Submitted Version of PhD Thesis

Grain boundary structure in minerals and

analogues during recrystallization in the

presence of a fluid phase

Submitted version of PhD Thesis

“Grain boundary structure in minerals and analogues during recrystallization

in the presence of a fluid phase”

Der Fakultät für Georessourcen und Materialtechnik

der Rheinisch-Westfälischen Technischen Hochschule Aachen

vorgelegte Dissertation zur Erlangung

des akademischen Grades eines Doktors der Naturwissenschaften

von

Dipl. Geol. Oliver Schenk

aus Aachen

i

Table of contents

Zusammenfassung 1

Abstract 3

1 Introduction 5

1.1 General 5

1.2 Brief review of recrystallization processes 5

1.3 The grain-scale distribution of fluids 11

1.4 Aim of this study 24

1.5 Overview of this thesis 25

1.6 Parts of the thesis which have been published 26

1.7 References 27

2 The effect of water on recrystallization behavior and grain

boundary morphology in calcite – observations of natural

marble mylonites 32

2.1 Introduction 32

2.2 Geological setting and sampling 36

2.3 Observations 41

2.4 Discussion 53

2.5 Conclusions 57

2.6 References 58

ii

3 Microstructural evolution and grain boundary structure

during static recrystallization in synthetic polycrystals of

Sodium Chloride containing saturated brine 63

3.1 Introduction 64

3.2 Aims of this study 67

3.3 Methods 67

3.4 Results 71

3.5 Discussion 83

3.6 References 88

3.7 Appendix 91

4 Structure of grain boundaries in wet, synthetic

polycrystalline, statically recrystallizing halite – evidence

from cryo-SEM observations 96

4.1 Introduction 97

4.2 Methods 100

4.3 Observations 102

4.4 Discussion 116

4.5 Conclusions 119

4.6 References 119

5 The migration of fluid-filled grain boundaries in

recrystallizing synthetic bischofite – observations from in-

situ deformation experiments in transmitted light 123

5.1 Introduction 124

5.2 Experimental techniques 125

5.3 Observations 127

iii

5.4 Discussion 139

5.5 Conclusions 145

5.6 Appendix 146

5.7 References 148

1

Zusammenfassung:

Thema der vorliegenden Arbeit ist die Struktur und Morphologie von Korngrenzen

während der Rekristallisation in Gegenwart von Fluiden. Die fluidunterstützte

Korngrenzenbewegung wird untersucht anhand einer Kombination aus in-situ-

Experimenten unter dem Durchlichtmikroskop und mikrostrukturellen Studien an

natürlichen Proben und Analogmaterialien.

Die Anwesenheit von Fluiden kann das rheologischen Verhalten von Mineralien und

Gesteinen beträchtlich beeinflussen. Ziel dieser Studie ist es, ein besseres Verständnis

der Eigenschaften von Fluiden in Korngrenzen – wie zum Beispiel Morphologie,

Permeabilität oder ihr Einfluss auf das Rekristallisationsverhalten – zu erlangen.

Es wird angenommen, dass fluidenthaltende Korngrenzen als schnelle intergranulare

Diffusionswege dienen und somit Prozesse des spannungsabhängigen Massentransfers

wie Drucklösung oder Korngrenzenwanderung beschleunigen. Detailliertes Wissen

über die kleinmaßstäbliche Fluidverteilung unter Ungleichgewichtsbedingungen ist

daher erforderlich, um die mechanischen Eigenschaften und Transportcharakeristika

von Gesteinen in Erdkruste und -mantel vorherzusagen.

Um Informationen über den Einfluss von Fluiden auf das Rekristallisationsverhalten

und die Korngrenzstruktur in natürlichem Kalzit zu erlangen, wurden

Marmormylonite vom Schneeberg-Komplex (italienische/österreichische Alpen) mit

denen des metamorphen Kernkomplexes von Naxos (Griechenland) verglichen. Beide

Arbeitsgebiete zeichnen sich durch eine ähnliche geologische Geschichte aus. Jedoch

sind sie verschieden in der Menge und Beschaffenheit der Fluide, die während der

Mylonitisierung anwesend waren. Während die mikrostrukturelle Entwicklung der

Schneeberg-Mylonite erheblich von den anwesenden Fluiden beeinflusst wurde, wie

Kalzitklüfte, dilatante Korngrenze und intragranulare Brüche belegen, wurden keine

dieser Merkmale in den Naxos-Myloniten beobachtet. Dennoch sind sowohl die

fluidreichen (Schneeberg) als auch die fluidarmen (Naxos) Marmormylonite durch

ähnliche Korngrenzstrukturen gekennzeichnet. Dies deutet darauf hin, dass die Fluide

2

keine gravierenden Auswirkungen auf das Rekristallisationsverhalten von Kalzit

gehabt haben.

Dagegen beeinflussen Fluide in beträchtlichem Maße das Rekristallisationsverhalten

von Steinsalz. Dies belegen die experimentellen Untersuchungen an statisch

rekristallisierendem synthetischem polykristallinem Halit, der geringe Mengen an

gesättigter wässriger Lösung enthält. Die kompaktierten feuchten Proben zweier

Korngrößenklassen (<10 µm und 200-355 µm) wurden ohne externe Spannung bei

Raumtemperatur „geglüht“. Beobachtungen an Rasterelektronen- und Auflicht-

mikroskopen zeigen, dass die grobkörnigen Proben primärer Rekristallisation

unterliegen. Dies ist deutlich zu erkennen an den idiomorphen verformungsfreien

Kristallen, die mit Korngrenzgeschwindigkeiten von bis zu 6 nm/s in die umgebenden

deformierten Körner wachsen. Innerhalb der feinkörnigen Proben folgt der primären

Rekristallisation das normale Kornwachstum. Nach wenigen Stunden sind die

Kontakte geheilt und das Kornwachstum beendet. Während dieses Stadiums beginnt

das außerordentliche Kornwachstum (sekundäre Rekristallisation). Das Wachstum

von primär und sekundär rekristallisierten Körnern wird als Folge der Anwesenheit

von Fluiden innerhalb der Korngrenzen interpretiert. Es kann i) entweder aus einer

signifikant anisotropen Korngrenzenenergie, ii) als Folge eines Wachstumstyps,

welcher auf der Bewegung von Stufen an Grenzflächen von Festkörper und Fluid

beruht, resultieren, oder iii) einer Kombination von beiden unterliegen

Die ganze Breite der eingebundenen Prozesse von fluidunterstützter

Korngrenzenbewegung kann direkt und kontinuierlich mit Hilfe von in-situ-

Experimenten im Durchlichtmikroskop beabachtet werden. Bischofit mit geringen

Mengen an gesättigter wässriger Lösung wurde bei Temperaturen zwischen 50 und

90 °C und einem Fluiddruck von etwa 1 MPa deformiert. Die Verformungsraten lagen

zwischen 5 · 10-6 und 1 · 10-4 s-1. Detaillierte Beobachtungen während und nach der

Deformation dokumentieren die Entwicklung und Bewegung von fluidgefüllten

Korngrenzen in rekristallisierndem Bischofit. Die Ergebnisse zeigen, dass während

der Korngrenzenwanderung Fluideinschlüsse übergangen, aufgenommen und/oder

zurückgelassen werden. Sie stehen in direktem Zusammenhang mit den Parametern

Korngrenzengeschwindigkeit, Dicke der fluidgefüllten Korngrenze und Größe und

Gestalt der Fluideinschlüsse. Zusätzlich geben die experimentellen Untersuchungen

einen direkten Hinweis darauf, dass die Korngrenzfluide sich in isolierte Einschlüsse

kontrahieren, nachdem sich Gleichgewichtsbedingungen eingestellt haben.

3

Abstract:

This thesis deals with the structure of grain boundaries during recrystallization in the

presence of fluids. It combines direct observations of fluid-assisted grain boundary

migration from in-situ experiments in transmitted light with microstructural studies on

naturally and experimentally recrystallized samples.

The aim is to get a better insight into the properties of fluids in grain boundaries such

as morphology, connectivity and their influence on recrystallization processes. One of

the major effects on rheology results from the presence of fluids in grain boundaries.

Such fluid-bearing grain boundaries serve as fast intergranular diffusional pathways

that allow processes of stress-driven mass transfer such as solution-precipitation creep

or fluid-assisted grain boundary migration. Detailed knowledge of the micro-scale

non-equilibrium fluid distribution is required to predict the mechanical and transport

properties of rocks deforming under metamorphic conditions.

To derive information on the effect of water-rich fluids on the recrystallization

behavior and grain boundary morphology in natural calcite, marble mylonites from

the Schneeberg complex (Italian/Austrian Alps) and the high grade core of Naxos

metamorphic core complex (Greece) were compared. Both settings have similar

geologic histories, but they are different in the nature of the fluids present during

mylonitization. Both the fluid-rich (Schneeburg) and fluid-poor (Naxos) marble

mylonites have similar grain boundary microstructures. The microstructural evolution

inside the Schneeberg mylonites was affected by the presence of fluids as shown by

the presence of syndeformational calcite veins, dilatant grain boundaries and

intragranular cracks. None of these features are present in the Naxos samples.

However, the fluids did not have a major influence on recrystallization behavior and

grain boundary morphology in calcite, at least for these two marble mylonites. The

fluids were interpreted to be dragged as isolated pores by the grain boundary during

migration.

In contrast the presence of fluids strongly affects the recrystallization behaviour in

halite, as shown by the experiments on statically recrystallizing synthetic

4

polycrystalline halite containing small amounts of brine. The compacted wet samples

of two different grain size classes (<10 µm and 200-355 µm) annealed at room

temperature without an external stress field. Observations from conventional scanning

electron and reflected light microscopy show that the coarse grained samples undergo

primary recrystallization indicated by large, euhedral strain-free grains that grow into

the deformed old grains. The rates of fluid-assisted grain boundary migration are

measured to be up to 6 nm/s. Inside the fine grained samples primary recrystallization

was followed by normal grain growth, but stopped after a few hours due to contact

healing, while exaggerated grain growth (secondary recrystallization) initiated at this

stage. The growth of the primary and secondary recrystallized grains is interpreted to

be due to the presence of fluid films on the grain boundaries as a consequence of

either significant anisotropic grain boundary energy and/or a solid-brine type growth

mechanism with a ledge jump mechanism. Direct evidence of the fluid-filled grain

boundaries were obtained from cryo-SEM observations and thicknesses of less than

30 nm have been measured.

The whole range of processes involved during fluid-assisted grain boundary migration

is continuously and directly observed by carrying out in-situ deformation experiments

in transmitted light microscopy. Bischofite containing small amounts of aqueous fluid

is deformed at temperatures between 50 and 90 °C, with the fluid pressure being

around 1 MPa. The strain rates range from 5 · 10-6 to 1 · 10-4 s-1. Detailed observations

during and after deformation document the development and migration of fluid-filled

grain boundaries in recrystallizing wet bischofite. The results show that during grain

boundary migration fluid inclusions are swept, incorporated and/or left behind by the

mobile grain boundaries. This is interpreted to depend on the grain boundary velocity,

the thickness of the fluid-filled grain boundary and the size and shape of the fluid

inclusions. Additionally the experiments present direct evidence for the contraction of

grain boundary fluids into isolated inclusions after equilibrium conditions are attained.

5

Chapter 1:

Introduction

1.0 General The dynamics of geological processes such as mantle convection, subduction,

mountain building or basin formation are highly influenced by the presence of fluids

(melt or volatile phases) as they affect the mechanical and transport properties of the

rocks inside the Earth’s mantle and crust (e.g. Fyfe et al., 1978). The rate of

deformation is significantly controlled by the rate of fluid flow (e.g. Ferry, 1994).

Deformation in turn has a major effect on the transport properties of rocks and on the

fluid distribution (e.g. de Meer et al., 2002), as e.g. increasing fluid pressures may

cause hydraulic microfractures that could lead to at least transiently high

permeabilities and associated high fluid flux (Vrolijk, 1987; Cox & Etheridge, 1989).

Additionally, high pore fluid pressure can attribute to a reduction of the maximum

strength of the lithosphere (Fig. 1.1).

The effect of fluids on deformation has been demonstrated i) both in field and

laboratory, ii) on all scales ranging from grain to plate boundaries, and iii) from the

inside of the Earth’s mantle up to crustal fault zones, i.e. from ductile via semi-brittle

towards brittle dominated deformation regimes.

Inside the oceanic mantle, the depth distribution of seismic anisotropy is interpreted to

be caused by the incorporation of water into the point defects of this nominally

anhydrous mineral (Karato & Jung, 1998) as deformation experiments in the high-

pressure/high-temperature regime show that the presence of water reduces the

strength of olivine aggregates (e.g. Hirth & Kohlstedt, 1996) and that the viscosity of

olivine is reduced with increasing water content.

This phenomenon is also inferred to be found in subduction zones, where fluids from

the subducting slab migrate upwards into the mantle wedge, decrease the viscosity of

the latter and promote the onset of melting. Whether the fluids migrate by porous

6

flow, channelized flow (through hydrofractures) or diapiric advection is determined

by rheology, permeability of the slab and the distribution of the fluids (Hirth &

Kohlstedt, 2003).

Figure 1.1: Example of a strength envelope for the crust showing brittle/frictional

behavior dominating at upper crustal levels (according to Byerlee’s law) and dislocation creep determining the crustal strength at deeper levels (solid curves). The dashed line represents the hypothesized weakening at the brittle-ductile transition probably related to effects of fluid-assisted deformation mechanisms (from Bos & Spiers, 2002).

Inside the Earth’s lithosphere, deformation is predominantly localized in ductile shear

zones or (semi-) brittle faults or fault zones. These are commonly permeable pathways

for crustal fluids. The involvement of fluids in shear zone development and faulting

processes is widely recognized (e.g. Hickman et al., 1995; Alsop & Holdsworth,

2004).

Physical indicators help to identify pathways for fluid flow as increased fluid pressure

reduces the strength of crustal rocks. In addition a variety of chemical effects

influences the rocks’ mechanical properties, i.e. mineral reactions and the transfer of

heat and mass (solutes) that facilitate deformation.

Besides these large-scale pathways for pervasive fluid flow, the micro-scale structure

of rocks or minerals has to be considered. Here the presence of fluids in grain

boundaries has a major impact on the rheology. Fluid-bearing grain boundaries serve

as fast intergranular diffusion path and allow processes of stress-driven mass transfer

as pressure solution (dissolution-precipitation creep) (e.g. Rutter, 1976; Spiers et al.

7

1990) or fluid-assisted grain boundary migration (e.g. Urai et al., 1986a). The

structure of such mobile fluid-filled grain boundaries and their influence on

recrystallization processes are still poorly understood.

1.1 Brief review of recrystallization processes In order to prevent confusion, the (metallurgical) terminology of some processes and

mechanisms that are involved during and after plastic deformation of aggregates is

summarized (Fig. 1.2):

During plastic deformation dislocations and interfaces are generated inside a

crystalline aggregate resulting in a raised stored energy (Humphreys & Hatherly,

1996). As the accumulated dislocations together with curved interfaces or chemical

potential gradients (Fig. 1.2a) are thermodynamically unstable, there is always a

driving force to approach equilibrium conditions by reducing the free energy. Hereby

the processes recrystallization, recovery and grain coarsening are involved.

Recrystallization is, in a general sense, defined as formation and migration of high

angle boundaries driven by the stored energy of deformation, whereas recovery

involves all processes of annihilation and rearrangement of dislocations to lower the

energy (Fig 1.2b) (Doherty et al., 1997).

Under static conditions, i.e. in the absence of concurrent deformation the modification

of the grain structure is termed static recrystallization.

Here, primary recrystallization takes place during annealing of a sufficiently

deformed material (Figs. 1.2c & d): here, very small, new, strain-free grains are

nucleated preferentially at old grain boundaries and grow at the expense of the

surrounding deformed material (Gottstein & Mecking, 1985). Once nuclei exist, this

process is predominantly driven by the difference in dislocation density. Further

annealing results in grain coarsening which lowers the energy configuration of the

grain boundaries. The corresponding driving force is the reduction in grain boundary

area and curved boundaries become straight during migration. During normal grain

growth the average grain size increases (Fig. 1.2e): small grains are consumed while

larger ones grow resulting in maintenance of a narrow distribution of both grain size

and shape (Fig. 1.2f) (Atkinson, 1988; Evans et al., 2001). In some circumstances a

few grains grow selectively to very large grains; a process known as exaggerated

grain growth (Fig. 1.2g). It is characterized by a bimodal grain size distribution with

8

one curve emerging with time compared to the primary distribution (Fig. 1.2h). This

process is also termed abnormal grain growth or secondary recrystallization.

Figure 1.2: Schematic illustration of the main annealing processes (after Humphreys

& Hatherly, 1996). a) deformed state, b) recovered, c) partially recrystallized, d) completely recrystallized, e) normal grain growth, g) abnormal grain growth. f) and h) schematically represent the grain size distribution during normal grain growth and abnormal grain growth, respectively.

9

In contrast, dynamic recrystallization is a process that can lower the free energy of a

crystalline aggregate during deformation. After reaching a critical strain dynamic

recrystallization initiates by nucleation of new primary recrystallizing grains,

preferentially at pre-existing high angle grain boundaries. Again, such strain-free

grains grow into the deformed microstructure driven by the stored energy of

deformation. However due to continuing deformation the dislocation density in the

new grains increases and reduces the driving force for further growth, eventually until

its cessation. The growth can also be limited by the nucleation of further grains at the

migrating grain boundaries, which then grow into both deformed old and deformed

recrystallized grains. With continuous deformation the recrystallized grains attain a

steady state grain size which is strongly dependent on the flow stress, but only minor

to the deformation temperature (Humphreys & Hatherly, 1996).

Upon cessation of deformation the nucleation of new grains is stopped. However

during further annealing the already existing nuclei grow with no incubation period

into the partly or fully recrystallized matrix, a process known as metadynamic

recrystallization.

Subsequently recovery and static recrystallization processes follow. After complete

recrystallization the material may be subject to further grain growth.

In geosciences dynamic (= syntectonic) recrystallization is a common process and two

major types are distinguished: subgrain rotation recrystallization and grain boundary

migration recrystallization.

A fundamental process that occurs during recrystallization and grain growth in

crystalline aggregates is grain boundary migration. In Earth sciences it is of

enormous significance as it influences mechanical and transport properties of rocks,

especially if a fluid-phase is present in grain boundaries.

Although grain boundary migration was subject of extensive investigations over the

last decades in material sciences, the details of the mechanism of migration of high

angle grain boundaries (grain boundaries with misorientations larger than 10 to 15 °)

is still under debate; detailed information on the atomic mechanism of grain boundary

migration is lacking (Gottstein & Shvindlerman, 1999). In the following some

metallurgical grain boundary migration models are briefly summarized.

The general concept of grain boundary migration is based upon the theory of reaction

rates (Haessner & Hofmann, 1978): as a result of thermal activation, atoms are

detached from the grains, move into the grain boundary lying in between and then re-

10

attach to one of the grains again. In the presence of a driving force, i.e. if a free energy

difference exists between two lattice regions, the grain boundary migrates

unidirectional. Detailed models focus on the material transport and one distinguish

between single-process and group-process theories. In the early single-process theory

(Turnbull, 1951) the boundary migration is controlled by single movements of atoms:

after being detached they jump across the narrow boundary before being attached to

the other grain. In contrast, Mott (1948) suggested that groups (islands) of atoms

move from one grain into the grain boundary, and that similar groups attach

themselves to the other lattice region and thus provided an explanation of the

experimentally derived high thermal activation energies (Gleiter, 1969) introduced the

crystallography to the grain boundary migration models. He pointed to the importance

of vacancies at the boundary edges with respect to the diffusion process and proposed

a model based on the motion of steps. Crystal surfaces consist of steps from which

atoms are removed. In the presence of a driving force the atoms diffuse for short

distances inside the grain boundary, before they are added again at steps of the

adjacent crystal lattice. This process is analogous to the growth of crystals from vapor.

In this model the mobility depends on both misorientation and boundary plane (that is

characterized by the step density) and explains the observed anisotropy of grain

boundary migration.

Other geometric models focus on the defect structure of the grain boundary, e.g. the

motion of steps by the movement of dislocations (King and Smith, 1980), or the

motion by a cooperative shuffling of atoms in groups (Babcock & Balluffi, 1989).

Recently, Merkle et al. (2002) provided evidence of the existence of such cooperative

atomic motion in grain boundary migration by in-situ experiments observed in high-

resolution transmission electron microscopy. Even if the metallurgists investigate

materials of exactly known composition, a lot of questions still remain: e.g. the role of

atoms inside the grain boundary such as the distance of diffusion inside the boundary

or the time atoms remain or move inside the grain boundary region.

In geomaterials grain boundary migration is complex due to the large variability that

range from pure solid-state grain boundaries to micron-sized, wide grain boundaries

that might be filled with fluids or melt.

11

1.2 The grain-scale distribution of fluids The grain-scale fluid distribution and connectivity have important consequences for

rheology and mass transfer processes. The majority of previous work has concentrated

on characterizing the fluid distribution under purely hydrostatic conditions.

Here, the fluid topology in a low porosity monophase polycrystalline aggregate is

controlled by the balance between solid-solid and solid-fluid interfacial energies, and

hence the dihedral angle θ (Fig. 1.3).

Figure 1.3: Schematic drawings showing the dihedral angle and its significance for

connectivity of fluids in texturally equilibrated porous rocks. a) the geometry of the dihedral angle θ results from balancing of grain boundary interfacial energy (γss) and solid-fluid interfacial energies (γsf); b) representation of idealized fluid distribution with cross-sections: the left grain shows a dihedral angle of less than 60° resulting in fluid distribution along a connected triple junction network; for θ > 60° (right side) however, the fluid is restricted to isolated inclusions in triple junctions or along grain boundaries (after Smith, 1964).

12

In the case of θ > 60° the fluids will be present inside isolated inclusions, whereas for

0° < θ < 60° the fluid forms an interconnected network on grain boundary triple

junctions (Holness, 1997). A dihedral angle θ = 0° results in complete wetting of grain

boundaries, but is interpreted unlikely to occur under these equilibrium conditions

(Watson & Brenan, 1987). Hess (1994) proposed that very thin films may exist even

on grain faces in equilibrium fluid topologies characterized by non-zero dihedral

angles.

The stress state of much of the solid Earth however, is governed deviatoric (= non-

hydrostatic) stresses. This affects the dihedral angle and so the fluid distribution (e.g.

Urai et al., 1986b; Tullis et al., 1996) and is supported by theoretical considerations

that show that the concept of equilibrium fluid distribution may not be applicable or

relevant during deformation (Heidug, 1991).

Several ductile deformation processes are affected by the presence of fluids as e.g.

pressure solution and recrystallization processes.

Under relatively low stresses at which dislocation creep mechanisms are slow,

pressure solution creep (dissolution-precipitation or fluid-phase diffusional creep) is

important. It is characterized by the dissolution of material at interfaces with a high

differential stress, diffusion through a fluid phase provided by stress/strain induced

gradients in solubility and precipitation at interfaces under low stress (e.g. Schutjens,

1991). The geometry of the fluid residing on the grain boundaries is still

controversially discussed. Three different models that have been proposed (den Brok

et al., 2002): i) the thin film model (Rutter, 1976; Hickman & Evans, 1991; Renard &

Ortoleva, 1997), ii) the island-channel model (Lehner, 1990; Spiers & Schutjens,

1990; Schutjens & Spiers, 1999) and iii) the island-crack model (Gratz, 1991; den

Brok, 1998) (Fig. 1.4).

In the thin film boundary model (Fig. 1.4a) the grains are separated by a thin,

structured water film with a thickness of a few nanometers. This film is proposed to

transmit the contact stress and diffusion is the process of the transport of dissolved

material. Hickman & Evans (1995) studied the morphology of a grain to grain contact

of a convex halite lens against a silica plate in the presence of brine under stress and

did not find a relief in the contact area at least at the resolution of the interferometer

(100nm). Additionally the film thickness was observed to increase with decreasing

normal stress onto the contact.

13

Figure 1.4: Schematic illustration of the grain boundary structure according to the

different models for pressure solution (after den Brok et al., 2002): a) thin film model; b) island-channel model; c) island-crack model. See text for a detailed description.

The island-channel boundary model (Fig. 1.4b) is based on the assumption that –

during pressure solution – the fluids residing in thin films are squeezed out between

the grains resulting in solid-solid contact (islands) through which the contact stresses

are transmitted and water-filled channels through which the material transport takes

place by diffusion. This microscopically rough island-channel structure is

dynamically stable.

The island-crack boundary model (Fig. 1.4c) proposes static islands that are separated

by microfracture-controlled fluid channels. In contrast to the solid-solid contact of the

island-channel model, the islands in this model contain thin films comparable to the

earlier proposed thin film boundary model. Due to the low thickness of these thin

films diffusion through the latter is rate-controlling. However compared to the thin

film model, the total diffusivity in the island-crack model is increased by the presence

of the microcracks.

