Apophyllite Alteration in Aqueous Solutions

Home , Apophyllite, Carletonite, Cavansite

Apophyllite alteration in aqueous solutions

A nano-scale study of phyllosilicate reactions

Dissertation zur Erlangung des Grades Doktor der Naturwissenschaften (Dr. rer. nat.) an der Fakultät für Geowissenschaften der Ruhr-Universität Bochum

vorgelegt von

Kirill Aldushin

aus Moskau, Rußland

Bochum 2004 ANKNOWLEDGEMENTS

First of all I would like to thank my supervisor Priv.-Doz. Dr. Guntram Jordan for the opportunity to do my PhD thesis at the Ruhr-University of Bochum. I am grateful for his valuable advices and educational discussions. I also want to express my gratitude for his patience and continuous support during the last three and a half years.

I would like to thank Prof. Dr. Wolfgang W. Schmahl for the constructive criticism and aiding of this work. I thank Prof. Dr. Werner Rammensee (Köln) for his help at the early stage of the work. I would also like to acknowledge the review by Prof. Dr. Michael Alber.

Special thanks to Dr. Hans-Werner Becker (Institut für Physik mit Ionenstrahlung, RUB) for his friendly assistance in working with RBS and XPS. Priv.-Doz. Dr. Michael Fechtelkord for his kindly support in NMR is gratefully acknowledged. I am grateful to Dr. Heinz-Jürgen Bernhardt for his help with EPM analysis. The aid provided by Dr. Ralf Dohmen in handling RBS spectra is greatly appreciated.

I am grateful to Dr. Thomas Lohkämper for his help with PSIM measurements and to Dr. Thomas Reinecke and Dr. Bernd Marler for the aid with XRD.

Many thanks to Sandra Grabowski, Astrid Michelle, and Udo Trombach for their help in laboratory work; thanks to Achim Schlieper for the computer support.

I also appreciate very much various kinds of help provided by all members of the Mineralogy department of Ruhr-University Bochum. Thanks a lot for your openness and friendly atmosphere.

Funding for the construction of the HAFM and for conducting this project was provided by the Deutsche Forschungsgemeinschaft (DFG) and is gratefully acknowledged.

Finally, I warmly thank my family, especially my wife Elena and my parents, who made an inestimable contribution into this thesis. Without their continuos encouragement and moral support it would have been not possible to accomplish this work. Thank you for everything!

Bochum, October 2004 Kirill Aldushin

2 Contents

1. INTRODUCTION…....………………………………………………………….. 5

2. MINERAL DATA AND PREVIOUS STUDIES OF APOPHYLLITE…....…. 11

2.1 Definition of apophyllite, its properties, and occurrence…...……………... 11 2.2 Apophyllite structure and its relation to other minerals…………………... 13 2.3 Apophyllite alteration……………………..…………………………………. 16

3. EXPERIMENTAL…………………………………..………………………….… 19

3.1 Experimental methods………………..……………………………………… 19 3.1.1 Hydrothermal atomic force microscopy (HAFM)………………….…….. 19 3.1.2 Rutherford backscattering spectrometry (RBS)………………………...... 23 3.1.3 X-ray photoelectron spectroscopy (XPS)………………………….……... 24 3.1.4 Nuclear magnetic resonance (NMR)………….………………………….. 24

3.2 Sample preparation……………..………………………………………...…. 25

4. RESULTS AND DISCUSSION……….…………………………………………. 27

4.1 Introduction into apophyllite alteration at acidic conditions……………... 27 4.1.1 HAFM…………………...... …………………………………………. 27 4.1.2 RBS………………………………....……………………………………. 37 4.1.3 XPS………………………………………………………………………. 39 4.1.4 Discussion…………………………………………....…………………... 41

4.2 Protonation stages……….…………………………………………………... 45 4.2.1 AFM……………………………………………………………………… 45 4.2.2 NMR……………………………..……….……………………………… 51 1H MAS NMR………………………………..………………………….….... 51

3 29Si MAS NMR and {1H} 29Si CPMAS NMR………………………………… 52 4.2.3 Discussion……………………………...... ……………………………... 54 Transformation P → T……………………………………………….…….. 55 Transformation T → A……………………………………………….…….. 56 Transformation T → H and A → H, amorphous structures……….……….. 58

4.3 Alteration at pH 4 - 5.6………….……………..…………………………….. 59 4.3.1 Results……………………………………………………………………. 59 4.3.2 Discussion………………………………………………………………... 64

4.4 Hillock rotation……………..………………………………………………... 69 4.4.1 Results……………………………………………………………………. 69 4.4.2 Discussion………………………………………………………………... 71

5. CONCLUSION…...... …………………………………………………………… 81

REFERENCES…..………………………………………………………………….. 85

Erklärung ………………………………...... ……...... ……….... 92

Curriculum Vitae…………………………………………...... …………………. 93

4 1. INTRODUCTION

Detailed studies of reactions of silicates with acidic solutions provide insight into fundamental natural processes, such as weathering, formation of minerals, or global cycling of chemical elements (see e.g. Nagy et al., 1991; Wieland and Stumm, 1992; Walther, 1996). Among the diversity of silicate minerals, sheet silicates can be distinguished as one of the most important subclass. Phyllosilicates are known as important rock-forming minerals present in all types of rocks (magmatic, metamorphic, and sedimentary) and they play a key-role in soils as nutrient components. Furthermore, phyllosilicate minerals are widely used in many industrial fields. These facts may explain the huge interest of researchers in these minerals. Many investigations have been conducted to study the alteration of phyllosilicates (generally clays and micas) in aqueous solutions (e.g. Newman, 1970; Sposito, 1984; Kalinowski and Schweda, 1996; Zysset and Schindler, 1996; Malmström and Banwart, 1997; Huertas et al., 1999). Aside from the significance for environmental science, these studies are also important for technical applications e. g., production of sorbents, catalyst carriers, decontaminants (Corma and Perez-Pariente, 1987; Ravichandran and Sivasankar, 1997; Saito et al., 1997; Temuujin et al. 2001). Most sheet silicates can be composed by two modular units: a sheet of corner- linked tetrahedra and a sheet of edge-linked octahedra (e.g. Moore and Reynolds, 1997). In the tetrahedral sheet (Fig. 1a), the dominant cation T is Si4+, but Al3+ substitutes it 3+ frequently and Fe occasionally; the cation/oxygen ratio is T2O5. The octahedral sheet can be thought of as two planes of closest-packed oxygen and hydroxyl ions with cations occupying the resulting octahedral sites between the two planes (Fig. 1b). The cations are basically Al3+, Mg2+, Fe2+, or Fe3+. Phyllosilicates in which the cations occupy all available octahedral positions are called trioctahedral; dioctahedral phyllosilicates are those in which only two-thirds of octahedral positions are occupied by cations. The assemblage of one tetrahedral sheet and one octahedral sheet is called a 1:1 layer silicate structure. The structures with one octahedral sheet sandwiched between two tetrahedral sheets are called 2:1 layer silicate. The structure of chlorites consists of alternating 2:1 layers and octahedral layers, thus making a 2:1:1 structure. In some

5 a c

Al3+, Mg2+, Fe2+, Fe3+

O2- or OH-

Figure 1. Structure of sheet silicates: (a) sketch of sheet structure of silica tetrahedra arranged in a hexagonal network; (b) sketch showing the sheet structure of the octahedral units; (c) an example of phyllosilicate structure: TOT-layers alternate with the layers of interlayer cations.

minerals, e.g. talk, pyrophyllite, or kaolinite the layers are connected together by Van der Vaals bonds. In other phyllosilicates, such as most clays and micas, a cation substitution within the layers (basically Si4+ → Al3+) causes a positive charge deficiency. This deficiency is compensated by interlayer cations (usually K+, Mg2+, Na+, or Fe2+) which couple the silicate layers together. An example of such structure is shown in Figure 1c. Phyllosilicates show a variety of reactions while being in contact with aqueous solutions. The type of the reaction may depend on inherent mineral properties (e.g. mineral structure, composition, grain size and texture) and conditions, such as temperature, pressure, or solution composition. One of the reactions is ion-exchange:

6 the uptake of ions by the surface and interlayers coupled with a release of ions from the surface and interlayers. In acidic solutions phyllosilicates usually undergo selective leaching of interlayer cations and cations located in the octahedral sheet. In some cases this leaching leads to the formation of an amorphous phase or at least to a significant loss of crystallinity of the product (Frondel, 1979; Kaviratna and Pinnavaia, 1994; Aznar et al., 1996). In other cases, cation depletion does not cause decomposition of the structure and the product retains the structural features of the parental mineral (Lagaly et al., 1975; Frondel, 1979; Pabst, 1958; Rodriguez et al., 1994; Kosuge et al., 1995). Another interesting property of phyllosilicates is their ability to swell while being exposed to aqueous solutions. This property is basically attributed to the minerals of the smectite group (saponite, hectorite, montmorillonite, beidellite, and nontronite) and vermiculite. The swelling effect together with the ion exchange reactions are probably the most extensively studied processes of phyllosilicates (e.g. Norrish, 1954; Walker, 1960; van Olphen, 1965; Barshad and Kishk, 1968; Newman, 1970; Sposito and Prost, 1982; Laird, 1996). Such a great interest is caused by the importance of these processes in environmental and engineering science, as well as in industrial applications. Generally, mineral reactions in aqueous solutions have been studied e.g., by X- ray diffraction (XRD), nuclear magnetic resonance (NMR), infra-red spectroscopy (IR), or high resolution transmission electron microscopy (HRTEM). Mineral reactions can be also studied by using surface-sensitive techniques. Since any reaction of a mineral starts from its surface, investigation of interface processes is crucial for understanding mineral reactivity. However, it needs to be taken into account, that chemical and physical properties of minerals at the surface are different to those in the bulk mineral (Hochella, 1990). Surfaces can be studied by various methods, e.g. X-ray photoelectron spectroscopy (XPS), low energy electron diffraction (LEED), BET (Brunauer - Emmett - Teller) surface area analysis, Rutherford backscattering spectroscopy (RBS), extended X-ray absorption fine structure (EXAFS), crystal truncation rod (CTR), scanning electron microscopy (SEM) and atomic force microscopy (AFM). AFM is a powerful in situ method for surface investigation. This method allows the direct observation of mineral surfaces in air, solution or vacuum and has been

7 proven to be a highly effective technique for the investigation of morphological changes during mineral dissolution, precipitation, and growth at nano-scale (e. g. Nagy and Blum, 1994; Stipp et al., 1994; Teng and Dove, 1997; Bosbach et al., 1998; Astilleros et al., 2002). By using this method, information about the molecular mechanisms and kinetics of mineral surface reactions can be obtained. According to Dove and Platt (1996), reaction rates within the range of 10-10-10-6 mol/m2⋅s can be detected by AFM. The original design of atomic force microscope limited studies to be performed at ambient conditions. For this reason, most AFM studies of surface reactivity so far concern minerals which exhibit reaction rates fast enough to be detectable within an appropriate time range of AFM imaging (i.e. faster than 10-10 mol/m2⋅s), such as many carbonates and sulphates (e.g. Hillner et al., 1992; Hall and Cullen, 1996; Jordan and Rammensee, 1998; Shiraki et al., 2000). The development of hydrothermal AFM (HAFM, Higgins et al., 1998a) helped to solve many difficulties which are typical for room-temperature AFM studies. This device enables in-situ AFM experiments at hydrothermal conditions, i.e. above the ambient boiling point of water. A successful application of this technique advanced considerably the understanding of processes taking place on mineral-water interfaces (Jordan et al., 1999; Jordan et al., 2001; Higgins et al., 2002). Studying phyllosilicate surfaces at nanoscale attracts particular attention because the mechanisms of reactions in aqueous solutions differ from those of other minerals (e.g. Hochella et al., 1998; Bickmore et al., 2001; Aldushin et al., 2004b), and still are largely unknown. Since the rates of most silicate-water interface reactions are in the range of 10-10 mol/m2⋅s and less at ambient conditions, these reactions can barely be investigated in situ by AFM. There are, however, a number of room-temperature AFM studies on the behaviour of phyllosilicates in acidic solution. For example, Hochella et al. (1998) observed the delamination and/or recombining of clay-size phlogopite particles while establishing an equilibrium thickness in Na+ containing solutions. Rufe and Hochella (1999) investigated the alteration behaviour of phlogopite in the pH range between 2 and 5.7. For in situ AFM experiments the samples were preliminary etched by HF; the duration of experiments was up to 127 hours. Under acidic conditions the basal surface of phlogopite was shown to leach, hydrate, and expand to an amorphous silica-enriched film.

8 Dissolution of two smectite minerals – hectorite and nontronite – has been investigated under acidic conditions (Bosbach et al., 2000; Bickmore et al., 2001). It has been found that these minerals dissolve at pH 2 at the crystal edges, with the basal surface remaining unreactive during the timescale of the experiment. In these studies a specific preparation technique has been used, which allows in situ imaging of small clay particles under moderate acidic conditions. Brandt et al. (2003) investigated the dissolution of chlorite by using a combination of AFM and mixed flow reactor. In situ microscopic investigations of samples preconditioned at pH 2 for several months suggested that chlorite dissolution is defect controlled. The authors show that chlorite dissolves via the formation and spreading of shallow etch pits at the basal surface, thus indicating a layer-by-layer dissolution mechanism. Thus, dissolution mechanism of chlorite differs from other phyllosilicates which have been shown to dissolve preferentially from crystal edges (e.g. Bosbach et al., 2000; Bickmore et al., 2001). As can be seen, the alteration mechanisms of phyllosilicates are very different, but most of them are scarcely reacting at room temperature and require prolonged duration of experiments, preferential pretreatment in order to initiate alteration reactions, or require a large reactive surface area e.g. by grinding the sample to a submicron size. Some phyllosilicates, however, exhibit a surprising reactivity at room temperature. For example, the hydrous phyllosilicate apophyllite, which is subject of the present study, shows a remarkable reactivity even at room temperature (Aldushin et al., 2004 a, b). Although this hydrous sheet silicate is less common than micas or clays, the study of this mineral can give very useful insights in the interpretation of processes taking place in many other silicates. Although apophyllite is a sheet silicate and structurally and chemically very close to clays and micas, it also resembles the minerals of the zeolite group, e.g. in structure, conditions of formation, low specific gravity and the ability to release water under heating (Marriner et al., 1990). Hence, apophyllite is expected to exhibit reactive properties similar to those of both zeolites and phyllosilicates. Apophyllite has been shown as a promising material in terms of structural intercalation of various organic compounds into its structure (e.g. Sogo et al., 2000). In

9 addition, apophyllite has been detected as a source of high fluoride concentrations in the groundwater (Cave, 2002), that stimulates dissolution studies of this mineral. The primary goals of this study were the investigation of the mechanisms and kinetics of the surface reactions of apophyllite in aqueous solutions and the comparison of these mechanisms with the alteration mechanisms of clays and micas at appropriate conditions. The apophyllite alteration in aqueous solutions has been studied by different microscopic and spectroscopic methods. In situ investigation of apophyllite (001) surface alteration in aqueous solutions has been conducted by hydrothermal AFM at temperatures from 20 up to 130 °C and pressures from 1 up to 35 bar in aqueous solutions ranging from pH 1.5 to 10. X-ray photoelectron spectroscopy (XPS) and Rutherford backscattering spectroscopy (RBS) analysis were conducted to analyse the alteration of the chemical composition of the near-surface region of the apophyllite (001) surface caused by the reaction with acidic solution. In order to detect possible structural changes caused by the acidic attack, nuclear magnetic resonance spectroscopy (NMR) has been applied. The combination of these techniques allowed to obtain reliable information on the mechanism and kinetic of apophyllite alteration.

10 2. MINERAL DATA AND PREVIOUS STUDIES OF APOPHYLLITE

2.1 Definition of apophyllite, its properties and occurrence

Apophyllite was first identified and confirmed as a distinct mineral at the beginning of nineteenth century by the French mineralogist Rene-Just Hauy. Apophyllite (that roughly means "to leaf apart" in Greek) has got its name due to tendency to flake apart when heated in a blowpipe. Further findings have shown that apophyllite is rather a name of a mineral group which includes several chemically and structurally similar minerals: fluorapophyllite, hydroxyapophyllite, ammonian fluorapophyllite, ammonian hydroxyapophyllite, natroapophyllite, and carletonite (table 1).

Table 1. Nomenclature, chemical composition and structural characteristics of the apophyllite group minerals.

Mineral Formula Symmetry; Reference parameters, Å

Fluorapophyllite KCa4Si8O20F · 8H2O P 4/mnc; Colville et al. a=8.963, c=15.804 (1971)

Hydroxyapophyllite KCa4Si8O20(OH) · 8H2O P 4/mnc; Dunn et al. a=8.978, c=15.830 (1978)

Ammonian (K, NH4)Ca4Si8O20F · 8H2O P 4/mnc; Marriner et al. fluorapophyllite a=8.978, c=15.781 (1990)

Ammonian (K, NH4)Ca4Si8O20(OH) · 8H2O P 4/mnc; Marriner et al. hydroxyapophyllite a=8.988, c=15.888 (1990)

Natroapophyllite NaCa4Si8O20F · 8H2O P nnm; a=8.875, Matsueda et al. b=8.881, c=15.79 (1981)

Carletonite KNa4Ca4Si8O18[CO3]4(F,OH)·H2O P 4/mbm; Chao (1971a) a=13.178, c=16.695

As can be seen from Table 1, all minerals (except carletonite) are closely related to each other, although a substitution of potassium by sodium causes a symmetry reduction in the case of natroapophyllite. Fluorapophyllite and hydroxyapophyllite are the most abundant group members and represent end members of a solid solution series

11 of minerals with different F/OH ratio. Carletonite significantly differs from the other minerals, but is also often considered to be a member of the apophyllite group due to the unique structural motive which is exclusive for the minerals of this group. All minerals of the group possess almost identical physicochemical properties, therefore the name “apophyllite” is commonly used to describe these minerals (except carletonite). The samples used in the present study are fluorapophyllite (see Section 3.2), but for simplicity the term “apophyllite” in its general meaning will be used.

