRZQLQ7DEOH PDVVVSHFWURPHWU\ /$,&306QRDQGIUDJPHQWQRE 7KHHOHFWURQPL $QLVRVWURSLFGLVSODFHPHQWSDUDPHWHUVDUHVXPPDUL]HGLQ7DEOH FURSUREH (03 DQDO\VHVZHUHGRQHZLWKD&$0(&$6;LQVWUXPHQWXVLQJ QDWXUDOR[LGHVDVVWDQGDUGV7KHRQO\PLQRUHOHPHQWGHWHFWHGZDVWLWDQLXP7KH ',6&866,21 IROORZLQJWUDFHHOHPHQWVKDYHFRQFHQWUDWLRQVFORVHWRWKHGHWHFWLRQOLPLWRIWKHV\V

WHP&U+IDQG6F1RUDUHHDUWKHOHPHQWVZHUHGHWHFWHG7KHHOHFWURQPLFURSUREH ,QLWLDOWHVWUHÀQHPHQWVLQVSDFHJURXS3OHGWRVWURQJFRUUH DQDO\VHVZHUHXVHGWRQRUPDOL]HWKH/$,&306PHDVXUHPHQWV 7DEOH ODWLRQVDPRQJVLWHVUHODWHGE\FHQWURV\PPHWU\DQGWKHDFFXUDF\ 7KH/$,&306V\VWHPFRQVLVWVRIDQP$U)H[FLPHUODVHU *HR/DV& 0LFUR/DV *|WWLQJHQ *HUPDQ\ FRXSOHG WR DQ ,&3PDVV VSHFWURPHWHU (ODQ RIWKHPRGHOFRXOGEHGUDPDWLFDOO\LPSURYHGE\UHÀ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²QDPHO\&UDQG9³UHVSRQVLEOHIRU EDQGVRI$O2RFWDKHGUDH[WHQGLQJDORQJ>@ZKLFKDUHODWHU WKHUHGRUDQJHFRORURISDLQLWH DOO\OLQNHGE\VKDUHGHGJHVDQGFRUQHUV7KHSDLQLWHVWUXFWXUH 7KHUHIUDFWLYHLQGLFHVRISDLQLWHZHUHGHWHUPLQHGZLWKDFRQYHQWLRQDOUHIUDF )LJ IHDWXUHVWZRW\SHVRIFKDQQHOVSDUDOOHOWRWKHFD[LV7KH WRPHWHUDQGD%UHZVWHUDQJOHUHIUDFWRPHWHU7KHODWWHULQVWUXPHQWZDVQHFHVVDU\ EHFDXVHWKHKLJKHVWUHIUDFWLYHLQGH[LVJUHDWHUWKDQ7KHRSWLFDOFKDUDFWHUZDV 5"#-&%BUBDPMMFDUJPOBOESFmOFNFOUQBSBNFUFSTGPSQBJOJUF FRQÀUPHGWREHXQLD[LDOQHJDWLYHXVLQJDSRODULVFRSHDQGSURMHFWLRQVSKHUHIRU DSZTUBMTJ[F NN tt SDLQLWHQRDQGE\LPPHUVLRQPLFURFRS\IRUVDPSOHE%RWKFU\VWDOVVKRZVWURQJ SFmOFEDPNQPTJUJPO $B ;S#<"M 0 > SOHRFKURLVPIURPEURZQLVKUHGWRRUDQJH\HOORZ7KH%UHZVWHUDQJOHVIRUSDLQLWHV TQBDFHSPVQ 1N QRVDQGDZHUHPHDVXUHGDV DQG UHVSHF B ¯ WLYHO\7KHUHIUDFWLYHLQGLFHVIRUQRDUHF PHDVXUHGZLWKDFRQYHQWLRQDO D ¯ UHIUDFWRPHWHU DQGX FDOFXODWHGIURP%UHZVWHUDQJOHPHDVXUHPHQWV +LJK RNBY ¡ ELUHIULQJHQFHZDVFRQÀUPHGE\PLFURVFRSLFLQYHVWLJDWLRQV IDFHWDQGLQFOXVLRQ ILM NJO NBY ocIc GRXEOLQJ ,QVSLWHRIWKHLQFUHDVHG1DDQG7LFRQWHQWRIWKHVDPSOHXVHG 7DEOH cLc c c WKHPHDVXUHGUHIUDFWLYHLQGLFHVDUHQRWVLJQLÀFDQWO\GLIIHUHQWIURPOLWHUDWXUHGDWDIRU o M TDBOUZQF ¡X UBOR SDLQLWHZLWKOHVVRUQR1DDQG7L 6KLJOH\HWDO 6DPSOHQRVKRZHGVWURQJ .FBTVSFESFnFDUJPOT JUHHQLVKEOXHÁXRUHVFHQFHXQGHUVKRUWZDYH89OLJKW7KHGHQVLW\RIVDPSOHVQRV PCTFSWFESFnFDUJPOT<* T * > DQGEZHUHGHWHUPLQHGK\GURVWDWLFDOO\\LHOGLQJ JFP VOJRVFSFnFDUJPOT 7KHVWUXFWXUHRIDIUDJPHQW ttNPLQGLDPHWHU RISDLQLWH OVNCFSPGQBSBNFUFST