At crustal levels, at which dislocation creep mechanisms are dominant, the fluids have

a significant effect on the recrystallization behaviour and microstructural evolution of

14

many minerals such as quartz (Griggs, 1974; Tullis & Yund, 1982; Jaoul et al., 1984;

Kronenberg & Tullis, 1984; Hirth & Tullis, 1992; Post & Tullis, 1998), feldspar

(Tullis et al., 1996; Dimanov et al., 1999; Post & Tullis, 1999), olivine (Karato et al.,

1986; Mei & Kohlstedt, 2000a, b) or ionic salts (Urai, 1983a, 1985; Urai et al., 1986a;

Peach et al., 2001; Watanabe & Peach, 2002, ter Heege et al., 2005).

It is argued that – in the presence of fluids – the grain boundary structure changes

during dynamic recrystallization, such that grain boundary migration is assisted by

thin fluid films residing on the grain boundaries (Urai et al., 1986b; Drury & Urai,

1990). Such fluid-filled grain boundaries were observed in natural and experimentally

deformed rocks:

One method to show the presence of such fluid films was the application of the ether

test in water containing halite samples (Spiers et al., 1986): during evaporation of the

ether the fluid film was disrupted into isolated non-volatile droplets. Urai et al.

(1986a) showed fluid films by SEM observations on deformed water-containing halite

samples: 1 month after the experiment grain boundaries showed smooth surfaces,

whereas samples annealed for one year showed grain boundaries with isolated

bubbles. The authors interpreted these results as evidence for the presence of brine

films that shrink into isolated fluid inclusions after grain boundary migration stopped.

Similar observations were shown by in-situ experiments conducted on wet bischofite,

during which water-filled grain boundaries neck down after grain boundary migration

stopped (Urai, 1987). Additionally, in these experiments some cigar-shaped fluid

inclusions were left behind the migrating grain boundary supporting the hypothesis of

the presence of brine films.

Analogous microstructures were observed in experiments on the melt topology in

partially molten mantle peridotites (Jin et al., 1994; Bai et al., 1997). During

deformation under low stresses the melt spread onto grain boundaries. After cessation

of deformation and successive reannealing the melt film pulled back to form isolated

melt inclusions. Such stress-induced grain boundary wetting was also reported from

experiments on feldspar (Tullis et al., 1996) and is suggested to be related to surface

energy anisotropies.

Indications of fluid films on migrating grain boundaries in nature were found in quartz

mylonites (Mancktelow et al., 1998; Mancktelow & Pennacchioni, 2004). Scanning

electron microscopy (SEM) observations of ‘worm-like’ structures with the tendency

15

of forming ridges along triple junction tubes were proposed to result from necking

down of former fluid films.

The phenomenon of grain boundary wetting was also observed in material sciences.

Most commonly impurities or second phases on grain boundaries inhibit or decelerate

migration. However experiments on grain boundaries in Al doped with small amounts

of Pb suggest that above a transient temperature the aluminium grain boundaries are

wetted by liquid lead being responsible for the enhanced migration rates (Molodov et

al., 1997). After deformation has stopped Pb is found in isolated inclusions on the Al

grain boundaries and support the interpretation that the liquid layer may have pulled

back into its low energy configuration.

According to a the fluid film model (Urai et al., 1986b), the fluid-filled grain

boundaries are interpreted to migrate by i) dissolution of the deformed grains, ii)

diffusion through the brine film and iii) precipitation on the low-index facets of the

recrystallized grains forming smooth grain surfaces (Fig. 1.5), comparable to the step

model of Gleiter (1969). The driving force is provided by the difference in dislocation

stored energy across the boundary. The rate of grain boundary migration assisted by

fluid films is much higher than that for dry grain boundaries (Urai et al., 1986b),

conceivably because the diffusivity in a fluid phase is dramatically higher than in a

dry grain boundary.

The migration rate is determined by whichever of the three steps of dissolution,

diffusion or precipitation is the slowest (Lehner & Bataille, 1984).

The migration is interpreted to be dependent on the film thickness (Fig. 1.6). Due to

the presence of fluids on grain boundaries the migration rate is proposed to increase,

however gradually as the assumption of equal diffusion coefficients does not hold for

very thin films compared to the bulk fluid (Rutter, 1976).

To derive the migration kinetics of fluid-filled grain boundaries quantitatively, Urai et

al. (1986b) considered two crystals of identical composition and structure but

arbitrary orientation and different dislocation density that are separated by a fluid

phase of solution of the crystals (see Fig. 1.5). The crystal with the higher dislocation

density is not in equilibrium with the fluid, which causes a local supersaturation at this

interface and results in a concentration gradient across the fluid layer.

16

Figure 1.5: Schematic illustration of fluid-assisted grain boundary migration (see text

a detailed description; after Peach et al., 2001).

Figure 1.6: Diffusion coefficient of the grain boundary fluid, based on the data of

Watanabe and Peach (2002) as a function of film thickness and the effect of the decrease in diffusion coefficient on diffusion-controlled grain boundary migration (cf. Urai et al., 1986b).

17

With the assumptions that i) the diffusion across the fluid film is rate controlling

(neglecting interface kinetics) and ii) the diffusion coefficient is identical to that of the

bulk fluid, these authors proposed the grain boundary migration to be

( )0

sgb

sol

D c c MV

t ρ−

=

with c0: concentration in the fluid in equilibrium with the unstrained crystal [mol/m3] cs: concentration in the fluid in equilibrium with the strained crystal [mol/m3] t: thickness of the fluid film [m] M: molecular weight of the solid [kg/mol] D: diffusion coefficient of the migrating fluid [m2/s] ρsol: density of the solids [kg/m3].

This fluid film model was further developed by Peach et al. (2001) and Watanabe &

Peach (2002). Besides diffusion control they included interface reaction control of the

migration rate into their calculations, as dissolution and precipitation may also be rate

controlling.

For diffusion to be rate limiting the grain boundary migration rate Vgb [m/s] is

gb

D c fVt R TΩ ∆

=

with Ω: molar volume of solid [m3/mol] ∆f: difference in elastic strain energy [J/mol] (=∆µ: difference in chemical

potential) R: gas constant [J/(mol·K)] T: temperature [K].

In the case of dissolution or precipitation reaction to be rate limiting, these authors

propose for Vgb:

gbfV l

R T∆⎛ ⎞= Ω ⎜ ⎟

⎝ ⎠

with l: appropriate reaction rate coefficient [mol/(m2·s)].

18

Dependent on conditions which allow or limit recovery processes and considering that

the parameters D, C and l are temperature dependent, Peach et al. (2001) summarized

their calculations for grain boundary migration in a general equation, that is similar to

the power low creep:

with A: constant, dependent on rate-controlling mechanism and on relative importance of subboundary vs. dislocation stored energy

E: constant of apparent activation energy, dependent on rate-controlling mechanism [J/mol]

σ: applied stress [Pa] m: constant [-], dependent on importance of subgrains (=1) vs. free

dislocations (=2).

When grain boundary movement is stopped, surface energy related forces reassert

themselves and the fluid film is interpreted to contract into rolls and bubbles (Fig.

1.7). Such a development towards equilibrium conditions is also reported form crack

healing processes (e.g. Lemmlein, 1956; Brantley et al., 1990).

The influence of surface energy related forces was also shown in uniaxial

densification tests on fine grained sodium nitrate aggregates in the presence of sodium

nitrate saturated solution (Visser, 1999). For grain sizes smaller than 20 µm in the low

stress regime the strain rates decreased with decreasing grain size, and these

observations were attributed to the effect of surface energy forces.

Additionally Visser (1999) developed models for mass transfer processes of

crystalline aggregates containing a fluid phase that include both stress and surface

related driving forces.

From the models it was concluded that the relative driving forces are strongly

dependent on amongst others grain size, applied stress and interfacial energies. Based

on these observations distinct fields in stressed systems were delineated in which

either stress-driven solution-precipitation creep may occur or surface-energy

controlled neck growth or grain boundary healing processes may predominate.

exp

mgb

EV AR T

σ−⎛ ⎞= ⎜ ⎟⎝ ⎠

19

Figure 1.7: Schematic drawing illustrating the formation of arrays of fluid inclusions

in a grain boundary containing a fluid film during migration (from Urai, 1983b).

In the following the criteria are briefly presented together with the diagrams Visser

plotted for the halite-brine system (Fig. 1.8):

i) Criterion 1, for the condition of zero driving force for solution-precipitation creep,

ii) Criterion 2, for the condition of zero driving force for neck growth and

iii) Criterion 3, for grain boundary healing versus a dynamically wetted island-

channel structure.

20

Figure 1.8: Maps of material transport mechanism fields derived by the application

of the Visser criteria for the NaCl-brine system (from Visser, 1999). a) criterion 1 separates the fields for pressure solution and contact healing for conditions with zero driving force for solution-precipitation creep; b) criteria 2a and 2b distinguish the fields for marginal dissolution and neck growth for conditions of zero driving force for neck growth; c) criteria 3a and 3b separate fields of dynamically wetted island-channel structure and grain boundary healing.

21

Criterion 1:

Here the driving force is described by

4e sl

v

ae dσ γ

− ,

with e

v

aeσ = stress related term and

4 sl

dγ = surface energy related term.

with σe: surface-energy-controlled yield stress [Pa] γsl: fluid-solid surface energy [J/m2] ev: volumetric strain [-] a: packing constant [-] d: grain size [m]

, i.e. the driving force for solution-precipitation creep can be decreased by surface

energy effects.

If the surface related energy term ( 4 sl

dγ ) exceeds the stress energy related one ( e

v

aeσ ),

the grain-to-grain contact will cement-up; however if e

v

aeσ > 4 sl

dγ , densification will

continue, and for the case of e

v

aeσ >>> 4 sl

dγ pressure solution will proceed according

to the conventional models (e.g. Spiers et al., 1990).

By replacing d with 2rgr and e

v

aeσ with σn, the dimensionless criterion is derived, that

separates the contact healing from the pressure solution field (Fig. 1.8a):

gr

ln ln ln 2 r

s sn sl

R T R Tσ γ⎛ ⎞⎛ ⎞Ω Ω

= +⎜ ⎟⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠

with rgr: radius of curvature of grains at the pore wall [m] σn: normal stress [Pa] Ωs: molar volume of solid phase [m3/mol] T: absolute temperature [K] R: molar gas constant [J/(K·mol)]

22

Criterion 2:

The driving force for neck growth in turn, can be reduced due to the effects of stress.

In this criterion it is considered that the dihedral angle (θ) at the contact margin may

be different from the equilibrium wetting angle (θeq). This results in a local driving

force that changes the contact geometry and is expressed by the term

cos cos2 2

eqθθ⎛ ⎞−⎜ ⎟

⎝ ⎠ and opposed to the term scm

ssl

r fγ⎛ ⎞

∆⎜ ⎟Ω⎝ ⎠

with rcm: radius of curvature at contact margin [m] θ: dihedral angle [°] θeq: equilibrium dihedral angle [°] ∆fs: excess (Helmholtz) free energy at the neck region [J/mol]

When both terms balance each other (equilibrium conditions), the zero driving force

condition for neck growth is

cos cos2 2

seqssl

cm

fr RT RT

θγ θ⎛ ⎞ ∆Ω − =⎜ ⎟

⎝ ⎠

This condition separates the neck growth field from the marginal dissolution field

(Fig. 1.8b).

As both elastic and plastic energy can attribute to the excess free energy (∆fs), Visser

distinguishes two extreme situations:

Criterion 2a, in which purely elastic strain energy exists at the contact margin: 2

2s snf

Eσ

∆ = Ω

with E: Youngs modulus [Pa]

The dimensionless equation is then

1 1 2ln ln ln ln cos cos2 2 2 2

s s seqn sl

cm

ERT r RT RT

θσ γ θ⎡ ⎤⎛ ⎞⎛ ⎞ ⎛ ⎞⎛ ⎞Ω Ω Ω= + + −⎢ ⎥⎜ ⎟⎜ ⎟ ⎜ ⎟⎜ ⎟

⎝ ⎠ ⎝ ⎠⎝ ⎠ ⎝ ⎠ ⎣ ⎦

23

Criterion 2b, for dislocation stored energy to be the dominant contribution at the

contact margin:

( )s sdislf Wρ∆ = Ω

with 2

2 2

2 ndisl b G

σρ ≈ and 212

W Gb=

with ρdisl: local dislocation density at the neck margin [m-2] W: dislocation line energy [J/m] b: Burgers vector [m] G: shear modulus [Pa]

This results in

( )22 sns

s

RTfGRT

σ Ω∆ =

Ω

And the respective criterion is expressed by

1 1ln ln ln ln cos cos2 2 2 2

s s seqn sl

cm

GRT r RT RT

θσ γ θ⎡ ⎤⎛ ⎞⎛ ⎞ ⎛ ⎞⎛ ⎞Ω Ω Ω= + + −⎢ ⎥⎜ ⎟⎜ ⎟ ⎜ ⎟⎜ ⎟

⎝ ⎠ ⎝ ⎠⎝ ⎠ ⎝ ⎠ ⎣ ⎦

Criterion 3:

However, not only the driving forces at the contact margins need to be considered.

Neck growth arguments may also be applicable within grain boundaries with an

island-channel structure, describing the growth of islands and hence grain boundary

healing. Here the island-island contacts are treated as small necks over an area

fraction α. Thus the driving force is balanced by the terms: cos cos2 2

eqθθ⎛ ⎞−⎜ ⎟

⎝ ⎠ and

sims

sl

r fγ⎛ ⎞

∆⎜ ⎟Ω⎝ ⎠.

with α: area fraction occupied by small necks [-] rim: radius of curvature of island margin [m]

The only difference to criterion 2 is the parameter ρisl which here describes the radius

of an island margin. The respective diagram (Fig. 1.8c) separates the pressure solution

24

field with an active dynamically stable island-channel structure from the contact

healing field.

Analogous to criterion 2, this criterion can be described for the two extreme

situations, i.e that the excess free energy (∆fs) is contributed by purely elastic or

dislocation stored energies.

Criterion 3a, for elastic energy to be the dominant contribution at the island margin:

1 1 2ln ln ln ln cos cos2 2 2 2

s s seqn sl

im

ERT r RT RT

θσ γ θα

⎡ ⎤⎛ ⎞⎛ ⎞ ⎛ ⎞⎛ ⎞Ω Ω Ω= + + −⎢ ⎥⎜ ⎟⎜ ⎟ ⎜ ⎟⎜ ⎟

⎝ ⎠ ⎝ ⎠⎝ ⎠ ⎝ ⎠ ⎣ ⎦

Criterion 3b, in which purely dislocation stored energy exists at the island margin:

1 1ln ln ln ln cos cos2 2 2 2

s s seqn sl

im

GRT r RT RT

θσ γ θα

⎡ ⎤⎛ ⎞⎛ ⎞ ⎛ ⎞⎛ ⎞Ω Ω Ω= + + −⎢ ⎥⎜ ⎟⎜ ⎟ ⎜ ⎟⎜ ⎟

⎝ ⎠ ⎝ ⎠⎝ ⎠ ⎝ ⎠ ⎣ ⎦

Postulating a dynamically stable island-channel structure from which stress is

released, Visser interpreted that only the surface energy drives the system to its

equilibrium conditions. This results in growth of islands and shrinking of channels

and hence in sealing the boundary by internal redistribution of fluid and solid in the

contact region. Such islands are proposed to grow preferentially close to the contact

rims and promote the sealing of the grain-to-grain contact which in turn prevents

transport in our out the boundary region and thus traps the fluids in isolated

inclusions. Finally, Visser (1999) concluded that pressure solution will slow down

and finally stop after attaining a strain and grain size dependent ‘yield stress’.

1.3 Aim of this study The principal aim of this study is to get a better insight into the properties of fluids in

grain boundaries, such as morphology, connectivity and their effect on

recrystallization processes. Such an understanding is required for modelling and

predicting transport and mechanical properties of rocks deforming under metamorphic

conditions. Although there is strong evidence for the presence of fluid-filled grain

25

boundaries during recrystallization under non-equilibrium conditions, their detailed

nature is still poorly understood. Therefore in this study the structure of fluid-filled

grain boundaries was observed in experimentally and naturally recrystallized samples.

As the detailed morphology of fluid-bearing boundaries has been only inferred

indirectly so far, direct observations on their geometry were conducted by using the

newly developed see-through apparatus that enables deformation under constant fluid

pressure.

1.4 Overview of this thesis In chapter 2 the effect of water on recrystallization processes and grain boundary

morphology in naturally deformed marble mylonites during shear zone evolution is

presented. Calcite marble mylonites from the Schneeberg Complex (Southern Tyrole,

Italy) are compared to those from the Naxos Metamorphic Core Complex (Greece).

While the two settings are characterized by similar lithology and geological history,

they are different in the nature of fluids present during late-stage deformation. The

respective shear zone profiles were sampled and chemical data was derived from

XRF, microprobe and stable isotopes. The microstructures were studied by both

optical microscopy of thin sections and scanning electron microscopy of broken and

polished & etched samples and finally compared with the recrystallization behavior of

other rock forming minerals in the presence of fluids.

Chapter 3 describes the experimental investigation into the role of brine on the

microstructural evolution and grain boundary morphology in halite. Although there is

evidence for fluid-filled grain boundaries in recrystallizing halite from experiments

and natural samples, their detailed nature is under debate. One problem of the fluid

film model was that the observations were made after removal of the stress, which

could have led to a redistribution of the fluid by viscous flow. To avoid such a

redistribution of fluids after stress release, in this study compacted, synthetic,

polycrystalline wet halite samples were annealed under atmospheric conditions,

without deformation. The microstuctural evolution and grain boundary morphology of

the statically recrystallizing samples was studied by scanning electron and reflected

light microscopy of surfaces that were either polished and etched or broken.

As these interpretations of grain boundary structure in experimentally deformed wet

salt samples annealed statically at room temperature were based on indirect evidence

from reflected light microscopy and conventional SEM, chapter 4 presents direct

26

observations of fluid-filled grain boundaries using the cryo-scanning electron

microscope (cryo-SEM) in which the grain boundary fluids were frozen before

breaking the samples. In order to characterize the crystallographic nature of grain

boundaries, detailed Electron Backscatter Diffraction (EBSD) analysis was used. The

results were compared with existing models of recrystallization and grain growth

processes.

Chapter 5 presents in-situ deformation experiments in transmitted light microscopy

as they allow continuous and real-time observation of the whole range of processes

involved during fluid assisted grain boundary migration. Therefore, a new see-through

deformation apparatus with an integrated controlled pore fluid pressure system was

developed. The experiments were carried out on bischofite in the presence of

saturated solution at different temperatures. The microstructural evolution was studied

during and after deformation, with a special focus on migrating fluid-filled grain

boundaries and their break-up into arrays of isolated inclusions. The rates of grain

boundary migration were measured, assigned to the different temperatures and strain

rates and compared with observations and models from previous studies.

Chapters 2 to 5 are written as self-standing entities, published, submitted or intended

to be published as separate papers. As a result some repetition was unavoidable.

1.5 Parts of the thesis which have been published

Articles:

Schenk, O. & Urai, J.L. submitted. The migration of fluid-filled grain boundaries in recrystallizing synthetic bischofite – first results of in-situ HPHT deformation experiments in transmitted light. Journal Of Metamorphic Geology.

Schenk, O., Urai, J. L. & Piazolo, S. submitted. Structure of grain boundaries in wet, synthetic polycrystalline, statically recrystallizing halite – evidence from cryo-SEM observations. Geofluids.

Schenk, O., Urai, J. L. & Evans, B. submitted. The effect of water on recrystallization behavior and grain boundary morphology in calcite – observations of natural marble mylonites. Journal Of Structural Geology.

Schenk, O. & Urai, J. L. 2004. Microstructural evolution and grain boundary structure during static recrystallization in synthetic polycrystals of sodium chloride containing saturated brine. Contributions to Mineralogy and Petrology 146, 671-682.

27

Conference proceedings:

Schenk, O. & Urai, J. L. 2004. The effect of brine on grain boundary structure in synthetic halite polycrystals recrystallized under static conditions: observation from conventional and cryogenic SEM. TSK X. Aachen, Germany, 81.

Schenk, O., Urai, J. L. & Evans, B. 2003. The influence of grain boundary fluids on the recrystallization behavior in calcite: a comparison of "dry" and "wet" marble mylonites. EOS Trans. AGU 84 (46), Fall Meet. Suppl., Abstract T41B-07. San Francisco, USA.

Urai, J. L. & Schenk, O. 2003. Microstructural evolution and grain boundary structure during static recrystallization in synthetic polycrystals of sodium chloride containing saturated brine. EOS Trans. AGU 84 (46), Fall Meet. Suppl., Abstract T42A-0284. San Francisco, USA.

Schenk, O. & Urai, J. L. 2003. Microstructural evolution and grain boundary structure during static recrystallization of synthetic polycrystals of sodium chloride containing a saturated salt solution. DRT 2003. Geosciences Rennes, St. Malo, France, 138.

de Bresser, H., Urai, J. L., Schenk, O. & Olgaard, D. 2003. Strength and microstructure of wet and dry recrystallizing marble. In: DRT 2003. Geosciences Rennes, St. Malo, France, 33.

Urai, J. L., De Bresser, H., Schenk, O. & Olgaard, D. 2002. Effect of water on the strength and microstructure of recrystallizing marble. VMSG Symposium 2002 "Dynamics of the Earth and Mars", Universiteit Utrecht, The Netherlands.

1.6 References Alsop, G. I. & Holdsworth, R. E. 2004. Shear zones - an introduction and overview.

In: Flow Processes in Faults and Shear Zones (edited by Alsop, G. I., Holdworth, R. E., McCaffrey, K. J. W. & Hand, M.) 224. Geological Society Special Publications, 1-9.

Atkinson, H. V. 1988. Overview no. 65 : Theories of normal grain growth in pure single phase systems. Acta Metallurgica 36(3), 469-491.

Babcock, S. E. & Balluffi, R. W. 1989. Grain boundary kinetics--I. In situ observations of coupled grain boundary dislocation motion, crystal translation and boundary displacement. Acta Metallurgica 37(9), 2357-2365.

Bai, Q., Jin, Z.-M. & Green II, H. W. 1997. Experimentally investigation of the rheology of partially molten peridotite at upper mantle pressures and temperatures. In: Deformation-enhanced Fluid Transport in the Earth's Crust and Mantle (edited by Holness, M. B.). Chapman & Hall, London.

Bos, B. & Spiers, C. J. 2002. Frictional-viscous flow of phyllosilicate-bearing fault rock: Microphysical model and implications for crustal strength profiles. Journal of Geophysical Research 107(B2), ECV 1-1 - 1-13.

Brantley, S. L., Evans, B., Hickman, S. H. & Crerar, D. A. 1990. Healing of microcracks in quartz: Implications for fluid flow. Geology 18, 136-139.

de Meer, S., Drury, M. R., de Bresser, J. H. P. & Pennock, G. M. 2002. Current issues and new developments in deformation mechanisms, rheology and tectonics. In: Geological Society Special Publications (edited by de Meer, S., Drury, M. R., de Bresser, J. H. P. & Pennock, G. M.) 200. Geological Society of London; United Kingdom, 1-27.

28

den Brok, S. W. J. 1998. Effect of microcracking on pressure-solution strain rate; the Gratz grain-boundary model. Geology 26(10), 915-918.

den Brok, B., Morel, J. & Zahid, M. 2002. In situ experimental study of roughness development at a stressed solid/fluid interface. In: Geological Society Special Publications (edited by de Meer, S., Drury, M. R., de Bresser, H. & Pennock, G. M.) 200. Geological Society of London : London, United Kingdom, United Kingdom, 73-83.

Cox, S. F. & Etheridge, M. A. 1989. Coupled grain-scale dilatancy and mass transfer during deformation at high fluid pressures: examples from Mount Lyell, Tasmania. Journal of Structural Geology 11(1-2), 147-162.

Dimanov, A., Dresen, G., Xiao, X. & Wirth, R. 1999. Grain boundary diffusion creep of synthetic anorthite aggregates: The effect of water. Journal of Geophysical Research 104(B5), 10,483-10,497.

Doherty, R. D., Hughes, D. A., Humphreys, F. J., Jonas, J. J., Juul Jensen, D., Kassner, M. E., King, W. E., McNelley, T. R., McQueen, H. J. & Rollett, A. D. 1997. Current issues in recrystallization: a review. Material Science and Engineering A 238, 219-274.

Drury, M. R. & Urai, J. L. 1990. Deformation-related recrystallization processes. Tectonophysics 172(3-4), 235-253.

Evans, B., Renner, J. & Hirth, G. 2001. A few remarks on the kinetics of static grain growth in rocks. International Journal of Earth Science (Geologische Rundschau) 90, 88-103.

Ferry, J. M. 1994. A historical review of metamorphic fluid flow. Journal of Geophysical Research 99(B8), 15,487-15498.