Figure 2. Apophyllite crystals (from Poona, India; longest about 1.5 cm) showing well-developed tetragonal prism terminated by the dipyramid {101}. After www.mhhe.com/earthsci/geology/hibbard/apophyllite.mhtml

Apophyllite often forms well-shaped transparent crystals with a size of up to 5-6 cm (Fig. 2). The crystals are usually colourless, white or greenish, rarely yellow, pink or violet; they possess a perfect cleavage parallel to (001); Mohs hardness is about 4.5-5. The crystals typically have a prismatic habit and show combinations of the forms {110}, {101}, and {001} (Fig. 3).

001 101

110

Figure 3. Principal habit types and habit-modifying forms of apophyllite crystals.

12 The main occurrence of apophyllite is as a secondary mineral in druses in basalts where it is often accompanied by zeolites, datolite, pectolite, and calcite. It also can be found in cavities in granites, in fissures in metamorphic rocks bordering granite, or in limestones. Among associated minerals there are prehnite, quartz, heulandite, stilbite, natrolite, analcime, datolite, cavansite, calcite, wollastonite, kinoite, gyrolite, and many other zeolites. Notable occurrences include Poona in India; Christmas Mine, Arizona; Fairfax, Virginia; Upper Peninsula, Michigan; Oregon, Pennsylvania; Paterson, New Jersey and North Carolina, USA; Rio Grande do Sul, Brazil; Isle of Skye, Scotland; Collinward, Northern Ireland; Mexico; Nova Scotia and Mont Saint-Hilaire, Canada; Kongsberg, Norway; Harz Mountains, Germany; Sampo Mine, Takahashi, Okayama, Honshu, Japan.

2.2 Apophyllite structure and its relation to other minerals

The structure of apophyllite was first determined by Taylor and Náray-Szabó (1931) and refined by later investigations (Colville et al., 1971; Chao, 1971b; Prince, 1971). It has been found that apophyllite has a layered structure (Fig. 4), where the tetrahedral silicate layers alternate with layers of Ca, K, F and OH. In contrast to many other phyllosilicates which have six-membered silicate rings, the silicate layers of apophyllite are composed of interconnected four- and eight-membered rings, with the terminal, non-bridging tetrahedral apexes of the four-membered rings alternatingly pointing up and down along the c-direction (Fig. 5a). The sheets are considerably distorted: the bases of four-membered rings are not co-planar and the average level of one ring is app. 1Å above or below the bases of an adjacent ring. Moreover, the eight- membered rings are distorted laterally, so the entire silicate sheet is puckered, although individual tetrahedra remain almost undistorted. The terminal tetrahedral apexes of adjoining layers oppose each other and form via the interim Ca-ions a ≡Si-O-Ca-O-Si≡ type bonding (Fig. 5b). Each calcium ion is surrounded by one fluorine ion, four oxygens, and two water molecules; each potassium ion is surrounded by eight water molecules. It has been suggested that the water molecule is strongly polarized in the apophyllite structure and can be considered as a hydroxyl ion which is linked to one Ca-ion, one K-ion, and a proton. The latter is

13 z

K+2 Ca + F-, OH- H+ O2-

Figure 4. Structure of apophyllite: the layers of silica tetrahedra alternate with the layers of cations, fluorine, hydroxyls, and water molecules (a = 8.96 Å; c = 15.8 Å). located between the OH- and the apex O2- as illustrated schematically in Figure 5c. Thus the water molecules play a considerable role in the structure, linking cations within the interlayer and participating in the silicate layers coupling (Taylor and Náray-Szabó, 1931). Although apophyllite is a sheet silicate, its structure resembles that of zeolites since silicate layers linked by Ca-ions make a kind of framework with alternating voids. However, the three dimensional ≡Si-O-Si≡ type bonding in zeolites provides a much stronger linkage than the linkage via Ca2+ in apophyllite. At the same time apophyllite structure can be compared with that of clays and micas. The structure of these minerals consists of tetrahedral silicate sheets, which are strongly joined by octahedrally coordinated divalent or trivalent cations into TOT layers. Between these layers are the layers of exchangeable cations (see Fig. 1). As can

14 be seen, apophyllite actually has a similar structure: silicate sheets alternate with cation layers, although the alternating tetrahedra together with the Ca-ions in apophyllite provide a stronger bonding of silicate sheets in comparison to the linkage by interlayer cations in other phyllosilicates. In this respect the linkage via Ca-ions may be rather compared with the bonds within the TOT layers in clays and micas.

a [100] b [100]

a a

up down K+2 Ca + O2- F-,OH- OH- H+ HO c Ca2+ Ca2+ 2

OH- H+ O2- Si4+

K+ Ca2+ Figure 5. Structure of apophyllite: (a) the projection along c to the silica layers shows four- and eight-member rings in (001), with 4-member rings alternatingly pointing up (dark tetrahedra) and down (light tetrahedra) the c-direction; (b) (001) projection

showing apex oxygens of SiO42 tetrahedra, and Ca, K, F, OH, and H O which link the silica sheets; oxygen atoms and water molecules lie 1 Å above and below the plane of other ions. The silica sheets are linked together by Ca ions (bonded to two oxygens from the adjacent layers) and two water molecules. Each F (OH) ion is coordinated by four Ca ions; each K ion is surrounded by eight water molecules; (c) diagram illustrating OH - H - O linkage: OH ion is linked to a Ca, K, and H ions; the proton lies between the OH-2 and O - which is also held by two calcium ions and one silicon ion (after Taylor and Náray-Szabó, 1931).

15 2.3 Apophyllite alteration

Due to its unique structure and close relation to zeolites and clays, apophyllite has been extensively studied since its first discovery about 200 years ago. Most studies concern the abnormal optical properties (Akizuki and Terada, 1998, and references therein) and the artificial dehydration behaviour (Marriner et al., 1990, and references therein) of this mineral. The latter phenomenon has attracted particular attention. It has been found that apophyllite exhibits two dehydration stages at about 310 °C and 420 °C; the exact temperature of dehydration stages depends on the chemical composition of the sample (Marriner at al., 1990). Refinement of the crystal structure has shown that all sites occupied by water molecules are equivalent (Colville et al., 1971; Prince et al., 1971), therefore an existence of several independent dehydration stages can not be attributed to structurally different sites. Chao (1971b) suggested that losing a half of the water molecules at the first dehydration stage causes the formation of a new phase which differs to apophyllite only in the environment of water molecules. Lacy (1965) has shown that the first dehydration stage causes the removal of 5/8 of all structural water, which can be released without significant changes of apophyllite structure. Activation energies for the first and second stages have been determined to be 14 and 63 kcal/mole, respectively. It is noticeable, that the first value is close to the activation energy of “zeolitic water” loss (Marriner at al., 1990). The second value, in contrast, rather indicates a rupture of Si-O-Si bonds. Therefore it was concluded that the rest of apophyllite water release is accompanied by the structural decomposition of the mineral (Lacy, 1965). These observations confirm the suggestion made by Taylor and Náray- Szabó (1931) about the important role of water in apophyllite structure. There is also a number of studies describing apophyllite alteration caused by interaction with aqueous solutions. One of the first attempts in this research area has been undertaken by Joshi and Ittyachen (1967, 1968) who etched the basal surface of apophyllite crystals using ammonium bifluoride solutions. The treatment caused a formation of square-shaped and octagonal etch pits, which had different orientation on the surface depending on the concentration of the etchant in the solution, as detected by optical microscopy. A similar study has been performed by Pande and Vadrabade (1990), but in contrast to the previous authors they used dilute HF as etchant containing

16 different concentrations of ammonium fluoride as a poisoning salt. These authors also detected the rotation of etch pits due to different concentrations of the etching compound (NH4F). In addition, they observed that lateral growth of the pits depends on the pH of the etching solution. Structural peculiarities of apophyllite suggest possible technical applications of this mineral. Although the apophyllite structure considerably differs from zeolites, its layered structure may be also used for intercalation of various compounds, e.g. large organic molecules (Lagaly, 1979). In contrast to zeolites, which may accommodate guest material without a structural change of the framework, intercalation in apophyllite requires a treatment of the mineral in order to obtain a crystalline silicic acid suitable for intercalation. This reaction involves a rupture of interlayer bonds and splitting of silicate layers accompanied with a removal of interlayer cations. There are several studies describing the formation of crystalline silicic acid from apophyllite due to the reaction with acidic aqueous solutions. Frondel (1979) reported a formation of crystalline silicic acid (Silica-AP – H2Si4O9) from fluorapophyllite by treating the powder in hydrochloric, sulphuric, and acetic acids at room temperature. A freshly-prepared acidic residue obtained shows weak diffraction patterns, which become stronger and sharper after a prolonged washing in water. By considering leaching products of several other minerals, Frondel concluded that “crystalline silica hydrates are afforded only by phyllosilicates lacking tetrahedral Al”. Lagaly and Matouschek (1980) observed the formation of crystalline silicic acid

(H8Si8O20·xH2O – “H-apophyllite”) from apophyllite due to the reaction with HCl at 0 - 4 °C. It has been suggested that the reaction involves an exchange of interlayer cation and protonation of terminal oxygens. The experiments at low temperature have been performed because of the tendency of apophyllite to form amorphous products at higher temperature. According to studies of Frondel (1979), Lagaly and Matouschek (1980) and Sogo et al. (1998) both of these hydrates have the fundamental silica sheet structure of apophyllite, although the material obtained at lower temperature possesses a higher crystallinity. Heating above 200 °C has been found to destroy the crystalline structure due to the condensation of silanol groups. In addition, Theodossiu et al. (2001) reported

17 surface amorphization of apophyllite sample due to acidic attack and therefore a total collapse of the local crystal structure. A kinetic study of apophyllite dissolution was conducted by Cave (2002), who explored dissolution of this mineral in the pH-range from 2 to 10 at room temperature. This author found that apophyllite dissolves non-stoichiometrically in acidic solutions with a preferential loss of interlayer cations and that dissolution approaches congruency in the neutral pH range. The dissolution rates above pH 4 have been found to be almost 3 times higher for samples with a high OH/F ratio. In other words, at pH 4 the apophyllite with high OH content dissolves faster than the fluorine-containing apophyllite. At neutral pH, dissolution rates of apophyllite are faster than those of kaolinite or muscovite; this has been attributed to an increased strain on the bonds in the four-membered silicate rings (Cave, 2002).

18 3. EXPERIMENTAL

3.1 Experimental methods

3.1.1 Hydrothermal atomic force microscopy (HAFM)

Atomic force microscopy is a powerful method for the investigation of the morphology, reactivity, and local properties of solid surfaces with high resolution. This technique has been invented by Binning et al. (1986). It allows imaging of surfaces in air, liquid, and vacuum. The main parts of AFM are: a probe, a piezo-actuator for probe (or sample) moving, a feedback system for controlling the forces between the probe and the sample, and a computer obtaining and processing the data. The probe consists of a base, a cantilever, and a fine sharp tip (Fig. 6), which stays in physical contact to the sample surface (contact mode). The piezo-actuator tube moves the sample in lateral and

photodiode mirror laser

base cantilever tip

sample

Figure 6. A principal scheme of atomic force microscope: the sample is scanned by the probe, which consists of the base, cantilever and a sharp tip. The beam induced by the laser is reflected from the cantilever to the photodiode, which detects cantilever deflections caused by surface morphology.

19 vertical directions by the voltage applied to different segments of the tube, so the probe scans the sample surface line by line. The microscope can be also constructed in an opposite way: the probe is mounted on the piezo actuator and moves relative to the affixed sample. The forces between the tip and the surface cause the cantilever to bend. This bending is recorded by a detection system which consists of a laser and a four-sectional photodiode (Fig. 6). The displacement of the beam reflected from the opposite side of the tip causes different illumination of the photodiode sections. The computer processes the signal obtained from the photodiode, and the feed back loop causes the piezo- actuator to adjust the sample in order to set the forces between the tip and the sample to a constant pre-set value. The output voltages of the feedback loop represent a 3-dimensional data set of the surface morphology. Images created by this data set are referred to as “height mode” images. Images, created by the photodiode output signal are referred to as “deflection mode” images. The “deflection mode” images often provide a better contrast than the “height mode” images. In AFM experiments presented here uncoated Si-cantilevers with integrated tips (spring constant: 0.1-0.3 N/m) were used. Ex-situ AFM images were obtained using a TopoMetrix TMX 1010 AFM operating in contact mode. However, the major part of this study has been performed by using the hydrothermal atomic force microscope. Since this technique is newly introduced (Higgins et al, 1998a), a brief description of the HAFM follows. This method has been developed in order to overcome the inherent constrains of commercial AFM to operate at room temperature exclusively. The microscope first presented by Higgins et al. (1998a) allows imaging of surfaces in aqueous solutions up to 150 °C and 6 atm. The method is simple in use and does not require special preparation of samples. Nevertheless, imaging under hydrothermal conditions requires considerable innovations in usual AFM design. In order to conduct experiments above the ambient boiling point of water, the solution in the fluid cell has to be pressurized; the pressure must exceed than the vapour pressure of water at a given temperature to prevent formation of bubbles which interfere with tip and laser beam. The fluid cell has to be reliably separated from all mechanical and electromechanical parts of the

20 microscope. Moreover, due to constructional peculiarities of the AFM, the pressure in all parts of the microscope should roughly be the same to provide a proper functioning of the piezo-actuator. Figure 7a shows a scheme of the main part of HAFM used in the experiments; the microscope was developed in our laboratory and is similar to that presented by Higgins et al. (1998a). The device used allows in-situ measurements of the solid-liquid interface at pressures up to 50 bars and temperatures up to 170 °C. The piezoelectric actuator (1) is mounted on the steel plate (2) and can be moved in vertical direction by a stepper motor (3). All these elements are contained in the microscope base (4) which is separated from the fluid cell (5) by the elastic chemically inert membrane (6). The piezoelectric tube ends with a thermal insulating spacer (7). The microscope base is

ab 9 d f a e g

h 5 c b 6 7 1 2

4 8 3

Figure 7. A principal scheme of the hydrothermal atomic force microscope: (a) 1 - piezoelectric actuator; 2 - steel plate; 3 - stepper motor; 4 - microscope base; 5 - fluid cell; 6 - elastic chemically inert membrane; 7 - thermal insulating spacer; 8 - gas inlet port; 9 - optical head with laser, mirror, and photodiode; (b) fluid cell: a - sapphire window; b - fluid inlet; c - fluid outlet; d - sample; e - titanium wire for sample fixation; f - AFM-cantilever; g - cell cover; h - ring heater.

21 pressurised through the gas inlet port (8). The laser, mirror, and photodiode are mounted in the optical head (9). A detailed sketch of the fluid cell is shown in Figure 7b. Most parts of the fluid cell which are in contact to the solution are made of passivated Ti; the membrane is made of Kalrez (Dupont), and the window (a) is made of sapphire. The fluid flows through the cell by means of fluid inlet (b) and outlet (c). A sample (d) can be mechanically affixed to the sample holder by a fine titanium wire (e). The wire allows fixation without any adhesive. AFM-cantilever (f) is mounted to the cell cover (g) by a fine clip with a screw. The cell is heated by a ring heater (h). The sealing of the cell is provided by O-rings between all adjoining parts. Figure 8 demonstrates a simplified principle design of the microscope, except the electrical equipment and the gas tubing (Teflon) interconnecting all microscope parts for pressure equalisation which are omitted for clarity. The bladders (1) made of

1 2

6 5

4 3

8 7

Figure 8. A simplified principle design of the microscope: 1 - bladders with solutions; 2 - stainless steel vessels; 3 - central valve for fluid flow control; 4 - preheater; 5 - main heater; 6 - cooler with circulating cold water; 7 - collector; 8 - test-tubes.

22 elastic chemically inert material (Viton) serve as liquid reservoirs and are placed in pressurized stainless steel vessels (2). The use of several reservoirs allows to exchange the solution in the fluid cell (e.g. from one pH to another) or to mix two or three solutions almost directly before entering to the cell. Solution supply is provided by gravity (height difference between the bladders and fluid cell). Fluid flow is controlled by the central valve (3) from 0 up to 10 µL/s. In front of the fluid cell the solution flows through a titanium tubing, where it is heated by a preheater (4). In the cell the solution is heated by the main heater (5). At the outlet from the cell the solution is cooled down by a cooler with circulating cold water (6) to prevent tubing damage by the hot fluid. Fluid from the cell can be either collected in the collector (7) or retrieved by test-tubes (8) for further chemical analysis.

3.1.2 Rutherford backscattering spectrometry (RBS)

Rutherford backscattering spectrometry (RBS) analysis was conducted to find out the chemical composition of the near-surface region of the samples before and after their treatment in acidic solution. The energy of the backscattered particles depends on the masses of the sample elements according to the kinematics of the scattering process. For layers below the surface, the signals are shifted towards lower energies due to the energy loss of the particles in the sample causing a step-like pattern of the spectrum. Depth information can be obtained, using known energy loss data for ions in matter and the density of the investigated material. The relative sensitivity for the elements is proportional to the square of the atomic number. The method, thus, is more sensitive to heavier elements for which traces of less than one atomic layer at the surface can be detected. The RBS spectra were measured with the 2 MeV single charged He-beam of the 4MV Dynamitron-Tandem of Ruhr-University Bochum with a beam intensity of about 10 nA. The backscattered particles were measured at an angle of 170º by a Si-detector with a resolution of 15 keV. The spectra were analysed using the program RBX (Kotai, 1994). This program calculates the shape of the spectra based on the experimental conditions and the stoichiometry of the sample as input parameters. The sample can consist of several layers and a changing stoichiometry within a layer can be described

23 by an error function representing the concentration of a species varying with depth. The simulated spectra are then compared to the measurement until a sufficient fit is achieved.