FU\VWDOQREZDVVWXGLHGE\VLQJOHFU\VWDO;UD\GLIIUDFWLRQZLWKDQ(QUDI1RQLXV 3JOU &$'GLIIUDFWRPHWHUZLWKJUDSKLWHPRQRFKURPDWHG0R.B;UDGLDWLRQDW. 3 'DWDUHGXFWLRQLQFOXGLQJEDFNJURXQGDQG/RUHQW]SRODUL]DWLRQFRUUHFWLRQVDQGDQ X3 HPSLULFDODEVRUSWLRQFRUUHFWLRQEDVHGRQZVFDQVZHUHSHUIRUPHGXVLQJWKH6'3 (PP'

5"#-& 1PTJUJPOBMQBSBNFUFSTBOE%FR ¯ XJUITUBOEBSEEFWJBUJPOT 5"#-&-"*$1.4BOBMZTFTPGQBJOJUF JOQBSFOUIFTFTGPSQBJOJUF 4BNQMF /P /PC "UPN Y Z [ #FR ¯ XU ;S $B0 $B /B /B0 "M "M0 "M 5J0 0 #0 0 o ;S0 0 o o )G0 0 o 70 # o $S0 5PUBM 5"#-& "OJTPUSPQJDEJTQMBDFNFOUQBSBNFUFSTXJUITUBOEBSEEFWJB BQGV UJPOTJOQBSFOUIFTFTGPSQBJOJUF $B /B BUPN 6 6 6 6 6 6 "M ;S 5J $B # "M ;S "M )G 0 7 0 o $S 0 5PUBMDBUJPOT 0 $B /B #

159 8. The crystal structure of painite CaZrB[Al9O18] revisited

612 ARMBRUSTER ET AL.: CRYSTAL STRUCTURE OF PAINITE ARMBRUSTER ET AL.: CRYSTAL STRUCTURE OF PAINITE 613

FIGURE 1. The structure of painite is composed of corner- and edge 2+ sharing AlO6 octahedra. The hexagonal channels contain Ca cations (large spheres), whereas the six-membered triangular channels contain B3+ (small spheres) and Zr4+ atoms. The dark triangles represent the relatively rare triangular-prismatic coordination of Zr by O atoms. Ca2+ can also be regarded as the central atom of a strongly distorted Ca-O octahedron.

2+ FIGURE 2. The Ca site is located at 0, 0, 0 and lies on the 63 screw larger channel, at the origin, has a hexagonal cross-section and axis. It is surrounded by six symmetrically equivalent O2 sites at an 2+ is occupied by Ca at z = 0 and 0.5, whereas the second type, almost ideal bonding distance (Ca-O2 = 2.403 Å). This arrangement located at 1/3, 2/3, z, has a triangular cross-section and contains can either be regarded as an extremely distorted octahedron, or as a B3+ at z = 0.25 and Zr4+ at z = 0.75, respectively. The Zr4+ site distorted hexagon. The central cation is relatively free to move parallel is coordinated by six symmetry equivalent O1 sites, which to the c axis, which is reflected in the strongly anisotropic atomic are related by the threefold axis and the perpendicular mirror displacement parameters (probability ellipsoids are arbitrarily scaled plane, leading to trigonal prismatic arrangement. The bonding for better illustration). distance (2.121 Å) corresponds to the ideal Zr-O distance in sixfold coordination as reported by Shannon (1976). There are three additional O sites at a distance of 2.593 Å. Thus the cor- leads to the expected electron density of 18.3 e/Å3, which is in responding coordination polyhedron could also be interpreted excellent agreement with the refined value. The four-valent cation as a tri-capped trigonal prism. necessary for charge compensation is titanium, which replaces The B site is located at the intersection of the threefold axis aluminum. Ti is distributed over two Al sites and contributes only and the mirror plane (z = 0.25) in the center of the narrow, trian- with 2% to the Al occupancy. Thus this rather low concentration gular channel. The nearest neighboring atoms are three symmetry could not be resolved from the diffraction data. equivalent O4 sites forming a regular triangle around B with a The octahedral framework is composed of two types of AlO6 bonding distance 1.374 Å. Along [001] the sites located at 1/3, octahedra with central sites Al1 and Al2. The bond angles in 2/3, z are alternately occupied by B and Zr, whereas the sequence both polyhedra are distorted and show values between 77.52° in the laterally adjacent channel at 2/3, 1/3, z is inverted. and 102.57°, whereas the bonding distances vary between 1.821 The Ca sites at 0, 0, 0 and 0, 0, 1/2 are located in the larger and 2.082 Å (Table 5) and the averages are slightly shorter than channel on the 63 screw axis. They are bonded to six symmetry- the mean Al-O bonding distance of 1.935 Å (Shannon 1976). equivalent O atoms (O2 at 2.403 Å) of the octahedral frame- Octahedral distortions are triggered by the different ionic radii of work. The Ca coordination may be described by an octahedron Zr and B in the triangular channel (Fig. 3). The smaller B atom strongly flattened alongc or by a distorted hexagon (Fig. 2). In moves the connected octahedron vertices toward the center of this configuration the Ca site is laterally strongly restricted, but the channel slightly reducing the aperture, whereas the larger is relatively free to move along [001], normal to the coordina- Zr ion pushes the surrounding O atoms away from the center. tion polygon. This is reflected in the strongly anisotropic atomic This causes the undulating character of the edge-sharing bands displacement parameters for Ca (Table 3, Fig. 2), which have a of octahedra forming the framework. significantly larger component along [001]. One of the surprising In contrast, in the structurally related mineral fluoborite results of the structure refinement was that the observed electron B3[Mg9(F,OH)9O9] (e.g., Cámara and Ottolini 2000) the edge- density (18.6 e/Å3) at the Ca position was too low to be caused sharing bands of Mg octahedra are straight because only B oc- by complete Ca occupancy. The strong negative correlation be- cupies the trigonal channels. Fluoroborite has for this reason c/3 tween Ca and Na content observed in the LA-ICP-MS analyses periodicity relative to painite. Furthermore, the interior surface of of fragment no. 6b suggests a partial replacement of Ca by Na, the hexagonal channel in fluoborite is lined with F, OH and the which would explain the deficient electron density relative to a channels are therefore empty. Jeremejevite B5[ 3Al6(OH)3O15] site occupied only by Ca. Actually, 19% Na substitution for Ca has three octahedral vacancies pfu replaced by three triangular