Fyfe, W. S., Price, N. J. & Thompson, A. B. 1978. Fluids in the Earth's crust. Elsevier Publishing, Amsterdam, 383 pp.

Gleiter, H. 1969. Theory of grain boundary migration rate. Acta Metallurgica 17(7), 853-862.

Gottstein, G. & Mecking, H. 1985. Recrystallization. In: Preferred Orientations in Deformed Metals and Rocks (edited by Wenk, H. R.). Academic Press, United States, 183-218.

Gottstein, G. & Shvindlerman, L. S. 1999. Grain boundary migration in metals: thermodynamics, kinetics, applications. CRC Press, 385 pp.

Gratz, A. J. 1991. Solution-transfer compaction of quartzites; progress toward a rate law. Geology 19(9), 901-904.

Griggs, D. 1974. A model of hydrolytic weakening in quartz. Journal of Geophysical Research 79(11), 1653-1661.

Haessner, F. & Hofmann, S. 1978. Migration of High Angle Grain Boundaries. In: Recrystallization of Metallic Materials (edited by Haessner, F.). Dr. Riederer Verlag GmbH, Stuttgart, 63-95.

Heidug, W. K. 1991. A thermodynamic analysis of the conditions of equilibrium at nonhydrostatically stressed and curved solid-fluid interfaces. Journal of Geophysical Research 96(B13), 21,909-21,921.

Hess, P. C. 1994. Thermodynamics of thin fluid films. Journal of Geophysical Research 99(B4), 7219-7229.

Hickman, S. H. & Evans, B. 1991. Experimental pressure solution in halite; the effect of grain/interphase boundary structure. Journal of the Geological Society 148, 549-560.

Hickman, S. H. & Evans, B. 1995. Kinetics of pressure solution at halite-silica interfaces and intergranular clay films. Journal of Geophysical Research, B, Solid Earth and Planets 100(7), 13,113-13,132.

29

Hickman, S., Sibson, R. & Bruhn, R. 1995. Introduction to special section: Mechanical involvement of fluids in faulting. Journal of Geophysical Research B 100(7), 12,831-12,840.

Hirth, G. & Tullis, J. 1992. Dislocation creep regimes in quartz aggregates. Journal of Structural Geology 14(2), 145-159.

Hirth, G. & Kohlstedt, D. L. 1996. Water in the oceanic upper mantle: implications for rheology, melt extraction and the evolution of the lithosphere. Earth and Planetary Science Letters 144(1-2), 93-108.

Hirth, G. & Kohlstedt, D. L. 2003. Rheology of the upper mantle and the mantle wedge: a view from the experimentalists. In: Inside the Subduction Factory (edited by Eiler, J.) 138. Geophysical Monograph, 83-105.

Holness, M. B. (Editor), 1997. Deformation-enhanced fluid transport in the earth's crust and mantle. The Mineralogical Society Series, 8. Chapman & Hall, London, 333 pp.

Humphreys, F. J. & Hatherly, M. 1996. Recrystallization and related annealing phenomena. Pergamon, 497 pp.

Jackson, M. P. A. & Talbot, C. J. 1986. External shapes, strain rates, and dynamics of salt structures. Geological society of america bulletin 97(March), 305-323.

Jin, Z.-M., Green II, H. W. & Zhou, Y. 1994. Melt topology in partially molten mantle peridotite during ductile deformation. Nature 372, 164-167.

Karato, S.-I., Paterson, M. S. & FitzGerald, J. D. 1986. Rheology of Synthetic Olivine Aggregates: Influence of Grain Size and Water. Journal of Geophysical Research 91(B8), 8151-8176.

Karato, S.-i. & Jung, H. 1998. Water, partial melting and the origin of the seismic low velocity and high attenuation zone in the upper mantle. Earth and Planetary Science Letters 157(3-4), 193-207.

King, A. H. & Smith, D. A. 1980. The effects on grain-boundary processes of the steps in the boundary plane associated with the cores of grain-boundary dislocations. Acta Crystallographica A 36, 335-343.

Kronenberg, A. K. & Tullis, J. 1984. Flow strengths of quartz aggregates: grain size and pressure effects due to hydrolytic weakening. Journal of Geophysical Research 89(B6), 4281-4297.

Lehner, F. K. & Bataille, J. 1984. Non-equilibrium thermodynamics of pressure solution. Pure and Applied Geophysics 122, 53-85.

Lehner, F. K. 1990. Thermodynamics of rock deformation by pressure solution. In: Deformation processes in minerals, ceramics and rocks (edited by Barber, D. J. & Meredith, P. D.). Unwin Hyman, 296-333.

Lemmlein, G. G. 1956. Formation of fluid inclusions and their use in geological thermometry. Geochemistry, 630-642.

Mancktelow, N. S., Grujic, D. & Johnson, E. L. 1998. An SEM study of porosity and grain boundary microstructure in quartz mylonites, Simplon fault zone, Central Alps. Contributions to Mineralogy and Petrology 131(1), 71-85.

Mancktelow, N. S. & Pennacchioni, G. 2004. The influence of grain boundary fluids on the microstructure of quartz-feldspar mylonites. Journal of Structural Geology 26(1), 47-69.

Mei, S. & Kohlstedt, D. L. 2000a. Influence of water on plastic deformation of olivine aggregates 1. Diffusion creep regime. Journal of Geophysical Research, B, Solid Earth and Planets 105(9), 21,457-21,469.

Mei, S. & Kohlstedt, D. L. 2000b. Influence of water on plastic deformation of olivine aggregates 2. Dislocation creep regime. Journal of Geophysical Research, B, Solid Earth and Planets 105(9), 21,471-21,481.

30

Merkle, K. L., Thompson, L. J. & Philipp, F. 2002. Collective Effects in Grain Boundary Migration. Physical Review Letters 88(22), 225501-1 - 225501-4.

Molodov, D. A., Czubayko, U., Gottstein, G. & Shvindlerman, L. S. 1997. Thermal extraction of Pb from the bulk to the surface in Al doped with Pb. Scripta Materialia 37(1), 111-117.

Mott, N. F. 1948. Slip at grain boundaries and grain growth in metals. Proceedings of the Physical Society 60, 391-394.

Peach, C. J., Spiers, C. J. & Trimby, P. W. 2001. Effect of confining pressure on dilatation, recrystallization, and flow of rock salt at 150°C. Journal of Geophysical Research 106(B7), 13,315-13,328.

Post, A. & Tullis, J. 1998. The rate of water penetration in experimentally deformed quartzite: implications for hydrolytic weakening. Tectonophysics 295(1-2), 117-137.

Renard, F. & Ortoleva, P. 1997. Water films at grain-grain contacts: Debye-Huckel, osmotic model of stress, salinity, and mineralogy dependence. Geochimica et Cosmochimica Acta 61(10), 1963-1970.

Rutter, E. H. 1976. The kinetics of rock deformation by pressure solution. Philosophical Transactions of the Royal Society of London A 283, 203-219.

Schutjens, P. 1991. Intergranular pressure solution in halite aggregates and quartz sands: an experimental investigation. PhD thesis, Universiteit Utrecht, 233 pp.

Schutjens, P. & Spiers, C. J. 1999. Intergranular pressure solution in NaCl: Grain-to-grain contact experiments under the optical microscope. Oil & Gas Science and Technology - Rev 54(6), 729-750.

Smith, C. S. 1964. Some elementary principles of polycrystalline microstructure. Metallurgical Reviews 9, 1-48.

Spiers, C. J., Urai, J., Lister, G. S., Boland, J. N. & Zwart, H. J., 1986. The influence of fluid-rock interaction on the rheology of salt rock and on ionic transport in the salt, University of Utrecht, 132 pp.

Spiers, C. J. & Schutjens, P. 1990. Densification of crystalline aggregates by fluid-phase diffusional creep. In: Deformation processes in minerals, ceramics and rocks (edited by Barber, D. J. & Meredith, P. D.). Unwin Hyman, 334-353.

Spiers, C. J., Schutjens, P. M. T. M., Brzesowsky, R. H., Peach, C. J., Liezenberg, J. L. & Zwart, H. J. 1990. Experimental determination of constitutive parameters governing creep of rocksalt by pressure solution. In: Deformation mechanisms, rheology and tectonics (edited by Knipe, R. J. & Rutter, E. H.) 54, Leeds, United Kingdom, 215-227.

ter Heege, J. H., De Bresser, J. H. P. & Spiers, C. J. 2005. Dynamic recrystallization of wet synthetic polycrystalline halite: dependence of grain size distribution on flow stress, temperature and strain. Tectonophysics 396(1-2), 35-57.

Tullis, J. & Yund, R. A. 1982. Grain growth kinetics of quartz and calcite aggregates. Journal of Geology 90, 301-318.

Tullis, J., Yund, R. A. & Farver, J. 1996. Deformation-enhanced fluid distribution in feldspar aggregates and implications for ductile shear zones. Geology 24(1), 63-66.

Turnbull, D. 1951. Theory of grain boundary migration rates. Transactions of the Metallurgical Society AIME 191, 1-7.

Urai, J. L. 1983a. Water assisted dynamic recrystallization and weakening in polycrystalline bischofite. Tectonophysics 96(1-2), 125-157.

Urai, J. L. 1983b. Deformation of Wet Salt Rocks. Unpublished PhD thesis, Rijksuniversiteit Utrecht, 223 pp.

31

Urai, J. L. 1985. Water-enhanced dynamic recrystallization and solution transfer in experimentally deformed carnallite. Tectonophysics 120(3-4), 285-317.

Urai, J. L., Means, W. D. & Lister, G. S. 1986a. Dynamic recrystallization of minerals. In: Mineral and rock deformation; laboratory studies; the Paterson volume. AGU Geophysical Monograph (edited by Hobbs, B. E. & Heard, H. C.) 36, 161-199.

Urai, J. L., Spiers, C. J., Zwart, H. J. & Lister, G. S. 1986b. Weakening of rock salt by water during long-term creep. Nature (London) 324(6097), 554-557.

Urai, J. L., Spiers, C. J., Peach, C., Franssen, R. C. M. W. & Liezenberg, J. L. 1987. Deformation mechanisms operating in naturally deformed halite rocks as deduced from microstructural investigations. Geologie en Mijnbouw 66, 165-176.

Visser, H. J. M. 1999. Mass transfer processes in crystalline aggregates containing a fluid phase. PhD thesis, Universiteit Utrecht, 244 pp.

Vrolijk, P. 1987. Tectonically driven fluid flow in the Kodiak accretionary complex, Alaska. Geology 15, 466-469.

Watanabe, T. & Peach, C. J. 2002. Electrical impedance measurement of plastically deforming halite rocks at 125°C and 50 MPa. Journal of Geophysical Research 107(B1), ECV 2-1 - 2-12.

Watson, E. B. & Brenan, J. M. 1987. Fluids in the lithosphere; 1, Experimentally-determined wetting characteristics of CO (sub 2) -H (sub 2) O fluids and their implications for fluid transport, host-rock physical properties, and fluid inclusion formation. Earth and Planetary Science Letters 85(4), 497.

32

Chapter 2:

The effect of water on recrystallization behavior and grain boundary morphology in calcite – observations of natural marble mylonites1

2.0 Abstract

Fluids are inferred to play a major role in the deformation and recrystallization of

many minerals (e.g. quartz, olivine, halite, feldspar). In this study we sought to

identify the effect of fluids on grain boundary morphology and recrystallization

processes in marble mylonites during shear zone evolution. We compared the

chemistry, microstructure and mesostructure of calcite marble mylonites from the

Schneeberg Complex, Southern Tyrole, Italy, to that from the Naxos Metamorphic

Core Complex, Greece. These two areas were selected for comparison because they

have similar lithology and resemble each other in chemical composition. In addition,

calcite-dolomite geothermometry indicates similar temperatures for shear zone

formation: 279 ±25 (Schneeberg Complex) and 271 ±15 °C (Naxos high-grade core).

However, the two settings are different in the nature of the fluids present during the

shear zone evolution: In the Schneeberg mylonites, both the alteration of minerals

during retrograde metamorphism in the neighboring micaschists and the existence of

veins suggest that aqueous fluids were present during mylonitization. The absence of

these features in the Naxos samples indicates that aqueous fluids were not as prevalent

1 Schenk, O., Urai, J.L. & Evans, B., in press. The effect of water on recrystallization behavior and

grain boundary morphology in calcite – observations of natural marble mylonites. Journal of Structural Geology

33

during deformation. This conclusion is also supported by the stable isotope signature.

Observations of broken and planar surfaces using optical and scanning electron

microscopes did not indicate major differences between the two mylonites: Grain

boundaries in both settings display pores with shapes controlled by crystallography,

and have pore morphologies that are similar to observations from crack and grain-

boundary healing experiments. Grain size reduction was predominantly the result of

subgrain rotation recrystallization. However, the coarse grains inside the wet

protomylonites (Schneeberg) are characterized by intracrystalline shear zones.

2.1 Introduction

In many orogenic belts, including, for example, in the Alps (Pfiffner, 1982;

Heitzmann, 1987; Burkhard, 1993), Spain (Behrmann, 1983), or Canada (Busch &

Van der Pluijm, 1995), marbles often accumulate large amounts of strain in localized

shear zones that involve deformation by crystal plastic processes (e.g. Bestmann et al.

(2000); Ulrich et al. (2002)). Such late-stage shear zones are formed under a variety of

thermal regimes and tectonic settings, but often record deformation at relatively low

pressure and temperatures (Bestmann et al., 2000). Owing to the extreme localization

of strain, such marble sequences are thought to play a key role in crustal deformation

processes, and have often been a subject of field studies (Schmid et al., 1977; Pfiffner,

1982; Behrmann, 1983; Heitzmann, 1987; Burkhard, 1993; Busch & Van der Pluijm,

1995; Badertscher & Burkhard, 2000; Bestmann et al., 2000; Badertscher et al., 2002;

Ulrich et al., 2002). The microstructures within these shear zones contain important

information on sense of shear, recrystallization mechanisms, and final grain size. The

stress conditions can be estimated by applying various flow laws, derived from

experimental studies (Schmid et al., 1980; Rowe & Rutter, 1990; Walker et al., 1990;

Rutter, 1995; Covey-Crump, 1998; de Bresser, 2002; Renner & Evans, 2002;

Herwegh et al., 2003). However, despite extensive field and laboratory investigation,

many questions remain concerning the mechanical behavior of carbonates, the exact

rheology appropriate to describe natural deformation, particularly at very large strains

(Pieri et al., 2001) and the influence of such variables as pore fluids, second phases

and chemical solutes.

One significant consideration for crustal deformation is the presence or absence of

water. Experimental deformation of wet aggregates indicate that fluids have a major

34

effect on mechanical properties and microstructural evolution of many minerals,

including quartz (Griggs, 1974; Tullis & Yund, 1982; Jaoul et al., 1984; Kronenberg

& Tullis, 1984; Hirth & Tullis, 1992; Post & Tullis, 1998), ionic salts (Skrotzki &

Welch, 1983; Urai, 1983, 1985; Urai et al., 1986b; Spiers et al., 1990; Spiers &

Brzesowsky, 1993; Peach et al., 2001; Watanabe & Peach, 2002; Schenk & Urai,

2004), feldspar (Dimanov et al., 1999), olivine (Mei & Kohlstedt, 2000b, a) or

clinopyroxene (Chen & Kohlstedt, 2003; Kohlstedt et al., 2003). In these minerals

fluids affect point defect concentrations and diffusion rates, enhance grain boundary

mobility, or alter dislocation dynamics. At low temperatures, solution transfer

processes may operate.

Considering the importance of deformation of marble formations in the processes of

mountain building, it is surprising that only a few laboratory studies have been

conducted to investigate the influence of fluids on the mechanical properties and

recrystallization behavior of calcite.

The results of comparisons between wet and dry samples are somewhat equivocal.

(Adams & Nicolson, 1900) deformed Carrara marble at a temperature of 300°C in the

presence of water, but did not observe differences with similar tests under dry

conditions at 300 and 400°C. Rutter (1974) deformed coarse-grained Carrara marble

and fine-grained Solnhofen marble, with and without interstitial water in the range of

20 to 500°C. He concluded that the presence of fluids did not significantly affect the

mechanical (rheological) behavior of the coarse-grained marble, but that the strength

of the fine-grained marble was reduced, at least at low temperatures. The presence of

porosity in the Solnhofen marble is significant, and the strength reductions in the

Solnhofen may have been related to weakening caused by compaction with an

associated increase in pore fluid pressure.

At higher temperatures, in dense, coarse-grained marbles, the presence of water seems

to weaken calcite rocks only by a small amount. In a few experiments done on

nominally drained samples of synthetic marble at 700 °C, confining pressure of 100

MPa; and strain rates ranging from 10-3 to 2 · 10-5 s-1, Olgaard (1985) found that wet

samples were only a few percent weaker than dry ones. In the most recent study, de

Bresser et al. (2005) compared the mechanical properties of pre-dried Carrara marble

deformed in axial compression at temperatures between 600 – 1000°C; confining

pressure of 300 MPa, and strain rate of 10-5 s-1 with that of wet, undrained Carrara

35

samples, deformed under the same conditions. At almost all conditions the wet

samples were only somewhat weaker than the dry ones. The final grain shape in the

wet samples was somewhat rounder than the dry samples.

Tullis & Yund (1982) and Olgaard (1985) studied the influence of water on the grain

growth kinetics in calcite aggregates. The annealing experiments on Solnhofen

limestone of Tullis & Yund (1982) at temperatures of 650 – 1000 °C and confining

pressures of 200 to 1500 MPa indicated faster growth rates of wet compared to the dry

samples (but the effect of water is not as pronounced as it is for novaculite (quartz). In

grain growth experiments on fine-grained synthetic calcite, Olgaard & Evans (1988)

concluded that normal grain growth in synthetic marble with added water was faster

than that in samples containing carbon dioxide inclusions, but slower than that in very

pure synthetic marbles with few or no fluid inclusions.

In this study we compare natural marble mylonites that recrystallized under varying

pore-fluid conditions. The extent of the interaction between the marble rocks and the

pore fluids contained within them has been characterized mainly by stable isotope

studies (Burkhard & Kerrich, 1988; Baker & Matthews, 1995; Lewis et al., 1998;

Matthews et al., 1999; Kirschner & Kennedy, 2001). We focus on the effect of fluids

on the grain boundary morphology and recrystallization behavior. We sampled marble

shear zones from the Schneeberg complex (Italian/Austrian Alps) and the high-grade

core of Naxos, Greece, both having a similar geological history, but with different

amounts of fluids present during late-stage deformation. Across the respective shear

zone profiles we looked at the microstructures in thin sections together with chemical

data derived from XRF and stable isotopes. In addition the mylonite samples were

broken and investigated under the SEM, as done previously, on hot-isostatically

pressed synthetic marble (Olgaard & FitzGerald, 1993) and on natural quartz

mylonites from the Simplon Fault Zone (Mancktelow et al., 1998). In a more recent

paper, Mancktelow & Pennacchioni (2004) compared natural quartz-feldspar

mylonites with variable amounts of water present during mylonitization with respect

to the grain boundary structure. They conclude that water-rich fluids enhance the

grain boundary mobility in quartz significantly.

36

2.2 Geological setting and sampling

2.2.1 Schneeberg complex

The Schneeberg Complex is a Paleozoic subunit of the Ötztal group (Ötztal-Stubai

Complex, Eastern Alps), NNW of Meran (Italy) (Fig. 2.1), and located between the

rest of the Ötztal and the Texel group, which lies to the south (Hoinkes et al., 1987;

Schmid & Haas, 1989). From NW towards SE, the Schneeberg Complex is composed

of the Monotonous Series, the Heterogeneous Series and the Laaser Series (Sölva et

al., in press). The rocks of the Laaser Series are pure, white marbles and garnet-mica

schists intercalated by layers of quartzite, amphibolite and calc-schist.

Whereas the dominant deformation structures of the Ötztal basement are Variscan, the

most southeastern part (Schneeberg Complex and Texel group) is overprinted by

strong Alpine deformation (Sölva et al., 2001) that formed a well-known set of fold

interference patterns with steep fold axes called ‘Schlingen’. Pre-alpine deformation is

preserved only as relicts in garnets (Sölva et al., 2001).

The Alpine deformational history is related to the eo-Alpine collision, with the

evolution from high-grade to lower greenschist facies metamorphic conditions

indicating crustal uplift (Spalla, 1990). Five deformation stages (D1-D5) can be

distinguished. Kinematic and geometric shear sense indicators are in agreement with

observations of (Sölva et al., in press) who assume that the high pressure rocks of the

Texel Complex were continuously exhumed relative to the Ötztal-Stubai Complex, by

motion along the NW dipping Schneeberg Normal Fault Zone (SNFZ). This fault

zone was active from amphibolite to lower greenschist facies (D1-D4). The

microstructure indicates that deformation was accommodated by a mixture of crystal

plasticity and cataclastic mechanisms. As temperature decreased, cataclasis became

more dominant (particularly in stage D5) (Sölva et al., in press), resulting in

progressive strain localization in marble units of the Laaser Series.

As consequence of this localization, the initially coarse-grained marble underwent

dramatic grain size reduction.

The area sampled in this study is located in the Schneeberg Complex between the

L’Altissima (Hohe Wilde) and Cima Fiammante (Lodner) within the Laaser Series.

37

We focused on D4 shear zones, together with their respective marble host rocks and

adjacent garnet-mica schists. Sample locations are shown in Figure 2.1. See Table 2.1

for GPS-data and additional observations.

Table 2.1: Overview of samples described in this paper.

GPS a samples

R H lithology recrystallization XRF ICP isotopes thermometry

S-1 b 06 55 372 51 79 058 marble protolith S

S-2 b 06 55 372 51 79 058 micaschist -

N-1a1 06 54 960 51 80 444 marble protolith x

N-1b1 06 54 960 51 80 444 marble protolith x x N-1

N-1c1 06 54 960 51 80 444 marble protolith x

N-2a2a 06 54 942 51 80 486 marble mylonite x

N-2a2b 06 54 942 51 80 486 marble protomylonite x

N-2d 06 54 942 51 80 486 marble mylonite x N-2

N-2h 06 54 942 51 80 486 marble mylonite x x x

E-1-1 06 54 867 51 80 170 marble mylonite x x x x

E-1-2 06 54 867 51 80 170 marble protomylonite x x

E-1-3 06 54 867 51 80 170 marble protomylonite x

E-1-4 06 54 867 51 80 170 marble protolith x

E-2 06 54 867 51 80 170 micaschist -

E-3 c 06 54 867 51 80 170 marble mylonite x x

E

E-4 06 54 867 51 80 170 marble mylonite x x x

M-8-1a 06 54 567 51 78 999 marble mylonite x x

M-8-1b 06 54 567 51 78 999 marble protomylonite x

M-8-2a 06 54 567 51 78 999 marble mylonite x x

M-8-2b 06 54 567 51 78 999 marble protomylonite x

M-8-3a 06 54 567 51 78 999 marble protomylonite x x

M-8-3b 06 54 567 51 78 999 marble mylonite x

M-8-3c 06 54 567 51 78 999 marble protomylonite x

M-8-4 06 54 567 51 78 999 marble protolith x

Schn

eebe

rg C

ompl

ex

M-8

M-8-5 06 54 567 51 78 999 marble protolith x

8a1a 37°05'36.4" 025°28'53.0" marble mylonite x x x x

8a1b 37°05'36.4" 025°28'53.0" marble protomylonite x 8

8a1c 37°05'36.4" 025°28'53.0" marble protolith x x

11a1a 37°05'12.6" 025°28'20.0" marble mylonite x x x

11a1b 37°05'12.6" 025°28'20.0" marble protomylonite x

11a1c 37°05'12.6" 025°28'20.0" marble protolith x x

11a2 37°05'12.6" 025°28'20.0" marble mylonite x

11a3 37°05'12.6" 025°28'20.0" marble protolith x Nax

os h

igh

grad

e co

re

11

11b 37°05'12.6" 025°28'20.0" micaschist -

a map datums: UTM European Datum 1950 (Schneeberg Comple); WGS84 (Naxos high grade core) b in distance to the late stage shear zones c several generations of syndeformational veins; thin section used for cathodoluminescence

38

Figure 2.1: Geological map of the Southwestern tip of the Schneeberg Complex

(coordinate system: UTM European Datum 1950). The letters contained in boxes indicate the collection locations for samples listed in Tables 2.1-2.5.

39

2.2.2 Naxos high-grade core

Naxos, Greece, the largest Cycladic island in the Aegean Sea (Fig. 2.2), belongs to the

Attic Cycladic Metamorphic Belt (Lister et al., 1984; Feenstra, 1985), and was

affected by at least two Alpine regional tectono-metamorphic events (Urai et al.,

1990). The first of the two, the main Alpine orogeny (Eocene) was due to the closure

of the Mesozoic Pindos Ocean (Hansen & Heide, 1999). This early compressional

tectonic phase (D1) ended 50 to 40 Ma ago and involved subduction of continental

margin material, generation of a nappe pile and regional high-pressure-low-

temperature (HPLT) metamorphism (M1).