3.1.3 X-ray photoelectron spectroscopy (XPS)

Similar to RBS, X-ray photoelectron spectroscopy (XPS) also gives chemical information about the near-surface region. But in contrast to RBS, one cannot obtain a chemical depth profile from XPS spectrum - depth information is usually limited to several nanometers. However, this method allows to obtain much more precise chemical data of the investigated region. The method is based on the photoelectric effect, which occurs when an X-ray beam causes a release of valence or inner shell electrons from an atom. The kinetic energy of the ejected electron equals the difference between the photon energy and the energy required to escape from the atom (binding energy). The binding energies of electrons are characteristics of each element. The integral intensities can be used for a quantitative estimation of the mineral surface composition. In addition, XPS provides chemical bonding information. By this method all elements except H and He can be detected. The measurements were performed in a surface analysis VG / Fisons system, which is equipped with a CLAM2 analyser. The base pressure in the system is 5 · 10-9 mbar. During measurements samples were oriented in 45º with respect to the x-ray gun as well as to the analyser. The Al-Kα line was used for the emission of the photoelectrons, which were analysed with a pass-energy of 100 eV. A survey-spectrum was taken from each sample as well as more detailed spectra of the regions of interest.

3.1.4 Nuclear magnetic resonance (NMR)

NMR is an advanced technique for the determination of various molecular features, such as the molecule structure and arrangement. The method is based on the adsorption of electromagnetic energy by a substance caused by reorientation of nucleus magnetic moments. This is observed due to a combination of a strong constant magnetic field and radio waves. The frequencies of resonation are different for different

24 nucleuses. In addition, surroundings of the atom also influence the resonance frequency, thus giving information about the molecule structure. The NMR spectra were recorded on a Bruker ASX 400 NMR spectrometer at a magnetic field strength of 9.34 T. 29Si MAS NMR measurements were carried out at 79.49 MHz using a standard Bruker 7 mm MAS probe with a sample spinning rate of 3.5 kHz, a single pulse duration of 2 µs (90° pulse length 5.8 µs) and 1,000 - 1,600 scans were accumulated with a 60 s recycle delay. {1H} 29Si CPMAS NMR experiments were carried out with a 1H 90° pulse length of 9.0 µs and contact times of 250 µs, 1.0 ms, and 3.0 ms. Shifts were referenced to tetramethylsilane for 1H and 29Si. The 1H MAS NMR spectra were obtained at 400.13 MHz using a standard Bruker 4 mm MAS probe. Typical conditions were pulse lengths of 2.0 µs pulse length (90 degree pulse length 5.0 µs) and a 60 s recycle delay. 32 scans were accumulated at a MAS rotation frequency of 12 kHz for the 1H MAS NMR spectra.

3.2 Sample preparation

In the experiments, apophyllite from Poona (India) was used where it is often accompanied by heulandite. Electron microprobe analysis of the samples (Table 2) revealed that the crystals are fluorapophyllites.

Table 2. Electron microprobe analysis of the apophyllite samples.

Compound, wt. % Atomic % average (n = 100) std. dev. average std. dev.

SiO2 51.451 0.263 Si 13.168 0.123

Al2O3 0.016 0.014 Al 0.005 0.004

K2O 5.032 0.097 K 1.643 0.035 CaO 24.456 0.207 Ca 6.706 0.081 F 2.371 0.067 F 1.919 0.055

H2O 16.669 0.282 H 28.458 0.331 O 48.101 0.138 Total 99.995 100.000

25 Transparent, colourless crystals were cleaved by a knife immediately before affixing them in the fluid cell and immersing into the solution. The samples used were 0.2 - 1.5 mm thick, the size of the (001) surface was about 5-15 mm2. After the fluid cell was filled with solution, the cell was then closed, pressurized, and heated for hydrothermal experiments. The duration of experiments was up to 50 hours. Solutions were prepared using high purity deionized water (resistivity: 18 MΩ cm); the pH values of solutions were adjusted to pH 1.5 – 5.6 at room temperature by adding HCl or HNO3 and to pH 9 - 10 by adding NaOH. The flow rate of the solution through the fluid cell was set to values between 0 and 10 µl/s. For XPS measurements freshly cleaved apophyllite crystals were immersed into

HNO3-solutions at pH 2.5 for 1, 3, and 5 hours. After the treatment, the crystals were air-dried for 30 minutes and mounted into the spectrometer together with a freshly cleaved, untreated sample as a reference. For RBS experiments freshly cleaved crystals were immersed into pH 2 solution (HCl) for 10 hours and then air dried. For NMR analysis, apophyllite crystals were ground in a mortar and immersed into HNO3-solutions at pH 1 - 2.5 (initial pH) for 1 - 11 hours. The increase of pH of the solution by the progressing reaction was kept within one pH unit by choosing proper solution to sample ratios.

26 4. RESULTS AND DISCUSSION

4.1 Introduction into apophyllite alteration at acidic conditions

4.1.1 HAFM

In order to study the kinetics and morphological changes caused by the reaction of apophyllite in aqueous solution in situ AFM experiments have been performed. On the (001) surface of apophyllite, the formation of small hillocks can be observed a few minutes after the injection of the acidic solution into the AFM cell (Fig. 9). The hillocks are usually square shaped (reflecting the tetragonal symmetry), they have a flat surface,

Figure 9. Formation of square hillocks on the surface and their subsequent peeling (pH 3, T = 20 °C, P = 1 bar): (a) after 180 min exposure to acidic solution: first generation of hillocks; (b) 60 min later: formation of a second generation; (c) 70 min after the first image: start of peeling of the uppermost surface layer; (d) 85 min; and (e) 140 min after the first image: peeling progresses; (f) 150 min after the first image: new surface only with hillocks of the second and third generations. Scan field 2 × 2 µm2. All AFM images are presented in "deflection mode” and can be perceived like morphological images illuminated from the left.

27 distinct boundaries, and an identical orientation (Fig. 9a). The areal density of the hillocks is 2 - 20 µm-2. The hillocks spread with time in lateral directions and finally form a continuous layer on the surface. At pH 3, the spreading rate measured perpendicular to the straight boundaries of hillocks is 1.8 ± 0.3 nm/min. In Figure 9b, the formation of a second generation of hillocks can be observed. The second generation of hillocks preferentially occurs in the centres of the hillocks of the first generation. They have the same height, orientation, and lateral spreading rate as the first generation. Later, the formation of a new, third generation of hillocks can be observed. The sequence of the formation of hillock generations is limited by a significant additional process. After the formation of the 2nd to 4th generation, the upper layer peels off the surface by the continuously scanning tip (Fig. 9c-e); the process is often accompanied by an accumulation of the peeled material at the margins of the scan field. This shows that the upper layer becomes unstable after some period of time. The height of this peeled-off layer is about 0.7 - 0.9 nm in areas without hillocks and 1.0 - 1.2 nm in areas where the hillocks have already formed (measured at steps of partially peeled- off layers). After the removal of the upper layer (Fig. 9e-f), it can be seen that the second generation hillocks still remain and spread on the surface, whereas the hillocks belonging to the first generation have been removed along with the peeled off layer. In Figure 9f the formation of the third generation of hillocks can be observed. The next peeling process will remove the upper layer along with the hillocks of the second generation and the hillocks of the third generation will remain on the surface. These observations show that the formation of hillocks is not a growth process taking place on the surface. If a formation of a new phase was taking place on the original surface, one would expect that the peeling process would remove the youngest generation of hillocks. Here, the opposite behaviour can be observed: peeling of the surface layer removes the oldest generation. Therefore, the formation of hillocks occurs not on the surface, but underneath or, to be more exact, under the first one or two silicate layers. After the formation of the first generation, the formation of the second is initiated underneath the first generation - the younger the hillock generation, the deeper it actually is located. This process is illustrated schematically in Figure 10. Since the peeled, hillock-free layer has a height of about 0.7 - 0.9 nm (approximately

28 corresponding to one half of the c-lattice constant of apophyllite), each layer in the sketch corresponds to one silicate layer (Fig. 10a). A process taking place underneath the upper silicate layer provokes rising of the silicate layer and, as a result, we observe a laterally spreading hillock on the surface (Fig. 10b). The same process recurs under the second silicate layer (Fig. 10c). Since the size of the later formed hillock is smaller, the second generation hillock is apparently above the first one. Finally, peeling of the upper layer removes the first generation and only the hillock of the second generation remains (Fig. 10d-e).

Figure 10. Scheme of the hillock formation on the surface: (a) atomic layers of the mineral; (b) formation of the first generation hillock; (c) the same process recurs under the second silicate layer; (d) peeling removes the upper layer; (e) only the second generation hillock remains on the surface.

The hillocks are softer than the pristine apophyllite surface. Applying even medium loading forces (such as 2.5 ± 0.5 nN) causes a considerable decrease in the

29 hillock’s height. The surface of the compressed hillocks is almost on the level of the unaltered surface, leaving visible only the outline of the square shaped hillocks (Fig. 11a). However, scanning with high loading forces usually induces peeling of the upper surface layer. Decreasing the loading force to 1 ± 0.3 nN causes an increase in the height of the hillocks (Fig. 11b). By further reducing the loading forces to the limit (0.4 ± 0.2 nN), the hillocks show their maximum height (Fig. 11c).

Figure 11. Influence of different loading

forces (FL) on the hillocks height:

(a) FL = 2.5 ± 0.5 nN: decrease in the height of hillocks;

(b) FL = 1 ± 0.3 nN: increasing height of the hillocks;

Decreasing the pH-value of the solution from 3 to 2.5 - 1.5 causes further interesting features of the apophyllite-solution interface. As in the experiments at pH 3,

30 all hillocks of one generation have roughly a square shaped outline in the same orientation, and have a tendency to form a uniform layer (Fig. 12a). Also, the formation of a second generation hillock can be detected (marked by an arrow). However at lower pH values, the hillocks spread more rapidly (pH 2: 8.7 ± 1.3 nm/min). Additionally, as can be seen from Figures 12b-c, the second generation hillocks are rotated relative to the hillocks of the first generation by an angle of about 26 ± 3°. Further observations show that the hillocks of third generation have the orientation identical to that of hillocks of the first generation.

Figure 12. Effect of pH values < 3 (pH 2, T = 20 °C, P = 1 bar): (a) hillocks of the first generation; formation of the second generation hillock (marked by an arrow); (b) 13 min later and (c) 31 min after the first image: spreading of the hillocks; the hillock of the second generation is rotated by an angle of about 26 ±× 3°. Scan field 2 2 µm2.

Exposure of the sample to pH 2 for 5 hours without scanning leads to the formation of an unstable and soft layer with a thickness of about 15 - 20 nm. Scanning

31 of this surface results in a removal of this thick soft layer and uncovers a hard surface on which the formation of the hillocks continues. The observation further revealed that the rate of lateral spreading of the hillocks does not depend on their size. The rate remains roughly constant from the time when the hillock becomes visible until its coalescence with adjacent hillocks (Fig. 13).

Apart from the formation of hillocks, rising of the silicate layer could also be found at cleavage steps on the basal surface at all pH values from 1.5 to 3. The spreading rate of this process is comparable to the hillock spreading (Fig. 14).

Figure 14. Swelling of cleavage step edges and formation of square hillocks: (a) unaltered surface with cleavage steps after 15 minutes in water (pH = 5.6); (b) the surface area (slightly shifted) after exposure for 20 minutes in acidic solution (pH = 1.5); the swelling along the cleavage steps as well as hillock formation (marked by arrows) can be seen. Scan field 4.5 × 4.5 µm2.

32 The hillocks also develop at screw dislocation sites. Figure 15a shows a step, emanating at a screw dislocation on the (001) surface of apophyllite. The reaction at this step causes a fast peeling of the layer, therefore no swelling zone can be detected. In addition, small hillocks develop on the surface (Fig. 15b, c), however, their appearance is limited due to the rapid peeling of the topmost layer. The peeling off process uncovers the pristine surface, which starts to react by the formation of the small

Figure 15. Surface alteration at screw dislocation sites (pH 2): (a) a step, emanating at a screw dislocation; scan field 3.4 × 3.4 µm2.; (b) 30 min and (c) 100 min after the first image: the surface is peeled off directly after the swelling; also, small hillocks develop on the surface; scan field 8 ×µ 8 m2; (d) the layer retreats in square-spiral fashion, revealing stabilized step orientations at these conditions. Scan field 25 ×µ 25 m2.

33 hillocks. The layer retreats in square-spiral fashion, demonstrating stabilised step orientations at these conditions (Fig. 15d). As discussed in section 4.4 (Hillock rotation), the occurring step orientation depends on the solution pH. The alteration also affects the centre of the dislocation. The reaction causes the formation of a square hillock with rounded corners (Fig. 15c) at the position of the dislocation outcrop. The height of the hillock is approx. 10 nm. An excessive height of the hillock is probably due to simultaneous swelling of several layers along the dislocation line. Thus, the hillocks can be distinguished into monolayer hillocks, which affect only one layer, and multilayer hillocks, which affect several layers simultaneously. The spreading rate of the multilayer hillock is approx. 4.5 nm/min at pH 2. This is 1.5 - 2 times slower than the spreading rate of the monolayer hillocks at these conditions. The solution flowing through the fluid cell of the microscope can be exchanged in-situ. A replacement of low pH solutions by high pH solutions significantly changes the reactions taking place on the apophyllite (001) surface. Increasing the pH to 9 – 10 causes existing hillocks to decompose within seconds. However, the pristine surface remains stable within the time frame of experiments. Figure 16a shows the hillocks on apophyllite (001) surface at pH 2.5. After the exchange of solution to pH 10, the hillocks are decomposed leaving behind a negative morphology of the swollen areas (Fig. 16b-c). The replacement of the solution also affects the hillocks developing at screw dislocations. Figure 17a shows the hillock formation at a screw dislocation site on apophyllite surface at pH 2. The multilayer hillock developing at the dislocation line and monolayer hillocks at the right side of the image can be observed. The height of the multilayer hillock is about 15 nm. The reaction also affects two steps emanating at the screw dislocation; the layers are swollen and peel off just after the swelling. Each step corresponds to one silicate layer, therefore the length of the Burgers vector of the dislocation equals to one c-parameter of apophyllite. The exchange of the solution to high pH causes the decomposition of all swollen sites (Fig. 17b). The monolayer hillocks decompose leaving shallow pits on the surface with the depth corresponding to one silicate layer. The decomposition of the multilayer hillock causes a formation of a deep pit. The pyramidal shape of the pit indicates that the deeper layers were less

34 affected by the alteration. The depth of the pit is about 30 nm that corresponds to approx. 38 silicate layers.

Figure 16. In-situ exchange of solution - low pH → high pH: (a) the hillocks developing at pH 2.5; two generations of hillocks can be observed; (b) 5 min later and (c) 10 min after the first image at pH 10: the hillocks decompose rapidly, leaving behind an inverse morphology. Scan field 2.5 ×µ 2.5 m2.

The formation of hillocks can also be observed at higher temperatures (40 – 100 °C) and higher pressures (2 - 35 bar). In experiments at high temperatures solutions with pH values of 2.9 - 3 were used. As can be seen in Fig. 18, the temperature significantly influences the spreading rate: at pH 3 and 20 °C the rate is about 1.5 nm/min, at 100 °C it increases to about 200 nm/min. In spite of such high rates the hillocks generally keep their square forms, usually with rounded corners, although in some cases, especially at 100 °C, they may appear almost completely round. At these

35 Figure 17. In-situ exchange of solution (low pH → high pH) at screw dislocation sites: (a) multilayer hillock at a dislocation line and monolayer hillocks (pH 2); (b) the replacement of the solution (pH 2 → pH 10) causes a decomposition of monolayer hillocks and the multilayer hillock; a deep pit is formed. The depth of the pit is about 30 nm, that corresponds to approx. 38 silicate layers. Scan field 3.5×µ 3.5 m2. conditions, scanning with even minimum loading forces leads to an almost immediate peeling of the surface. Therefore, the observation of the development of hillocks is hampered. From the temperature dependence of the spreading rate an apparent activation energy for the growth of hillocks can be calculated. Based on the data shown in Fig. 18, this activation energy is 57 ± 4 kJ/mol.

Figure 18. Influence of temperature on the rate of hillock spreading at pH = 3, P = 35 bar. The error bars in the diagram depict the standard deviation.

36 The rate of hillock spreading also depends on pressure. The pressure dependence is weaker than the temperature dependence and has the inverse correlation: increasing pressure causes the rate of spreading to decrease (Fig. 19). For solutions with pH 2.9 at T = 20 °C, the spreading rate is about 3 nm/min at atmospheric pressure and decreases to about 1.5 nm/min at P = 35 bar.

Figure 19. Rate of lateral spreading of hillocks versus pressure in the HAFM at T = 20°C and pH = 2.9. The error bars show the simple standard deviation of the measurements.

4.1.2 RBS

In order to obtain information about the chemical composition of the near surface region of apophyllite after acidic treatment and to estimate the depth of alteration, RBS analysis has been performed. Two kinds of samples have been used: (i) apophyllite (001) surface pretreated by immersing the crystal in a pH 2 HCl-solution for 10 hours at room temperature and (ii) freshly cleaved apophyllite (001) surface. Figure 20a shows the RBS spectrum of the freshly cleaved surface of apophyllite and a superimposed simulated spectrum based on the atomic ratios as obtained by electron microprobe analysis (see Table 2). Figure 20b shows the comparison between the RBS surface spectrum of the pretreated sample and the simulated spectrum of the pristine surface of the freshly cleaved crystal. The content of F- in apophyllite is too low to definitely judge any changes in its amount due to the treatment. However, the comparison of the spectra clearly reveals a depletion of the elements calcium and potassium in the near surface region of the pretreated crystal. Simulated spectra can be fitted to this experimentally obtained data by varying the atomic ratios of the elements

37 5000 (a) 4500 Spectra 4000 Simulated Spectra

3500 O 3000 s F

unt 2500 o

C 2000 Si

1500 Ca + K 1000

500

0 300 500 700 900 1100 1300 1500 Energy of scattered α-particles (keV)

20000 Spectra 18000 (b) 16000 Simulated Spectra

14000 O Simulated spectra fresh 12000 ld s F 10000 Si

Count 8000 6000 4000 Ca + K 2000 0 300 500 700 900 1100 1300 1500 Energy of scattered α-particles (keV)

0,09 (c)

it 0,08 n 0,07 0,06 rmula u

fo 0,05 0,04 0,03 0,02

Atomic Ca+K/ 0,01 0 0 50 100 150 200 250 300 350 Depth, nm

Figure 20. RBS-spectra of the near surface region on the (001) surface of apophyllite: (a) RBS spectrum of a freshly cleaved surface superimposed by a simulated spectrum (red line) based on the electron microprobe analysis (see Table 2). (b) RBS surface spectrum of a sample immersed in pH 2 at room temperature for 10 hours superimposed by a simulated spectrum of the reacted surface (blue line). Additionally for comparison, a simulated spectrum of a freshly cleaved surface (red line) is inserted. (c) Change of Ca + K concentration with increasing depth. The depth of the altered layer is about 250 ± 20 nm. as a function of depth. Figure 20c shows the concentration of Ca + K (in formula units) with increasing depth in the near surface region of the pretreated crystal. Depletion of Ca + K is observed to a depth of 250 ± 20 nm. The error is estimated from the uncertainty in determining the depth at which the concentration reaches a constant value in Figure 20c, other experimental errors are smaller and here negligible. Below 250 nm the concentration of Ca + K is about 8 atomic percents, which is in accordance with the electron microprobe analysis of untreated apophyllite (Ca + K = 6.71 % + 1.64 % = 8.35 %). With decreasing depth this value decreases to 1.53 % at the surface. While Ca + K concentrations are clearly reduced in the near surface region, Si does not show such a depletion, indicating more or less intact silicate layers.