160 8.20. Acknowledgments

%2JURXSVOLQLQJWKHZDOORIWKHKH[DJRQDOFKDQQHOV7ZRDG

GLWLRQDO%2JURXSVSIXSOXJWKHWULJRQDOFKDQQHOVDVLQSDLQLWH DQGÁXRERULWH,QFRQWUDVWWRWKHVXJJHVWLRQE\0RRUHDQG$UDNL DOOWKUHHPLQHUDOVSDLQLWH WKLVVWXG\ ÁXRERULWH HJ&i PDUDDQG2WWROLQL DQGMHUHPHMHYLWH 5RGHOODVHWDO

6RNRORYDHWDO VKDUHWKHVDPHVSDFHJURXS3P

$&.12:/('*0(176 :HWKDQN5.DPSIDQG6.ULYRYLFKHYIRUWKHLUFRQVWUXFWLYHUHYLHZV7KLV ZRUNZDVVXSSRUWHGE\WKH6ZLVV1DWLRQDO6FLHQFH)RXQGDWLRQ *UDQW WR7$FU\VWDOFKHPLVWU\RIPLQHUDOV ZKLFKLVKLJKO\DSSUHFLDWHG

5()(5(1&(6&,7(' &iPDUD)DQG2WWROLQL/ 1HZGDWDRQWKHFU\VWDOFKHPLVWU\RIÁXRERULWH E\PHDQVRI65()6,06DQG(03DQDO\VHV$PHULFDQ0LQHUDORJLVW ² ),*85(7KHHGJHFRQQHFWHGFKDLQVDORQJ>@DUHDIIHFWHGE\WKH &ODULQJEXOO*RUGRQ)+H\0+DQG3D\QH&- 3DLQLWHDQHZPLQHUDO IURP0RJRN%XUPD0LQHUDORJLFDO0DJD]LQH² VL]HRIWKHFDWLRQLQVLGHWKHFKDQQHO7KHVPDOO%DWRPV VSKHUHV PRYH (QUDI1RQLXV 6WUXFWXUHGHWHUPLQDWLRQSDFNDJH 6'3 FRPSXWHUSURJUDP WKHFRQQHFWHGYHUWLFHVRIRFWDKHGUDWRZDUGWKHFHQWHURIWKHFKDQQHO (QUDIQRQLXV'HOIWWKH1HWKHUODQGV ZKHUHDVWKHODUJHU=UDWRPV GDUNVTXDUHVUHSUHVHQWLQJWKHSULVPDWLF *RORYDVWLNRY 1, %HORYD -1 DQG %HORY 19 &U\VWDO VWUXFWXUH RI FRRUGLQDWLRQSRO\KHGUD SXVKWKHRFWDKHGUDDZD\IURPWKHFHQWHU7KLV MHUHPHMHYLWH=DSLVNL9VHVR\X]QRJR0LQHUDORJLFKHVNRJR2EVFKHVWYD ² FDXVHVGLVWRUWLRQVLQWKHRFWDKHGUDDQGXQGXODWLQJFKDUDFWHULVWLFRIWKH *QWKHU')ULVFKNQHFKW5DQG+HLQULFK&$ &DSDELOLWLHVRID IUDPHZRUNFKDLQV QP$U)H[FLPHUODVHUIRU/$,&306PLFURDQDO\VLVRIJHRORJLFDOPDWHULDOV -RXUQDORI$QDO\WLFDO$WRPLF6SHFWURPHWU\² +DUORZ*( 7KH0RJRN6WRQH7UDFW0\DQPDU0LQHUDOVZLWKFRPSOH[ SUDJHQHVHV3URFHHGLQJVRIWKHWKFRQIHUHQFHRQ´0LQHUDOVDQG0XVHXPVµ 0HOERXUQH$XVWUDOLD 5"#-&4FMFDUFEJOUFSBUPNJDEJTUBODFT ¯ GPSQBJOJUF ,\HU/$1 7KHJHRORJ\DQGJHPVWRQHVRIWKH0RJRN6WRQH7UDFW%XUPD ;So0 t 0HPRLUVRIWKH*HRORJLFDO6XUYH\RI,QGLD ;So0 t /RQJHULFK+3-DFNVRQ6(DQG*QWKHU' /DVHUDEODWLRQLQGXFWLYHO\ FRXSOHGSODVPDPDVVVSHFWURPHWU\WUDQVLHQWVLJQDOGDWDDFTXLVLWLRQDQGDQD $Bo0 t O\WHFRQFHQWUDWLRQFDOFXODWLRQ-RXUQDORI$QDO\WLFDO$WRPLF6SHFWURPHWU\ ² "Mo0 0RRUH 3% DQG$UDNL7 3DLQLWH &D=U%>$O2@ ,WV FU\VWDO VWUXF WXUH DQG UHODWLRQ WR MHUHPHMHYLWH % >%2; $O 2+ 2 @ DQG ÁXRERULWH "Mo0 % >0J )2+ 2 @$PHULFDQ0LQHUDORJLVW² "Mo0 3HUHWWL$ 1HZÀQGLQJVRISDLQLWH&RQWULEXWLRQVWR*HPRORJ\² "Mo0 5RGHOODV&*DUFLD%ODQFR6DQG9HJDV$ &U\VWDOVWUXFWXUHUHÀQHPHQW "Mo0 RIMHUHPHMHYLWH $O%)2 =HLWVFKULIWIU.ULVWDOORJUDSKLH² "Mo0 6KDQQRQ 5' 5HYLVHG HIIHFWLYH LRQLF UDGLL DQG V\VWHPDWLF VWXGLHV RI .FBO LQWHUDWRPLFGLVWDQFHVLQKDOLGHVDQGFKDOFRJHQLGHV$FWD&U\VWDOORJUDSKLFD $² "Mo0 t 6KHOGULFN*0 6+(/; FRPSXWHUSURJUDP 8QLYHUVLW\RI*|WWLQJHQ "Mo0 *HUPDQ\ "Mo0 t 6KLJOH\-(.DPSI$5DQG5RVVPDQ*5 1HZGDWDRQSDLQLWH0LQ "Mo0 HUDORJLFDO0DJD]LQH² .FBO 6RNRORYD(9(JRURY7LVPHQNR<..DUJDO·WVHY69.O\DNKLQ9$DQG 8UQVRY96 5HÀQHPHQWRIWKHFU\VWDOVWUXFWXUHRIV\QWKHWLFÁXRULDQ