Figure 2.2: Simplified geological map of Naxos (after Urai et al., 1990) with a

detailed overview of the studied outcrops (coordinate system: WGS 84). That compressional phase was followed by a period of extensional tectonics (D2) in

Early Miocene (Urai et al., 1990). The extensional phase probably accompanied a

40

southward retreat of the N-dipping subduction zone south of Crete (Urai & Feenstra,

2001), during which a back-arc basin formed with thinned crust, high heat flow, rapid

uplift of lower crustal rocks, and intrusion of granitoid magmas (Pe-Piper et al.,

1997). Also associated with this extensional phase was a regional greenschist facies

metamorphic event (M2a; ~25Ma) followed by localized deformation under high-

temperature-medium-pressure (HTMP; ~700 °C and 0.6 GPa) metamorphism (M2b;

~20-16Ma) (Buick & Holland, 1989). The elongate thermal dome that resulted

dominates the structure of the island. The metamorphic zonation is nearly concentric

with the dome, as shown in Figure 2.2 by six metamorphic isograds mapped in pelitic

and bauxitic units (Jansen & Schuiling, 1976). The core of the dome consists of

migmatized Hercynian basement, inside of which rafts of coarse-grained, pure, white

marble exist. They contain synmetamorphic pegmatites and thin boudinaged

amphibolites showing N-S extension during M2 metamorphism. The core is

surrounded by a cover of marble and schist (Hansen & Heide, 1999); together these

units are described as metamorphic core complex. This entire core complex is

strongly deformed due to a major crustal shear zone that was active during M2

metamorphism (Urai & Feenstra, 2001). The deformation that resulted is visible as

km-scale isoclinal folds with fold axes trending N-S. These folds are coaxially

refolded by open, upright folds (Urai et al., 1990). This regional fabric forms the

structural dome with its foliation warping over the migmatite core (Urai & Feenstra,

2001).

With decreasing temperatures during further extension and uplift, the deformation was

strongly localized in mylonite zones (D3) inside the marble units and mica schists.

These post-M2b mylonites are characterized by extreme grain size reduction (Urai et

al., 1990) and are parallel to the local orientation of bedding or to older high-grade

schistosity.

The samples of the Naxos core complex for this study were taken inside a marble raft

in the high-grade core close to Kinidaros, where very fine-grained post-M2b marble

mylonites are embedded in the very coarse-grained calcite marble of high purity (see

Figure 2.2 and Table 2.1 for details). As was done for the Schneeberg rocks, oriented

samples were taken from the marble mylonites, as well as from the respective host

rocks and adjacent mica schists.

41

2.3 Observations

2.3.1 Schneeberg complex

Mesostructures

The Laaser Series rocks consist of marbles, garnet-mica schists and minor intercalated

layers of amphibolite, quartzite, dolomite marble and calc-schist. At a distance of ~ 10

m from the D4 shear zones, the calcite marble is white, pure, and coarse-grained. The

marble is dominantly calcite, but also contains small amounts of quartz and rarely

mica. Inside the shear zones, the fine-grained marble mylonites are white, porcelain-

like layers alternating with yellowish bands of ferrous compounds formed by

alteration of thin mica layers. Thin (mm scale) calcite veins inside the mylonites were

observed in some outcrops.

The marble shear zones are up to 5m thick and often have alternating layers with

different degrees of recrystallization. The D4 mylonitic lineation trends top towards

WNW, i.e., the shear sense is sinistral, agreeing with (Sölva et al., 2001).

The mica schists in the sampling area contain euhedral garnets with sizes up to

10 mm. Away from the shear zones, the garnets are brownish-red, but close to the

shear zones the mica schists are often intercalated as 3 to 10 cm thick layers with

garnets often being greenish. The mylonitization is restricted to the weak calcite

marble units.

Microstructures

Grains in the calcite marble host rocks are coarse (up to 2 mm) with lobate grain

boundaries, suggesting dynamic recrystallization at high temperatures (Fig. 2.3a). The

host rock is predominantly calcite, but also contains small amounts of randomly

distributed second phases with a volume fraction of less than ~ 2 %. Quartz grains are

commonly rounded with sizes up to 100 µm, while muscovite occurs as flakes up to

200 µm in size.

Inside the shear zones, the protomylonites show the typical core and mantle structure.

Judging from optical extinctions, the subgrains in the core structures have the same

size as the fine mantle grains, suggesting that subgrain rotation recrystallization was

the dominant recrystallization process.

42

Figure 2.3: Optical micrographs (transmitted light, crossed nicols) of marble samples

from the Schneeberg area (a-c) and the Naxos high-grade core (d-f). a) coarse-grained, dynamically recrystallized marble host rock in the Schneeberg complex, not affected by late-stage D4 shear zones. b) typical D4-protomylonite of the Schneeberg Complex; subsequent recrystallization by subgrain rotation results in the core and mantle structure; recrystallization commonly starts at twin boundaries (tb);in addition, the coarse, old grains are often cut by intragranular microcracks (im). c) typical mylonitic microstructure due to complete recrystallization during strain localization. d) dynamically recrystallized, coarse-grained, marble host rock inside the high-grade core of Naxos, with subgrain rotation recrystallization as dominant recrystallization process and thin twins being slightly curved. e) protomylonite with the typical core and mantle structure, presumably resulting from subgrain rotation recrystallization. f) mylonitic microstructure due to complete recrystallization during strain localization.

43

The coarse grains are characterized by undulatory extinction and kinking, both

indications of strong plastic deformation. The presence of thick curved twins (type III

after Burkhard (1993)) inside the coarse grains, and the fact that twin boundaries

appear to have migrated suggests that mylonitization temperatures were higher than

250 °C (Burkhard, 1993; Ferrill et al., 2004). The coarse grains are often cut by linear

arrays of fine grains. These features are likely intragranular microcracks that have, in

turn, been recrystallized. In addition, recrystallized grains are often found along twin

boundaries, forming intragranular shear zones (Fig. 2.3b). Within the mylonite zones,

the marbles are completely recrystallized (Fig. 2.3c) with final grain sizes of 5 to 20

µm. The grain size data is derived by measuring the equidimensional circular

diameter on polished and etched surfaces (see Herwegh (2000)). Applying the mean

square root grain size of 10 µm to Rutter’s (1995) sub-grain rotation piezometer

differential stresses of 107 MPa are calculated. In some outcrops the mylonites

contain calcite veins (Fig. 2.4a) which are deformed and recrystallized again.

The garnet-mica schists of the Laaser Series consist of quartz, mica, feldspar and

almandine-rich garnet. Away from the D4 zones, backscattered electron micrographs

of the garnets do not show any alteration (Fig. 2.5a). However, close to the marble

shear zones the garnets are highly altered to chlorite pods that preserve the garnet’s

original shape (Fig. 2.5b). Other retrograde reactions are common, including

sericitizion of plagioclase pointing to the activity of fluids under lower greenschist

facies.

Hot cathodoluminescence (CL) was used to obtain qualitative information on the

chemical distribution inside the Schneeberg marble mylonites. Whereas the fine-

grained matrix is characterized by a dull (brown-orange) luminescence, the veins can

be distinguished by a brighter color (yellow-orange) (Fig. 2.4b). The bright color is

due to substitution of the Ca2+ site by Mn2+ (Machel & Burton, 1991; Lewis et al.,

1998; Barbin, 2000) and suggests the presence of externally derived fluids with a

different chemical composition (open system). Crosscutting relationships of the veins

and different degrees of diffusion of the luminescence intensity point to several

generations of fracturing and crystallization (fracture-sealing) during mylonitization.

To study the grain boundary morphology of the recrystallized calcite and to minimize

the influence of possible late-stage fluid infiltration on the grain boundary

44

morphology, the samples were taken in a reasonable distance from the surface (tens of

cm). The sections with a thickness of ~2mm were broken after several cycles of

heating (~220 °C) and cooling (~25 °C). The temperature cycling promoted

intergranular fractures. The samples were sputtered with Au-Pd and observed in SEM.

Most grain boundaries of the Schneeberg mylonites are characterized by isolated,

triangular pores. They differ in size, but similar in shape and orientation, indicating

that they are crystallography controlled (Fig. 2.6a-c). There is also evidence of a

connected network of triple-junction tubes, with the dihedral angles being controlled

by crystallographic orientation of the respective grains (Fig. 2.6c).

Figure 2.4: Syndeformational calcite veins inside the D4 marble mylonites. a) optical

micrograph (transmitted light, crossed nicols). b) cathodoluminescence micrograph showing different generations of calcite veins; the color of the veins is brighter than that of the fine-grained marble mylonite matrix, probably owing to Mn2+substitution and suggesting that the fluids were externally derived.

45

Figure 2.5: Backscattered electron images of garnets in mica schists from the

Schneeberg area (a & b) and from the Naxos high-grade core (c). a) garnet in a mica schist located in distance of the D4 shear zones in the Schneeberg area showing no evidence of alteration to chlorite or any other retrograde reaction of garnet or mica. b) garnet in a mica schist located just next to a D4 marble shear zone in the Schneeberg complex; the garnet’s original shape is still visible, but it is highly altered to chlorite by retrograde reactions due to the involvement of fluids. c) garnet inside mica schist just next to a marble mylonite inside the high-grade core of Naxos; the garnet is not affected by retrograde reactions and its appearance is similar to the garnets sampled in distance of the late-stage shear zones.

46

Figure 2.6: SEM micrographs of broken surfaces of marble mylonites from the

Schneeberg Complex (a-c) and from the high-grade core in Naxos (d-f) showing the grain boundary morphology. a) & b) grain boundaries contain triangular pores controlled by crystallography. c) the pores differ in volume, but their orientation and shape is similar; note also the tubes along triple grain junctions and the dihedral angle. d) In the Naxos rocks, grain boundaries are smoother and contain only a few small pores; note the 3D topography of the heavily thick twinned grain in the lower left corner. e) rarely, triple junctions contain tubular pores. f) in a few cases, isolated, crystallography controlled pores are observed on a grain boundary; suggesting that a small amount of fluids was present in the Naxos mylonites.

47

Chemistry

The chemical composition of the mylonites and their respective host rocks was

analyzed by XRF and ICP-OES to investigate the influence of fluids during shear

zone evolution (Tables 2.2 & 2.3). Additionally EDX analysis on polished and etched

surfaces and roentgen diffractometry on insoluble residue of the dissolved marble

samples was used to gain additional information on the chemistry of those mineral

phases.

Table 2.2: XRF data for some elements of selected samples (values in wt %). Schneeberg Complex Naxos high grade core

rel. accu-

racy [%] N-1b1 N-2h E-1-1 E-1-2 E-4 M-8-1 M-8-3 M-8-4 8a 11a-1 11a-3

Fe2O3 (T) 0.39 0.19 0.12 0.12 0.07 0.14 0.22 0.16 0.12 0.02 0.02 0.03

TiO2 0.03 0.01 0.01 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01

Al2O3 0.24 0.01 0.11 0.40 0.12 0.09 0.17 0.12 0.15 0.33 0.06 0.04

MnO 0.02 0.01 0.02 0.01 0.01 0.02 0.17 0.08 0.03 0.01 0.01 0.01

MgO 0.16 0.26 0.45 0.33 0.28 0.28 0.14 0.32 0.56 0.36 0.48 0.44

Na2O 0.08 0.14 0.18 0.18 0.25 0.17 0.18 0.13 0.32 0.29 0.19 0.14

Table 2.3: ICP-OES data for some elements of selected samples (values in ppm). Schneeberg Complex Naxos high grade core

E 8a 11a

h.r. a myl. b myl. h.r. myl. h.r. myl.

E-1-4 E-1-1 E-4 8a1c 8a1a 11a1c 11a1a

Mn 24.0 46.2 125.2 11.5 8.6 6.6 20.4

Fe 320.0 493.0 665.3 22.2 130.5 34.6 40.2

Ni 0.2 0.3 0.4 0.3 0.1 0.7 0.2

Cu 0.1 0.2 0.1 0.2 0.1 0.2 0.3

Zn 0.6 1.5 0.8 0.3 0.4 0.7 1.1

Cr 0.0 0.2 0.0 0.4 0.4 0.2 0.1

Ti 1.0 2.1 1.1 1.6 1.3 1.1 1.0

Mg 1343.5 1087.5 1190.3 1421.8 1209.1 2063.9 1558.4

Sr 96.4 96.1 384.7 66.5 52.1 80.8 69.7

Al 13.5 24.9 26.8 1.6 1.5 15.6 2.2

Si 14.9 20.9 21.3 3.3 2.8 4.2 4.2

Na 28.6 19.9 20.2 10.0 10.4 11.2 17.2

K 7.2 6.2 7.9 2.3 2.0 3.3 5.1

P 9.9 26.9 29.8 24.6 26.9 41.9 24.3

a host rock b mylonite

48

The marble host rock of the Laaser Series is very pure (Tables 2.2 & 2.3). RDA and

EDX analyses indicate muscovite and quartz being second phase minerals. Inside the

shear zones the mylonites are enriched in some elements, especially Mn and Al (Fig.

2.7a). Some of the elements can be attributed to the second phase minerals muscovite,

biotite and chlorite. However the ICP analysis shows that the calcite composition of

the mylonites is commonly enriched in several other elements, including Na, Mg, and

Ti (Fig. 2.7b). Inside the mylonites elements as Mn and Mg are incorporated into the

calcite lattice and are presumably derived from fluids that also promoted the alteration

towards chlorite.

Figure 2.7: Plots of chemical analyses of the selected mylonites normalized to their

respective host rocks derived by a) XRF and b) ICP-OES (circles: Schneeberg Complex; squares: Naxos high-grade core; thick line at one: host rock composition). The Schneeberg mylonites show a larger variation in some elements compared to their respective host rocks, whereas the Naxos mylonites do not show these significant compositional changes.

49

In addition to the chemical composition, the presence of fluids during shear zone

evolution can be detected by measurements of stable isotopes. We selected rock chip

samples of homogeneous marble with different degrees of recrystallization, traversing

from the host rock to center of the mylonite zone. The stable isotopes C13 and O18

were analyzed at the Mineralogical Department of the University Bonn, Germany (R.

Hoffbauer) (Table 2.4) and are displayed in PDB and plotted as a function of the

degree of recrystallization (Fig. 2.8a). Whereas the respective host rock data range

between -12 and -9 ‰ for δ18O and between 0.95 and 1.8 ‰ for δ13C, there is far

greater scatter of values in the recrystallized parts, ranging from -14 to -6 ‰ for δ18O

and from 0.3 to 1.6 ‰ for δ13C. In addition, isotope data derived from a

syndeformational vein that formed inside a mylonite (E-3; with values of δ18O = -

13.69‰ and δ13C = -0.42‰) indicates that the fluids were derived from an external

source.

Figure 2.8: Diagrams of stable isotopes C13 and O18 (both: PDB standard) as cross-

sections from host rock (0% RX) towards mylonite (100% RX) for several shear zones in the Schneeberg Complex (a & b) and the Naxos high-grade core (c & d). The recrystallized parts in the Schneeberg marble shear zones show a large scatter of data for both C13 and O18, whereas, there are only minor changes throughout the profiles in the Naxos marble shear zones. The asterisk represents the vein data inside a Schneeberg marble mylonite (E-3).

50

Table 2.4: Data of the analyses of the stable isotopes C13 and O18 displayed in PDB standard. Sample E-3 represents data of a syndeformational vein.

samples % (Recr.) δ18O [°/°°] δ18O [°/°°] δ13C [°/°°]

(SMOW) (PDB) (PDB)

M-8-1a 100 18.42 -12.98 1.18

M-8-1b 50 18.46 -12.94 1.30

M-8-2a 100 18.53 -12.87 1.39

M-8-2b 50 18.64 -12.76 1.53

M-8-3a 50 18.65 -12.75 1.48

M-8-3b 100 18.66 -12.74 1.45

M-8-3c 50 18.66 -12.74 1.44

M-8-4 5 19.97 -11.47 1.51

M-8-5 0 20.19 -11.26 1.53

E-1-1 100 18.56 -12.84 1.62

E-1-2 50 21.35 -10.13 1.23

E-1-3 10 20.09 -11.36 1.36

E-3 0 (vein) 17.69 -13.69 -0.42

E-4 100 17.64 -13.73 0.34

N-2d1 100 22.51 -9.01 1.67

N-2h 100 20.65 -10.81 1.16

N-2a2a 100 25.08 -6.51 0.63

N-2a2b 50 25.03 -6.56 0.67

N-1a1 0 22.49 -9.03 1.81

N-1b1 0 22.13 -9.37 1.21

Schn

eebe

rg C

ompl

ex

N-1c1 0 19.42 -12.01 0.95

8a1a 100 23.42 -8.12 2.06

8a1b 50 23.38 -8.16 2.06

8a1c 0 23.40 -8.14 2.07

11a1a 100 25.65 -5.96 2.19

11a1b 50 25.80 -5.81 2.09

11a1c 0 26.15 -5.47 2.06 Nax

os h

igh

grad

e co

re

51

To constrain the temperature at which D4-mylonitization took place, calcite-dolomite

solvus geothermometry was applied. According to Matthews et al. (1999) calcite-

dolomite geothermometry can be applied in recrystallized calcite (due to cation

equilibration) even at temperatures below 400 °C. The analyses for Ca and Mg were

made by electron microprobe using (Table 2.5). The temperatures, calculated

according to the equation of Lieberman & Rice (1986) indicate mylonitization

temperatures of 279 ± 25 °C, in agreement with the presence of type III calcite twins,

the existence of twin boundary migration and with the studies of Sölva et al. (in press)

who proposed lower greenschist facies conditions for the D4-Laaser Series-shear zone.

Table 2.5: Microprobe data of several marble mylonites used to calculate the temperature during mylonitization from calcite-dolomite solvus geothermometry (Lieberman & Rice, 1986). The entire data set is available from the authors upon request.

wt% CaO wt% MgO XMgCO3 T [°C] a

sample n b

mean std dev mean std dev mean std dev mean std dev

N-2h 23 55.01 0.4104 0.39 0.1106 0.0083 0.00234 300 24

E-1-1 27 55.55 0.2839 0.25 0.0400 0.0052 0.00084 262 13Schneeberg

E-3 8 54.33 0.3066 0.29 0.0443 0.0062 0.00096 276 13

8a1a 20 55.61 0.3287 0.23 0.0235 0.0049 0.00049 258 8Naxos

11a2 19 54.30 0.3418 0.32 0.0251 0.0068 0.00052 285 6

a calculated according to the equation of Lieberman & Rice (1986) (T[°C] = 3685.7 / (1.6145 - ln XMgCO3) – 273)

b number of measurements

2.3.2 Naxos high-grade core

Mesostructures

Inside the high-grade core of Naxos, the marble rafts are predominantly of calcite

with rare quartz grains. The calcite marble, famous for its purity and white color and

mined since ancient times, is very coarse-grained with grain sizes up to 15 mm due to

peak M2b conditions (~ 700 °C and ~ 0.6 GPa). The marble rafts are intercalated by

pegmatite intrusions and layers of amphibolite and mica schist. Mylonite zones are

restricted to the marble units and are easily detected owing to the striking difference

in grain size and the milky, porcelain-like appearance. The shear zones have a

52

thickness of up to one meter with alternating degrees of recrystallization; the

stretching lineations trend N-S (Urai et al., 1990).

The intercalated layers of mica schist and amphibolite are up to 10 cm thick. Due to

the complex deformation history that includes N-S extension during D2, the layers are

boudinaged and folded with fold axes trending N-S.

The mica schists contain a small number of euhedral garnets. Their grain size is up to

2mm, and they are -reddish-brown, both near to and far from the shear zones.

Microstructures

The marble far from the shear zones is coarse-grained with lobate grain boundaries,

pointing to high-temperature dynamic recrystallization (at peak M2b conditions; Urai

et al., 1990) (Fig. 2.3d). As in the Schneeberg Complex, the Naxos protomylonites are

also characterized by core and mantle structures, indicating subgrain rotation

recrystallization (Fig. 2.3e). In contrast to the Schneeberg samples, the Naxos

protomylonites do not contain intragranular microcracks or shear zones.

Subsequent strain localization is accompanied by complete mylonitization to grain

sizes ranging between 20 to 50 µm (Fig. 2.3c). A final grain size of 25 µm (mean

square root) corresponds to a stress of 48 MPa, using the equation for rotation

recrystallization of Rutter (1995). We did not observe any veins inside the Naxos

mylonites.

Similar to the mica schists of the Laaser Series in the Italian Alps, the intercalated

mica schists inside the marble rafts of the Naxos high-grade core consist of quartz,

mica, feldspar and almandine, but have lower garnet content. Backscattered images of

garnets in mica schists, both far from and near to the shear zones are very similar and

do not show any chloritization (alteration) of the garnets or sericitization of

plagioclase (Fig. 2.5c). Thus, we infer that the mica schists were not affected by late-

stage retrograde metamorphic reactions.

Most grain boundaries of the Naxos mylonites are slightly curved and occasionally

contain pores (Fig. 2.6d). Minor porosity is present at triple junctions, rarely within

triple junction tubes (Fig. 2.6e). Uncommonly, isolated crystallography-controlled

pores are present on the grain boundaries of these samples (Fig. 2.6f) similar in

appearance to those observed in the Schneeberg.

53

Chemistry

As in the Schneeberg marbles, the Naxos high-grade marble is characterized by its

purity (Tables 2.2 & 2.3). RDA and EDX analyses show that quartz and muscovite

together with very rare amounts of feldspar are present as randomly distributed

second phases with contents of less than 0.5 %. The quartz grains are rounded and are

characterized by sizes of commonly 10 µm. The two mylonites are very similar in

chemical composition compared to their host rock (Fig. 2.7a & b). However,

compared to the Schneeberg shear zones, the chemical composition of the Naxos

mylonites deviates only slightly from that in the respective host rock.

We analyzed carbon and oxygen isotopes along cross-sections of two different shear

zones inside the Naxos high-grade core (Table 2.4). The host rock data is about 2.1 ‰

for δ13C and ranges between –8 and –5.5 ‰ for δ18O, the latter being in accordance

with data from Baker & Matthews (1995). The isotopic signature of the recrystallized

mylonites is very similar to the host rock (Fig. 2.8b). Compared to the Schneeberg

mylonites, the Naxos samples show such smaller changes in composition. Applying

the calcite-dolomite solvus geothermometry for the post-M2b mylonites inside the

high-grade core of Naxos indicates temperatures of 271 ± 15 °C (Table 2.5), i.e., close

to those calculated for the Schneeberg mylonites.

2.4 Discussion

The host rocks and the marble mylonites from the two study areas are very similar.

Both of the host formations consist of calcite marbles of high purity, intercalated

between mica schist, and both have an early history of polyphase, high-grade

metamorphism, accompanied by deformation that produced massive, coarse-grained

rocks. Both hosts suffered a second deformation episode that took place during uplift,

and that resulted in localized deformation in shear zones, at temperatures around 270

– 280 °C.

The shear zones in the two study areas are similar, too: strain localization resulted in

progressive recrystallization of the coarse-grained host into a fine-grained marble

mylonite. Intermediate stage protomylonites have core and mantle structure evolving

by a combination of subgrain rotation and grain boundary migration. The

recrystallized grain size of the mylonites is slightly different, indicating higher

54

differential stress in Schneeberg (~100 MPa) than Naxos (~50 MPa). If the two

marbles had the same rheology, the difference in grain size would suggest that strain

rates in Schneeberg were higher, but experiments indicate that there are large

differences in rheology of calcite marbles of only slightly varying composition (e.g.

de Bresser et al., 2002).

The main difference between the two study areas is the fugacity of water in the fluids

present during mylonite formation. The role of CO2 within the pore fluid must not be

neglected, as calcite needs CO2 to be stable. Clearly, detailed knowledge of the fluid

chemistry would contribute to a better understanding of the mass transfer processes

that were active during shear zone evolution. From our present data we propose that

the difference of the presence of fluids, and the fact that most other parameters are

similar, forms the basis for evaluating the role of fluids in calcite shear zones at

temperatures around 270 – 280 °C. A schematic model summarizing the discussion

below is shown in Figure 2.9.

Figure 2.9: Sketches of the microstructures of the marble mylonites recrystallized

under a) wet and b) less wet conditions. The pre-mylonitic structure, sketched in the lower two diagrams, is assumed to be similar in both regimes, except for the amount of fluids present as pores on the grain boundaries. During shear zone evolution, veins, intragranular cracks and dilatant grain boundaries, result from the activity of fluids inside the wet system; features that are absent, or at least much less numerous in the drier (less-wet) system. However, in both systems, any porosity that is present is probably dragged by slow grain boundary migration during mylonitization (indicated by small arrows), resulting in similar grain boundary morphology (as observed in SEM).