4.1.3 XPS

The (001) surfaces of a freshly cleaved apophyllite crystal and of crystals immersed in acidic solution were examined by XPS. In contrast to RBS, XPS is a much more surface-sensitive method and allows to analyse the chemical composition of minerals within the topmost 1 - 2 nm. The XPS survey-spectra are shown in Figure 21. In accordance to RBS experiments, XPS shows that acidic solution causes a depletion of K+ and Ca2+. Moreover, a depletion of F- can clearly be discerned. Examining the relative intensities in XPS-spectra allows to quantify the change in the chemical composition of the topmost layers of the apophyllite (001) surface relative to the untreated sample. The data, which are deduced from detailed measurements in the particular regions of interest in the spectra are shown in Table 3 (for calculations Si 2p, K 2p, Ca 2p and F 1s peaks were used).

Table 3. XPS analysis of the (001) surface of apophyllite treated in pH 2.5 solution.

Ctreated/Cuntreated [%] 1 hour 3 hours 5 hours Ca 19.8 ± 0.8 12.4 ± 0.6 4.7 ± 0.3 K 10.6 ± 0.4 8.5 ± 0.4 < 0.3 F 12.1 ± 0.5 7.5 ± 0.4 < 0.3

39 Figure 21. XPS spectra of the near-surface region on the (001) surface of apophyllite: (a) freshly cleaved surface (black); (b) sample immersed in pH 2.5 solution at room temperature for 1 hour (blue), (c) 3 hours (green) , and (d) 5 hours (red). The spectra indicate gradual depletion of K, Ca and F ions due to acidic treatment.

Comparing the cation-concentrations, Table 3 shows that the depletion of Ca2+ is significantly slower than the depletion of K+. After a treatment in pH 2.5 for 5 hours, the concentration of calcium in the topmost layers decreases to about 4.7 % (with respect to the pristine calcium concentration), whereas the concentration of potassium falls below the detection limit. As it has been shown above, RBS depth profile analysis revealed that after a treatment of apophyllite samples in pH 2 solution for 10 hours the concentration of Ca + K decreases to about 18 % at the surface. The lower cation- concentration measured by XPS can be explained by the different sensitivities of the methods: for RBS the depth of information is about 10 - 15 nm, whereas XPS analyses the chemical composition of minerals within the topmost 1 - 2 nm.

40 4.1.4 Discussion

In order to understand the behaviour of the apophyllite (001) surface in acidic solution, three basic observations have to be considered: (i) RBS and XPS indicate a change in chemical composition in the near surface region. The results show a decrease in the Ca2+ and K+ concentration. Release of these cations requires charge compensation, which may only be partially achieved by removal of F-. (ii) The observed spreading rate of the hillocks increases with decreasing pH. Therefore, the surface reaction leading to the formation and spreading of the hillocks is clearly proton promoted, making it reasonable to expect protons compensating for the loss of interlayer cations, and, moreover, making it likely that protonation drives the removal of interlayer cations into the solution. (iii) The formation of hillocks, or “localized swelling”, along with the fact that the surface of the hillocks becomes unstable and is easily removable by the AFM tip, suggests that there is a profound change in the nature of interlayer coupling. This suggestion is also supported by pH-jump experiments, indicating the decomposition of the hillocks in high pH solution. The change may possibly be related to an attachment of protons to the terminal oxygens of the silicate sheets. Directly at the surface of the pretreated sample, RBS indicates a concentration of Ca2+ + K+ of 1.53 atomic %. Due to the close atomic weights of Ca and K, it is not possible to distinguish between these two elements by RBS. The XPS-results show, however, a preferential release of K+ in comparison to Ca2+. Experiments investigating the long term reaction of the surface region of apophyllite with aqueous solutions were conducted by Cave (2002). As it was shown by this author, SEM-EDX analyses of apophyllite immersed in pH 2 solution for 4 weeks reveal only Si and O and no detectable K, Ca, or F in the outer layers of the mineral residue. Other studies of the system apophyllite - acidic solutions (batch dissolution and X-ray experiments) further revealed that the reaction leads to the formation of a crystalline silica hydrate residue by selectively releasing the Ca, F, OH and K ions from the interlayer space (Frondel, 1979; Sogo et al., 1998).

41 Thus both the long term and the present short term investigations show that Ca2+-ions are released from the surface region. Since in the apophyllite structure the silicate layers are linked together by Ca2+-ions (other ions play an inferior role in 2+ + interlayer linkage), release of Ca -ions and the formation of silanol groups by H3O entering the interlayer space is likely to cause a reduction of bonding forces between the silicate layers. Removal/replacement of Ca2+-ions, therefore, is likely to cause an increase in the distance between the layers, which may explain the hillock formation. It is further likely that additional H2O molecules enter the interlayer space as the swelling occurs, although there is no direct evidence for hydration. This reaction is schematically shown on Figure 22. The swelling can reversibly be suppressed by applying loading forces higher than ~ 3 nN on the AFM tip.

Figure 22. Scheme of alteration process of apophyllite at acidic conditions. Silica layers are linked together by Ca ions; other ions are omitted for clarity. Protonation of apex oxygens causes the removal of interlayer ions and a reduction of attractive forces

between the layers; H2O molecules may enter the interlayer space. Further assessments of the swollen areas can be made by performing in-situ pH- jump experiments (acidic pH → basic pH). The experiments reveal that the hillocks become unstable at high pH conditions and decompose rapidly, leaving behind a negative morphology of the swollen areas. The pristine surface, however, remains stable after the solution exchange. These observations support the suggestion about protonation of the apex oxygens of the silicate layers in the course of swelling at low pH. The exchange of the solution causes deprotonation of silanol groups and the decomposition of destabilised silicate layer.

42 The observations revealed that the spreading rate of the hillocks (at least of several uppermost generations) does not depend on their size. This indicates that the proton diffusion from the nucleation position to the reaction front plays a minor role of in the process of hillock spreading. This may imply that cation replacement takes place directly at the strained sites of the hillock trough the silicate layer. + The active role of H3O is emphasised by the observation that in the range of pH values studied (pH 1.5 - 3) the rate of hillock spreading rises by an order of 0.6 from approximately 1.8 nm/min at pH 3 to 19 nm/min at pH 1.5 at room temperature (Fig. 23). Furthermore, an increasing proton concentration continuously changes the orientation of the straight reaction front in subsequent layers. This observation will be discussed later in more detail.

Starting from a nucleation point, the hillocks spread laterally at straight boundaries until a complete altered layer is formed by coinciding hillocks. The same process further affects successive layers. Since the hillocks are often nucleating at the same lateral position in successive layers, it can be assumed that the point of nucleation is related to anomalies in the crystal structure like linear defects. In the case of screw dislocations, the most receptive site to acidic attack is the line of the dislocation. The line of the dislocation allows protons to penetrate inward the crystal and to attack the structure in several layers almost at the same time. Swelling of several layers at the same position results in hillocks with excessive height. The lateral spreading rate of the multilayer hillocks is lower than that of monolayer hillocks. The

43 spreading rate of the multilayer hillocks is likely to be constrained by the strain appearing due to almost simultaneous swelling of several layers. Another factor limiting the rate is diffusion of protons inward and interlayer cations outward either through the strained layers at the reaction front exclusively or additionally through the layers along the dislocation line. The replacement of the solution also affects the hillocks developing at screw dislocations, causing the decomposition of all protonated sites. While the decomposition of the monolayer hillocks causes pits with a depth corresponding to one silicate layer, the decomposition of the multilayer hillock leads to a formation of a deeper pit. By measuring the pit depth, the extent of the alteration process can be estimated. It is also possible to correlate the height of the hillock developing at low pH and the depth of the resulting pit by decomposition at high pH. The monolayer hillock is approx. 0.4 nm higher than the pristine surface. Therefore, protonation of several layers causes a hillock with the height h = n·0.4 nm, where n is the number of layers affected by the protonation. The depth of the pit (d) c caused by the decomposition of the hillock is: d = n , where c is the c-parameter of 2 apophyllite (1.58 nm). Thus, the depth of the pit and the height of the precursor hillock formed at low pH are correlated by: d ≈ 2h. As mentioned above, the depth of the altered layer, as determined by RBS is about 250 nm after 10 hours at pH 2, while according to the AFM in situ observations at similar conditions, the depth of this layer should be about 40 - 50 nm. This discrepancy can be explained by the presence of steps on the cleaved surface. Due to additional ion replacement progressing from cleavage steps, a rough stepped surface reacts as a whole much faster than the flat surface. The AFM measurements require an atomically flat surface and cover an area of not more than 40 × 40 µm2, whereas the RBS beam in the experiments covered an area of 0.5 × 0.5 mm2. According to long term experiments presented by Cave (2002) the thickness of the layer at pH 2 can reach about 1000 nm after 4 weeks. This implies that on a long term diffusion processes may play a substantial role in alteration kinetics of apophyllite. At high temperature and low pH, the silicate layers become more unstable than at room temperature, causing the silicate layers to peel off more rapidly. This process

44 impedes the detailed acquisition of quantitative kinetic data of the protonation reaction at these conditions by AFM.

4.2 Protonation stages

4.2.1 AFM

In the previous part of the work a general conception of the kinetics and mechanisms of apophyllite alteration in acidic solution was given. However, the reaction mechanisms of the apophyllite alteration at these conditions require a more detailed consideration. Figure 24a shows hillocks developing on the pristine (001) surface of apophyllite due to the interaction with acidic solution (pristine surface = area P on scheme in Fig. 24b). A thorough examination of the morphology of the hillocks reveals that the hillocks are not uniform in height but can be differentiated into a lower surrounding terrace (area T) and the higher main hillock (area H). This can be also seen on Figure 24d, showing a cross section along the line indicated on Figure 24c. The height of the terrace is 1.5 ± 0.5 Å whereas the height of the main hillock is 3.5 ± 0.5 Å. The terrace width reaches values of up to 25 nm. Based on the cross section in Figure 24d, Figure 24e shows a schematic model representing the locations of excess thickness in the apophyllite layer structure. The step between the pristine surface (P) and the terrace (T) is straight, whereas the step between the terrace (T) and the main hillock (H) is jagged. The entire hillocks (T + H) are spreading laterally, coalescing, and forming a uniform layer (Fig. 24c). The coalescence first takes place by merging of the lower terrace (T) and then by merging of the main hillocks (H). However, there is not just a single mechanism of hillock merger. Figure 25a shows two merging hillocks. After the coalescence of the terraces (T) of the two hillocks (Fig. 25b), the combined area T separates from the inner corner and moves away from the main hillocks (H). The separating area T forms a bridge-like conjunction (marked by an arrow on Fig. 25b) between the two hillocks. This conjunction moves rapidly leaving behind an apparently unaltered surface area (A) with a height similar to the pristine surface. The rapid movement of the conjunction stops when the position

45 1 2

d 1 2

H e H T T T P P

Figure 24. Formation of hillocks on the (001) pristine apophyllite surface (area P) at pH 2.5: (a) mature hillocks (area HT) are surrounded by lower terraces (area ). In the left side of the image a second generation of hillocks can be observed; scan field 2 ×µ 2 m2; (b) schematic image of (a) showing the different areas in different gray-scales; (c) 8 min later: the hillocks coalesce first by merging of the lower terrace (T) and then by merging of the main hillocks (H); (d) cross section as labeled in (c); (e) schematic model based on the cross section showing the actual location of swelling interlayers, colorless stripes: silicate layers; gray regions: protonated interlayers.

parallel to the hillock orientation has been reached, thus enveloping the two merging hillocks (Fig. 25c-e). Upon ceasing rapid movement and acquiring the enveloping position, the conjunction begins to swell (i.e., to transform into state H) and to develop its own surrounding terrace (area T) at its outer perimeter. Then, the conjunction begins

46 Figure 25. Formation of conjunctions between two hillocks at pH 2.5 (image acquisition: b, c, d, e, f = 6, 9, 12, 15, 20 min after a, respectively): (a) two merging hillocks; pristine surface PT, lower terraces and mature hillocks H can be observed; (b) in the lower inner corner of the hillocks the combined terrace separates from the mature hillocks forming a conjunction (marked by an arrow) between the hillocks; (c-e) the conjunction moves rapidly leaving behind an apparently unaltered surface area (A); a second conjunction forms in the upper inner corner between the hillocks; (f) the conjunctions stop moving when the position parallel to the hillock step orientation has been reached. The conjunctions now begin to widen in lateral dimension and develop an own surrounding terrace at the outer perimeter. Scan field 0.9 ×µ 0.9 m2. to slowly spread laterally (Fig. 25f). In the case of a merger with another hillock, the surrounding terrace again can separate and form a new conjunction. However, at the inner perimeter of a conjunction, which has been transformed into state H, no new surrounding terrace develops. This clearly indicates that the area A, through which the rapid moving conjunction has passed, behaves differently than the pristine surface (area P). The direct transition of area A into a mature hillock (state of area H) was found to be 7-10 times slower than the transformation of the pristine surface (P) into state H via state T. The step at the transition A → H is jagged and in some cases has a branched or

47 dendrite-like shape. Furthermore, within area A subsequent hillock formation is impeded even within lower layers (beneath the areas transformed to state A). The formation of conjunctions can also be observed at cleavage steps (Fig. 26).

Figure 26. Formation of conjunctions at cleavage steps (pH 2): (a) from right to left:

pristine lower layer (state PHL) - mature hillock on upper layer (state ) - upper layer in

state TP - pristine upper layer (state U); (b) 17 min later: parallel pattern of mature stripes of state H; two conjunctions (marked by arrows) can be seen approaching each other; scan field 2.7 ×µ 2.7 m2; (c) 35 min after the first image: the two conjunctions have been coalesced and stopped moving. Since on both sides of the united conjunction area A has been formed, no new conjunction can emanate from this stripe; (d) 60 min after the first image: formation of hillocks of the second generation (upper part of the scan field); area A still remains stable. Scan field 3.7 ×µ 3.7 m2.

48 The reaction initiates at the edges of the step forming a uniform sequence of states across the step (Fig. 26a; from right to left): pristine lower layer (state PL) → mature hillock on upper layer (state H) → upper layer in state T → pristine upper layer (state

PU). Then on the upper layer, state T separates and rapidly moves away from state H leaving behind an area in state A (Fig. 26b). As in the case of conjunctions at merging hillocks, new conjunctions can separate from the conjunction which has stopped moving and has differentiated into state T and H. In this way, the parallel pattern of mature stripes of state H (Figure 26b, c) is caused by a sequence of conjunctions which are separating, moving, stopping, and differentiating. Additionally in Figure 26b two conjunctions can be seen approaching each other. In Figure 26c, these conjunctions merged and stopped moving. Since on both sides of the united conjunction state A has been formed, no new conjunction can emanate from this stripe. The angles within the pattern are dominated by 90° reflecting that conjunctions stop rapid movement when they have reached directions parallel to the hillock orientation. Therefore, the stripe- pattern matches the orientation of the hillocks within the respective layer. The discrepancy between the orientation of the stripes and the hillocks in Figure 26c is caused by the different hillock orientation in successive layers at pH 2 conditions. In figure 26d the formation of hillocks of the second generation can be seen (upper part of the scan field); it is noticeable, that area A still remains stable. As the velocity of transformation P → T → H at spreading hillocks depends on the pH of the solution so does the velocity of the rapidly moving conjunctions. However, the velocity of moving conjunctions is approximately five to eight times higher than the velocity of hillock spreading. As shown in Figure 27, the rate of moving conjunctions increases from about 15 nm/min at pH 3 to about 150 nm/min at pH 1.5. The data in Figure 27 are based on the highest rates obtained during the course of conjunction movement. The movement of conjunctions towards the final (enveloping) position is often hampered and temporarily pinned (cf. the zigzag shape of the conjunction in Fig. 25c) causing the rates to vary strongly. Therefore only maximum values reflect the uninfluenced movement of conjunctions. Also as shown in Figure 24, in some cases area T is permanently attached to the main hillock (area H) and does not separate and move at all (see also Fig. 28).

49 Figure 27. Maximum observed rates of conjunction movement vs. pH of solution.

A replacement of acidic solution by basic solution causes existing hillocks, conjunctions, and the areas through which the conjunctions have passed (area A) to decompose within seconds. The pristine surface (area P) remains stable within the time frame of experiments. Figure 28a-d shows the apophyllite (001) surface at low pH with

Figure 28. In-situ exchange of solution - low pH → high pH (image acquisition: b, c, d, e, f = 3, 7, 14, 20, 24 min after a, respectively): (a-d) pH 2.5: two merging hillocks and development of the conjunction. Note that only in the inner corner to the upper left a conjunction is formed, whereas in the opposite corner the formation of a conjunction is impeded by area T being pinned to the mature hillock; (e-f) in situ exchange of the solution (pH 2.5 → pH 10): hillocks, conjunction, and area A decompose within seconds, leaving behind an inverse morphology. Scan field 2 ×µ 2 m2.