#o0 t MHUHPHMHYLWH$O>%2@)*HRORJL\D² :HEVWHU5 *HPVWKHLUVRXUFHVGHVFULSWLRQVDQGLGHQWLÀFDWLRQS %XWWHUZRUWKV6HYHQRDNV8.

0$186&5,375(&(,9('-8/< 0$186&5,37$&&(37('2&72%(5 0$186&5,37+$1'/('%<3(7(5%8516

161 A. Appendix

A.1. A sample Fullprof input ﬁle

The following listing shows the FULLPROF input file used for the Rietveld refinement of TiS11-Rb at 25 ◦C. Two phases were considered: Phase number 1 was Rb-exchanged AM- 2 refined in Rietveld mode, and phase number 2 was K-bearing AM-2 refined in LeBail pattern matching mode. In the first part all parameters related to this particular refinement are commented. Detailed information about all parameters is given in the FULL- PROF documentation shipped with the FULLPROF software package or available online (http://www-llb.cea.fr/fullweb/fp2k/fp2k.htm). The file was used with FULLPROF version 3.20, Feb2005-LLB JRC, Multi Pattern: Linux-version.

A.1.1. Input PCR ﬁle with comments Comments always refer to the preceding example code.

The comment following the COMM flag is shown as title on the diffraction pattern plot. Lines starting with ! are comment lines which are set automatically by FULLPROF. The global χ2 value should converge to 1.0 for an optimum ratio of observed and refined data. The parameter Npr selects the peak shape function to be used. No. 7 is the Thompson- Cox-Hastings pseudo-Voigt with refinable Gaussian and Lorentzian components. Nph describes the number of phases in the input file. Nba is the number of sampled background points. If the number is positive, the points are interpolated linearly, for negative numbers a qubic spline is used for interpolation. Nex specifies the number of excluded regions.

! !Ipr Ppl Ioc Mat Pcr Ls1 Ls2 Ls3 NLI Prf Ins Rpa Sym Hkl Fou Sho Ana 1 0 0 1 2 1 0 0 0 3 10 1 1 0 3 0 1 !

One must specify the format of the output plot file and of the measured profile (Prf and Ins). These values depend on the frontend program used to display the profile plot, and on the conversion tools used to create the input data file.

162 A.1. A sample Fullprof input ﬁle

! lambda1 Lambda2 Ratio Bkpos Wdt Cthm muR AsyLim Rpolarz ->Patt# 1 1.540600 1.544300 0.5000 40.000 4.0000 0.7998 0.0000 75.00 0.5000 ! !NCY Eps R_at R_an R_pr R_gl Thmin Step Thmax PSD Sent0 5 0.20 0.99 0.99 0.99 0.99 9.0150 0.017001 99.9703 0.000 0.000 !

These values describe the device parameters set during data collection. AsyLim is the upper limit for asymmetric peak shapes. It must not be lower than the first peak in the diffraction pattern. 75◦ is an approved value for diffraction patterns collected to 100◦ 2θ on the Philips X’Pert Pro diffractometer used in this work. The scan parameters Thmin, Step and Thmax must be adjusted for each data set. NCY sets the number of least squares cycles for the refinement. 5 is a good value for rough profile optimisations, 15 was used for the structure refinement, and 25 for the final cycles of the refinement, when R values converged slowly.

!2Theta/TOF/E(Kev) Background for Pattern# 1 9.951 1070.100 12.448 881.259 15.880 708.154 23.370 566.524 34.448 487.840 45.215 448.498 58.867 448.498 67.059 424.893 78.137 409.156 89.450 385.551 94.911 385.551 98.968 401.288 ! ! Excluded regions (LowT HighT) for Pattern# 1 0.00 9.95 99.00 180.00 !

Background points can be specified as pairs of 2θ and intensities. If the number of background points given above is positive, intensities can be refined. Excluded regions must be specified as pairs of 2θ 2θ values. It is a common (and maybe redundant) technique to exclude the regions from 0◦ to the beginning of the pattern, and from the end of the pattern to 180◦, in order to ensure that the refinement only consideres values in the measured range.