55

We infer that the Schneeberg D4 mylonites were formed in the presence of an

abundance of aqueous fluids based on the following: The calcite composition of the

mylonites was altered during shear zone evolution; mica flakes within these rocks

were altered to chlorite; stable isotopes within the mylonites were depleted with

respect to their host rocks; and finally chemical changes were suggested by

cathodoluminescence observations. In addition, syndeformational calcite veins and

the retrograde garnets in adjacent mica schists provide additional evidence for fluids.

A late syndeformational vein in the Schneeberg mylonite has the isotopic signature of

a metamorphic fluid of an external source.

These changes in mineralization, solid-solution impurities, and isotope concentrations

require advective transport by a fluid, combined with local redistribution on the grain

scale. The details of this process are apparently complex: In most mylonite samples

calcite is enriched in e.g. manganese and titanium as shown by ICP data. The

incorporation of Mn into the calcite lattice can occur during recrystallization

(Olgaard, 1985; McCaig et al., 1999). Similar Mn enrichments in marble mylonites,

documented by cathodoluminescence, have led other authors to argue for the presence

of externally derived fluids (Busch & Van der Pluijm, 1995; Badertscher & Burkhard,

2000). Other studies (e.g. Bestmann et al. (2000)) document the formation of mylonite

without pervasive fluid flow.

In contrast to the Schneeberg mylonites, the Naxos mylonites do not show significant

differences with respect to the host rocks neither in bulk chemical composition, nor in

stable isotopes. The Naxos marbles do not contain calcite veins, and the garnets in the

adjacent mica schists were not affected by retrograde reactions. The lack of

differences between host and mylonite suggests that the Naxos mylonites were formed

under much less “wet” conditions than the Schneeberg mylonites.

Because of leakage during uplift (associated with grain boundary cracking owing to

the high thermal expansion anisotropy of calcite) we do not know what the speciation

of the fluid in the grain boundary pores was during recrystallization. Based on the

chemical changes and pore morphology we interpret these to have been fluid-filled.

Such triangular pores have been observed during crack-healing in single-crystal

calcite when water is present (Hickman and Evans, 1987) and on grain boundaries in

Cararra marble annealed with water at 800 °C (see Figure 5e in de Bresser et al.,

56

2005). Such experiments are, of course, done under conditions at which late-stage

fluid infiltration (e.g. rain water) can be excluded.

We now ask the question how the presence of fluids in the Schneeberg D4 shear zones

has influenced microstructural evolution. Multiple generations of dynamically

recrystallized calcite veins indicate that the veins were formed and deformed during

mylonitization, probably by periodic influx of high-pressure fluid with subsequent

precipitation of calcite, as the rocks were exhumed along the Schneeberg Normal

Fault Zone (SNFZ). Therefore, the changes in calcite composition may have been

caused by two processes: 1.) new calcite crystallized from the fluid and then

subsequently recrystallized (see also Kennedy & Logan, 1997; Bardetscher et al.,

2002; Herwegh & Kunze, 2002), or 2.) impurities were incorporated during dynamic

grain boundary migration recrystallization. In fact, recrystallized intragranular

fractures inside the coarse grains point to the combined activity of these processes.

Grain boundaries containing many isolated pores and connected tubes along triple

grain junctions are also apparently associated with recrystallization processes. Under

conditions of low effective pressure, grain boundaries may have dilated and formed

new porosity, some of which may have subsequently been removed by compaction

when the fluid pressure dropped.

In the Naxos mylonites, most of the processes described above are absent. Triple

junction porosity exists, but such porosity is most commonly isolated, that is,

unconnected. The permeability of these mylonites is probably very low. Rare grain

boundaries with many pores may have been associated with local fluid enrichment.

Such pores probably hindered grain boundary migration by Zener drag, particularly if

they were filled with a non-aqueous fluid (Fig. 2.9) (Stüwe, 1978; Olgaard, 1985; Urai

et al., 1986a; Evans et al., 2001; Herwegh & Kunze, 2002; Herwegh & Berger, 2004).

Returning to the Schneeberg mylonites, it is then interesting to note that the presence

of fluids did not have a major effect on grain boundary morphology and

recrystallization behavior. Although the amount of pores on grain boundaries is

different, other morphological characteristics are similar. The recrystallization

mechanisms are similar, too: subgrain rotation recrystallization followed by slow

grain boundary migration. Thus we infer that drag of the grain boundary fluid in pores

occurred in the Schneeberg mylonites as well.

57

2.5 Conclusions

To assess the effect of water-rich fluids on the recrystallization behavior and grain

boundary morphology in natural calcite, we compared marble mylonites from the

Schneeberg Complex in the Italian/Austrian Alps to mylonites from the high-grade

core of Naxos, Greece. Both settings have similar geologic histories, but they differed

in the nature of the fluids present during mylonitization. Both the fluid-rich

(Schneeberg) and fluid-poor (Naxos) marble mylonites have similar grain boundary

microstructures. The microstructural evolution inside the Schneeberg mylonites was

affected by the presence of fluids as shown by (1) the introduction of

syndeformational calcite veins with slightly different chemical composition, (2) the

presence of intragranular cracks inside the protomylonites, (3) evidence for dilatant

grain boundaries and (4) the existence of triple junction tube porosity inside the

mylonites. None of these features are present in the Naxos samples. However, this

presence of fluids apparently did not affect the recrystallization behavior, at least as

can be judged by the microstructure: The dominant recrystallization process of both

types of mylonites was probably subgrain rotation recrystallization followed by slow

grain boundary migration. In addition the grain boundary morphology is roughly

similar, even if the amount of porosity residing on the grain boundaries is different.

Apparently, the fluids did not have a major influence on recrystallization behavior and

grain boundary morphology in calcite, at least for these two marble mylonites.

Such a conclusion is in agreement with deformation experiments of Carrara marble at

high temperature and high strain rates (de Bresser et al., 2005), but is in strong

contrast to inferences concerning the effect of water on recrystallization and grain

boundary migration and morphology in other minerals such as quartz, halite, olivine

or feldspar.

Acknowledgements

The authors would like to thank M. Burkhard and M. Herwegh for the critical and

constructive reviews and comments. We are grateful to R. Hoffbauer for carrying out

the stable isotope measurements. A. Heimann is thanked for his assistance with

sample preparation and analytical work. M. Frongillo, J. Kallinna and U. Wollenberg

are thanked for their help with the SEM. We also thank N. Chatterjee for his essential

assistance with the microprobe and S. Sindern and R. Neef for carrying out the ICP-

58

OES analyses. We acknowledge W. Kraus for preparing thin sections, H. Sölva for

providing an early version of his paper and H. de Bresser for his valuable comments

on the recrystallization of marble. JLU thanks M.I. Spalla for discussions on

Schneeberg geology. OS acknowledges the Prof. Dr. K. Heitfeld Stiftung for the

financial support during his stay at MIT. This project is funded by the Deutsche

Forschungsgemeinschaft (UR 64/4-1). BE gratefully acknowledges funding from NSF

EAR 0309510.

2.6 References

Adams, F. D. & Nicolson, J. T. 1900. An experimental investigation into the flow of marble. Philosophical Transactions of the Royal Society of London, Series A 195, 363-401.

Badertscher, N. P. & Burkhard, M. 2000. Brittle-ductile deformation in the Glarus thrust Lochseiten (LK) calc-mylonite. Terra Nova 12, 281-288.

Badertscher, N. P., Beaudoin, G., Therrien, R. & Burkhard, M. 2002. Glarus overthrust; a major pathway for the escape of fluids out of the Alpine Orogen. Geology 30(10), 875-878.

Baker, J. & Matthews, A. 1995. The stable isotopic evolution of a metamorphic complex, Naxos, Greece. Contributions to Mineralogy and Petrology 120, 391-403.

Barbin, V. 2000. Cathodoluminescence of carbonate shells; biochemical vs diagenetic process. In: Cathodoluminescence in Geosciences (edited by Pagel, M., Barbin, V., Blanc, P. & Ohnenstetter, D.). Springer, Berlin, 303-329.

Behrmann, J. H. 1983. Microstructure and fabric transitions in calcite tectonites from the Sierra Alhamilla (Spain). Geologische Rundschau 72(2), 605-618.

Bestmann, M., Kunze, K. & Matthews, A. 2000. Evolution of a calcite marble shear zone complex on Thassos Island, Greece; microstructural and textural fabrics and their kinematic significance. Journal of Structural Geology 22(11-12), 1789-1807.

Buick, I. S. & Holland, T. J. B. 1989. The P-T-t Path associated with crustal extension, Naxos. In: Evolution of Metamorphic Belts (edited by Daly, J. S., Cliff, R. A. & Yardley, B. W. D.) 43. Geological Society Special Publication, London, 365-369.

Burkhard, M. & Kerrich, R. 1988. Fluid regimes in the deformation of the Helvetic nappes, Switzerland, as inferred from stable isotope data. Contributions to Mineralogy and Petrology 99, 416-429.

Burkhard, M. 1993. Calcite twins, their geometry, appearance and significance as stress-strain markers and indicators of tectonic regime; a review. Journal of Structural Geology 15(3-5), 351-368.

Busch, J. P. & Van der Pluijm, B. A. 1995. Calcite textures, microstructures and rheological properties of marble mylonites in the Bancroft shear zone, Ontario, Canada. Journal of Structural Geology 17(5), 677-688.

Chen, S. & Kohlstedt, D. L. 2003. Water Weakening of Clinopyroxene in the Dislocation Creep Regime. In:. EOS Trans. AGU 84 (46), Fall Meet. Suppl., Abstract S22E-03. San Francisco, USA

59

Covey-Crump, S. J. 1998. Evolution of mechanical state in Carrara marble during deformation at 400° to 700°C. Journal of Geophysical Research B: Solid Earth 103(12), 29,781-29,794.

de Bresser, J. H. P. 2002. On the mechanism of dislocation creep in calcite at high temperature: inferences from experimentally measured pressure sensitivity and strain rate sensitivity of flow stress. Journal of Geophysical Research 107(B12), ECV 4-1 - 4-16.

de Bresser, J. H. P., Evans, B. & Renner, J. 2002. On estimating the strength of calcite rocks under natural conditions. In: Geological Society Special Publications (edited by de Meer, S., Drury, M. R., de Bresser, J. H. P. & Pennock, G. M.) 200. Geological Society of London : London, United Kingdom, United Kingdom, 309-329.

de Bresser, J. H. P., Urai, J. L. & Olgaard, D. L. 2005. Effect of water on the strength and microstructure of Carrara marble axially deformed at high temperatures. Journal of Structural Geology 27(2), 265-281.

Dimanov, A., Dresen, G., Xiao, X. & Wirth, R. 1999. Grain boundary diffusion creep of synthetic anorthite aggregates: The effect of water. Journal of Geophysical Research 104(B5), 10,483-10,497.

Evans, B., Renner, J. & Hirth, G. 2001. A few remarks on the kinetics of static grain growth in rocks. International Journal of Earth Science (Geologische Rundschau) 90, 88-103.

Feenstra, A. 1985. Metamorphism of Bauxites on Naxos, Greece. PhD thesis, Rijksuniversiteit Utrecht, 206p.

Ferrill, D. A., Morris, A. P., Evans, M. A., Burkhard, M., Groshong, J., Richard H. & Onasch, C. M. 2004. Calcite twin morphology: a low-temperature deformation geothermometer. Journal of Structural Geology 26(8), 1521-1529.

Griggs, D. 1974. A model of hydrolytic weakening in quartz. Journal of Geophysical Research 79(11), 1653-1661.

Hansen, M. & Heide, A. 1999. Deformation of metabauxites and surrounding marbles on Naxos, Greece. Unpublished Master thesis, RWTH Aachen, 121p.

Heitzmann, P. 1987. Calcite mylonites in the Central Alpine "root zone". Tectonophysics 135(1-3), 207-215.

Herwegh, M. 2000. A new technique to automatically quantify microstructures of fine-grained carbonate mylonites; two-step etching combined with SEM imaging and image analysis. Journal of Structural Geology 22(4), 391-400.

Herwegh, M. & Kunze, K. 2002. The influence of nano-scale second-phase particles on deformation of fine-grained calcite mylonites. Journal of Structural Geology 24(9), 1463-1478.

Herwegh, M., Xiao, X. & Evans, B. 2003. The effect of dissolved magnesium on diffusion creep in calcite. Earth and Planetary Science Letters 212(3-4), 457-470.

Herwegh, M. & Berger, A. 2004. Deformation mechanisms in second-phase affected microstructures and their energy balance. Journal of Structural Geology 26(8), 1483-1498.

Hickman, S. H. & Evans, B. 1987. Influence of geometry upon crack healing rate in calcite. Physics and Chemistry of Minerals 15(1), 91-102.

Hirth, G. & Tullis, J. 1992. Dislocation creep regimes in quartz aggregates. Journal of Structural Geology 14(2), 145-159.

Hoinkes, G., Frank, W., Mauracher, J., Peschel, R., Purtscheller, F. & Tessadri, R. 1987. Petrography of the Schneeberg Complex. In: Geodynamics of the Eastern Alps (edited by Fluegel, H. W. & Faupl, P.), 190-199.

60

Jansen, J. B. H. & Schuiling, R. D. 1976. Metamorphism on Naxos: Petrology and Geothermal Gradients. American Journal of Science 276, 1225-1253.

Jaoul, O., Tullis, J. & Kronenberg, A. 1984. The effect of varying water contents on the creep behavior of heavitree quartzite. Journal of Geophysical Research 89(B6), 4298-4312.

Kennedy, L. A. & Logan, J. M. 1997. The role of veining and dissolution in the evolution of fine-grained mylonites: the McConnell thrust, Alberta. Journal of Structural Geology 19(6), 785-797.

Kirschner, D. L. & Kennedy, L. A. 2001. Limited syntectonic fluid flow in carbonate-hosted thrust faults of the Front Ranges, Canadian Rockies, inferred from stable isotope data and structures. Journal of Geophysical Research 106(B5), 8827-8840.

Kohlstedt, D. L., Majumder, S. & Mei, S. 2003. Water Weakening of Clinopyroxene in the Diffusion Creep Regime. In: EOS Trans. AGU 84 (46), Fall Meet. Suppl., Abstract S22A-0428. San Francisco, USA

Kronenberg, A. K. & Tullis, J. 1984. Flow strengths of quartz aggregates: grain size and pressure effects due to hydrolytic weakening. Journal of Geophysical Research 89(B6), 4281-4297.

Lewis, S., Holness, M. & Graham, C. M. 1998. Ion microprobe study of marble from Naxos, Greece; grain-scale fluid pathways and stable isotope equilibration during metamorphism. Geology Boulder 26(10), 935-938.

Lieberman, J. E. & Rice, M. 1986. Petrology of marble and peridotite in the Seiad ultramafic complex, northern California, USA. Journal of Metamorphic Geology 4, 179-199.

Lister, G. S., Banga, G. & Feenstra, A. 1984. Metamorphic core complexes of Cordilleran type in the Cyclades, Aegean Sea, Greece. Geology 12, 221-225.

Machel, H. G. & Burton, E. A. 1991. Factors governing cathodoluminescence in calcite and dolomite and their implications for studies of carbonate diagenesis. In: Luminescence Microscopy: Quantitative and Qualitative Aspects (SEPM-1991) (edited by Barker, C. E. & Kopp, O. C.), 37-57.

Mancktelow, N. S., Grujic, D. & Johnson, E. L. 1998. An SEM study of porosity and grain boundary microstructure in quartz mylonites, Simplon fault zone, Central Alps. Contributions to Mineralogy and Petrology 131(1), 71-85.

Mancktelow, N. S. & Pennacchioni, G. 2004. The influence of grain boundary fluids on the microstructure of quartz-feldspar mylonites. Journal of Structural Geology 26(1), 47-69.

Matthews, A., Lieberman, J., Avigad, D. & Garfunkel, Z. 1999. Fluid-rock interaction and thermal evolution during thrusting of an Alpine metamorphic complex (Tinos island, Greece). Contributions to Mineralogy and Petrology 135, 212-224.

McCaig, A. M., Rutter, E. H. & Lloyd, G. E. 1999. Experimental studies of the incorporation of Mn and Sr into calcite during grain boundary migration. In: AGU Fall Meeting 80. Eos Trans. AGU, San Francisco, 1027.

Mei, S. & Kohlstedt, D. L. 2000a. Influence of water on plastic deformation of olivine aggregates 1. Diffusion creep regime. Journal of Geophysical Research, B, Solid Earth and Planets 105(9), 21,457-21,469.

Mei, S. & Kohlstedt, D. L. 2000b. Influence of water on plastic deformation of olivine aggregates 2. Dislocation creep regime. Journal of Geophysical Research, B, Solid Earth and Planets 105(9), 21,471-21,481.

61

Olgaard, D. L. 1985. Grain Growth and Mechanical Processes in two-phased Synthetic Marbles and Natural Fault Gouge. Unpublished PhD thesis, Massachusetts Institut of Technology, 204p.

Olgaard, D. L. & Evans, B. 1988. Grain growth in synthetic marbles with added mica and water. Contributions to Mineralogy and Petrology 100(2), 246-260.

Olgaard, D. L. & FitzGerald, J. D. 1993. Evolution of pore microstructures during healing of grain boundaries in synthetic calcite rocks. Contributions to Mineralogy and Petrology 115(2), 138-154.

Peach, C. J., Spiers, C. J. & Trimby, P. W. 2001. Effect of confining pressure on dilatation, recrystallization, and flow of rock salt at 150°C. Journal of Geophysical Research 106(B7), 13,315-13,328.

Pe-Piper, G., Kotopouli, C. N. & Piper, D. J. W. 1997. Granitoid rocks of Naxos, Greece: Regional geology and petrology. Geological Journal 32, 153-177.

Pfiffner, O. A. 1982. Deformation mechanisms and flow regimes in limestones from the Helvetic Zone of the Swiss Alps. Journal of Structural Geology 4(4), 429-442.

Pieri, M., Burlini, L., Kunze, K., Stretton, I. & Olgaard, D. L. 2001. Rheological and microstructural evolution of Carrara marble with high shear strain: results from high temperature torsion experiments. Journal of Structural Geology 23(9), 1393-1413.

Post, A. & Tullis, J. 1998. The rate of water penetration in experimentally deformed quartzite: implications for hydrolytic weakening. Tectonophysics 295(1-2), 117-137.

Renner, J. & Evans, B. 2002. Do calcite rocks obey the power-law creep equation? In: Geological Society Special Publications (edited by de Meer, S., Drury, M. R., de Bresser, J. H. P. & Pennock, G. M.) 200. Geological Society of London, United Kingdom, 293-307.

Rowe, K. J. & Rutter, E. H. 1990. Palaeostress estimation using calcite twinning; experimental calibration and application to nature. Journal of Structural Geology 12(1), 1-17.

Rutter, E. H. 1974. The influence of temperature, strain rate and interstitial water in the experimental deformation of calcite rocks. Tectonophysics 22(3-4), 311-334.

Rutter, E. H. 1995. Experimental study of the influence of stress, temperature, and strain on the dynamic recrystallization of Carrara Marble. Journal of Geophysical Research, B, Solid Earth and Planets 100(12), 24,651-24,663.

Schenk, O. & Urai, J. L. 2004. Microstructural evolution and grain boundary structure during static recrystallization in synthetic polycrystals of Sodium Chloride containing saturated brine. Contributions to Mineralogy and Petrology 146(6), 671-682.

Schmid, S. M., Boland, J. N. & Paterson, M. S. 1977. Superplastic flow in finegrained limestone. Tectonophysics 43, 257-291.

Schmid, S. M., Paterson, M. S. & Boland, J. N. 1980. High temperature flow and dynamic recrystallization in Carrara Marble. Tectonophysics 65(3-4), 245-280.

Schmid, S. M. & Haas, R. 1989. Transition from near-surface thrusting to intrabasement decollement, Schlinig Thrust, Eastern Alps. Tectonics 8(4), 697-718.

Skrotzki, W. & Welch, P. 1983. Development of texture and microstructure in extruded ionic polycrystalline aggregates. Tectonophysics 99, 47-61.

Sölva, H., Thöni, M., Grasemann, B. & Linner, M. 2001. Emplacement of eo-Alpine high-pressure rocks in the Austroalpine Oetztal complex (Texel group, Italy/Austria). Geodinamica Acta 14, 345-360.

62

Sölva, H., Grasemann, B., Thöni, M., Thiede, R. & Habler, G. in press. The Schneeberg Normal Fault Zone: normal faulting associated with Cretaceous SE-directed extrusion in the Eastern Alps (Italy/Austria). Tectonophysics.

Spalla, M. I. 1990. Polyphased deformation during uplifting of metamorphic rocks; the example of the deformational history of the Texel Gruppe (Central-Western Austroalpine domain of the Italian Eastern Alps). Memorie della Societa Geologica Italiana 45, 125-134.

Spiers, C. J., Schutjens, P. M. T. M., Brzesowsky, R. H., Peach, C. J., Liezenberg, J. L. & Zwart, H. J. 1990. Experimental determination of constitutive parameters governing creep of rocksalt by pressure solution. In: Deformation mechanisms, rheology and tectonics (edited by Knipe, R. J. & Rutter, E. H.). Geological Society Special Publications 54. Geological Society of London, United Kingdom, 215-227.

Spiers, C. J. & Brzesowsky, R. H. 1993. Densification behaviour of wet granular salt: Theory versus experiment. In: Seventh Symposium on Salt. Elsevier Science Publishers B.V., Amsterdam, 83-92.

Stüwe, H. P. 1978. Driving and dragging forces in recrystallization. In: Recrystallization of metallic materials (edited by Haessner, F.). Dr Rieder Verlag, Stuttgart, 11-21.

Tullis, J. & Yund, R. A. 1982. Grain growth kinetics of quartz and calcite aggregates. Journal of Geology 90, 301-318.

Ulrich, S., Schulmann, K. & Casey, M. 2002. Microstructural evolution and rheological behaviour of marbles deformed at different crustal levels. Journal of Structural Geology 24(5), 979-995.

Urai, J. L. 1983. Water assisted dynamic recrystallization and weakening in polycrystalline bischofite. Tectonophysics 96(1-2), 125-157.

Urai, J. L. 1985. Water-enhanced dynamic recrystallization and solution transfer in experimentally deformed carnallite. Tectonophysics 120(3-4), 285-317.

Urai, J. L., Means, W. D. & Lister, G. S. 1986a. Dynamic recrystallization of minerals. In: Mineral and rock deformation; laboratory studies; the Paterson volume. AGU Geophysical Monograph (edited by Hobbs, B. E. & Heard, H. C.) 36, 161-199.

Urai, J. L., Spiers, C. J., Zwart, H. J. & Lister, G. S. 1986b. Weakening of rock salt by water during long-term creep. Nature (London) 324(6097), 554-557.

Urai, J. L., Schuiling, R. D. & Jansen, J. B. H. 1990. Alpine deformation on Naxos (Greece). In: Deformation mechanisms, rheology and tectonics (edited by Knipe, R. J. & Rutter, E. H.). Geological Society Special Publications 54. Geological Society of London, United Kingdom, 509-522.

Urai, J. L. & Feenstra, A. 2001. Weakening associated with the diaspore-corundum dehydration reaction in metabauxites; an example from Naxos (Greece). Journal of Structural Geology 23(6-7), 941-950.

Walker, A. N., Rutter, E. H. & Brodie, K. H. 1990. Experimental study of grain-size sensitive flow of synthetic, hot-pressed calcite rocks. In: Deformation mechanisms, rheology and tectonics (edited by Knipe, R. J. & Rutter, E. H.). Geological Society Special Publications 54. Geological Society of London, United Kingdom, 259-284.

Watanabe, T. & Peach, C. J. 2002. Electrical impedance measurement of plastically deforming halite rocks at 125°C and 50 MPa. Journal of Geophysical Research 107(B1), ECV 2-1 - 2-12.

63

Chapter 3:

Microstructural evolution and grain boundary structure during static recrystallization in synthetic polycrystals of Sodium Chloride containing saturated brine2

3.0 Abstract

The effects of brine on recrystallization in halite are well known. However, properties

of brine such as morphology, connectivity, diffusivity and the resulting influences on

deformation mechanisms are still a matter of debate. This paper presents a

microstructural study of dense, statically recrystallizing synthetic polycrystalline

halite containing small amounts of brine. We used powders of two different grain size

classes: <10 µm and 200-355 µm. The aggregates were compacted to brine-filled

porosities less than about 2% and annealed at room temperature, without an external

stress field.

Coarse-grained samples undergo recrystallization manifested by the growth of large

(up to 300 µm) strain-free grains into the deformed old grains. The new grains are

frequently euhedral, with mobile grain boundaries moving at rates up to 6 nm/s. Their

mobility is interpreted to be high due to the presence of water. Grain surfaces are

smooth and the width of the water-rich zones is usually below the resolution of the

SEM (less than 50 nm).

2 Schenk, O. & Urai, J.L. 2004. Microstructural evolution and grain boundary structure during static

recrystallization in synthetic polycrystals of Sodium Chloride containing saturated brine. Contributions to Mineralogy and Petrology 146(6), 671-682 (published without Appendix).