50 two merging hillocks at the left side of the images. In the upper inner corner of the hillocks area T separates and forms a conjunction (in the other corner area T is clearly pinned to the mature hillock). After the exchange of solution to high pH, the surface layer in area T, H, and A is decomposed leaving behind a negative morphology of all the areas that have been subject to alteration by interaction with acidic aqueous solution (Fig. 28e). Here it is important to note that the decomposition of area A along with area T and H clearly shows that area A is not behaving like the pristine area P and therefore is clearly altered by the rapidly passing conjunction. However, as shown by the still visible remains of area A in Figure 28f, area A is more stable than T and H.

4.2.2 NMR

1H MAS NMR

The 1H MAS NMR spectrum of untreated apophyllite in Figure 29a shows a single signal at 6.1 ppm and an extensive rotational side band pattern which implies a strong homonuclear dipolar interaction for the 1H nucleus. The signal can be assigned to water which is located in the octahedral sheets of the apophyllite (Yesinowski et al., 1988; Fechtelkord et al., 2003). Due to the small intermolecular distance of the water molecules a strong homonuclear dipolar interaction between the water molecule protons arises. The acid-treated apophyllite shows a shift of the 1H signal from 6.1 ppm to 5.2 ppm with decreasing pH value as shown in Figure 29. Moreover, the resulting linewidth decreases and the intensity of the sidebands gets lower. However, the sidebands did not change in lineshape. This can be interpreted in such a way that the original water signal is superimposed by a new narrow signal which gets stronger in intensity with decreasing pH-value but shows nearly no sideband intensity. The signal can be assigned to silanol groups (≡Si-OH); the protons of these groups have no direct dipolar interaction and thus no sideband intensity. Additionally, in contrast to water, there is no hydrogen bonding to other OH groups causing a narrower signal.

51 2 5.

c) * * * * * * 5.5

* * * * * * * * b) * 6.1 * * * * a) * * * * * 100 50 0 -50 -100 (ppm)

Figure 29. 1H MAS NMR spectra of (a) untreated apophyllite, (b) treated at pH 2.5 for 1 hour and (c) at pH 1.0 for 1 hour. Asterisks mark spinning sidebands.

29Si MAS NMR and {1H} 29Si CPMAS NMR

The 29Si MAS NMR spectrum of untreated apophyllite in Figure 30a shows a 29 - single Si MAS NMR signal at -92.9 ppm caused by [Si(OSi)3O ] groups (SiO4-groups with three bridging and one terminal oxygen). After the acid treatment three additional signals appear at -99.0, -101.7, and -103.4 ppm (Fig. 30b). The three signals can be assigned to three different protonated silicon sites [Si(OSi)3OH] (the terminal oxygen is protonated) with different tetrahedral bonding angles. This interpretation is supported by {1H} 29Si CPMAS NMR experiments with varying contact times as shown in Figure 31. For a short contact time of 250 µs (Fig. 31a) the 3 new components show a strong relative increase in signal intensity compared to the original apophyllite signal. With

Figure 30. 29Si MAS NMR spectra of (a) untreated apophyllite, and (b) treated at pH 1 for 1 hour. short contact times only those protons which are in a short distance to silicon sites (i.e., silanol protons) can transfer magnetisation. Therefore, the signal increase can be considered as a clear proof that these three signals are caused by silanol groups. With increasing contact time (Fig 31b, c) the signals at -98.9 and -101.2 ppm increase in relative intensity, showing that in this case also close-by water molecules transfer magnetisation. The intensity distribution at large contact times indicates that the signal at -103.0 ppm refers to the smallest amount of water molecules in the neighbourhood. According to the 29Si MAS NMR spectrum (Fig. 31b), the most frequent silanol groups are those with the least amount of water in their vicinity. After a prolonged acidic treatment of the apophyllite sample (11 hours in pH 1.6 solution), the 29Si MAS NMR spectrum additionally exhibits a broad signal with a maximum at -111.0 ppm. This signal indicates Si(OSi)4-type silica with varying bonding parameters comparable to poorly crystalline or even amorphous silica. In {1H} 29 Si CPMAS NMR experiments the Si(OSi)4-type signal is completely absent, indicating that this signal is not related to silanol groups. It is further important to note, that a signal of [Si(OSi)2(OH)2] – groups forming due to protonation of silicate tetrahedra at sheets edges was too weak to be detected by NMR.

53 Figure 31. {12H} 9Si CPMAS MAS NMR spectra of apophyllite treated at pH 1 for 1 hour with contact times of (a) 250 sm, (b) 1 s and (c) 3 ms.

4.2.3 Discussion

The hillocks nucleate at defect sites or steps on the apophyllite surface. At these + positions, H3O -molecules start to penetrate into the interlayer region and trigger an ion- replacement reaction. The reaction proceeds by spreading hillocks with straight steps. Therefore, alteration of the apophyllite (001) surface in acidic solution can be described by an ion-replacement reaction causing a protonated and cation-depleted residue. However, the AFM experiments reveal that this replacement reaction takes place in two or three successive stages that cause different morphological patterns. Thus, it can be inferred that the overall-replacement reaction consists of a dual-step or triple-step process. The dual-step process is represented by the transformation of the pristine

54 surface (area P) to terrace (area T) and by the successive transformation of area T to the mature hillock (area H). The triple-step process comprises the transformations P → T → A → H. Each of these four transformations (P → T, T → H, T → A, A → H) has distinct properties and, therefore, corresponds to a separate reaction. NMR experiments reveal the mechanism of reaction as a protonation of the terminal apex-position of the silicate tetrahedra forming silanol groups and, as we have seen in the previous chapter, RBS and XPS indicate a removal of interlayer cations. The NMR results further show that three different protonation sites develop in apophyllite crystals in acidic solution. The three types of ≡O3SiOH-groups show differences in their amount of close-by hydrogen and structural differences, which, however, are scarcely related to the crystal structure itself due to equivalence of the structural units in apophyllite. The AFM experiments show that the formation of mature hillocks (i.e. the formation of maximum protonated silicate layers and fully cation-depleted interlayers – area H) takes places discontinuously in up to 3 transformation steps. This discontinuity and the fact that all three transformations are proton promoted makes it reasonable to assume that the transformations are associated with distinct protonation steps and types of silanol groups. If a mixture of all three protonation types was taking place initially, a continuous morphological transformation of the pristine surface to the mature hillocks would be expected.

Transformation P → T

+ Since protonation via H3O implies the uptake of three hydrogen atoms in total, silanol groups with the least amount of close-by hydrogen can be expected to be found in an early stage of apophyllite protonation. Therefore, the type of protonation taking place in the initial transformation process P → T is possibly causing the silanol group giving rise to the 29Si MAS NMR signal at -103.0 ppm (i.e., the signal associated to the least amount of close-by hydrogen). The 29Si MAS NMR signals of the other two silanol groups indicate higher amount of close-by hydrogen and thus are more likely associated to silanol groups formed by subsequent transformations, as more water molecules diffuse into the interlayer space. The XPS-results show a preferential release of K+ in comparison to Ca2+. Therefore, it is reasonable to assume that the release of K+

55 is rather part of the initial transformation P → T than part of subsequent transformations. When terraces coalesce (area T) i.e., when the transformation fronts P → T of two hillocks meet (Fig. 25), the reaction rate tremendously increases until the front reaches the position marked by the shape of an enveloping hillock. The rapid P → T transformation also takes place at steps which are not directed parallel to the according hillock orientation (Fig. 26). This behaviour has a remarkable likeness to the behaviour of straight steps at coalescing etch pits during calcite and magnesite dissolution (Jordan and Rammensee, 1998; Jordan et al., 2001). At these coalescing pits, jagged and unbounded steps are generated which retreat at high rates until the pit has reached the shape of an enveloping pit with straight steps. The requirements for straight shape steps can be described by the terrace-ledge-kink-model (e.g., Liang et al., 1996; Jordan et al., 2001). At straight steps, the removal of units parallel to the step direction (i.e., removal at kinks) is much faster than the removal of units perpendicular to the step direction (i.e., nucleation of kink sites or “double kink sites”). At jagged steps formed by coalescing straight steps, material can dissolve without the slow nucleation of double kinks sites. Therefore, the step retreat rate increases and the steps become jagged or zigzag-shaped. The retreat rate of jagged steps is not hampered by the slow nucleation rate of double kink sites. Therefore, coalescing jagged steps cannot generate steps with an enhanced retreat rate. The remarkable likeness of the behaviour of transformation reaction P → T on the (001) surface of apophyllite to carbonate dissolution suggests analogy. Since protonation of the silicate apexes in apophyllite is the process corresponding to the removal of material at carbonate steps, it can be inferred that protonation parallel to the direction of the reaction front is much faster than perpendicular (Fig. 32). Also, impurities or defects suspected to cause pinning of step retreat on calcite (Jordan & Rammensee, 1998) may likely be responsible for the pinning of the rapid P → T transformation.

Transformation T → A

The transformation T → A exclusively takes place subsequent to the rapid P → T transformation. In analogy to carbonate dissolution, the jagged transformation front

56 Rs T Rf H Rs A

H Rs

Figure 32. Scheme showing a merger of two hillocks: at the coalescing hillocks, jagged and unbounded conjunctions are generated which advance at high rates (Rf- fast rate, PT →→ A reaction) tending to reach the position of an enveloping hillock with straight steps. The Rf is faster than Rs (slow rate), the rate of the P →→ TH transformation at straight fronts.

T → H was never observed to move as rapid as the straight P → T transformation front. Therefore, in the case of rapid P → T transformation the width of area T should increase noticeably, since the T → H transformation cannot attain the same rate. This increased width of area T would correspond to a longer residence time of a certain location in the protonated state T. However according to the observations, the residence time of the protonation state T induced by the rapid P → T transformation does not increase, i.e. the terrace always keeps roughly the same width. Since the T → H transformation does not take place in time, a transformation into state A follows. The state A is morphologically characterised by a decreased height (in comparison with state T), approximately at the level of the pristine surface. In contrast to state T, state A exhibits a surprisingly high stability: it even has better resistivity against the transformation into the state of maximum protonation than the pristine surface, i.e. A → H is slower than P → T → H. Further assessments of state A can be made by in-situ pH-jump experiments (acidic pH → basic pH). In the high pH solution not only the hillock areas T and H decompose quickly but also area A. The pristine surface (area P) remains stable in high pH solutions within the time scale of experiments. This clearly shows that area A is

57 altered with respect to the pristine surface. Also, the sensitivity of state A to high pH- solutions makes it reasonable to assume that state A still has protonated silicate groups despite the strong decrease in surface height. However, there is little information whether the transformation T → A involves a further protonation, i.e., an increased density of protonated silicate apexes within the apophyllite interlayer. Even if no further protonation was associated with the T → A transformation, the structural modifications induced by T → A likely cause modifications of the Si-OH bonding parameters and of the amount of close-by hydrogen. Considering surface silanol groups, even slight structural modifications have shown to exert considerable influence on bonding parameters (Bickmore et al., 2003, 2004). Thus, the transformation T → A can not be excluded of being responsible for the formation of the second type of silanol-groups detected by NMR.

Transformation T → H and A → H, amorphous structures

The final transformation into the mature hillock (state H) can take place either from the highly stable state A or from terrace state T. Both transformations are proton promoted reactions, but the transformation rates are noticeably different. After the transformation fronts T → H or A → H have passed on the surface, further changes in surface morphology or layer height cannot be detected by AFM. Since state H is highly sensitive to the forces exerted by the scanning tip, it does not bear a prolonged probing and, therefore, an assessment of a progressive alteration of the structure beyond state H is not possible. In this respect, NMR analysis points towards a progressive structural degradation manifested in the broad signal around -111.0 ppm after a prolonged acidic treatment (11 hours at pH 1.6) of the apophyllite sample. This structural degradation is in accordance with the results reported by Lagaly and Matouschek (1980) and Theodossiu et al. (2001).

58 4.3. Alteration at pH 4 - 5.6

4.3.1 Results

In the two previous sections the kinetics and reaction mechanisms of apophyllite have been elucidated in the pH range from 1.5 to 3. In the present section the alteration of apophyllite at pH 4 - 5.6 is considered. Since the concentration of protons at these conditions is lower than at pH 1.5 – 3, the protonation rate (P → T → H) is considerably slower. Therefore the reacted layers are not peeled off so readily. This allows a prolonged probing of the surface and the detection of further alteration processes of apophyllite layers is possible. Figure 33a shows the apophyllite surface at pH 4. Initially, the hillocks which develop on the surface are similar to those obtained at lower pH values (i.e. pH 1.5 - 3). However, subsequent morphological changes indicate that the protonated surface in state H undergoes further alteration. Due to the fast delamination of the layers at lower pH, this further alteration stage could not be observed at pH < 4. As it can be seen in Figure 33b, the central area of the hillocks (H) becomes rough and increases in height by almost one order of magnitude (up to 2 - 3 nm). Thus, the hillocks eventually are transforming into a further alteration state, here called high hillocks (HH). The surface in this state exhibits a higher resistance to mechanical alteration by the scanning probe in comparison to the states H and T. Although a mechanical alteration of HH-surface requires higher loading forces of the scanning tip than in the case of state H, it is in general possible to partially peel off the HH-surface by scanning tip (Figure 33c). It should also be noted, that at pH 4 the transformation into HH-state was never detected without the precursor protonation of pristine surface (P → T → H transition). Stopping the flow of pH 4 solution through the fluid cell causes the rate of protonation reaction (P → T → H transition) to slow down after several hours. However, the reaction H → HH advances independently with the same rate. Thus the reaction front H → HH approaches the P → T → H reaction steps. At pH values close to neutral (pH 5.6) and room temperature any alteration can barely be detected within several hours. Therefore, most experiments at this pH were carried out at elevated temperature that allows to increase the alteration rates

59 Figure 33. Alteration of apophyllite surface at pH 4: (a) formation of hillocks on the surface (PT →→ H transformation); (b) 46 min later and (c) 105 min after the first image: central area of the hillocks (H) becomes rough and significantly increases in height (up to 2 - 3 nm) and transforms into another alteration state: high hillocks (HH). Scan field 2 ×µ 2 m2.

significantly. Figure 34 shows hillocks developing on apophyllite surface at 50 °C. The square shaped hillocks spread slowly (4 ± 0.5 nm/min) in lateral directions, coalesce, and finally cover the surface. The height of the hillocks is about 2.5 ± 0.5 nm. The morphological transformation of pristine surface into this hillock state takes place directly: during the transition from P to HH the intermediate reaction states T and H could not be resolved within the resolution limits of AFM. Further information about the alteration processes at the apophyllite surface can be gained from pH-jump experiments. Figure 35a shows hillocks generated by protonation at pH 2.5 (P → T → H transition). In-situ exchange of the solution (pH 2.5 → pH 5.6) causes a deceleration of the protonation reaction and a transformation of the

60 Figure 34. Formation of high hillocks (HH) on apophyllite surface at pH 5.6, T = 50 °C, P = 9 bar: (a) the height of the hillocks is about 2.5 ± 0.5 nm; no low-hillock rim can be discerned; (b) 20 min and (c) 43 min after the first image: the hillocks spread laterally and cover the surface. Scan field 1.35 ×µ 1.35 m2.

states T and H into HH (Fig. 35b). Thus, the single protonated layer is transformed into HH-state with a height of 2.5 ± 0.5 nm. Further alteration of this surface area caused by reverse in-situ solution exchange can be followed on Figure 36. The solution exchange pH 5.6 → pH 2.5 causes a recommencement of fast protonation. In Figs. 36a-h again a protonation reaction can be resolved that precedes the HH formation. Also new hillocks develop on the pristine surface. Within the preceding reaction individual reaction steps can be resolved. The hillocks coalesce (indicated by arrows) and form conjunctions and area A. Recurring exchange of the solution (pH 2.5 → pH 5.6, Fig. 36i-k) causes the transformation of the protonated areas (except area A) into the HH-state. A pH jump from 5.6 to basic

61 Figure 35. In-situ exchange of solution - low pH → pH 5.6: (a) hillocks developing at pH 2.5; areas T and H can be discerned; (b) in situ exchange of the solution (pH 2.5 → pH 5.6) caused areas TH and to transform into high hillocks (HH). Scan field 1.7 × 1.7 µm.2 conditions (e.g. pH 10) causes the high hillocks to decompose, leaving square etch pits with a depth of one half of the c-parameter ~ 8 Å (Fig. 36l). Once the surface layer is completely transformed into the HH state, no further morphological changes can be observed on the surface. Therefore alteration progressing into the bulk crystal cannot be followed by the number of hillock generations visible on the surface, as it was in the case of the low hillocks. However, pH-jump experiments (pH 5.6 → basic pH) allow to estimate the extent of alteration. The solution exchange after a prolonged treatment at pH 5.6 reveals that the alteration process not only affects the topmost layer but also proceeds deeper into the crystal. By measuring the depth of the pit that forms after the jump to high pH the number of affected layers can be measured. The height of the altered zone and the number of affected layers can also be measured at high hillocks which emanate from cleavage steps on the surface. For example at cleavage steps consisting of three silicate layers, HH-heights of even about 7 – 8 nm could be measured. Based on the data shown in Fig. 37, the activation energy of the lateral spreading rate of high hillocks at pH 5.6 is 42 ± 5 kJ/mol. The formation of high hillocks could be

62 Figure 36. In-situ exchange of solutions - pH 5.6 →→ low pH pH 5.6 → high pH: (a) apophyllite surface with high hillocks at pH 5.6 (zoom out of the Fig. 35); (b - h) in-situ exchange of solution (pH 5.6 →→ pH 2.5) causes the PT → H or P → T → A transformations recommences around the HH; arrows indicate the formation of conjunctions at coalescing hillocks; area HH remains stable within this period (about 105 min); (i - k) reverse exchange of the solution (pH 2.5 → pH 5.6) causes areas TH and transform to the state HH, while area A remains stable within this period (about 20 min); (l) exchange of the solution (pH 5.6 → pH 10) leads to decomposition of areas HH and A. Scan field 2.7 × 2.7 µm2.