! 62 !Number of refined parameters ! ! Zero Code SyCos Code SySin Code Lambda Code MORE ->Patt# 1 0.06040 11.00 0.00000 0.00 0.00000 0.00 0.000000 0.00 0

Linear pattern shift (Zero) is usually the ﬁrst reﬁned parameter with parameter code 11.00.

!------! Data for PHASE number: 1 ==> Current R_Bragg for Pattern# 1: 3.48 !------

163 A. Appendix

TiS11-Rb, X-ray, 25 C, 1.5406 A ! !Nat Dis Ang Pr1 Pr2 Pr3 Jbt Irf Isy Str Furth ATZ Nvk Npr More 18 18 0 1.0 0.0 0.0 0 0 0 0 0 1481.390 0 7 1

Number of atoms (Nat), distance constraints (Dis) and angle constraints (Ang) must be given here. Jbt = 0 is used for Rietveld structure reﬁnement mode. Npr again deﬁnes the peak shape function and is normally identical to the number set above.

! !Jvi Jdi Hel Sol Mom Ter Brind RMua RMub RMuc Jtyp Nsp_Ref Ph_Shift 1 3 0 0 0 0 1.0000 0.0000 0.0000 0.0000 0 0 0 ! ! Max_dst(dist) (angles) Bond-Valence Calc. 3.6000 120.0000 1

These lines only appear if More is set to 1 in the block above. Jdi > 0 creates a ﬁle with interatomic distances to a limit of Max dst(dist) A.˚

P 21 21 21 <--Space group symbol !Atom Typ X Y Z Biso Occ In Fin N_t Spc /Codes Ti TI 0.05450 0.28476 0.74425 0.00000 1.00000 0 0 0 0 161.00 171.00 181.00 0.00 0.00 Rb1 RB 0.43546 0.58275 0.76936 2.88146 0.99488 0 0 0 0 101.00 111.00 121.00 621.00 581.00 K1 K 0.43546 0.58275 0.76936 2.88146 0.00512 0 0 0 0 101.00 111.00 121.00 621.00 -581.00 Rb2 RB 0.20344 0.13082 0.31055 2.88146 0.82592 0 0 0 0 131.00 141.00 151.00 621.00 591.00 K2 K 0.20344 0.13082 0.31055 2.88146 0.17408 0 0 0 0 131.00 141.00 151.00 621.00 -591.00 Si1 SI 0.04158 0.54851 0.75380 0.00000 1.00000 0 0 0 0 191.00 201.00 211.00 0.00 0.00 Si2 SI 0.33305 0.32306 0.97069 0.00000 1.00000 0 0 0 0 221.00 231.00 241.00 0.00 0.00 Si3 SI 0.35249 0.32507 0.56562 0.00000 1.00000 0 0 0 0 251.00 261.00 271.00 0.00 0.00 O1 O 0.17649 0.29690 0.95561 0.00000 1.00000 0 0 0 0 281.00 291.00 301.00 0.00 0.00 O2 O 0.40468 0.34388 0.77405 0.00000 1.00000 0 0 0 0 311.00 321.00 331.00 0.00 0.00 O3 O 0.19389 0.30933 0.56059 0.00000 1.00000 0 0 0 0 341.00 351.00 361.00 0.00 0.00 O4 O 0.15351 0.57364 0.59860 0.00000 1.00000 0 0 0 0 371.00 381.00 391.00 0.00 0.00 O5 O 0.94805 0.26383 0.51908 0.00000 1.00000 0 0 0 0 401.00 411.00 421.00 0.00 0.00 O6 O 0.92558 0.63364 0.74079 0.00000 1.00000 0 0 0 0 431.00 441.00 451.00 0.00 0.00 O7 O 0.11549 0.57023 0.95026 0.00000 1.00000 0 0 0 0 461.00 471.00 481.00 0.00 0.00 O8 O 0.90674 0.27040 0.92110 0.00000 1.00000 0 0 0 0 491.00 501.00 511.00 0.00 0.00 O9 O 1.00351 0.42941 0.71938 0.00000 1.00000 0 0 0 0

164 A.1. A sample Fullprof input ﬁle

521.00 531.00 541.00 0.00 0.00 Ow1 O 0.19712 0.93572 0.40543 2.88146 1.00000 0 0 0 0 551.00 561.00 571.00 621.00 600.00

Structural data of the ﬁrst phase. The number of atoms must correspond to the number speciﬁed in Nat.

!------> Profile Parameters for Pattern # 1 ! Scale Shape1 Bov Str1 Str2 Str3 Strain-Model 0.10505E-03 0.00000 0.27508 0.00000 0.00000 0.00000 0 21.00000 0.000 611.000 0.000 0.000 0.000 ! U V W X Y GauSiz LorSiz Size-Model 0.102452 -0.064479 0.013693 0.108989 0.046079 0.000000 0.000000 0 0.000 0.000 0.000 0.000 0.000 0.000 0.000 ! a b c alpha beta gamma #Cell Info 10.085133 13.015672 7.225446 90.000000 90.000000 90.000000 31.00000 41.00000 51.00000 0.00000 0.00000 0.00000 ! Pref1 Pref2 Asy1 Asy2 Asy3 Asy4 S_L D_L 0.00000 1.00000 0.04396 0.03048 0.00000 0.00000 0.00000 0.00000 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Profile parameters for the first phase. A scale factor must be refined in Rietveld mode.

Depending on the quality of the diffraction pattern a Boverall parameter (Bov) may be refined rather than individual displacement parameters for each atom. Peak shape parameters U, V, W, X, and Y were measured with a corundum standard sample for the Philips X’Pert Pro device and were fixed in this refinement. The same accounts for the asymmetry parameters Asy1 and Asy2. Unit cell parameters are refined according to the space group. Correlating parameters (e.g. a, b, and c in cubic symmetry) should be set to the same value and refined with the same parameter number.