64

The evolution of fine-grained samples starts with primary recrystallization and a

reorganization of grain boundaries. After this stage which lasts a few hours, normal

grain growth effectively stops, and no significant increase of grain size is observed

even after several months. Microstructural observations indicate contact healing at the

grain boundaries, with dihedral angles ranging between 20 and 110°. We interpret

these boundaries to be fluid-free, with the brine residing in a network of triple

junction tubes. This system of triple junctions is interconnected and associated with

significant permeability.

While grain growth is inhibited in the fine-grained samples, after a few hours of

annealing exaggerated grain growth is commonly initiated. This is manifested by the

growth of large, euhedral grains replacing the fine-grained matrix. These grains also

grow with low-index facets and their boundaries are also interpreted to be mobile due

to the existence of a water-rich phase.

3.1 Introduction

The presence of brine in halite strongly affects its mechanical and transport properties.

Observations from nature and experiments show that in deforming wet halite solution

transfer creep and dynamic recrystallization are dramatically enhanced, and that fluid

distribution is complex, both during and after deformation. Understanding the fluid

distribution is important for modeling and prediction of processes in a wide range of

geological environments, such as the seasonal movement of salt glaciers (Talbot and

Rogers, 1980; Wenkert, 1979), strain localization in shear zones in salt (Miralles et

al., 2001), and the flow of salt during diapiric movement (Jackson and Talbot, 1986).

There is also interest in applied fields e.g. to predict the behavior of underground

caverns (Fokker et al., 1996) or modeling properties of salt as seals for gas or oil

(Holness and Lewis, 1997; Lewis and Holness, 1996; Peach et al., 2001).

The reason for the complexity in brine distribution is the high solubility and

diffusivity of sodium chloride in water, even at room temperature. Therefore, in

addition to the redistribution of the fluid under stress by viscous flow (with high

viscosities in very thin fluid films (Peach et al., 2001)) during grain boundary sliding

and microcracking, this high solubility and diffusivity may lead to solution and

crystallization processes in grain boundaries, driven by local gradients in chemical

65

potential, due to the presence of external stress, dislocations, grain boundaries, solid

solution impurities or other defects.

Two end members of this grain boundary process are pressure solution (Spiers et al.,

1990) and fluid-assisted grain boundary migration (Urai et al., 1986b; Watanabe and

Peach, 2002). Although these two end members are quite different, the process of

dissolution and precipitation is common to both, and several transitional processes can

be envisaged (Urai et al., 1986a). This is further illustrated in Figure 3.1. Here we

consider two crystals A and B, with two marker points a and b, respectively, separated

by a boundary. We describe this bicrystal in a coordinate system of which the origin is

fixed at the boundary, the y-axis is along the boundary and the x-axis is at 90° to the

boundary. We then define Va and Vb as the x-components of velocity of the two

marker points a and b on both crystals. We now plot Va vs Vb. For the case of a

migrating, conservative grain boundary, the velocities will plot on the line Va = Vb,

i.e. both crystals move to the right or left at equal velocity. For pressure solution, the

velocities will plot in the top left quadrant: on the line Va = -Vb both crystals will

dissolve at equal rates. The diagonal in the bottom right quadrant represents a crack-

seal accretion process where the two crystals both move apart and grow at the same

rate (Hilgers et al., 2001; Hilgers and Urai, 2002).

The off-diagonal positions represent general grain boundaries at which dissolution or

accretion rates are unequal on both sides. For the case of grain boundary migration

this means that the motion is non-conservative (involving addition or removal of

material). For the case of pressure solution this means that material on both sides is

not removed at equal rates.

Urai et al. (1986a) proposed briefly the possibility of such general boundaries. Their

importance has become clearer based on the work of de Bresser et al. (2001), who

argued that dynamic recrystallization brings a polycrystal to the transition region

between grain size sensitive and grain size insensitive creep fields, so that grain

boundary migration, dissolution and precipitation processes occur quite naturally

together, all at significant rates. In such a material general grain boundaries should be

the rule rather than exception.

66

Figure 3.1: Schematic diagram illustrating the relationship between different

processes in general, non-conservative grain boundaries. See text for detailed description.

The large effect of water on the recrystallization process in sodium chloride is well

documented. Drury and Urai (1990), Peach et al. (2001), Skrotzki and Welch (1983),

Spiers et al. (1990), Urai et al. (1986a), Urai et al. (1986b), Watanabe and Peach

(2002), have shown that between room temperature and 150 ºC wet, polycrystalline,

natural samples deformed in the non-dilatant field recrystallize readily during and

after deformation, and have presented observations which can be interpreted as that

the boundaries contain thin brine films. In comparison, the experiments of Guillopé

and Poirier (1979) and Franssen (1993), with dry sodium chloride showed that grain

boundaries are essentially immobile at temperatures below 400 ºC. A model for the

migration of water-containing boundaries, based on the serial processes of dissolution,

diffusion through the fluid and crystallization was proposed by Urai et al. (1986a) and

Garcia Celma et al. (1988). This model was further expanded by Peach et al. (2001)

and Watanabe and Peach (2002), who also included interface reaction control of the

migration rate into their calculations.

One of the criticisms of the model of fluid films on mobile grain boundaries was

based on the fact that the samples, which recrystallized in an anisotropic stress state,

67

were examined after removal of the stress. This could have led to a redistribution of

the fluid by viscous flow, possibly to evaporation of the fluid, and thus to errors of

interpretation.

Another argument is based on consideration of thermodynamic equilibrium, where at

room temperature and a pressure of 1 bar the contact angle between sodium chloride

and brine is about 70°. This should lead to contact healing (Holness and Lewis, 1997;

Lewis and Holness, 1996; Smith, 1964). If fluid films exist at some stage, it is not

clear how and how long they can be maintained in such a system.

Pressure solution experiments on wet halite (de Meer et al., 2002; Hickman and

Evans, 1991; Martin et al., 1999; Peach, 1991; Schutjens, 1991; Spiers and Schutjens,

1990; Spiers et al., 1990) have shown that this process is dramatically enhanced by

the presence of brine, and that the grain boundaries often have an island-channel

structure at length scales of micrometers. This is quite different to the structures

observed in mobile grain boundaries, where the grain surfaces are smooth (Urai et al.,

1986b). Transitional behavior may have played a role in the experiments of Spiers and

Brzesowsky (1993), who found recrystallization in compaction experiments

dominated by pressure solution at pressures larger than 4 MPa.

Under suitable conditions, contact healing was also shown to occur in pressure

solution experiments (Hickman and Evans, 1991; Visser, 1999). The theoretical basis

of grain contact healing under stress was further presented by Visser (1999), who

considered surface energy terms in addition to stress-related driving forces and

predicted the fields for pressure solution, contact healing and neck growth.

3.2 Aims of this study

The aim of this study was to better understand the effect of small amounts of brine on

the migration of grain boundaries in halite, and to obtain constraints on the properties

of the brine in these grain boundaries.

To avoid redistribution of the fluid by viscous flow after removal of stress, samples

were annealed under atmospheric pressure, without deformation.

3.3 Methods

Wet synthetic polycrystalline halite samples were cold-pressed from powders of

analytical grade NaCl (Roth, Art. 9265.3; NaCl content > 99.9%) or table salt. We

68

compacted wet sodium chloride powder of two different grain size classes (< 10 µm

and 200-355 µm) to dense aggregates with brine-filled porosities less than 2% which

were annealed at room temperature. The microstructural evolution was studied by

scanning electron- and reflected light microscopy of surfaces which were either

polished and etched or broken.

3.3.1 Preparation of powders

Two sets of samples were prepared, differing only in the grain size of the starting

material:

i. Coarse-grained starting material: the grain size of 200-355 µm was obtained by

sieving the as-received NaCl powder.

ii. Fine-grained starting material: dry NaCl powder was ground in a swingmill.

Remaining coarse grains were extracted by Stokes separation: after grinding again

in a saturated NaCl solution in an agate mortar, the slurry was poured into a

column containing saturated NaCl solution, and decanted after the coarse grains

settled on the bottom. The main problem with this method was that the grains in

the powder tended to cluster, preventing full dispersion of the fine grained

fraction. This procedure was repeated for a second time to produce grains of

dominantly 5-10 µm containing rare larger grains of about 15 µm, suspended in

the saturated solution.

3.3.2 Compaction

After treating for 30 seconds in an ultrasonic bath, the salt brine mixture was poured

into a cylindrical die (Fig. 3.2). An O-ring on the lower piston sealed the vessel from

below, while the weight of the upper piston (without O-ring) allowed most of the

brine to slowly drain out on the top. Subsequently the pressure on the upper piston

was raised to 150 MPa in a few minutes, and maintained for 5 minutes. Then the

pressure was removed, and the sample was extracted from the die by removing the

upper piston and pressing the lower piston upwards. The unconsolidated material in

the die had a height of 6-8 mm, producing dense cylindrical samples with a diameter

of 1cm and a height of 2-4 mm. This aspect ratio was chosen to avoid density

differences in the sample due to friction along the die's wall during pressing.

69

Figure 3.2: Schematic diagram of the uniaxial compaction apparatus.

The samples produced had 97.5 ± 0.5% theoretical density, containing a small amount

of saturated brine as a pore fluid. The procedure was developed to minimize the

possibility of introduction of gas bubbles in the samples, so that the pores were brine-

filled.

For comparison, an equivalent set of samples was prepared, with cyclohexane as pore

fluid of a dried (104 ºC for 3 days) coarse-grained powder.

3.3.3 Annealing

To study the evolution of the microstructure in combination with the influence of

fluids, the brine-filled samples were annealed in small, air-tight containers containing

a small amount of saturated NaCl solution, not in contact with the samples. This

ensured that the vapor pressure of H2O around the sample was buffered at the

equilibrium value. The annealing was carried out at room temperature and at 80 ºC, in

this paper we report the results of the experiments at a temperature of 24 ± 1 ºC.

70

3.3.4 Sample preparation for microscopy

The samples, which recrystallize at room temperature, cannot be quenched. When a

sample's outer surface has been prepared for microscopy, the sample may continue

recrystallizing and modifying its structure. This possibility was carefully evaluated by

observing the same surface as a function of time, until we were convinced that the

outer surface of the samples does not change with time.

Many samples showed some efflorescence after removal from the annealing cell

pointing to a network of connected, fluid-filled pores in the sample.

3.3.5 Polished and etched surfaces

The observation of grain boundary structure and grain size distribution was done by

reflected light- and scanning electron microscopy (SEM). Polished and etched thick

sections of the annealed samples were produced by using the method described in

Urai et al. (1987). The sample’s surface was carefully ground and polished. Then the

section was immersed and agitated in slightly undersaturated NaCl solution

(~5.5 molar) for 10 seconds. The solution was removed by rinsing the section with a

powerful jet of cyclohexane. It was quickly dried with a hot air blower and stored

under dry conditions.

3.3.6 Fractured surfaces

For observation of grain surfaces by SEM the samples were broken in tension directly

after removal from the annealing cell and stored under dry conditions.

3.3.7 Grain size measurements

Grain size was measured at different stages of annealing. The size of large grains

formed by primary recrystallization and exaggerated grain growth (see below) was

measured only on grains which were rectangular in the section plane (sectioned

normal to opposite cube faces), whereas the grain size during normal grain growth

was measured using the intercept method (Underwood, 1970).

71

3.4 Results

An overview of the samples described in this paper is given in Table 3.1.

Table 3.1: Overview of samples described in this paper.

sample type of salt starting

grainsize [µm]

annealing

time [d]

storagea

OS-003c table salt >355 2.79 S

OS-003d table salt >355 4.97 S

OS-023c Roth <50 98.18 S

OS-039a Roth <50 0.04 N

OS-039b Roth <50 0.09 N

OS-041a Roth <50 5.83 N

OS-048 Roth <50 O

OS-063c Roth <10 3.99 N

OS-063d Roth <10 33.15 N

OS-065a Roth <10 0.17 N

OS-065b Roth <10 0.90 N

OS-065c Roth <10 7.05 N

OS-065d Roth <10 29.02 N

OS-070a Roth <10 0.01 N

OS-070b Roth <10 0.17 N

OS-070c Roth <10 20.20 N

OS-071a Roth <10 0.80 N

OS-071b Roth <10 4.78 N

OS-072a Roth <10 0.02 N

OS-072b Roth <10 0.05 N

OS-073a Roth <10 0.07 N

OS-073b Roth <10 0.08 N

OS-074a Roth <10 0.11 N

OS-074b Roth <10 0.13 N

OS-075a Roth <10 19.00 N

OS-075b Roth <10 27.00 N

72

OS-085a Roth 200-355 0.02 N

OS-085b Roth 200-355 0.23 N

OS-086a Roth 200-355 1.05 N

OS-086b Roth 200-355 1.34 N

OS-088a Roth 200-355 4.08 N

OS-088b Roth 200-355 6.06 N

OS-089a Roth 200-355 7.98 N

OS-090a Roth 200-355 32.08 N

OS-090b Roth 200-355 39.10 N

OS-091b Roth 200-355 48.24 N

OS-092a Roth 200-355 59.36 N

OS-093a Roth 200-355 53.05 N

OS-096a Roth 200-355 2.15 N

OS-097a Roth 200-355 2.21 N

OS-111a Roth <10 1.17 N

OS-111b Roth <10 4.13 N

OS-114a Roth <10 11.05 N

OS-117a Roth <10 0.67 N

OS-118 Roth <10 O

OS-119a Roth <10 0.01 N

OS-120a Roth <10 0.56 N

OS-121 Roth <50 O

OS-122 Roth <50 O

OS-123 table salt >355 O

OS-124 table salt >355 O

a S: stored in air with silica gel N: stored in air with humidity buffered by saturated NaCl solution

O: dried at 104 °C for 3 days

3.4.1 Water content

The water content of the samples was calculated from the weight loss due to

evaporation by drying. The brine-filled, connected porosity is 2.5 ± 0.5 % for both

fine- and coarse-grained samples.

73

In addition the water content was measured for one annealed, fine-grained sample by

infrared spectroscopy (A. Kronenberg, pers. communication) which indicated

~660ppm H2O. The discrepancy with the higher values found by drying can be

explained by the failure of infrared spectroscopy to detect fluid-filled pores much

larger than 1µm, as these are going to absorb the IR photons at the OH band

wavenumber extremely effectively (Kronenberg, pers. communication).

3.4.2 Comparison with dry samples

It is generally accepted that completely water-free samples of NaCl do not

recrystallize at room temperature. Franssen (1993) and Guillopé and Poirier (1979)

have shown that dry grain boundaries become mobile at temperatures above 450 ºC.

To validate this for our samples, we examined the nominally dry samples with

cyclohexane as pore fluid. Reflected light microscopy of these samples shows an

almost total absence of recrystallization, with a few local occurrences of small new

grains. We interpret these to be the consequence of small fluid inclusions in the

grains, which were not fully removed by the drying procedure. Another method to dry

the NaCl is to melt it, and grind the solidified mass into a fine powder. Samples

prepared by this method, and pressed dry into dense blocks show the total absence of

recrystallization (JLU, unpublished observations).

3.4.3 Microstructural evolution of coarse-grained samples ("primary recrystallization")

The deformation of the initial cube-shaped grains was clearly shown by the irregular

grain shapes, the abundant defect structure and by microcracks initiated at grain

boundaries (Fig. 3.3a). Nucleated in these high-strain zones, new grains start to grow

after a few hours of annealing (Fig. 3.3b & 3c). These grains continue to grow into the

surrounding deformed grains until they meet another recrystallized grain. Usually new

grains are pinned at old grain boundaries, but sometimes they grow across old grain

boundaries (Fig. 3.3c). Most new grains tend to be euhedral, but often the grain edges

are not sharp, and curved grain boundaries also occur.

74

Figure 3.3: Micrographs of coarse-grained samples. See Table 3.1 for a detailed

description of the respective samples (also relevant to all sample micrographs shown in this paper). a) SEM image of a broken surface of sample OS-003d showing microcracks (see arrow) and other damage on grain surfaces pointing to locally high defect density favorable for nucleation of primary recrystallization. b) SEM image of a broken surface of sample OS-003c illustrating primary recrystallized grains. Note the difference of the surface morphology of old and new grains. c) reflected light image of a polished and etched surface of sample OS-012c showing the growth of euhedral new grains

75

during primary recrystallization. Boundaries between two new grains do not show evidence of mobility. Note the abundant defect structure in the deformed old grains.

Figure 3.4 shows the size of these euhedral grains fully enclosed within deformed old

grains, as a function of annealing time. Initially grain boundary migration rates are of

the order of 2-6 nm/s, but with increasing time of annealing the growth rate decreases.

Although the grains are euhedral, it is noted that the grains’ corners are in fact not

sharp (Fig. 3.3c & Fig. 3.5). SEM observations show that these rounded corners

consist of smaller facets which have generally the same orientation as the large faces.

However, also facets with orientations other than 100 were found (Fig. 3.5).

Figure 3.4: Combined diagram of grainsize versus annealing time data for the three

different types of grain boundary migration described in this paper (see text for methods of measurements and Table 3.1 for further details). Note the decrease in growth rate during primary recrystallization, the large scatter of data in secondary recrystallization (exaggerated grain growth) and the essentially zero growth rate in normal grain growth.

76

Figure 3.5: SEM micrographs of a polished and etched surface of sample OS-092a

showing the euhedral growth of primary recrystallizing grains by consuming deformed material. The trace of the grain boundary is covered by efflorescence. The detail (b) illustrates the shape of the ‘rounded’ corners, with ‘stepped’ growth according to the crystallographic system. Efflorescence at the boundary of the recrystallized grain points to existence of fluids.

In cross section, indications of fluids in the boundaries of these primary recrystallized

grains are efflorescence (Fig. 3.5) and grain boundary voids with a thickness of less

than 100nm. Most commonly however, a separation between the two grains (which

could have contained a fluid film) could not be resolved at the limits of resolution of

the SEM. In plane view, some details of the internal structure of the grain boundaries

could be resolved, showing a regular network of slight depressions on the surface of

the grain being consumed (Fig. 3.6).

77

Figure 3.6: SEM micrograph of a broken surface of sample OS-097ab showing a

negative imprint of a recrystallized grain. Note the structured interface of the old grain: the parallel bands are interpreted as etch pit-like structures related to dislocation networks inside the old grain (see also sketch in Fig. 3.12a).

The occurrence of efflorescence on both polished and broken surfaces was commonly

patchy, with some parts showing strong efflorescence and others none at all. This

observation also holds for the fine-grained samples, and its significance will be

discussed further below.

3.4.4 Microstructural evolution of fine-grained samples

“Recrystallization and grain growth”

Compaction of the wet, fine-grained slurry also produces dense aggregates. The grain

size versus time data is shown in Figure 3.4. It can be seen that even for long

annealing times there is no significant grain growth. Closer observation of the

78

microstructure during the first hours of annealing does reveal a reduction of grains

with a visible internal defect structure, a reduction of irregularities in grain

boundaries, and some rearrangement of porosity (Fig. 3.7).

Figure 3.7: Series of SEM micrographs of polished and etched surfaces of the

samples (a) OS-070a, (b) OS-072b, (c) OS-074b, (d) OS-070b, (e) OS-071a and (f) OS-071b illustrating the evolution of grain structure with time. With increasing time the serrated grain boundaries straighten out and defect structure in the grains is reduced. After one hour of annealing a further increase in grain size ceases.

Using the grain size versus time data from primary recrystallization (Fig. 3.4), we

estimate the time required to completely replace deformed grains by primary

recrystallization in these fine-grained samples to be less than 5 hours. Therefore we

infer that primary recrystallization is largely completed after this period, in agreement

with our observations (Fig. 3.8).

79

Figure 3.8: SEM micrographs of a broken surface of sample OS-039b illustrating

primary recrystallization during the first hours of annealing of the fine-grained samples. The large grain in the central part of the image is a deformed old grain being consumed by a defect-free grain with generally low index faces, but also faces different from 100 are found at the recrystallized grain’s corner (b).

Efflorescence in large parts of the sample surface always formed during preparation

of the fine-grained samples (Fig. 3.9), regardless the time of annealing. If the samples

are oven-dried before polishing and etching, the surface structure is similar, but no

efflorescence is formed. However, there were always parts of the surface which

formed no efflorescence. These locations were usually small protrusions on the

broken surface (Fig. 3.9a).

80

Figure 3.9: SEM micrographs of samples OS-041a (a) (broken surface) and OS-114a

(b) (polished and etched surface) showing the effects of efflorescence. In (a) the central part develops no efflorescence: we infer that it is isolated from the fluid-filled network in the sample. b) shows the efflorescence in detail. c) is a sketch illustrating the capillary-driven transport of brine that leads to efflorescence resulting in deposition of larger amounts of halite around grain boundaries emerging at the sample surface.

These areas without efflorescence allow observation of the grain boundary structure in

broken samples. The main feature is the occurrence of angular grains with planar

boundaries. Porosity is located predominantly at triple junctions, and sometimes

between grains. The planar parts of the grain boundaries are smooth, showing only a

few irregularities resolvable at the resolution of SEM, but with a clearly defined rim,

connected with the triple junction porosity. This observation is equivalent to the

clearly defined dihedral angles observed in images showing boundaries in profile.

Dihedral angles are highly variable: they range between ~20 to 110°, associated with

the tendency of many grains to develop rectangular shapes (Fig. 3.10).

81

Figure 3.10: SEM micrographs of efflorescence-free broken surfaces of samples OS-

041a (a) and OS-114a (b,c) illustrating grains with their characteristic smooth planar surfaces, with sometimes minor elevations at the grains’ edges. These structures indicate contact healing. The porosity is predominantly in triple junctions channels. In these figures the grains are thought to have completely healed grain boundaries and grain growth is stopped. Note the tendency of the

82

grains to develop rectangular shapes and also note the broad range of contact angles in (a) (see arrow; small bumps in this section are artifacts).

“Exaggerated grain growth”

In most fine-grained samples exaggerated grain growth was observed. These grains

are much larger than the starting grain size of 10µm and consume the surrounding fine

grains. They are characterized by their rectangular shape (Fig. 3.11). As shown in

Figure 3.4 the size of these grains increases with time of annealing but with a

decreasing growth rate, similar to primary recrystallized grains. These grains mostly

appear in a fine-grained matrix (Fig. 3.11). They are predominantly free of inclusions,

but rarely cigar-shaped inclusions are found inside them.

Figure 3.11: SEM micrograph of a polished and etched surface of sample OS-117a

showing the grain boundary of an exaggerated grain characterized by predominantly straight grain boundaries. Along these boundaries it consumes the fine-sized grains of the matrix.

At low magnification the boundaries of these new grains appear straight, but high

magnification micrographs show that they are irregular at length scales of the small

matrix grains (Fig. 3.11). The grain boundary separating the large new grains from the

matrix is commonly free of pores visible at the resolution of SEM.

83

3.5 Discussion

In agreement with much previous work, these experiments show that the presence of

small amounts of brine has a major effect on grain boundary migration in sodium

chloride. This effect is absent in samples containing a non-polar fluid.

The wet samples have a fluid-filled porosity of 2.5 ± 0.5 % and very few gas bubbles

which were carefully avoided by the preparation procedure. Many pores are relatively

large and form a connected network, in agreement with microstructural observations

and the drying data.

The sample preparation process exposed the pore fluid in the samples to dry air. As

shown by the efflorescence, this resulted in migration of the fluid and crystallization

from the evaporating brine. This effect was surprising, because the results of Lewis

and Holness (1996) report at conditions of 1 bar and 25 °C contact angles at which

porosity should not be connected.

Evaporation of immobile brine from a pore produces a coherently grown sodium

chloride layer with a thickness of 16% of the brine’s volume. Evaporation in

combination with capillary forces and a mobile and connected fluid phase will have a

larger effect: the fluid locally drains towards the surface, where it evaporates,

producing the observed efflorescence structures (see Fig. 3.9). In these locations this

may strongly obscure the original grain boundary structure. On the other hand, this

flow of the brine will also effectively drain the pore fluid from some other areas of the

sample surface. This drainage of brine away from the external surface is interpreted to

be the reason for efflorescence-free regions which can be used to study the details of

the grain boundaries (compare Fig. 3.9a). Therefore we interpret the efflorescence-

free parts of the microstructure to represent the in-situ structure with the brine

removed, with only minor modification due to sample preparation.

3.5.1 Primary Recrystallization

Euhedral primary recrystallized grains are also observed by Skrotzki and Welch

(1983) from extrusion experiments at room temperature, pointing to brine on the grain

boundary probably derived from fluid inclusions inside the deformed matrix. In

experiments and in nature however, curved and euhedral grains are both common.

84

In agreement with extensive literature, we interpret the results of the coarse-grained

samples as follows.