63 Figure 37. Influence of temperature on the rate of high hillock spreading at pH 5.6, P = 35 bar. observed up to about 110 °C, although already at 80 °C the spreading rate is too high to be quantitatively measured: the surface is transformed to the final rough state within seconds, if more than 5 - 7 nucleations per µm2 take place. However, in the case there are only a few linear defects (e.g. screw dislocations) present, the reaction still can be followed. Figure 38a shows large high hillocks developing at 110 °C. The height of the hillocks reaches 150 ± 20 nm that means that several tens of layers were altered. As can be seen in Figure 38b, the hillocks rapidly spread in lateral directions (170 ± 10 nm/min). Due to the simultaneous swelling of many layers, stress accumulates within the reacted sheets that leads to warping and cracking of the hillocks. Due to the fast decomposition of the altered layers by applying high pH solution, large etch pits are formed on the surface (depth up to 400 nm, Fig. 38c). The decomposition of altered material can also be attained at pH 5.6 by increasing temperature to more than 110 °C. At these temperatures the dissolution proceeds by retreating mono- and multilayer steps on the surface. Also large etch pits could be observed. Figure 39 shows the apophyllite surface after a pretreatment in pH 5.6 at 130 °C and 10 bar for about 3 hours. The surface is covered by large square etch pits. Based on the etch pits developing at these conditions, the dissolution rate has been estimated to be about 5 × 10-7 mol/m2s (pH 5.6, T = 130 °C, P = 10 bars).

4.3.2 Discussion

As it has been shown in sections 4.1 and 4.2, at low pH conditions (pH 1.5 – 3) the apophyllite (001) surface reacts by the formation of morphologically different

64 Figure 38. Formation of multilayer high hillocks at pH 5.6, T = 110 °C, P = 7 bar: (a) multilayer high hillocks developing at linear defects; (b) 12 min later: spreading of the hillocks; (c) 44 min after the first image: exchange of the solution (pH 5.6 → pH 10) causes the decomposition of areas HH. Also heating to temperatures above 110 °C at pH 5.6 (without the exchange to high pH) would have caused the decomposition of the high hillocks. Scan field 35 ×µ 35 m2.

swelling patterns (terraces, mature hillocks, and metastable states). Increasing the pH of solution causes further structural and textural changes of the reacted layers: at pH > 3 the development of high hillocks was observed. A permanent fluid flow through the microscope cell safely excludes the possibility that the high hillocks merely are a precipitation of secondary phases on the surface. Furthermore, pH-jump experiments also exclude this possibility clearly. At low pH conditions (pH 1.5 - 3) apophyllite swelling is caused by the protonation of silicate layers and the leaching of interlayer ions. However, the existence of an additional alteration state at pH 4 indicates that at pH 1.5 - 3 the protonated silica residue just represents an intermediate reacting state and does undergo further

65 Figure 39. Ex-situ AFM image of etch pits developed on the (001) surface of apophyllite after 3 hours at 130 °C (pH = 5.6, P = 10 bar). Scan field 100 ×µ 100 m2.

alteration. According to Casey and Bunker (1990), the initial stage of silicate leaching comprises a diffusion of protons and water into the bulk material, which partially hydrolyse the mineral structure and exchange alkali and alkali-earth ions. In the next stage of reaction, the silica-enriched residue may repolymerise and form a porous silica network. A similar reaction in the case of phyllosilicates has been reported by Kaviratna and Pinnavaia (1994), who investigated the products of acidic treated fluorohectorite and phlogopite by XRD, MAS NMR and BET surface area analysis. Both minerals have been found to react similarly, however the difference in swelling ability caused different textural properties of the resulting amorphous product. According to Kaviratna and Pinnavaia (1994), acidic treatment of these minerals involves a separation of TOT layers that allows protons and water to penetrate between the tetrahedral sheets. A subsequent cross-linking of the solvated sheets produces an amorphous residue with different amounts of water and protons between the sheets. The surface area of the hydrolysis product of fluorohectorite was found to be about 7.5 times greater than the surface area of the phlogopite-derived residue, thus indicating that swelling characteristics play an important role in the textural properties of the resulting product. A similar reaction may take place in apophyllite. It can be assumed that the two first alteration stages (T and H) represent cation replacement and separation of silica

66 + sheets. During prolonged treatment, water and H3O -molecules attack the protonated layer and initiate the rupture of bridging Si-O-Si bonds. An increasing durability of HH- area in comparison to H-area (as detected by scanning with high loading forces) suggests a partial cross-linking of the layers probably due to condensation of two Si-OH groups. Thus, the formation of HH-area may be due to a partial depolymerization within the sheets and partial cross-linking of protonated silica layers. The high rough hillocks on apophyllite indicate the formation of a porous material with a high amount of water within the cages formed by the cross-linked silicate layers. This suggestion is supported by the NMR results, showing various protonated states and eventually Si(OSi)4-type silica with varying bonding parameters caused by a prolonged treatment. The surface alteration at pH 5.6 can be considered in a way that the formation of HH areas takes place directly after the ion-replacement reaction. The lower proton concentration at pH 5.6 reduces the protonation rate. Comparison of the spreading rates of high hillocks at different pH values reveals that the rate does not depend on pH of the solution. Therefore within the spatial resolution of the images, protonation and HH formation coincide. Thus, at pH 5.6 the protonation is becoming the rate determining step in apophyllite surface alteration. The experiments at pH 4 without flow also support this assumption. In these experiments, the rate of P → T → H transition slows down, while H → HH transition proceeds at the same rate resulting in a narrowing of the protonated rim around the high hillocks due to the advance of the H → HH reaction front. The transformation of protonated layers into HH state at low pH (pH 1.5 – 3) was not detected within a timeframe of experiment. As it has been shown in section 4.1, even scanning with low loading forces causes peeling of the surface at the low pH conditions. A prolonged treatment of the sample at these conditions (e.g. 5 – 6 hours at pH 2) without scanning may cause a transformation of up to 7 - 10 layers into state H without subsequent H → HH reaction within the timeframe of the treatment. This shows that at pH < 4 the H → HH reaction is slower than at pH ≥ 4. Therefore, it is very likely that a sufficient amount of protons in the solution at least temporarily slows the H → HH reaction and amorphization of the residual structure. There is, however, less doubt that a final product of acidic alteration of apophyllite is structurally amorphous. This was confirmed by NMR spectra of prolonged acidic-treated apophyllite samples,

67 indicating an amorphization of the final product. Also, Lagaly and Matouschek (1980) and Theodossiu et al. (2001) reported an amorphization of apophyllite treated in acid at room temperature. However the results presented here contradict to those reported by Frondel (1979). According to this author, severe acidic treatment of apophyllite powder results in a low-crystalline residue; but a prolonged washing of the residue with H2O causes an increasing crystallinity. The in-situ pH-jump experiments indicate that the opposite process takes place: an exchange of acidic solution by neutral solution leads to a rapid amorphization of protonated layers. The observed discrepancy may be explained by the preparation method used in the study conducted by Frondel (1979): during the prolonged washing most part of amorphous material might be removed from the residue and thus caused a relative increase of crystalline material. The AFM-observations allow to distinct two different morphological alteration pattern – hillock development below 110 °C and dissolution by detachment of material at the layers as well as spreading etch pits above this temperature. At 110 °C, two competing processes take place: protonation of apex oxygens and replacement of interlayer cations on the one hand and depolymerization of the silica sheet on the other hand. It has been noted that the spreading rate of etch pits at 130 °C is slower than the rate of high hillock spreading at 90 °C. This means that at higher temperature the protonation of apex oxygens (which is the key-step of high hillock formation) becomes less dominant. At 110 °C, the rate of attack to the Si-O-Si bonds increases and can compete with the proton attack to the apex oxygens. Therefore depolymerization of Si- O-Si bonds within the sheets becomes more and more the dominating process. Thus, the AFM-studies indicate that at close to neutral pH above 110 °C apophyllite dissolves congruently by the decomposition of the silicate layers without a preferential leaching of the interlayer cations and without a formation of an amorphous residue.

68 4.4 Hillock rotation

4.4.1 Results

It has been shown in Section 4.1, that the spreading rate of the hillocks depends on the pH of the solution. However, the pH value also affects the orientation of the hillocks developing on the apophyllite surface at acidic conditions. Figure 40a shows hillocks developing on the pristine (001) surface of apophyllite at pH 4. Two generations of hillocks can be observed. The hillocks in adjacent layers are rotated by 0 – 5°. Thus at these pH values, the hillocks in all layers have roughly the same orientation on the surface. Figure 40b shows an example of hillocks in adjacent layers under lower pH conditions (approx. pH 2). At this pH, the hillocks in successive generations are rotated by an angle of 26 ± 4°. Furthermore as it can be seen in Fig. 41, the hillocks of the third generation have the same orientation as the hillocks of the first generation. Also, hillocks of the second and forth generations have identical orientation (not shown). Thus at pH 2, the hillock orientation is toggling back and forth in successive layers by approximately 26°.

Figure 40. Hillock formation at different pH values: (a) at pH 4 the hillocks of different generations have roughly the same orientation (scan field 1.7 ×µ 1.7 m2); (b) at pH 2 the hillocks in successive generations are rotated by an angle of 26 ± 4° (scan field 2.4 ×µ 2.4 m2).

69 Figure 41. Hillock orientation at pH 2 (image acquisition: b, c, d, e, f = 20, 28, 35, 46, 60 min after a, respectively): (a) hillocks of the first generation; (b - d) formation of the second generation hillocks and their spreading; the second generation is rotated by an angle of 26 ± 3°; (e) formation of the third generation (marked by an arrow); (f) orientation of third generation hillocks is identical to the orientation of the first generation. Scan field 1.5 ×µ 1.5 m2.

Experiments at pH 1.5 show that the hillocks of different generations again have approximately identical orientation. It has been found, however, that at pH 1.5 the crystallographic hillock direction is approximately parallel to [110], whereas at pH > 3 the hillocks are oriented almost parallel to [100]. The hillocks at pH 2 are rotated from [110] by about 15°. Another example showing that orientation of hillocks depends on the solution pH is shown in Figure 42. The hillocks in Figure 42a (pH 3) lie approx. parallel to [110]. An exchange of the solution to pH 2 causes the hillocks to change their shape: they first become rounded (Fig. 42b) and eventually obtain a roughly square shape with an orientation rotated by about 26° ± 5° with respect to the initial hillocks at pH 3 (Fig. 42c).

70 Figure 42. In-situ exchange of solutions - pH 3 → pH 2: (a) hillock formation at pH 3; (b) 18 min later: the exchange of the solution (pH 3 → pH 2) causes the hillocks change their shape and become rounded; (c) 39 min after the first image: resulting hillocks at pH 2 possess roughly square shape with an orientation rotated by 26 ± 5° with respect to the initial hillocks at pH 3. Scan field 3.3 ×µ 3.3 m2.

4.4.2 Discussion

The hillocks developing on the apophyllite basal surface under low-pH conditions represent a swelling process which is due to an ion-replacement reaction between the interlayer ions and protons. The fundamental role of H+ can be emphasised by two facts: 1) the reaction rate (hillock spreading rate) increases with decreasing pH and 2) hillock orientation on the surface depends on the pH. The driving force of the reaction is the protonation of apex oxygens of silicate tetrahedra. The sides of the hillocks morphologically mark the reaction fronts of protonation which changes its orientation depending on the amount of protons in the solution. The first protonation

71 step (P → T transformation) determines the direction of the following reaction. The reaction fronts at different pH can be attributed to certain crystallographic directions. Since the protonation is a dynamic process and the reaction front is continuously moving, it is difficult to ascribe a protonation pathway precisely to distinct structural units. However, at close to equilibrium conditions (pH 3 – 5.6) the reaction front is oriented approximately parallel to [100] direction. The orientation of this front is schematically shown on Figure 43 by a solid line passing through the groups of tetrahedra pointing towards the interlayer of interest and thus marking the positions of interlayer linkage. The pale 4-membered rings of tetrahedra have their apex pointing towards the neighbouring interlayer. An arrow shows the direction of the reaction (front propagation).

[100]

pH 3 - 5.6 pH 2 pH 1.5

Figure 43. Structural relation of the reaction front orientation at different pH. The arrows show the direction of front propagation. The front can be perceived as a row of units consisting of 4-membered rings, which mark the positions of interlayer linkage; one unit is marked by a red dashed square. Green dotted lines indicate the probable course of the straight fronts at pH 1.5 and 2.

72 The reacting layer can be considered as a number of elementary units where each 4-membered ring represents a single unit. In the case of pH 3 - 5.6 the row of units is oriented along [100] direction; the distance between the units along the reaction front is closest at these conditions. Protonation of the first unit within a complete row (marked by a red dashed square in Figure 43, pH 3-5.6) represents a nucleation process. The units adjacent to the protonated unit are oriented differently with respect to the nucleation unit, therefore their protonation rates are likely to differ. Close to equilibrium (pH 3 – 5.6), nucleation rates generally are slow. Therefore, protonation rates of neighbouring units are very likely to be faster than the nucleation rate in complete rows, resulting in a straight reaction front. Thus at these conditions, the nucleation rate at complete rows is rate determining; the difference between the protonation rates of the neighbouring units does not play a substantial role, and the reaction front is oriented more or less parallel to [100]. Increasing of proton concentration in the solution causes a deviation of the front from the close-to-equilibrium orientation. As can be seen in Figure 43, decreasing of pH causes a gradual transition of the front orientation from [100] at pH 3 – 5.6 to [110] at pH 1.5 passing trough approx. [210] at pH 2. The rotation of the reaction front can be caused by the change of protonation rates, which increases at far-from-equilibrium conditions (pH < 3). Also, the difference in protonation rates of the neighbouring units becomes more substantial. Thus, under more acidic conditions, the rate of the P → T → H transformation increases considerably, showing a dynamic front, which does not necessarily reflect the closest distances. Further information about the orientation of the reaction fronts can be gained by considering the formation of conjunctions at coalescing hillocks. It has been shown in the section 4.2, that when terraces (area T) of two hillocks coalesce, the reaction rate considerably increases until the front reaches the position marked by the shape of an enveloping hillock. The scheme on Figure 44a shows the coalescing of two hillocks at pH 3 with regard to the apophyllite structure. The reaction front P → T → H is oriented parallel to [100]. When the hillocks coalesce, the combined terrace detaches from the H- area, forming a conjunction between the hillocks. The conjunction moves towards the enveloping position, leaving behind the area A. The reaction front P → T → A is oriented parallel to [110] (at least at the initial stage of the reaction, before the

73 a [100] b

H A A H H A A H T P T P

Figure 44. Scheme of the reaction at two coalescing hillocks: (a) pH 3: PT →→ H front is oriented parallel to [100]; PT →→ A front is oriented parallel to [110] (before the conjunctions begin to curve); arrows show the direction of PT →→ A reaction. (b) pH 1.5: PT →→ H front is oriented parallel to [110]; PT →→ A front is oriented parallel to [100]. conjunctions begin to curve). Thus, the orientation of P → T → A reaction front at pH 3 is close to the orientation of P → T → H transformation front at pH 1.5, which is also almost parallel to [110]. According to the results presented in section 4.2.1, the maximum rate of P → T → A reaction at pH 3 is about 15 nm/min. This value is close to the rate of P → T → H transformation at pH 1.5 which is 19 ± 4 nm/min (see section 4.1.4). At this point, it might be assumed that the rate of P → T transformation for pH 1.5 and 3 is constant for a certain direction. In other words, in the case of the P → T front oriented parallel to [110] the rate of the transformation is about 15 – 19 nm/min for both - pH 1.5 and 3. However, the reactions at coalescing hillocks at pH 1.5 - 2.5 reveal that the behaviour of the reaction fronts does not support this assumption. The scheme on Figure 44b shows the coalescing of two hillocks at pH 1.5. The reaction front P → T → H at these conditions is oriented parallel to [110], the reaction rate is approx. 19 nm/min. Similarly to the case of pH 3, the combined terrace detaches from the H-area, generating a conjunction between the hillocks, and causing the formation of area A (P → T → A reaction). The reaction front is oriented parallel to

74 [100]. Thus, the orientation of P → T → A reaction front at pH 1.5 is close to the orientation of P → T → H transformation front at pH 3, which is also almost parallel to [100]. However, the rate of the P → T → A reaction at pH 1.5 may reach up to 150 nm/min (see section 4.2.1). This is almost two orders of magnitude faster than the rate of P → T → H transformation at pH 3 (1.8 ± 0.3 nm/min, section 4.1.4). Thus, in the P → T → H case, the rate of the reaction increases while rotating from [100] to [110]; in the case of P → T → A reaction, an opposite effect takes place. At pH 3 – 5.6 the P → T → H front is orientated parallel to [100], while at pH 1.5 the reaction front lies parallel to [110]. It is very likely that these two orientations represent two stabilised fronts along the low-index directions, so their movement is governed by slow nucleation and fast protonation of units along the front. The orientation of fronts at pH values between 1.5 and 3 represent deviations from the two end-member orientations, caused by the changes of nucleation rates and the difference of protonation rates of the neighbouring units.

The back and forth rotation of hillocks in adjacent layers can be explained by the glide plane (c) in the apophyllite structure. Figure 45a shows schematically the orientation of the hillocks at pH 3 – 5.6 (almost parallel to [100] direction). Since the angle between the reaction front and the c-glide plane is almost 45°, the hillocks in adjacent layers have a very small discrepancy angle and appear to have approximately the same orientation. Decreasing pH causes the hillocks to deviate from [100] orientation, that in turn causes an increasing angle between the hillocks of different generations. This effect is schematically shown in Figure 45b. A blue solid square represents the orientation of the hillock of the first generation. It is drawn through the centres of groups of tetrahedra pointing into the same direction. A red dashed square shows the orientation of the second-generation hillock; it is drawn through the centres of groups of tetrahedra pointing into the opposite direction. The hillocks of both generations can be superposed by a c-glide plane (dotted line on the Figure 45b). The angle between the reaction fronts in adjacent generations is about 26°. However, approaching pH 1.5 the discrepancy becomes lower again, because at these conditions the reaction front is close to [110] and thus is almost in a 90° angle to

75 the glide c-glide plane (Fig. 45a). Therefore at these conditions hillocks of all generation again have approximately the same orientation.

a a-axis c

pH 3 - 5.6 pH 2 pH 1.5 b first generation second generation c

Figure 45. Scheme showing the mechanism of hillock rotation in adjacent layers: (a) at pH 3 - 5.6 the hillocks are orientated almost parallel to [100] direction and almost in 45° to the c-glide plane (dotted lines); different colours mark different hillock generations; at pH 2 the orientation of hillocks is in about 12 - 15° to the c-glide plane, causing different orientations in adjacent layers; at pH 1.5 the hillocks are almost parallel to [110] and thus are almost in 90° angle to the cc-glide plane; (b) the effect of - glide plane at pH 2: blue solid square - first-generation hillock, red dashed square - second-generation hillock; the hillocks of both generations can be superposed by a c- glide plane (dotted line). The angle between the hillocks is about 26°.