! Soft distance constraints: Ti O1 1 0.00000 0.00000 0.00000 1.96200 0.01000 Ti O3 1 0.00000 0.00000 0.00000 1.96200 0.01000 Ti O5 1 -1.00000 0.00000 0.00000 1.96200 0.01000 Ti O6 3 1.00000 -0.50000 1.50000 1.96200 0.01000 Ti O8 1 -1.00000 0.00000 0.00000 1.96200 0.01000 Ti O9 1 -1.00000 0.00000 0.00000 1.96200 0.01000 Si1 O4 1 0.00000 0.00000 0.00000 1.62590 0.01000 Si1 O6 1 -1.00000 0.00000 0.00000 1.62590 0.01000 Si1 O7 1 0.00000 0.00000 0.00000 1.62590 0.01000 Si1 O9 1 -1.00000 0.00000 0.00000 1.62590 0.01000 Si2 O1 1 0.00000 0.00000 0.00000 1.62590 0.01000 Si2 O2 1 0.00000 0.00000 0.00000 1.62590 0.01000 Si2 O4 2 0.50000 1.00000 0.50000 1.62590 0.01000 Si2 O8 4 -0.50000 0.50000 2.00000 1.62590 0.01000 Si3 O2 1 0.00000 0.00000 0.00000 1.62590 0.01000 Si3 O3 1 0.00000 0.00000 0.00000 1.62590 0.01000 Si3 O5 4 -0.50000 0.50000 1.00000 1.62590 0.01000 Si3 O7 2 0.50000 1.00000 -0.50000 1.62590 0.01000

Soft distance constraints for framework atoms. These values are taken from a file named CFML Restraints.tpcr, which is created in the same directory after the first refinement

165 A. Appendix cycles in Rietveld mode. Distances and standard deviations (7th and 8th column) were set to values taken from literature.

!------! Data for PHASE number: 2 ==> Current R_Bragg for Pattern# 1: 0.49 !------TiS11-K, X-ray, 25 C, 1.5406 A ! !Nat Dis Ang Pr1 Pr2 Pr3 Jbt Irf Isy Str Furth ATZ Nvk Npr More 0 0 0 1.0 0.0 0.0 2 0 0 0 0 1481.390 0 7 0 !

Data block for the second phase. Since this phase is only refined in LeBail pattern matching mode, no atoms and constraints are to be specified. Jbt is set to 2 in order to activate the pattern matching mode. For the first cycle Irf must be set to 0 to generate a reflection list from the space group and cell dimensions. During early stages of the refinement it may be necessary to Irf to 0 again to regenerate the reflection list.

P 21 21 21 <--Space group symbol !------> Profile Parameters for Pattern # 1 ! Scale Shape1 Bov Str1 Str2 Str3 Strain-Model 0.12023E-03 0.00000 0.00000 0.00000 0.00000 0.00000 0 0.00000 0.000 0.000 0.000 0.000 0.000 ! U V W X Y GauSiz LorSiz Size-Model 0.007675 -0.007898 0.013001 0.108989 0.046079 0.095594 0.000000 0 0.000 0.000 0.000 0.000 0.000 91.000 0.000 ! a b c alpha beta gamma #Cell Info 10.008316 13.011608 7.211586 90.000000 90.000000 90.000000 61.00000 71.00000 81.00000 0.00000 0.00000 0.00000 ! Pref1 Pref2 Asy1 Asy2 Asy3 Asy4 S_L D_L 0.00000 1.00000 0.04396 0.03048 0.00000 0.00000 0.00000 0.00000 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Profile parameters are treated the same way as in Rietveld mode, with the exception that no scale value can be refined. In this example GauSiz is refined, which takes into account isotropic peak broadening as effect of small crystallite size.

A.1.2. The input PCR ﬁle

COMM TiS-Rb 25 deg ! Current global Chi2 (Bragg contrib.) = 0.3852 ! Files => DAT-file: TiS-Rb03-025, PCR-file: TiS-Rb03-025 !Job Npr Nph Nba Nex Nsc Nor Dum Iwg Ilo Ias Res Ste Nre Cry Uni Cor Opt Aut 0 7 2 -12 2 0 0 1 0 0 1 0 0 0 0 0 0 0 0 ! !Ipr Ppl Ioc Mat Pcr Ls1 Ls2 Ls3 NLI Prf Ins Rpa Sym Hkl Fou Sho Ana 1 0 0 1 2 1 0 0 0 3 10 1 1 0 3 0 1 ! ! lambda1 Lambda2 Ratio Bkpos Wdt Cthm muR AsyLim Rpolarz ->Patt# 1 1.540600 1.544300 0.5000 40.000 4.0000 0.7998 0.0000 75.00 0.5000 ! !NCY Eps R_at R_an R_pr R_gl Thmin Step Thmax PSD Sent0