The pressing of the coarse-grained, cube-shaped powders caused deformation of the

grains, which is shown by their deformed shape and internal defect microstructure

containing dislocation networks and microcracks. Annealing at room temperature

does not allow recovery processes inside the deformed grains. The compaction-caused

defect density is highest in the old grain boundary regions, leading to nucleation of

strain-free grains at these sites (Humphreys and Hatherly, 1996). We note that

differences in dislocation density are expected inside the old grains, and thus driving

forces for grain boundary migration will vary. This is in agreement with the scatter in

initial growth rates of the new grains. Primary recrystallization then proceeds in an

isotropic stress field of 1 atmosphere.

The structure of the migrating boundaries could not be observed in detail using the

techniques of this study. Arguments in favor of the interpretation of the existence of

fluid-enriched zones on grain boundaries are (i) the frequent efflorescence from such

boundaries, (ii) euhedral shape of the growing grains, and (iii) the high migration

rates. In agreement with previous work, we interpret the boundaries to continue a

semi-continuous water-enriched zone with a thickness less than about 50 nm. If the

zone was thicker, we would have resolved a separation between the grains in the

SEM. It was somewhat surprising to find so little evidence of thicker migrating fluid

films in these samples, as a normal stress is absent and there is no a priori reason for

initially thicker boundary films to start migrating.

3.5.2 Normal grain growth

The presence of fluids also plays a major role for the fine-grained samples as it

enhances the mobility of the grain boundaries for primary recrystallization of the

deformed grains. In agreement with the rates observed in coarse-grained samples,

primary recrystallization is interpreted to be similar to that in the coarse-grained

samples, but come to completion much more rapidly. Both broken and polished &

etched SEM micrographs show the rearrangement of grains, grain boundaries and

porosity being completed after several hours.

After this time, the grain growth essentially stops. Considerations of the driving force

for grain growth suggest that this is not just a much slower process which might be

85

significant at even longer time scales (Fig. 3.4). We interpret the grain size versus

time data, in combination with the clear contact angles as an indication of a

fundamental change in grain boundary mobility: at conditions of 1 bar and 25 °C neck

growth is initiated and the mobile grain boundaries reorganize into fluid-free

boundaries connecting the network of brine channels in the triple junction network.

This essentially stops grain growth. Under a small external stress field a similar

process was observed by Visser (1999) in fine-grained sodium nitrate.

A final point worth considering is why in our fine-grained samples neck growth in

grain boundaries does not lead to the formation of arrays of fluid inclusions. This can

be explained by the commonly reported characteristic size of such inclusions in

coarse-grained halite (Urai et al., 1986b). This size is almost as large as the grain size

in our samples, so that almost without exception neck growth leads to redistribution of

the fluid into the triple junction network.

The morphologies of grain edges and triple junctions with a broad range of contact

angles (20 to 110°) are interpreted to be caused by the high surface anisotropy of the

initial wet grain boundaries. This leads to a change in the morphologies around the

triple junctions, and local large deviations from the theoretical contact angle based on

isotropic surface energies. This results in a mosaic of small-sized grains with

straightened grain boundaries and with a broad spectrum of contact angles (Fig. 3.10).

Efflorescence still found in samples after annealing for months is evidence for fluid

mobility through this network, in agreement with the wide range of contact angles.

This is in contrast to the observations found by Lewis and Holness (1996) who argued

that at similar conditions (1 bar and ~25 °C) the contact angles are about 70° resulting

in non-permeability of salt rocks.

3.5.3 Exaggerated grain growth

Because of its similarity to primary recrystallization, exaggerated (abnormal or

discontinuous) grain growth is also called secondary recrystallization. This process

requires an already recrystallized structure, in which grain growth is impeded, unless

some grains enjoy some advantages other than size over its neighbors (Humphreys

and Hatherly, 1996). According to these authors such advantages are second phase

particles, texture and surface effects. As for normal grain growth, this process is

86

driven by the reduction in grain boundary energy resulting in grain boundary area

reduction during coarsening (Detert, 1978; Evans et al., 2001).

As for the coarse-grained samples, exaggerated grain growth takes place under static

conditions. Its driving force is the stored energy inside the volume of fine-grained

matrix that is being consumed.

Secondary recrystallization starts after normal grain growth already stopped. The fine-

grained matrix consists of grains all with different orientation resulting in surface

energy variations between the grains (Humphreys and Hatherly, 1996). As the surface

energy of a grain itself is strongly dependent on the surface chemistry, fluids existing

on grain boundaries and in triple junction tubes have also a major effect on the onset

of secondary recrystallization. We interpret the observations as follows: the abnormal

growing grain consumes the neighboring fine-grained matrix and the grain boundary

fluid-rich zone incorporates fluids present in pores. As for primary recrystallization

the fluids are assumed to be distributed uniformly on the mobile grain boundaries, so

that the grain continues its growth into an euhedral shape indicating again the fluid-

dependent anisotropy of surface energies. In ceramics similar processes are well

known: exaggerated grain growth with large faceted grains is explained by anisotropy

of surface energies (Kingery, 1974).

3.5.4 Nature of grain boundary fluid

Although direct observation of the fluid in mobile grain boundaries was not possible

in this study, all our observations led us to the interpretation of their presence. If the

fluid-enriched zone has short range thickness variations, these should be commonly

less than 50 nm. Below this length scale the thickness of the fluid is not resolved.

There is however a clear difference in lateral structure of grain boundaries in primary

recrystallization and exaggerated grain growth, illustrated in Figure 3.12.

87

Figure 3.12: Schematic illustration of the hypothesis of the structure of migrating

brine-filled grain boundaries. a) Primary recrystallization: the strain-free grain consumes the deformed grain indicated by dislocations. Dissolution is interpreted to be favored at etch pit-like structured locations. b) Secondary recrystallization: the exaggerated grain consumes the fine-grained matrix by incorporating the fluids that are residing in a network of triple junction tubes. Similar to (a) these areas are preferred dissolution sites. For both processes, the thickness of the fluid-enriched zones is assumed to be less than 50 nm.

In primary recrystallization the new grain is growing into an old one with a high

dislocation density. Our observations suggest that the primary morphological factor

on the surface of the grain being dissolved is the array of etch pits around the

termination of dislocation arrays on the grain surface. At these sites the rate of

88

dissolution is slightly higher, and in combination with the euhedral new grains this is

proposed to lead to a thickness variation of the grain boundary fluid-rich zone.

In exaggerated grain growth the migrating grain boundaries are permanently

connected to the terminations of fluid-filled triple junction tubes (Fig. 12b). In

between these, there are the dry grain boundaries which are being consumed. We

propose that an etch-pit like structure is also present here, forming a system of fine

channels in the grain boundary which are connecting the triple junction tube

terminations. One possible explanation for the observation that exaggerated-grown

grains are free of inclusions is that the fluid which might be otherwise collected in the

moving grain boundary is removed from the system by flow along the triple junction

tubes.

To get more information on the properties as thickness and morphology of these

brine-filled grain boundaries we are starting direct observations in a cryogenic SEM,

in which the fluids will be frozen before breaking the sample.

Acknowledgements

We are grateful to Andreas Kronenberg for carrying out the infrared spectrometry.

Uwe Wollenberg and Jörg Kallinna provided essential assistance with SEM. This

project was funded by the Deutsche Forschungsgemeinschaft (UR 64/4-1).

3.6 References

de Bresser JHP, Ter Heege JH and Spiers CJ (2001) Grain size reduction by dynamic recrystallization: Can it result in major rheological weakening? International Journal of Earth Sciences (Geologische Rundschau) 90:28-45

de Meer S, Spiers CJ, Peach CJ and Watanabe T (2002) Diffusive properties of fluid-filled grain boundaries measuredelectrically during active pressure solution. Earth and Planetary Science Letters 200:147-157

Detert K (1978) Secondary Recrystallization. In: Haessner F (Ed), Recrystallization of Metallic Materials. Dr. Riederer Verlag GmbH, Stuttgart, pp 97-109

Drury MR and Urai JL (1990) Deformation-related recrystallization processes. Tectonophysics 172:235-253

Evans B, Renner J and Hirth G (2001) A few remarks on the kinetics of static grain growth in rocks. International Journal of Earth Science (Geologische Rundschau) 90:88-103

Fokker PA, Urai JL and Steeneken PV (1996) Production-induced convergence of solution mined caverns in Magnesium salts and associated subsidence, In: Proceedings Int. Conference on Land Subsidence, Den Haag, pp 281-289

89

Franssen RCMW (1993) Rheology of Synthetic Rocksalt with Emphasis on the Influence of Deformation History and Geometry on the Flow Behaviour. Geologica Ultraiectina, 113. Universiteit Utrecht, pp 221

Garcia Celma A, Urai JL and Spiers CJ (1988) A laboratory investigation into the interaction of recrystallization and radiation damage effects in polycrystalline salt rocks, Department of Structural and Apllied Geology, Institute of Earth Science, pp 125

Guillopé M and Poirier JP (1979) Dynamic recrystallization during creep of single-crystalline halite: An experimental study. Journal of Geophysical Research 84:5557-5567

Hickman SH and Evans B (1991) Experimental pressure solution in halite; the effect of grain/interphase boundary structure. Journal of the Geological Society of London 148:549-560

Hilgers C, Koehn D, Bons PD and Urai JL (2001) Development of crystal morphology during unitaxial growth in a progressively widening vein; II, Numerical simulations of the evolution of antitaxial fibrous veins. Journal of Structural Geology 23:873-885

Hilgers C and Urai JL (2002) Microstructural observations on natural syntectonic fibrous veins: implications for the growth process. Tectonophysics 352:257-274

Holness M and Lewis S (1997) The structure of the halite-brine interface inferred from pressure and temperature variations of equilibrium dihedral angles in the halite-H2O-CO2 system. Geochimica et Cosmochimica Acta 61:795-804

Humphreys FJ and Hatherly M (1996) Recrystallization and related annealing phenomena. Pergamon, pp 497

Jackson MPA and Talbot CJ (1986) External shapes, strain rates, and dynamics of salt structures. Geological society of america bulletin 97:305-323

Kingery WD (1974) Plausible Concepts Necessary and Sufficient for Interpretation of Ceramic Grain-Boundary Phenomena: II, Solute Segregation, Grain-Boundary Diffusion, and General Discussion. Journal of the American Ceramic Society 57:74-83

Lewis S and Holness M (1996) Equilibrium halite-H2O dihedral angles: High rock-salt permeability in the shallow crust? Geology 24:431-434

Martin B, Roeller K and Stoeckhert B (1999) Low-stress pressure solution experiments on halite single-crystals. Tectonophysics 308:299-310

Miralles L, Sans M, Gali S and Santanach P (2001) 3-D rock salt fabrics in a shear zone (Suria Anticline, South-Pyrenees). Journal of Structural Geology 23:675-691

Peach CJ (1991) Influence of deformation on the fluid transport properties of salt rocks. Geologica Ultraiectina, 77. Universiteit Utrecht, Utrecht, pp 238

Peach CJ, Spiers CJ and Trimby PW (2001) Effect of confining pressure on dilatation, recrystallization, and flow of rock salt at 150°C. Journal of Geophysical Research 106:13,315-13,328

Schutjens P (1991) Intergranular pressure solution in halite aggregates and quartz sands: an experimental investigation. Geologica Ultraiectina, 76. Universiteit Utrecht, pp 233

Skrotzki W and Welch P (1983) Development of texture and microstructure in extruded ionic polycrystalline aggregates. Tectonophysics 99:47-61

Smith CS (1964) Some elementary principles of polycrystalline microstructure. Metallurgical Reviews 9:1-48

90

Spiers CJ and Brzesowsky RH (1993) Densification behaviour of wet granular salt: Theory versus experiment, Seventh Symposium on Salt. Elsevier Science Publishers B.V., Amsterdam, pp 83-92

Spiers CJ and Schutjens P (1990) Densification of crystalline aggregates by fluid-phase diffusional creep. In: Meredith PD (Ed), Deformation processes in minerals, ceramics and rocks. Unwin Hyman, pp 334-353

Spiers CJ, Schutjens PMTM, Brzesowsky RH, Peach CJ, Liezenberg JL and Zwart HJ (1990) Experimental determination of constitutive parameters governing creep of rocksalt by pressure solution. In: Rutter EH (Ed), Deformation Mechanisms, Rheology and Tectonics. Geological Society Special Publication, pp 215-227

Talbot CJ and Rogers EA (1980) Seasonal movements in a salt glaciers in Iran. Science 208:395-396

Underwood EE (1970) Quantitative Stereology. Addison-Wesley Publishing Company, pp 274

Urai JL, Means WD and Lister GS (1986a) Dynamic recrystallization of minerals. In: Heard HC (Ed), Mineral and rock deformation; laboratory studies; the Paterson volume. AGU Geophysical Monograph, pp 161-199

Urai JL, Spiers CJ, Peach C, Franssen RCMW and Liezenberg JL (1987) Deformation mechanisms operating in naturally deformed halite rocks as deduced from microstructural investigations. Geologie en Mijnbouw 66:165-176

Urai JL, Spiers CJ, Zwart HJ and Lister GS (1986b) Weakening of rock salt by water during long-term creep. Nature 324:554-557

Visser HJM (1999) Mass transfer processes in crystalline aggregates containing a fluid phase, Geologica Ultraiectina. Universiteit Utrecht, pp 244

Watanabe T and Peach CJ (2002) Electrical impedance measurement of plastically deforming halite rocks at 125°C and 50 MPa. Journal of Geophysical Research 107:ECV 2-1 - 2-12

Wenkert DD (1979) The flow of salt glaciers. Geophysical Research Letters 6:523-525

91

3.7 Appendix3

3.7.1 Application of the data to the models of Visser (1999)

The phenomenon of contact healing was shown to occur in pressure solution

experiments. Hickman & Evans (1991) studied the morphology of grain-to-grain

contacts of two halite crystals in the presence of brine under stress and observed

healing of the contacts as the contact angle between the two salt lenses changed

towards its equilibrium value and the grain boundaries showed no relief at a scale

observable with reflected light interferometry. Visser (1999) conducted uniaxial

densification tests on fine grained sodium nitrate aggregates in the presence of sodium

nitrate saturated solution with volumetric strains of 10-20 %. For high stresses

(> 3 · 104 Pa) and grain sizes > 20 µm she observed a densification behavior that is

consistent with conventional pressure solution models, i.e. that with decreasing grain

size the strain rate increases. However for low stresses (< 3 · 104 Pa) and fine grain

sizes (< 20 µm), the opposite dependence was observed (the strain rates decreased

with decreasing grain size). These observations were attributed to the effect of surface

energy forces that exceed the stress related forces.

The theoretical basis for grain contact healing under stress was further given by Visser

(1999), who considered surface energy terms in addition to stress-related driving

forces and predicted the fields for pressure solution, contact healing and neck growth

(for a detailed description of the models the reader is referred to chapter 1 of this

thesis).

Our experiments on statically recrystallizing synthetic polycrystals of sodium chloride

containing saturated brine are characterized by a similar microstructural behavior.

Inside the coarse grained samples euhedral primary recrystallized grains are observed

that grow into the surrounding deformed matrix, pointing to the presence of fluids on

the grain boundary. However the experiments on the compacted fine grained samples

show that – after completion of primary recrystallization and rearrangement grain

boundaries – normal grain growth stops (see Fig. 3.4).

3 not included in the published article.

92

In the following, we apply our observations of fluid-assisted grain boundary migration

to the criteria Visser presented to describe the competition between the driving forces

for neck growth/contact healing type processes and solution/precipitation creep.

Coarse grained samples

The compaction of the coarse grained samples caused deformation and resulted in

nucleation and growth of strain-free grains. Due to our experimental conditions the

primary recrystallizion proceeded in an isotropic stress field of 1 atmosphere

(≈ 105 Pa). Although during compaction direct measurements of the differential

stresses at the grain scale are not available, these can be inferred to be less than

20MPa based on single crystal experiments on halite (Carter & Hansen, 1983).

Assuming a dislocation density (ρdisl) of 2.5 · 1013 m-2 (= 25 µm-2) corresponding to a

stress (τ) of 12.5 MPa (Kemter & Strunk, 1977), the driving force for grain boundary

migration is 1 · 105 J · m-3 (Urai et al., 1986).

According to Nicolas & Poirier (1976) the internal elastic strain in crystals can be

described:

by the stress field of an edge dislocation:

( )

( )

( )

sin2 1

sin 1

cos2 1

rr

zz

r

G bx

G bx

G bxθ

θσπ ν

θσπ ν

θσπ ν

=−

=−

=−

and by the stress field of a screw dislocation:

13

23

sin2

cos2

Gbx

Gbx

θσπ

θσπ

= −

= +

with σ: stress Pa] G: shear modulus [Pa] b: Burgers vector [m] θ: periodic angle for screw movement around cylinder axis [°] ν: Poisson ratio [-] x: distance from dislocation core [m]

93

For sodium chloride the shear modulus (G) is 14.7 GPa Visser (1999) and the Burgers

vector (b) is 2.8 Ǻ for slip on 110 planes at low temperature (Nicolas & Poirier,

1976). Assuming the Poisson ratio (ν) to be 0.25 with θ ranging between 0 and 90°,

the stress field (σ) of the dislocation is described by:

0.1 2x x

σ< <

Again considering the dislocation density (ρdisl) to be = 25 µm-2, the simplified

approximation of the distance between dislocations is 0.2 µm. Thus for the radius

from the dislocation core (x) = 0.1 µm, the relationship shows that the locked-in

elastic stress of the deformed grains is in the range between 1 and 20 MPa.

The radius of curvature of the euhedral strain-free grains is proposed to be in the order

of meters. If we now assume that the locked-in elastic stress has a similar effect as the

external stress field in the models predicted by Visser (1999), our data can be plotted

into her material transport mechanism maps (Fig. A1). This comparison suggests that

during primary recrystallization contact healing should be prevented by the stress

locked into the deformed grains, without an external stress field and may explain why

fluid-filled grain boundaries do not neck down into fluid inclusion arrays during

primary recrystallization. An additional effect here may be the fact that the growing

grains are euhedral and bound by F-faces with a strong surface anisotropy.

Fine grained samples

Inside the brine-containing fine grained samples (grain size < 10µm) normal grain

growth stopped after primary recrystallization and grain boundary rearrangement were

completed (see Fig. 3.7). This was interpreted to be due to contact healing processes

that lead to the redistribution of fluids from mobile grain boundaries into a connected

network of triple junction tubes leaving behind fluid-free contacts. Let us assume i)

the stress acting at the contacts being around 10 Pa, a very low value due to body

forces in the sample and perhaps some residual stresses, and ii) the radius of curvature

being in the range of 10 µm to 1 mm (compare Fig. 3.7). If we plot the data into the

map of material transport mechanisms (Fig. 3.A1) it falls into the neck growth/contact

healing field.

94

Figure 3.A1: Map of material transport mechanism fields (Visser, 1999) that are

applied to the fluid-assisted recrystallization processes observed in our experiments of coarse and fine grained wet halite samples. Here the Visser criteria are proposed to separate the field of contact healing from that of fluid-assisted grain boundary migration. The plotted data of our experiments show that the primary recrystallization of the coarse grained samples.

This means that surface energy related forces dominate and result in healing of grain-

to-grain contacts (criterion 1) and cessation of normal grain growth. Even under non-

stressed but also non-equilibrium conditions, the existence of surface energy related

forces may result in spreading of solid-solid contacts and finally in sealing of the

boundary (Visser, 1999). The absence of fluid inclusions on the grain boundaries is

interpreted to be due to the fine grain size, so that fluid redistribution leads to

accumulation in triple junctions and triple junction tubes.

3.7.2 References

Carter, N. L. & Hansen, F. D. 1983. Creep of rocksalt. Tectonophysics 92, 275-333. Hickman, S. H. & Evans, B. 1991. Experimental pressure solution in halite; the effect

of grain/interphase boundary structure. Journal of the Geological Society 148, 549-560.

95

Kemter, L. & Strunk, H. I. 1977. Dislocation density in deformed NaCl single crystals determined by transmission electron microscopy. Phys. Stat. Sol. A 40, 385-391.

Nicolas, A. & Poirier, J. P. 1976. Crystalline Plasticity and Solid State Flow in Metamorphic Rocks. John Wiley & Sons, London, 444 pp.

Urai, J. L., Means, W. D. & Lister, G. S. 1986. Dynamic recrystallization of minerals. In: Mineral and rock deformation; laboratory studies; the Paterson volume (edited by Hobbs, B. E. & Heard, H. C.). AGU Geophysical Monograph 36, 161-199.

Visser, H. J. M. 1999. Mass transfer processes in crystalline aggregates containing a fluid phase. PhD thesis, Universiteit Utrecht, 244 pp.

96

Chapter 4:

Structure of grain boundaries in wet, synthetic polycrystalline, statically recrystallizing halite – evidence from cryo-SEM observations4

4.0 Abstract It is well known from nature and experiments that the presence of brine strongly

affects the microstructural evolution and the mechanical and transport properties of

halite. Existing interpretations of grain boundary structure in deformed wet salt

samples annealed statically at room temperature were based on indirect evidence from

reflected light microscopy and conventional SEM.

This paper presents direct observations of fluid-filled grain boundaries using the cryo-

scanning electron microscope (cryo-SEM) in which the grain boundary fluids were

frozen before breaking the samples. The rapid cooling transforms the brine into the

phases ice and hydrohalite, which are easily recognized from typical segregation

patterns. We studied samples of wet, synthetic, polycrystalline halite annealed under

static conditions at room temperature. Inside coarse-grained samples thin segregation

patterns were observed at the boundaries of the primary recrystallizing grains. These

point to the existence of fluid films with a thickness in the range of 30 nm, but the

finer scale structure of the fluid remains unknown. Inside fine grained samples the

distribution and reorganization of fluids with annealing time is documented by the

combination of contact healing and successive accumulation of fluids in triple

junction tubes. The contact healing is attributed to the small initial grain size, such

4 Schenk, O., Urai, J.L. & Piazolo, S., submitted. Structure of grain boundaries in wet, synthetic polycrystalline, statically recrystallizing halite – evidence from cryo-SEM observations. Geofluids.

97

that the fluid film necks down by accumulating the fluids into previously existing

triple junctions.

Detailed EBSD measurements of both primary and secondary recrystallized grains

indicate that their growth is euhedral, i.e. that grain growth is controlled by the

anisotropy of grain boundary energy of the growing grain.

4.1 Introduction Recrystallization and grain coarsening are microscale transformations that have major

implications for the texture of rocks (Urai et al., 1986a; Evans et al., 2001). In

minerals such as quartz (Griggs, 1974; Tullis & Yund, 1982; Jaoul et al., 1984;

Kronenberg & Tullis, 1984; Hirth & Tullis, 1992; Post & Tullis, 1998), feldspar

(Tullis et al., 1996; Dimanov et al., 1999), olivine (Mei & Kohlstedt, 2003a, b),

bischofite (Urai, 1983) or carnallite (Urai, 1985) fluids play a significant role on

recrystallization and grain growth.

The effect of fluids on recrystallization in halite was shown by numerous observations

from nature and experiments during both fluid-phase diffusional creep (pressure

solution) (Spiers et al., 1990; Spiers & Schutjens, 1990; Hickman & Evans, 1991;

Schutjens, 1991; Peach, 1991; Martin et al., 1999; de Meer et al., 2002) and fluid-

assisted grain boundary migration (Urai et al, 1986b; Peach et al., 2001; Watanabe &

Peach, 2002; Schenk & Urai, 2004). The structure of the halite grain boundaries

which contain water is still a matter of debate. Firstly for pressure solution three

different models that do not exclude each other have been proposed (den Brok et al.,

2002): i) the thin film model (Rutter, 1976; Hickman & Evans, 1991; Renard &

Ortoleva, 1997), the island-channel model (Lehner, 1990; Spiers & Schutjens, 1990)

and iii) the island-crack model (Gratz, 1991; den Brok, 1998).

In the thin film boundary model the grains are separated by a thin, structured water

film with a thickness of a few nanometers. This film is proposed to transmit the

contact stress and diffusion is the process of the transport of dissolved material.

The island-channel boundary model is based on the assumption that – during pressure

solution – the fluids residing in thin films are squeezed out between the grains

resulting in solid-solid contact (islands) through which the contact stresses are

transmitted and water-filled channels through which the material transport takes place

by diffusion. This microscopically rough island-channel structure is dynamically

stable.

98

The island-crack boundary model proposes static islands that are separated by

microfracture-controlled fluid channels. In contrast to the solid-solid contact of the

island-channel model, the islands in this model contain thin films comparable to the

earlier proposed thin film boundary model. Due to the low thickness of these thin

films diffusion through the latter is rate-controlling. However, compared to the thin

film model, the total diffusivity in the island-crack model is increased by the presence

of the microcracks.

Secondly experiments on wet polycrystalline halite deformed at temperatures between

room temperature and 150 °C in the non-dilatant field indicate that the samples

recrystallize readily during and after deformation (Urai et al., 1986a, b; Drury & Urai,

1990; Spiers et al., 1990; Peach et al., 2001; Watanabe & Peach, 2002).