76 There are a few works related to the effect observed in present study. Jordan et al. (2001) observed the rotation of etch pit steps in the case of magnesite dissolution by using HAFM. It has been shown that straight step vicinality increases with decreasing pH. The results were explained by a terrace-ledge-kink-model (e.g., Liang et al., 1996): it was suggested that lowering of solution pH causes an increased anisotropy of kink detachment rates at steps and/or a rough equalisation of the rate of double-kink formation and the slower kink detachment rate. Joshi and Ittyachen (1967, 1968) have detected a change of shape and rotation of etch pits on the basal cleavages of apophyllite crystals due to etching at different concentrations of ammonium bifluoride solutions. The change of shape and orientation has been attributed to differences in etch resistivity: less densely packed (100) faces are less resistant to etching, and produce square etch pits with sides parallel to [100] at low concentration of etchant. At high concentrations square pits with sides parallel to [110] were observed, since (111) faces are more closely packed. An octagonal shape of the pits at medium concentration has been attributed to roughly equal reactivity of both (100) and (111) faces at these conditions. Pande and Vadrabade (1990) studied HF-etching of the basal surface of apophyllite using different concentrations of ammonium fluoride as an inhibitor and obtained similar results. They attributed the observed changes in the shape and orientation of the pits to different adsorption sites of aqueous complexes on apophyllite. These authors also observed a change in pit orientation and shape caused by different

NH4F concentrations: e.g. at 0.6 M the pits had an ideal square shape with sides parallel to [100], at 4 M the pits had octagonal shape, and at 6 M they obtained a square shape again with sides parallel to [110]. The pH of solutions in their experiments increased with increasing NH4F concentration (0.6 M - pH 1.6; 4 M – pH 5; 6 M – pH 5.5). It needs to be noted, that in all these studies (Joshi and Ittyachen (1967, 1968) and Pande and Vadrabade (1990) apophyllite crystal morphology was considered as a combination of the forms {001} and {111}. However, as has been suggested by Kostov (1975) and confirmed by Akizuki and Konno (1985), the principal habit forms of apophyllite are {110} and {101}. Therefore, the results of Joshi and Ittyachen (1967, 1968) and Pande and Vadrabade (1990) have to be reconsidered taking into account the data reported by Kostov (1975) and Akizuki and Konno (1985). By doing this the

77 results obtained by Pande and Vadrabade (1990) are consistent with the present results and show the same tendency for orientation of the reaction fronts with increasing pH. The results presented by Joshi and Ittyachen (1967, 1968) and Pande and Vadrabade (1990) can be also reconsidered in view of the present experimental data. The resolution of optical microscopes used in those studies did not allow to have a detailed insight into molecular etch pit morphology. However, by using AFM individual steps of the pits can be resolved. Figure 46 shows a pit on the apophyllite surface formed by the exchange of pH 5.6 solution for pH 10. It can be seen that the steps in adjacent layers are not parallel, but rotated by an angle of about 5°. This is caused by the effect of the c-glide plane, as discussed above. Thus, even at pH 5.6 the reaction front is not strictly parallel to [100].

Figure 46. A pit on the apophyllite (100) surface generated by a pH-jump from pH 5.6 to pH 10. The steps in adjacent layers are not parallel and rotated by an angle of about 5°. This rotation causes the corners of the square pits to be truncated. Scan field 9 ×µ 9 m2.

It can be also seen that even this slight rotation of the steps in adjacent layers results in a pit that has not a precisely square shape but rather an octagonal shape. It can be suggested therefore, that the octagonal shape pits obtained by Joshi and Ittyachen (1967, 1968) and Pande and Vadrabade (1990) are also due to the rotation of reaction

78 fronts in adjacent monolayers. At “higher” concentration of the etchant, the dissolution fronts increasingly deviate from low index orientations ([100] and [110]), similar to the rotation of the protonation front at pH 2 in present study. However, the dissolution at steps is restricted by the shape of the precursor pit. This effect is shown schematically on Figure 47. The first layer dissolves by formation of a monolayer square etch pit. The sides of the pit are not parallel or perpendicular to c- glide plane, but deviate from it to some degree, e.g. 15°. Therefore the next dissolving pit is rotated with respect to the first layer by 30° (Fig. 47a; different grey scales show the difference in depth of the pits: the lower pit is darker than the higher one). When the corners of the second pit reach the sides of the first pit (Fig. 47b, marked by an arrow), further spreading of the second pit is restricted by the sides of the first pit. Therefore the second pit obtains an octagonal shape (Fig. 47c). The third pit has the same orientation

a d

Figure 47. A schematic drawing showing the formation of a pseudo-octagonal pit on apophyllite surface due to the rotation of the reaction fronts in adjacent layers: (a) the pits in adjacent layers are rotated by 15° and -15° with respect to the c-glide plane; different grey scales show different layers (the lower layers are darker than higher layers); (b) the corners of the lower pit reach the sides of the first pit (marked by an arrow); (c) further spreading of the lower pit is restricted by the sides of the first pit, so the second pit obtains an octagonal shape; the corners of the third pit reach the sides of the second pit; (d) the resulting etch pit has a pseudo-octagonal shape, although the initial pit has the square outline.

79 as the first one, and its spreading will be restricted by the second pit. Thus, the spreading of the pits in each layer is restricted by the pit in precursor layer. The resulting etch pit will have a pseudo-octagonal shape (Fig. 47d). As can be seen, the initial pit (affecting the top-most layer) has the square outline. However, the resolution of optical microscopes used by Joshi and Ittyachen (1967, 1968) and Pande and Vadrabade (1990) was not sufficient to resolve monolayer steps, therefore the pits in their studies have been reported to be octagonal. A similar effect was observed in the case of barite dissolution (e.g. Putnis et al.,

1995; Higgins et al., 1998b). In this mineral, the 21 screw axis lying parallel to the crystallographic c-axis causes the inversion of triangular pits in adjacent layers, which results in the formation of macroscopic etch pits having a pseudo-hexagonal shape. In the case of apophyllite, the reaction fronts rotate by 45° depending on the solution composition. This causes two types of square pits and the formation of pseudo- octagonal pits.

80 5. CONCLUSION

In the present study, the combined HAFM, NMR, RBS, and XPS approach provides detailed information on the mechanisms and kinetics of apophyllite alteration in aqueous solutions. On the basis of the data some analogues to the behaviour of other phyllosilicates at appropriate conditions may be drawn.

The reactions of phyllosilicates with acidic solutions usually cause a selective leaching of the cations in the order: interlayer cations, octahedral cations, tetrahedral cations (e.g. Grim, 1953). The preferential release of interlayer cations has been attributed to a rapid ion-exchange reaction with protons (e.g. Newman, 1970; Malmström and Banwart, 1997). Furthermore the structural hydroxyls become protonated and cause a reduction of the positive charge deficiency within the layers (e.g. Gruner, 1934; Turpault and Trotignon, 1994). These considerable chemical changes can cause an inter-layer expansion under moderate acidic conditions (e.g. Newman, 1970; Kalinowski and Schweda, 1995; Malmström and Banwart, 1997; Ferrow et al., 1999) or - by a strong acidic attack - even a corrugation of the topmost layers (Turpault and Trotignon, 1994). For example, AFM studies on vermiculite (Aldushin et al. 2004c) could show that alteration under acidic conditions involves different types of inter-layer swelling besides depolymerization of the silicate layers and dissolution via etch pits formation. At low pH conditions, alteration of apophyllite shows similarities to the alteration of the other sheet silicates. The interaction of apophyllite with acidic solution + induces an ion replacement reaction between the interlayer bridging ions and H3O - ions. However, it is unlikely, that the mono-valent hydronium ions can provide the charge density that is provided by Ca2+-ions, which are grouped closely together in the apophyllite interlayer. Taking into account the NMR experiments (and also the data of Lagaly and Matouschek (1980) and Sogo et al. (1998), it can be concluded that the reaction involves a protonation of terminal oxygens of the silicate sheet to form silanol groups in the interlayer. This leads to a reduction of attractive forces between the silicate sheets and, therefore, to an increased distance between the silicate sheets.

81 The NMR results further revealed that the protonation sites are not equivalent, although they are of the same type i.e., Si(OSi)3(OH). According to NMR-data presented in other studies, for many clay minerals only one uniform [Si(OSi)3OH] protonation site has been reported (e.g. Kaviratna and Pinnavaia, 1994; Kosuge et al., 1995; Komadel et al., 1996; Aznar et. al., 1996; Okada et al., 2002). In this respect, the product of apophyllite acidic treatment is different from those of other phyllosilicates, indicating that structural differences between sheet silicates play a key-role in the mechanisms taking place due to acidic treatment. In the case of apophyllite, protonation takes place in a two- or three-step process. The protonation states significantly differ in their formation rate and stability. While in some cases the initial protonation can attain high rates, the rate of the subsequent rising of interlayer distance by a further protonation is constant. This rate difference along with the limited stability of the initially protonated state can cause the formation of an intermediate, metastable state that shows a remarkable stability against further protonation.

According to the published data, there are two basic pathways of phyllosilicate alteration in acidic aqueous solutions. In the first case, acidic attack on phyllosilicates begins at the edges of the crystals and works inward; the contribution of the crystal basal planes to the dissolution process is considered to be relatively small (e.g., Zysset and Schindler, 1996; Rufe and Hochella, 1999; Bosbach et al., 2000). In the second case, the basal surface plays a significant role in the dissolution process e.g., by 2- dimensional pit nucleation and etch pit formation (Kaviratna and Pinnavaia, 1994; Huertas et al., 1999; Brandt et. al., 2003). In the alteration process of apophyllite at low pH both edges and the basal surface participate. The protons penetrate through surface defects into the interlayer space and initiate the ion-replacement process which causes a cation depleted residue and, eventually, the Si-O-Si bonds of the residue to weaken and to break. In some cases, the alteration at crystal edges is even slower than on the basal surface because metastable states (state A) are readily generated at steps (and therefore at crystal edges) when the step orientation mismatches the directions of slowest protonation. The metastable state has been found to be even more stable against further alteration than the pristine surface; also the alteration rate of layers below metastable areas was found to be reduced.

82 The properties of the resulting residue depend on the chemical and structural characteristics of the parental mineral and the parameters of treatment. For example, Acker and Bricker (1992) report the formation of a vermiculite-type product due to the loss of octahedral cations of biotite at pH ≥ 4, whereas at pH 3 dissolution of biotite via destruction of the tetrahedral and octahedral layers takes place. Rufe and Hochella (1999) observed the formation of an expanded amorphous silica-enriched film on the basal surface of phlogopite under moderate acidic conditions. Kaviratna and Pinnavaia (1994) reported the formation of a porous amorphous product due to the depletion of Mg-ions from octahedral sites and subsequent crosslinking of the residual tetrahedral sheets during the acidic alteration of fluorohectorite and phlogopite. Kuwahara and Aoki (1995) detected a transformation of phlogopite to vermiculite and interstratified mica/vermiculite due to acidic treatment. According to Suquet et al. (1991) a mildly acidic attack to Llano vermiculite causes a formation of non-crystalline porous hydrated silica. Also, Ravichandran and Sivasankar (1997) reported an increase of surface area and acidity of the resulting residue by acidic alteration of vermiculite and montmorillonite. According to Ross (1967), sever acidic treatment of chlorite causes a formation of amorphous hydrated silica. Brandt et al. (2003) observed a non- stoichiometrical dissolution mechanism of chlorite at pH 2 - 4 with a preferential release of octahedral cations. They also suggested a transformation of chlorite to interstratified chlorite/vermiculite in the course of alteration. The comparison of these phyllosilicates to apophyllite reveals that the protonation of the terminating oxygen atoms of the silicate sheets of apophyllite and the resulting formation of hillocks has a close relation to the acid hydrolysis of the octahedral layer in 2:1 sheet silicates (e.g. Kaviratna and Pinnavaia, 1994). A progressive treatment causes further changes of the protonated residue by a partial depolymerization of the sheets and partial cross-linking of the protonated silica sheets. This leads to the formation of a porous material which probably contains a high portion of water within cages formed by the cross-linked silicate sheets. The AFM data are supported by the NMR results which showed that a prolonged treatment causes a structural amorphization. This eventual transformation of apophyllite into a silica- enriched amorphous residue is consistent with the transformations of many other phyllosilicates caused by an acidic treatment.

83 Thus, it could be shown that the combination of HAFM with several spectroscopic methods in the present study provided detailed knowledge on the mechanisms and kinetics of apophyllite alteration in aqueous acidic solutions. For instance using HAFM, it has been possible for the first time to observe crystalline swelling of phyllosilicates in-situ at a nano-scale resolution. Therefore, the application of these advanced techniques allowed to quantify the kinematics of individual reaction steps, enabled the in-situ observation of morphological surface modifications, and made it possible to link the observations to molecular mechanisms of the alteration process. The obtained results, therefore, provide a unique possibility to compare the mechanism and kinetics of apophyllite alteration with the alteration processes of other phyllosilicates and thus allow to assess the influences of structural singularities within the group phyllosilicates on alteration in aqueous acidic solution.

84 REFERENCES

Acker J. G. and Bricker O. P. (1992) The influence of pH on biotite dissolution and alteration kinetics at low temperature. Geochim. Cosmochim. Acta 56, 3073-3092.

Akizuki M. and Terada T. (1998) Origin of abnormal optical property of apophyllite. N. Jb. Miner. Mh., 5, 234-240.

Aldushin K., Jordan G., Rammensee W., Schmahl W. W., and Becker H.-W. (2004a) Apophyllite (001) surface alteration in aqueous solutions studied by HAFM. Geochim. Cosmochim. Acta, 68, 217-226.

Aldushin K., Jordan G., Fechtelkord M., Schmahl W. W., Becker H.-W., and Rammensee, W. (2004b) On the mechanism of apophyllite alteration in aqueous solutions. A combined AFM, XPS and MAS NMR study. Clays & Clay Miner. 52, 432-442.

Aldushin K., Jordan G., Schmahl W. W., and Rammensee W. (2004c) Swelling and dissolution of vermiculite studied in situ by hydrothermal AFM. Berichte der Deutschen Mineralogischen Gesellschaft, Beih. z. Eur. J. Mineral. 16, 4.

Astilleros J. M., Pina C. M., Fernández-Diáz L. and Putnis A. (2002) Molecular-scale surface processes during the growth of calcite in the presence of manganese. Geochim. Cosmochim. Acta 66, 3177-3189.

Aznar A.J., Gutierrez E., Diaz P., Alvarez A., and Poncelet G. (1996) Silica from sepiolite: Preparation, textural properties, and use as support to catalysts. Microporous Materials, 6, 105-114.

Barshad I. and Kishk F. M. (1968) Oxidation of ferrous iron in vermiculite and biotite. Alters fixation and replaceability of potassium. Science 162, 1401-1402.

Bickmore B. R., Bosbach D., Hochella, Jr., M. F., Charlet L. and Rufe E. (2001) In situ atomic force microscopy study of hectorite and nantronite dissolution: Implications for phyllosilicate edge surface structures and dissolution mechanisms. Amer. Mineral. 86, 411-423.

Bickmore B. R., Rosso K. M., Nagy K. L., Cygan R. T., and Tadanier C. J. (2003) Ab initio determination of edge surface structures for dioctahedral 2:1 phyllosilicates: Implications for acid-base reactivity. Clays & Clay Miner. 51, 359-371.

Bickmore B. R., Tadanier C. J., Rosso K. M., Monn W. D., and Eggett L. D. (2004) Bond-valence methods for pKa prediction: Critical reanalysis and a new approach. Geochim. Cosmochim. Acta 68, 2025-2042.

Binning G., Quate C. F., and Gerber C. (1986) Atomic Force Microscopy. Phys. Rev. Lett., 56, 930-933.

Barshad I. and Kishk F. M. (1968) Oxidation of ferrous iron in vermiculite and biotite. Alters fixation and replaceability of potassium. Science, 162, 1401-1402.

Bosbach D., Hall C., and Putnis A. (1998) Mineral precipitation and dissolution: in-situ microscopic observations on barite (001) with atomic force microscopy. Chem. Geol. 151, 143-160.

Bosbach D., Charlet L., Bickmore B. and Hochella, Jr., M. F., (2000) The dissolution of hectorite: In-situ, real-time observations using atomic force microscopy. Amer. Mineral. 85, 1209-1216.

Brandt F., Bosbach D., Krawczyk-Barsch E., Arnold T., and Bernhard G. (2003) Chlorite dissolution in the acid pH-range: A combined microscopic and macroscopic approach. Geochim. Cosmochim. Acta, 67, 1451-1461.

Casey W. H. and Bunker B. C. (1990) The leaching of mineral and glass surfaces during dissolution. In Reviews in Mineralogy: Mineral-Water Interface Geochemistry, 22 (eds. Hochella Jr., M. F. and White A.), 397-426.

Cave L. C. (2002) Apophyllite weathering and the aqueous geochemistry of a Karoo breccia pipe. PhD. thesis, Department of Geological Sciences, University of Cape Town.

Chao G. Y. (1971a) Carletonite, KNa4Ca4Si8O18(CO3)4(F,OH)·H2O, a new mineral from Mount St. Hilaire, Quebec. Amer. Mineral. 56, 1855-1866.

Chao G. Y. (1971b) Refinement of the crystal structure of apophyllite: II. Determination of the hydrogen positions by X-ray diffraction. Amer. Mineral. 56, 1234-1251.

Colville A. A., Anderson C. P. and Black P. M. (1971) Refinement of the crystal structure of apophyllite, I. X-ray diffraction and physical properties. Amer. Mineral. 56, 1222-1233.