166 A.1. A sample Fullprof input ﬁle

5 0.20 0.99 0.99 0.99 0.99 9.0150 0.017001 99.9703 0.000 0.000 ! !2Theta/TOF/E(Kev) Background for Pattern# 1 9.951 1070.100 12.448 881.259 15.880 708.154 23.370 566.524 34.448 487.840 45.215 448.498 58.867 448.498 67.059 424.893 78.137 409.156 89.450 385.551 94.911 385.551 98.968 401.288 ! ! Excluded regions (LowT HighT) for Pattern# 1 0.00 9.95 99.00 180.00 ! ! 62 !Number of refined parameters ! ! Zero Code SyCos Code SySin Code Lambda Code MORE ->Patt# 1 0.06040 11.00 0.00000 0.00 0.00000 0.00 0.000000 0.00 0 !------! Data for PHASE number: 1 ==> Current R_Bragg for Pattern# 1: 3.48 !------TiS11-Rb, X-ray, 25 C, 1.5406 A ! !Nat Dis Ang Pr1 Pr2 Pr3 Jbt Irf Isy Str Furth ATZ Nvk Npr More 18 18 0 1.0 0.0 0.0 0 0 0 0 0 1481.390 0 7 1 ! !Jvi Jdi Hel Sol Mom Ter Brind RMua RMub RMuc Jtyp Nsp_Ref Ph_Shift 1 3 0 0 0 0 1.0000 0.0000 0.0000 0.0000 0 0 0 ! ! Max_dst(dist) (angles) Bond-Valence Calc. 3.6000 120.0000 1 P 21 21 21 <--Space group symbol !Atom Typ X Y Z Biso Occ In Fin N_t Spc /Codes Ti TI 0.05450 0.28476 0.74425 0.00000 1.00000 0 0 0 0 161.00 171.00 181.00 0.00 0.00 Rb1 RB 0.43546 0.58275 0.76936 2.88146 0.99488 0 0 0 0 101.00 111.00 121.00 621.00 581.00 K1 K 0.43546 0.58275 0.76936 2.88146 0.00512 0 0 0 0 101.00 111.00 121.00 621.00 -581.00 Rb2 RB 0.20344 0.13082 0.31055 2.88146 0.82592 0 0 0 0 131.00 141.00 151.00 621.00 591.00 K2 K 0.20344 0.13082 0.31055 2.88146 0.17408 0 0 0 0 131.00 141.00 151.00 621.00 -591.00 Si1 SI 0.04158 0.54851 0.75380 0.00000 1.00000 0 0 0 0 191.00 201.00 211.00 0.00 0.00 Si2 SI 0.33305 0.32306 0.97069 0.00000 1.00000 0 0 0 0 221.00 231.00 241.00 0.00 0.00 Si3 SI 0.35249 0.32507 0.56562 0.00000 1.00000 0 0 0 0 251.00 261.00 271.00 0.00 0.00

167 A. Appendix

O1 O 0.17649 0.29690 0.95561 0.00000 1.00000 0 0 0 0 281.00 291.00 301.00 0.00 0.00 O2 O 0.40468 0.34388 0.77405 0.00000 1.00000 0 0 0 0 311.00 321.00 331.00 0.00 0.00 O3 O 0.19389 0.30933 0.56059 0.00000 1.00000 0 0 0 0 341.00 351.00 361.00 0.00 0.00 O4 O 0.15351 0.57364 0.59860 0.00000 1.00000 0 0 0 0 371.00 381.00 391.00 0.00 0.00 O5 O 0.94805 0.26383 0.51908 0.00000 1.00000 0 0 0 0 401.00 411.00 421.00 0.00 0.00 O6 O 0.92558 0.63364 0.74079 0.00000 1.00000 0 0 0 0 431.00 441.00 451.00 0.00 0.00 O7 O 0.11549 0.57023 0.95026 0.00000 1.00000 0 0 0 0 461.00 471.00 481.00 0.00 0.00 O8 O 0.90674 0.27040 0.92110 0.00000 1.00000 0 0 0 0 491.00 501.00 511.00 0.00 0.00 O9 O 1.00351 0.42941 0.71938 0.00000 1.00000 0 0 0 0 521.00 531.00 541.00 0.00 0.00 Ow1 O 0.19712 0.93572 0.40543 2.88146 1.00000 0 0 0 0 551.00 561.00 571.00 621.00 600.00 !------> Profile Parameters for Pattern # 1 ! Scale Shape1 Bov Str1 Str2 Str3 Strain-Model 0.10505E-03 0.00000 0.27508 0.00000 0.00000 0.00000 0 21.00000 0.000 611.000 0.000 0.000 0.000 ! U V W X Y GauSiz LorSiz Size-Model 0.102452 -0.064479 0.013693 0.108989 0.046079 0.000000 0.000000 0 0.000 0.000 0.000 0.000 0.000 0.000 0.000 ! a b c alpha beta gamma #Cell Info 10.085133 13.015672 7.225446 90.000000 90.000000 90.000000 31.00000 41.00000 51.00000 0.00000 0.00000 0.00000 ! Pref1 Pref2 Asy1 Asy2 Asy3 Asy4 S_L D_L 0.00000 1.00000 0.04396 0.03048 0.00000 0.00000 0.00000 0.00000 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 ! Soft distance constraints: Ti O1 1 0.00000 0.00000 0.00000 1.96200 0.01000 Ti O3 1 0.00000 0.00000 0.00000 1.96200 0.01000 Ti O5 1 -1.00000 0.00000 0.00000 1.96200 0.01000 Ti O6 3 1.00000 -0.50000 1.50000 1.96200 0.01000 Ti O8 1 -1.00000 0.00000 0.00000 1.96200 0.01000 Ti O9 1 -1.00000 0.00000 0.00000 1.96200 0.01000 Si1 O4 1 0.00000 0.00000 0.00000 1.62590 0.01000 Si1 O6 1 -1.00000 0.00000 0.00000 1.62590 0.01000 Si1 O7 1 0.00000 0.00000 0.00000 1.62590 0.01000 Si1 O9 1 -1.00000 0.00000 0.00000 1.62590 0.01000 Si2 O1 1 0.00000 0.00000 0.00000 1.62590 0.01000 Si2 O2 1 0.00000 0.00000 0.00000 1.62590 0.01000 Si2 O4 2 0.50000 1.00000 0.50000 1.62590 0.01000 Si2 O8 4 -0.50000 0.50000 2.00000 1.62590 0.01000 Si3 O2 1 0.00000 0.00000 0.00000 1.62590 0.01000 Si3 O3 1 0.00000 0.00000 0.00000 1.62590 0.01000 Si3 O5 4 -0.50000 0.50000 1.00000 1.62590 0.01000 Si3 O7 2 0.50000 1.00000 -0.50000 1.62590 0.01000 !------! Data for PHASE number: 2 ==> Current R_Bragg for Pattern# 1: 0.49 !------TiS11-K, X-ray, 25 C, 1.5406 A

168 A.1. A sample Fullprof input ﬁle

! !Nat Dis Ang Pr1 Pr2 Pr3 Jbt Irf Isy Str Furth ATZ Nvk Npr More 0 0 0 1.0 0.0 0.0 2 0 0 0 0 1481.390 0 7 0 ! P 21 21 21 <--Space group symbol !------> Profile Parameters for Pattern # 1 ! Scale Shape1 Bov Str1 Str2 Str3 Strain-Model 0.12023E-03 0.00000 0.00000 0.00000 0.00000 0.00000 0 0.00000 0.000 0.000 0.000 0.000 0.000 ! U V W X Y GauSiz LorSiz Size-Model 0.007675 -0.007898 0.013001 0.108989 0.046079 0.095594 0.000000 0 0.000 0.000 0.000 0.000 0.000 91.000 0.000 ! a b c alpha beta gamma #Cell Info 10.008316 13.011608 7.211586 90.000000 90.000000 90.000000 61.00000 71.00000 81.00000 0.00000 0.00000 0.00000 ! Pref1 Pref2 Asy1 Asy2 Asy3 Asy4 S_L D_L 0.00000 1.00000 0.04396 0.03048 0.00000 0.00000 0.00000 0.00000 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

169 B. Acknowledgements

The last lines of this work should be words of gratitude to all those who helped to make this thesis possible. First of all, I have to thank Prof. Dr. Thomas Armbruster for being a patient supervisor and for supporting this work with ideas and constructive criticism. I would also like to thank everybody from the Laboratorium fur¨ chemische und mineralogische Kristallographie, especially my room mates Dmitry Chernyshov and Thammarat Aree for bearing with me, my co-students Anna Malsy and Petra Simoncic for having good times in the lab and on conferences all over Europe, Jurg¨ Hauser for keeping the computers alive, Vladimir Malogajski for ﬁxing the exploded oven, and Marc Hostettler, Therese Luthi¨ Nyf- feler, Prof. Hans-Beat Burgi,¨ Margrit Hugli,¨ and Venugopalan Paloth for their help, support and encouragement over the last four years. I am very grateful to Kirsten Christensen and Dr. Xiaodong Zou from the Department of Structural Chemistry, Stockholm University, whose work provided the basis for my research. Their support and hospitality were a great boost of motivation and contributed to the success of this work. Special thanks go to Prof. Sergey Krivovichev for his collaboration in various projects, his moral and professional assistance during my work on this thesis, and for the good times we had in Bern, Grenoble, St. Petersburg and Rome. From the Institut fur¨ Geologie, University of Bern, I would like to thank all my friends, colleagues and former co-students for giving me a second home. I always had a lot of fun with Manuel ’Ape’ Eggimann, Bettina Flury, Alex De Gasparo, Ruth Maeder, Sarah ’Matete’ Antenen, and many more during lunch time. I’ve been knowing some of them for almost eight years. During that time we have become close friends, and I hope we will be for a very long time. It has been a pleasure for me to work with Dr. Urs Eggenberger from the powder XRD group. We always worked in a relaxed collaboration with constructive and motivating dis- cussions, and I will never forget his great support during my search for employment. Last but not least I am forever indebted to my family, especially to my parents Hans- Ruedi und Sylvia. They gave me the opportunity to enjoy the best education, and they always stood behind my decisions for a career in natural sciences and supported me by all means. Andri, Jann and Monica´ also encouraged me to realise my plans, which I highly appreciate.

Bern, December 5, 2005 Nicola Dobelin¨

170 C. Curriculum Vitae

C.1. Personal Details

Name: Nicola Dobelin¨ Gender: Male Date of birth: January 19, 1976 Place of birth: Bern Citizenship: Swiss

C.2. Education

1983–1989: Primary School, Lommiswil

1989–1991: Secondary School, Bezirksschule Selzach

1991–1996: Grammar School, Kantonsschule Solothurn

1996: Matura Typus C (natural sciences)

06/1996–10/1996: Military service, Romont (FR) Rank: Soldier Function: Truck driver

1996–1997: Studies of physics, University of Bern

1997–2002: Studies of Earth Sciences, University of Bern Specialisation: Mineralogy, Crystallography Thesis: Stepwise dehydration of Cd- and Sr-exchanged heulandite Supervisor: Prof. Th. Armbruster

07/2000–10/2000: Geological ﬁeld work in Gravelotte, Northern Province, South Africa

June 27, 2002: Degree of Master of Science in Earth Sciences

July 1, 2002–January 12, 2006: PhD thesis in Crystallography at the Laboratorium fur¨ chemische und mineralogische Kristallographie, University of Bern Supervision: Prof. Th. Armbruster Subject: Microporous titanosilicate AM-2: Synthesis, ion-exchange and dehydration

171