Corma A. and Perez-Periente J. (1987) Surface acidity and activity of modified sepiolite. Clay Miner. 22, 423-433.

Dove P. M. and Platt F. M. (1996) Compatible real-time rates of mineral dissolution by Atomic Force Microscope (AFM). Chem. Geol. 127, 331-338.

Dunn P. J., Rouse R. C. and Norberg J. A. (1978) Hydroxyapophyllite, a new mineral, and a redefinition of the apophyllite group, I. Description, occurrences, and nomenclature. Amer. Mineral. 63, 196-199.

Fechtelkord M., Behrens B., Holtz F., Fyfe C.A., Groat L.A., and Raudsepp M. (2003) Influence of F content on the composition of Al-rich synthetic phlogopite: Part I. New information on structure and phase-formation from 29Si, 1H, and 19F MAS NMR spectroscopies. Amer. Mineral. 88, 47-53.

Ferrow E. A., Kalinowski B. E., Veblen D. R., and Schweda P. (1999) Alteration products of experimentally weathered biotite studied by high-resolution TEM and Mössbauer spectroscopy. Eur. J. Mineral. 11, 999-1010.

Frondel C. (1979) Crystalline silica hydrates from leached silicates. Amer. Mineral. 64, 799-804.

Grim R. E. (1953) Clay mineralogy. The McGraw-Hill Book Company, Inc., New York, 384 pp.

Gruner J. W. (1934) The structure of vermiculite and their collapse by dehydration. Am. Mineral. 19, 557-575.

Hall C. and Cullen D. C. (1996) Scanning force microscopy of gypsum dissolution and crystal growth. Aiche J. 42, 232-238.

Higgins S. R., Eggleston C. M., Knauss K. G., and Boro C. O. (1998a) A hydrothermal atomic force microscope for imaging in aqueous solution up to 150°C. Rev. Sci. Instrum. 69, 2994-2998.

Higgins S. R., Jordan G., Eggleston, C. M., and Knauss K. G. (1998b) Dissolution kinetics of the barium sulphate (001) surface by hydrothermal atomic force microscopy. Langmuir, 14, 4967-4971.

Higgins S. R., Jordan G., and Eggleston C. M. (2002) Dissolution kinetics of magnesite in acidic aqueous solution: a hydrothermal atomic force microscopy study assessing step kinetics and dissolution flux. Geochim. Cosmochim. Acta 66, 3201-3210.

Hillner P. E., Gratz A. J., Manne S. and Hansma P. K. (1992) Atomic-scale imaging of calcite growth and dissolution in real time. Geology 20, 359-362.

Hochella Jr., M. F. (1990) Atomic structure, microtopography, composition, and reactivity of mineral surfaces. In Reviews in Mineralogy: Mineral – Water Interface Geochemistry (ed. Hochella Jr., M. F. and White A. F.) 23, 87-132.

Hochella Jr., M. F., Rakovan J., Rosso K., Bickmore B. R., and Rufe E. (1998) New directions in mineral surface geochemical research using scanning probe microscopy. In Mineral-Water Interfacial Reactions: Kinetics and Mechanisms, ACS Symposium Series, 715 (ed. Sparks D. L. and Grundl T. J. ) Washington, DC, 37-56.

Huertas F. J., Chou L., and Wollast R. (1999) Mechanism of kaolinite dissolution at room temperature and pressure. Part II: Kinetic study. Geochim. Cosmochim. Acta 63, 3261-3271.

Jordan G. and Rammensee W. (1998) Dissolution rates of calcite (10(1)over-bar4) obtained by scanning force microscopy: Microtopography-based dissolution kinetics on surfaces with anisotropic step velocities. Geochim. Cosmochim. Acta 62, 941-947.

Jordan G., Higgins S.R., Eggleston C.M., Swapp S.M., Janney D.E., and Knauss K.G. (1999) Acidic dissolution of plagioclase: In-situ observations by hydrothermal atomic force microscopy. Geochim. Cosmochim. Acta 63, 3183-3191.

Jordan G., Higgins S.R., Eggleston C.M., Knauss K.G. and Schmahl W. W. (2001) Dissolution kinetics of magnesite in acidic aqueous solution, a hydrothermal atomic force microscopy (HAFM) study: Step orientation and kink dynamics. Geochim. Cosmochim. Acta 65, 4257-4266.

Joshi M. S. and Ittyachen M. A. (1967) Rotation of etch pits on the basal cleavages of apophyllite crystals. Phyl. Mag. 16, 717-721.

Joshi M. S. and Ittyachen M. A. (1968) The growth mechanism of apophyllite crystals studied by surface microstructures. J. Cryst. Growth, 2, 47-50.

Kalinowski B. E. and Schweda P. (1996) Kinetics of muscovite, phlogopite and biotite dissolution and alteration at pH 1-4, room temperature. Geochim. Cosmochim. Acta 60, 367-385.

Kaviratna H. and Pinnavaia T.J. (1994) Acid hydrolysis of octahedral Mg2+ sites in 2:1 layered silicates: an assessment of edge attack and gallery access mechanisms. Clays & Clay Miner. 42, 717-723.

Komadel P., Madejova J., Janek M., Gates W. P., Kirkpatrick R. J., and Stucki J. W. (1996) Dissolution of hectorite in inorganic acids. Clays & Clay Miner., 44, 228-236.

Kostov I. (1975) Apophyllite morphology as en example of habit modification of planar crystals. N. Jb. Miner. Abh. 123, 128-137.

Kosuge K., Shimada K., and Tsunashima A. (1995) Micropore formation by acid treatment of antigorite. Chemical Materials, 7, 2241-2246.

Kotai E. (1994) Computer methods for analysis and simulation of RBS and ERDA spectra. Nucl. Instrum. Meth. Phys. Res. B 85, 588-596.

Kuwahara Y. and Aoki Y. (1995) Dissolution process of phlogopite in acidic solutions. Clays & Clay Miner. 43, 39-50.

Lacy E. D. (1965) Factors in the study of metamorphic reaction rates. In Controls of Metamorphism (Pitcher W. S. and Flinn G. W., eds.) 140-154, Oliver and Boyd, London.

Lagaly G. (1979) Crystalline silisic acids and their interface reactions. Adv. Coll. Interf. Sci. 11, 105-148.

Lagaly G., Beneke K., and Weiss A. (1975) Magadiite and H-Magadiite: II. H- Magadiite and its intercalation compounds. Amer. Mineral. 60, 650-658.

Lagaly G. and Matouschek R. (1980) The crystalline silicic acids from apophyllite, carletonite and gillespite. N. Jb. Mineral. / Abh. 138, 81-93.

Laird D. A. (1996) Model for crystalline swelling of 2:1 phyllosilicates. Clays & Clay Miner. 44, 553-559.

Liang Y., Baer D. R., McCoy J. M., Amonette J. E., and LaFemina J.P. (1996) Dissolution kinetics at the calcite-water interface. Geochim. Cosmochim. Acta 60, 4883-4887.

Malmström M. and Banwart S. (1997) Biotite dissolution at 25°C: The pH dependence of dissolution rate and stoichiometry. Geochim. Cosmochim. Acta 61, 2779-2795.

Marriner G. F., Tarney J. and Langford J. I. (1990) Apophyllite group: effects of chemical substitutions on the dehydration behaviour, recrystallization products and cell parameters. Mineral. Mag. 54, 567 - 577.

Matsueda H., Miura Y., and Rucklidge J. (1981) Natroapophyllite, a new orthorhombic sodium analog of apophyllite. I. Description, occurrence, and nomenclature. Amer. Mineral. 66, 410-415.

Moore D. M. and Reynolds Jr., R. C. (1997) X-ray Diffraction and the Identification and Analysis of Clay Minerals. Oxford university press, New York.

Nagy K. L., Blum A. E., and Lasaga A. C. (1991) Dissolution and precipitation of kaolinite at 80 °C and pH 3: The dependence on solution saturation state. Amer. J. Sci. 291, 649-686.

Nagy K. L. and Blum A. E., Eds., (1994) Scanning Probe Microscopy of Clay Minerals. The Clay Minerals Society, Boulder, USA.

Newman A. C. D. (1970) The synergetic effect of hydrogen ions on the cation exchange of potassium in micas. Clay Minerals 8, 361-373.

Norrish K. (1954) The swelling of montmorillonite. Faraday Disc. 18, 120-135.

Okada K., Nakazawa N., Kameshima Y., Yasumori A., Temuujin J., MacKenzie K. J. D., and Smith M. E. (2002) Preparation and porous properties of materials prepared by selective leaching of phlogopite. Clays & Clay Miner., 50, 624-632.

Pabst A. (1958) The crystal structure of leached gillespite, a sheet silicate. Amer. Mineral. 43, 970-980.

Pande D. R. and Vadrabade S. R. (1990) Etch pits on basal cleavage faces of apophyllite crystals. Mineral. Mag. 54, 559-565.

89 Putnis A., Junta-Rosso J. L., and Hochella, Jr., M. F. (1995) Dissolution of barite by chelating ligand: An atomic force microscopy study. Geochim. Cosmochim. Acta 59, 4623-4632.

Prince E. (1971) Refinement of the crystal structure of apophyllite III. Determination of the hydrogen positions by neutron diffraction. Amer. Mineral. 56, 1243-1251.

Ravichandran J. and Sivasankar B. (1997) Properties and catalytic activity of acid- modified montmorillonite and vermiculite. Clays & Clay Miner. 45, 854-358.

Rodriguez M. A. V., Gonzalez J. D. L., and Munoz M. A. B. (1994) Acid activation of a spanish sepiolite: physicochemical characterization, free silica content and surface area of products obtained. Clay Minerals, 29, 361-367.

Ross G. J. (1967) Acid dissolution of chlorites: release of magnesium, iron and aluminium and mode of acidic attack. Clays & Clay Miner. 17, 347-354.

Rufe E. and Hochella Jr., M. F. (1999) Quantitative assessment of reactive surface area of phlogopite during acid dissolution. Science 285, 874-876.

Saito Y., Motohashi T., Hayashi S., Yasumori A., and Okada K. (1997) High thermal resistance of gamma-Al2O3 prepared by the selective leaching of calcined kaolin minerals. J. Mater. Chem. 7, 1615-1621.

Shiraki R., Rock P. A. and Casey W. H. (2000) Dissolution kinetics of calcite in 0.1 m NaCl solution at room temperature: An atomic force microscopic (AFM) study. Aquatic Geochem. 6, 87-108.

Sogo Y., Iizuka F., and Yamazaki A. (1998) Preparation and properties of layered silica and layered alumino-silica hydrate from natural apophyllite. J. Ceram. Soc. Japan 106, 160-168.

Sogo Y., Kumazawa T., Okihara K., Yamazaki A. (2000) Hofmann degradation kinetics of n-octylamine adsorbed on layered aluminosilicates prepared from apophyllite. Termochim. Acta 364 193-201.

Sposito G. and Prost R. (1982) Structure of water adsorbed on smectites. Chem. Rev. 82, 553-573.

Sposito G. (1984) The surface chemistry of soils. Oxford University Press, Oxford, U.K.

Stipp S. L. S., Eggleston C. M., and Nielsen B. S. (1994) Calcite surface structure observed at micrographic and molecular scales with atomic force microscopy (AFM). Geochim. Cosmochim. Acta 58, 3023-3033.

Suquet H., Chevalier S., Marcilly C., and Barthomeuf D. (1991) Preparation of porous materials by chemical activation of the Llano vermiculite. Clay Minerals 26, 49-60.

Taylor and Náray-Szabó (1931) The structure of apophyllite. Zeits. Kristallogr. 77, 148- 156.

Teng H. H. and Dove P. M. (1997) Surface site-specific interactions of aspartate with calcite during dissolution: Implications for biomineralization. Amer. Mineral. 82, 878-887.

Temuujin J., Burmaa G., Amgalan J., Okada K., Jadambaa T., and MacKenzie K. J. D. (2001) Preparation of porous silica from mechanically activated kaolinite. J. Porous Mater. 8, 233-238.

Theodossiu W., Misaelides P. and Godelitsas A. (2001) Investigation of fluorine distribution on the surface of acid-treated apatite single crystals using nuclear resonant reaction analysis. Cryst. Res. Technol. 36, 1247-1251.

Turpault M.-P. and Trotignon L. (1994) The dissolution of biotite single crystals in dilute HNO3 at 24 °C: Evidence of an anisotropic corrosion process of micas in acidic solutions. Geochim. Cosmochim. Acta 58, 2761-2775.

Van Olphen H. (1965) Thermodynamics of interlayer adsorption of water in clays: I. sodium vermiculite. J. Coll. Sci. 20, 822-837.

Walker G. F. (1960) Macroscopic swelling of vermiculite crystals in water. Nature 187, 312-313.

Walther, J. V. (1996) Relation between rates of aluminosilicate mineral dissolution, pH, temperature, and surface charge. Amer. J. Sci., 296, 693-728.

Wieland E. and Stumm W. (1992) Dissolution kinetics of kaolinite in acidic aqueous solutions at 25 °C. Geochim. Cosmochim. Acta 56, 3339-3355.

Yesinowski J. P., Eckert H., and Rossman G. R. (1988) Characterization of Hydrous Species in Minerals by High-Speed 1H MAS-NMR. J. Amer. Chem. Soc. 110, 1367- 1375.

Zysset M. and Schindler P. W. (1996) The proton promoted dissolution kinetics of K- montmorillonite. Geochim. Cosmochim. Acta 60, 921-931.

91 Erklärung

Hiermit erkläre ich, daß ich die hier vorliegende Arbeit selbständig und ohne unerlaubte Hilfe ausgeführt und verfasst habe, und dass die Arbeit in dieser oder ähnlicher Form noch bei keiner Fakultät oder anderen Hochschule eingereicht wurde.

Bochum, 25. 10. 2004 Kirill Aldushin

92 CURRICULUM VITAE

Kirill Aldushin

21.12.1977 born in Chernogolovka, Russia

Elementary school and high school education:

1983 – 1994 Secondary School № 82, Chernogolovka, Russia

University education:

1994 – 1999 Master program in Geochemistry at Moscow State University

Specialisation: crystallography and crystallchemistry

Master’s thesis: “Experimental study of diamond crystallisation in new carbonate-carbon systems under high pressure”

Supervisors: Prof. Dr. Leonyuk N. I. (Moscow State University) Prof. Dr. Litvin Yu. A. (Institute of Experimental Mineralogy)

Scientific work:

1999 – 2001 Employment as scientist at Institute of Solid State Physics, Chernogolovka, Russia

Since July 2001 Employment as scientist at Ruhr-Universität Bochum and Universität zu Köln (DFG co-operative project Köln/Bochum)

93 List of publications:

1. Aldushin K., Jordan G., Schmahl W. W., and Rammensee W. (2004) Swelling and dissolution of vermiculite studied in situ by hydrothermal AFM. Berichte der DMG, Beih. z. Eur. J. Mineral. 16, 4.

2. Aldushin K., Jordan G., Schmahl W. W., and Rammensee W. (2004) Alteration of phyllosilicates studied in situ by hydrothermal AFM. Geochim. Cosmochim. Acta, Suppl. Issue 68, 170.

3. Jordan G., Aldushin K., Schmahl W. W., Rammensee W. (2004) On the alteration of sheet silicates in aqueous solutions. Lithos, Suppl. Issue 73, 53.

4. Aldushin K., Jordan G., Fechtelkord M., Schmahl W. W., Becker H.-W., and Rammensee, W. (2004) On the mechanism of apophyllite alteration in aqueous solutions. A combined AFM, XPS and MAS NMR study. Clays & Clay Miner. 52, 432-442.

5. Aldushin K., Jordan G., Rammensee W., Schmahl W.W. and Becker H-W. (2004) Apophyllite (001) surface alteration in aqueous solutions studied by HAFM. Geochim. Cosmochim. Acta 68, 217-226.

6. Aldushin K., Jordan G., Rammensee W., Schmahl W. W. and Becker H.-W. (2003) Nano-scale observations of apophyllite swelling: implications for clay mineralogy. Berichte der DMG, Beih. z. Eur. J. Mineral. 15, 4.

7. Aldushin K., Jordan G., Rammensee W. and Schmahl W. W. (2003) HAFM- Untersuchungen an alterierten Oberflächenschichten auf Apophyllit in wäßrigen Lösungen. Z. Krist., Suppl. Issue 20, 115.

8. Aldushin K., Jordan G., Rammensee W. and Schmahl W. W. (2002) Dissolution of apophyllite (001) in aqueous solution studied by HAFM. Berichte der DMG, Beih. z. Eur. J. Mineral. 14, 10.

9. Bazhenov A.V., Gorbunov A.V., Aldushin K.A., Masalov V.M. and Emelchenko G. A. (2002) Optical properties of thin films of closely packed SiO2 spheres. Physics of the Solid State 44, 1071-1076.

10. Emelchenko G.A., Aldushin K.A., Masalov V.M, Bazhenov A.V. and Gorbunov A.V. (2002) Growth and optical properties of self-ordering thin films of SiO2 microspheres. Physics of Low-Dimensional Structures 1/2, 99-112.

11. Masalov V.M., Aldushin K.A., Dolganov P.V. and Emelchenko G.A. (2001) SiO2- microspheres ordering in 2D structures. Physics of Low-Dimensional Structures 5/6, 45-54.

94 12. Aldushin K.A., Masalov V.M., Kulakov A.B., and Emelchenko G.A. (2001) About dissolution of aluminium oxide in Bi2O3- PbF2 melt. Materialovedenie 9, 9-11 (russian).

13. Litvin Yu.A., Aldushin K.A. and Zharikov V.A. (1999). Diamond synthesis at 8.5– 9.5 GPa in the K2Ca(CO3)2–Na2Ca(CO3)2–C system modelling compositions of fluid–carbonatite inclusions in kimberlitic diamonds. Doklady Earth Sciences, 367A (6), 813-816.

14. Chudinovskikh L.T., Litvin Yu.A. and Aldushin K.A. (1998) Experimental studies of diamond growth on the {111} and {100} faces of the seed crystals at 7-10 GPa. Experiment in Geosciences 7 (2), 49-51.