The grain boundaries are interpreted to contain thin fluid films. A method to show the

presence of such brine films in water containing halite samples is the application of

the ether test (Spiers et al., 1986): during evaporation of the ether the fluid film is

disrupted into isolated non-volatile droplets. Urai et al. (1986b) showed fluid films by

SEM observations on deformed water-containing halite samples: 1 month after the

experiment grain boundaries showed smooth surfaces, whereas samples annealed for

one year showed grain boundaries with isolated bubbles. The authors interpreted these

results as evidence for the presence of brine films that shrink into isolated fluid

inclusions after grain boundary migration stopped. Similar observations were shown

by in-situ experiments conducted on wet bischofite, during which water-filled grain

boundaries neck down after grain boundary migration stopped (Urai, 1987).

Additionally, in these experiments some cigar-shaped fluid inclusions were left

behind the migrating grain boundary supporting the hypothesis of the presence of

brine films.

The fluid-filled grain boundaries are interpreted to migrate by i) dissolution of the

deformed grains, ii) diffusion through the brine film and iii) precipitation on the low-

index facets of the recrystallized grains forming smooth grain surfaces, comparable to

the step model of Gleiter (1969). However so far the nature of such fluid films in

migrating boundaries has only be inferred indirectly. One problem was that the

observations were made after removal of the stress, which could have led to a

redistribution of the fluid by viscous flow.

99

This was avoided in a recent study in which the microstructural evolution of wet

compacted, statically recrystallizing halite samples with different initial grain sizes

was presented (Schenk & Urai, 2004).

The microstructural evolution and grain size data in Schenk & Urai (2004) are briefly

summarized in Figure 4.1: Inside the coarse grained samples primary recrystallization

occured by euhedral grains that grew into the old deformed grains (Fig. 4.1a). Inside

the fine grained samples primary recrystallization wais followed by normal grain

growth, but stopped after a few hours due to contact healing (Fig. 4.1b), while

exaggerated grain growth (secondary recrystallization) initiated at this stage (Fig.

4.1c).

Figure 4.1: Schematic illustration of the microstructural evolution of wet compacted,

statically recrystallizing halite samples described in Schenk & Urai (2004). The processes and grain size evolution of the samples described in this paper are identical. Note that normal grain growth is inhibited, while growth of primary recrystallized and exaggerated grains continue.

100

Schenk & Urai (2004) interpreted growth of the euhedral primary and secondary

recrystallized grains to be due to the presence of brine films on the grain boundaries.

Nevertheless, the detailed nature of the fluid distribution and its influence on grain

boundary migration was only partly resolved, because fluid is removed during sample

preparation.

In this paper we therefore set out to investigate directly and in detail the nature of the

fluid in grain boundaries during different stages of recrystallization within compacted,

polycrystalline halite samples. To do this we studied samples using the cryo-SEM

which recently has become an important tool in geosciences. Small rock chips are

shock-frozen to a very low temperature (~ –190 °C) and can be used for chemical

characterizations (Timofeeff, 2001; Samson, 2001) and for visualization of the

distribution of fluids in rocks, in particular to investigate qualitatively the fluid-rock

interfaces (Mann, 1994; Durand, 1998; Monma, 1997). In order to characterize the

crystallographic nature of grain boundaries we used detailed Electron Backscatter

Diffraction (EBSD) analysis.

4.2 Methods

4.2.1 Experimental methods

Preparation of compacted samples

The sample preparation is the same as in Schenk & Urai (2004), to which the reader is

referred for a detailed description. Here we summarize the procedure briefly: wet

sodium chloride powder of analytical grade (Roth, Art. 9265.3; NaCl content

> 99.9 %) of two different grain size classes (< 10 µm and 200-355 µm) were

compacted (cold-pressed) uniaxially with a pressure of p = 150 MPa for 5 minutes.

The resulting aggregates have brine-filled porosities less than 2 % together with very

minor occurrences of air-filled pores. The samples were annealed at room temperature

(24 ± 1 °C) over periods up to 9 months in small, air-tight containers with small

amounts of saturated salt solution, not in contact with the samples, except for samples

139 and 141 that were stored in a wet salt mush (see Table 4.1 for a detailed

description of the compacted samples).

101

Single crystal – brine setup

To interpret the frozen structure of brine inside the compacted samples, we compared

them with the structure of a thin film of frozen saturated sodium chloride solution

between halite plates. For this an industrial-grown halite single crystal was cleaved

along 100 cleavage facets into two thin wafers, which were bonded capillarily with

a droplet of saturated sodium chloride solution (Roth, Art. 9265.3; NaCl content

> 99.9 %) for 12 hours without applying any external stress. Subsequently, this setup

was inserted into a cryo-SEM holder and frozen at a temperature of T = -90 °C for

5 minutes.

4.2.2 Analytical methods

Preparation for cryo-SEM observations

To study the brine-filled grain boundaries directly, we investigated the synthetic halite

samples in a field emission scanning electron microscope (JSM-6300F, JEOL)

equipped with a dedicated cryo-preparation chamber (CT 1500 HF, Oxford

Instruments) at the Department of Plant Cell Biology, Wageningen University, the

Netherlands. If necessary, the samples were ground to the required thickness of

~1.6 mm very carefully to avoid any damage. Then the sample was placed into the

slot of the cryo-SEM holder and fixed with carbon conductive cement (Leit-C,

Neubauer chemicals). It was secured additionally by tightening carefully the screw of

the holder. Subsequently, the whole assembly (sample with holder) was immersed

into liquid nitrogen (-196 °C). Once frozen, the whole unit was transferred into the

cryo-preparation chamber at a temperature of T = -90 °C and p = 1.3*10-3 Pa (high

vacuum conditions), in which the sample was fractured by a cold knife (-196 °C) in

adequate distance from the tightened holder (see sketch in Figure 4.2a). After

sublimating for ~5 minutes, the sample was sputter coated with 8 nm Platinum and

subsequently transferred into the SEM on the sample holder with a temperature

between –170 and –190 °C. Images were recorded digitally.

102

EBSD

Detailed EBSD analysis was conducted on selected samples to investigate the nature

of grain boundaries of primary and secondary recrystallized grains in annealed

samples.

To obtain a high resolution EBSD pattern the samples (thick sections) were

mechanically polished using 1200-, 2400- and 4000 grade carborundum paper,

removing dust at regular intervals with a blast of dry compressed air. To remove

surface damage, the samples were chemically polished in pure analytical grade

methanol for 10 seconds, and then immediately and vigorously rinsed in a jet of

diethyl ether. Finally samples where carbon coated to reduce charge during EBSD

analysis.

Samples were analysed in a field-emission gun (FEG) CamScan X500 SEM at the

University of Liverpool. Full crystallographic orientation data were obtained from

electron backscatter diffraction (EBSD) patterns using a 20 kV acceleration voltage

and a beam current of 7 nA. EBSD patterns were auto-indexed using the CHANNEL

5.03 software of HKL Technology. The centre of 5-6 Kikuchi bands was detected

automatically whereby the solid angles calculated from the patterns were compared

with the calculated halite patterns originating from 47 reflectors. Data were obtained

by moving the beam at a fixed step size of 2 µm. The average percentage of EBSD

patterns that could not be indexed ranged between 30 and 35%; most of the un-

indexed analyses were at high angle grain boundaries. The maps were processed to

remove erroneous data in order to provide a more complete reconstruction of the

microstructure (Prior et al., 2002). The accuracy of individual EBSD orientation

measurements is better than 1°. The misorientation angle between grains was

calculated by selecting the minimum misorientation angle and its corresponding axis

from all possible symmetric variants (cf. Wheeler et al., 2001).

We present data in a combination of data displays: (1) maps showing the spatial

distribution of grains and their crystallography in different grayscales and (2) 3D

representation of the crystallographic orientation of individual grains of special

interest.

4.3 Observations The microstructural evolution of both the fine and coarse grained samples is identical

with the observations made in Schenk & Urai (2004). In this study grain boundary

103

morphology during recrystallization was studied on broken surfaces of samples

annealed for different periods using the cryo-FESEM.

An overview of the samples described in this paper and the main observations are

given in Table 4.1.

Table 4.1: Overview of samples described in this paper.

sample type of salt starting grainsize annealing time observations

[µm] [d]

1-XXs brine: Roth - 0,50 segregation pattern

085a Roth 200-355 252,23 prim RX

111a Roth <10 180,97 norm GG

138a Roth <10 29,93 exag GG

139a Roth <10 6,70 norm GG

141a Roth <10 1,10 norm GG

146 Roth 200-355 139,98 prim RX

152 Roth 200-355 0,50 prim RX a stored in salt mush

4.3.1 Segregation patterns of frozen brine

Setup with halite single crystals and brine

In order to correctly recognize the former presence of brine at grain boundaries we

compared the segregation pattern of boundaries of a sample in which brine was

enclosed between two single crystal wafers of halite (Fig. 4.2a). Due to the rapid

cooling the saturated brine shock-freezes and is transferred into the phases ice and

hydrohalite (NaCl *2 H2O) (Bodnar, 1993; Roedder, 1984). During sublimation for

5 minutes at –90 °C inside the sample stage the ice crystals evaporate leaving behind

negative imprints now seen as pores (Figs. 4.2b & c). Accordingly, if such pores are

observed in our samples, we suggest that they point directly to the presence of brine.

104

Figure 4.2: Cryo-SEM micrographs of the single crystal – brine setup. a) Two thin

wafers of halite single crystals (XX-1 and XX-2) were attached to each other parallel to the 100 facets with a droplet of saturated brine and broken with a cold knife inside the cryo-chamber. b) & c) show details of the segregation pattern of the frozen brine film with the phases hydrohalite and ice, the latter as negative imprint due to evaporation during sublimation.

105

General comparison of segregation patterns observed in

experimental samples

Brine inclusions are common in halite, and observed in a coarse-grained sample inside

an old, deformed grain (Fig. 4.3). This frozen, cubic-shaped fluid inclusion is

characterized by the typical segregation pattern, similar to those found in the single

crystal-brine setup, indicating that it represents frozen salt solution, i.e. with the

phases hydrohalite and (evaporated) ice. Inside both the coarse- and fine-grained

samples similar patterns were observed in pores, triple junction tubes and on grain

boundaries demonstrating that those patterns are also the result of brine in frozen

state. Thus, these patterns provide direct evidence on the distribution of brine inside

the compacted samples.

Figure 4.3: Cryo-SEM micrographs of a coarse deformed grain in sample 152 with a

frozen, cubic-shaped fluid inclusion. The segregation pattern is similar to that of the saturated brine in the setup shown in Figure 4.2. See Table 4.1 for a detailed description of the sample (also relevant to all sample micrographs shown in this paper).

106

The volume increase that is due to the transition of saturated brine into the phases

hydrohalite and ice might result in misinterpretations of the microstructure as the

previously brine-filled grain boundaries or triple junctions expand. However, Figure

4.3.b shows that the shock-freezing did not result in microstructural changes such that

the surrounding grain is fractured. Furthermore this micrograph illustrates that the size

of the evaporated ice is 5 times smaller than in the single crystal-brine setup (Fig.

4.2). This suggests that the size of the segregated components is related to the volume

of the fluid.

4.3.2 Microstructural evolution of wet grain boundaries in coarse-grained samples during primary recrystallization

Inside the coarse-grained aggregates (initial grain size: 200-355 µm) primary

recrystallization is the dominant process. Nucleation of primary recrystallized grains

occurs in high-strain zones close to the contact regions of old deformed grains. After

nucleation, these new grains are characterized by a cubic shape and grow into the

surrounding deformed material. This microstructural evolution is documented by the

samples that annealed 0.5 days, 4.5 and 8.5 months (samples 152, 146 and 085a,

respectively (see Table 4.1)). Segregation patterns of the frozen fluids indicate that in

all samples brine was present.

If two or more recrystallized, i.e. strain-free grains with a grain size smaller than ~20

µm are in contact to each other, the frozen brine is only present in triple junctions

(Fig. 4.4).

Grain boundary migration continues if a strain-free grain is in contact with a deformed

grain: Inside sample 146 (annealed for 4.5 months) some of these grains are larger

than 100 µm (Fig. 4.5). This contact zone between the new, recrystallized and the old,

107

deformed grain is characterized by a very thin segregation pattern and has a thickness

of less than 30 nm (Fig. 4.5c). Additionally some remnants of frozen fluids are visible

on the lower crystal face of the deformed crystals (see arrow in Fig. 4.5c).

Figure 4.4: Cryo-SEM micrographs of sample 152. At the contact of the two old,

deformed coarse grains new, recrystallized and cubic-shaped grains start to grow. Segregation pattern in a triple junction points to the presence of fluids.

Backscatter analysis indicates that there are no measurable differences in chemistry

between the deformed, the recrystallized grain and the solid phase of the segregation

pattern (cf. Heard & Ryerson, 1986), suggesting that second phases other than brine

did not influence this contact region. Inside the same sample, a different but rare

situation is displayed in Figure 4.6: Here, the porosity is predominantly air-filled, as

shown by the lack of segregation pattern. However, remnants of a frozen fluid phase

are observed at the contact zone of a recrystallized and a deformed grain. The

thickness is ~ 200 nm (Figs. 4.6b & c) with a segregation pattern being very similar to

that shown in the grain boundary region of Figure 4.5c.

108

Figure 4.5: Cryo-SEM micrographs of sample 146 showing thin segregation pattern

between the primary recrystallized and the deformed grain. This pattern (arrow in c)) is interpreted to represent a frozen fluid film with a thickness of less than 30 nm. Note that the deformed grain is characterized by complex cleavage pattern when compared to defect-free grain.

109

Figure 4.6: Cryo-SEM micrographs of sample 146 showing a primary recrystallizing

grain growing into an old, deformed grain. The arrow in c) points to the remnants of a frozen fluid film at the contact region. Note that this is a rare region in which porosity is predominantly air-filled.

110

4.3.3 Microstructural evolution of brine characteristics within fine-grained samples – Primary recrystallization, normal grain growth and exaggerated grain growth

The evolution inside the fine-grained samples (initial grain size: < 10 µm) starts with

primary recrystallization followed by a reduction of irregularities in grain boundaries

and rearrangement of porosity, that stops after only a few hours. There is no sign of

significant normal grain growth; however at this stage exaggerated grain growth takes

place in some samples.

Fluids are present in all the fine-grained samples whether they were stored in brine

saturated environment or salt mush regardless the annealing time of 1 day, 1 week, 1

month or 6 months (samples 141, 139, 138a and 111a, respectively) (see Table 4.1 for

the detailed sample description).

During the first days, the fluids – indicated by the typical segregation pattern – are

present on irregular and sometimes curved grain boundaries with a thickness of less

than 150 nm, but predominantly in triple junctions (Fig. 4.7).

After one week of annealing at room temperature the microstructure is characterized

by solid-solid contacts. (Fig. 4.8; see solid arrow). The triple junctions remain

irregular because of the lattice-dependent (euhedral) growth of the crystals into the

porosity (Fig. 4.8; see dashed arrow) with dihedral angles varying over a broad range.

An isolated brine-filled inclusion (250 * 50 nm) (Fig. 4.9) is interpreted to represent a

leftover-inclusion. However, this might also be explained by the presence of a fluid-

filled pore resting on the grain boundary.

111

Figure 4.7: Cryo-SEM micrographs of sample 141 showing that fluids are present in

triple junctions and in grain boundaries in the early stages of annealing. Close to large pores the grains grow normal to the 100 facets; however, at grain-grain contacts the grain boundaries are irregular and curved. The detail in b) is interpreted to represent the first step of contact healing (arrow).

112

Figure 4.8: Cryo-SEM micrograph of sample 139 with the typical segregation pattern

of brine-filled poristy. The fine-sized grains grow with low-index facets into the fluid-filled pore (dashed arrow). Note the high diversity of dihedral angles. The solid arrow points to a immobile solid-solid contact region of two recrystallized grains.

Figure 4.9: Cryo-SEM micrograph of sample 139. Some grain boundaries are fluid-

filled, irregular and curved, while others are already healed. The arrow points to fluid inclusion that was probably left behind during grain boundary migration.

The microstructure of sample 111a (annealed for 6 months) is characterized by

predominantly straightened grain boundaries without interactions of fluids and an

interconnected porosity with fluids present only along triple junctions and in triple

junction tubes (Fig. 4.10).

113

Figure 4.10: Cryo-SEM micrograph of sample 111a showing that after months of

annealing the microstructure is reorganized with commonly straightened, fluid-free grain boundaries (solid arrow) and fluids still present in triple junctions (dashed arrow).

Exaggerated grain growth is common inside these fine grained samples as shown by

reflected light microscopy and SEM (see EBSD pattern of sample 138a (annealed for

1 month; see Fig. 4.12). However, cryo-SEM did not allow detailed observations on

the contact of exaggerated grains with the fine grained matrix, probably due to the

plucking-out of the large grains during the low-temperature preparation.

4.3.4 Crystallographic nature of grain boundaries of primary and secondary recrystallized grains

After an annealing period of 8 months in a brine saturated humid environment we

observe in the coarse grained sample 085a (initial grain size: 200-355 µm) several

large euhedral grains. They show little to no lattice distortion within the individual

grain and a dominance of boundaries that form traces to 100 facets (Fig. 4.11). The

misorientation angle between the facetted boundaries of the primary recrystallized

grains and surrounding grains is always > 15º.

114

Figure 4.11: SEM and EBSD micrographs of the coarse-grained sample 085a

showing that (primary) recrystallizing grains grow with low-index facets (euhedral growth) into the old, deformed grain; a) shows the so-called band contrast as analysed by the EBSD technique. Dark areas are commonly deformed areas while light grey signifies undeformed grains. b) results from EBSD analysis depicting recrystallized grains with little internal lattice distortion (light grey in a)) and dominance of grain boundaries that are compatible with 100 facets. Insets show 3D representation of the crystallography of the repective grain.

Similar features are observed in the samples exhibiting exaggerated grain growth. We

investigated the crystallographic nature of the boundaries of such grains in the sample

138a (see Table 4.1 for experimental details). In this sample, most of the grains are

115

still very fine grained ( < 10 µm), only few significantly larger grains of up to 300 µm

are observed. The straight boundaries of these grains already described in the SEM

backscatter analysis are characterized by crystal orientation which are compatible

with 100 facets (Fig. 4.12).

Figure 4.12: SEM and EBSD micrographs of fine-grained sample 138a indicating

that exaggerated grown grains are in fact euhedral, i.e. that its growth is lattice-dependent. a) band contrast analysis showing exaggerated grains several orders of magnitude larger than the matrix grains. The fine grained matrix appears dark due to the abundance of grain boundaries which appear dark in a band contrast analysis image. Note the insets representing the crystallographic nature of the individiual exaggerated grains. These show that the boundaries of the exaggerated grains are 100 facets; b) same area as shown in a) this time showing grains in different grey shades.

116

4.4 Discussion Our cryo-SEM observations of segregation patterns along grain boundaries and triple

junctions show that small amounts of fluids are present in the majority of mobile grain

boundaries and in larger pores. In the cryo-SEM, these boundaries have a resolvable

structure which does not indicate a fluid film thicker than 30 nm. The small-scale

structure seen in Figure 4.5c, however, can be interpreted in two ways: it can be an

island-channel structure or a segregation pattern in a continuous fluid film. Thus,

although there is clear evidence of fluids in these mobile boundaries, we cannot obtain

conclusive information on the nano-scale structure from our observations (Fig. 4.13)

due to the unknown morphology of the segregation pattern in very thin frozen brine

films and to the limited resolution of the SEM for structures smaller than a few

nanometers.

The euhedral shape of the recrystallized grains is related to such fluid-filled grain

boundaries (Fig. 4.5c). The recrystallized grain’s surface is inferred to be an F-facet,

however if the fluid-filled grain boundary is a semi continuous fluid film (Fig. 4.13b)

or has an island-channel structure (Fig. 4.13c) remains unclear.

The presence of fluids in grain boundaries agrees with previous observations on

mobile grain boundaries in wet halite (Drury & Urai, 1990; Urai et al., 1986a, b;

Peach et al., 2001; Watanabe & Peach, 2002; Schenk & Urai, 2004), whereas

experiments on dry sodium chloride show that the grain boundaries are immobile

below temperatures of 400 °C (Guillopé & Poirier, 1979; Franssen, 1993). The fact

that recrystallized grains are characterized by an euhedral shape with a clear

crystallographic relationship in terms of facets (Figs. 4.5, 4.11 & 4.12) is in agreement

with observations of similar microstructures in other fluid-containing, recrystallizing

materials.

Such a preferred growth of primary or secondary recrystallized grains is interpreted to

be either i) a result from the high surface energy anisotropy of the wetted NaCl grain

boundaries or ii) a growth mechanism similar to that seen in crystal-melt systems

where ledge mechanism leads to the euhedral shape according to the step model of

Gleiter (1969) or iii) a combination of i) and ii).

117

Figure 4.13: Schematic illustration of mobile, fluid-filled grain boundaries as shown

by segregation pattern from cryo-SEM observations (a). The arrows indicate the euhedral growth of the primary recrystallizing grain (white) into the deformed grain. The true nature of these boundaries in terms of a semi-continuous fluid film (b) or an island-channel structure (c) cannot be resolved with the cryo-SEM.

High surface energy anisotropies are expected to play a major role in fully wetted

grain boundaries. Observations in olivine-ultramafic melt systems showed that

completely wetted grain boundaries are often found parallel to low-index facets

118

((010), (110) and (021)) (Jung and Waff, 1998). These are similar to our observations

of growing primary recrystallizing and exaggerated grains normal to the 100 faces.

However, Walte et al. (2003) questioned the importance of such effects of surface

energy anisotropy. They showed that completely wetted grain boundaries can simply

form by consumption of small grains during fluid-enhanced static recrystallization,

and they concluded that there is no need to relate the structures to surface energy

anisotropy, even though this might enhance the effect.

Another possibility for euhedral shape of the observed primary and secondary

recrystallized grains is the ledge jump grain boundary migration mechanism described

by Gleiter (1969). The assumption of a fluid-filled grain boundary with two solid-

fluid interfaces and a fluid layer in between is similar in terms of the sharp transition

of crystal lattice and adjacent grain boundary and the influence of the orientation of

the crystal on the migration rate. According to the step model the motion of the grain

boundary in the presence of a driving force proceeds by i) dissolution of ions from

favored sites (steps) of the shrinking old grain and from deformation-related

dislocations that reach the surface, ii) diffusion through the fluid layer and iii) re-

attachment at preferential steps of the growing strain-free grain. The euhedral shape

suggests that diffusion is not restricted to the shortest distance. However, the fluid

layer regulates (balances) the transport of ions, such that they are precipitated at

favored steps to preserve the character of the 100 facet.

Inside the coarse grained samples grain boundary migration stops if two or more

recrystallized grains get in contact to each other. As there is no difference in

dislocation density, a significant driving force is lacking. Only the grain boundary

(surface) energy can drive further grain boundary migration. In this situation, the

grain boundary fluid is accumulated along triple junction tubes leaving behind healed,

brine-free grain boundaries. These immobile solid-solid contacts could have

developed by boundary annealing, i.e. the surface energy driven attraction of grain

boundary fluids into the triple junction network, a process that is controlled by the

contact angle.

The cessation of normal grain growth inside the fine grained samples is also

interpreted to be caused by such boundary healing: below a critical grain size the

fluid-filled grain boundary contracts and accumulates in the triple junction network

(Visser, 1999).

119

4.5 Conclusions

• Cryo-SEM observations offer direct evidence of fluid-filled grain boundaries

in statically recrystallizing wet, polycrystalline sodium chloride samples. The

frozen fluid phase is represented by the segregation pattern composed of the

phases hydrohalite and evaporated ice.

• The thickness of such migrating brine-filled grain boundaries is usually less

than 30 nm. Finer scale structure is obscured by resolution of SEM and

segregation of brine during freezing.

• Primary recrystallized and exaggerated grown grains exhibit euhedral shapes

with 100 facets. We interpret this type of growth as a consequence of either

significant anisotropic grain boundary energy and/or a solid-melt/brine type

growth mechanism with a ledge jump mechanism.

• The results are in agreement with a model of brine-filled grain boundaries

during primary recrystallization and exaggerated grain growth, and healed

grain boundaries in normal grain growth.

Acknowledgements

We are grateful to A. van der Aelst (Department of Plant Cell Biology of Wageningen

University, The Netherlands) for his valuable assistance with the cryo-FESEM.

The comments on the phase conditions of the NaCl-H2O-system at low temperature

by R. Bodnar are greatly appreciated and H. Siemes is thanked for providing the NaCl

single crystals. This project is funded by the Deutsche Forschungsgemeinschaft (UR

64/4-1). SP acknowledges financial support by Marie Curie Fellowship HPMF-CT-

2001-01457, NERC grant NER/A/S/2001/01181 and HEFCE through the grant

JR98LIPR.

4.6 References Bodnar, R. J. 1993. Revised equation and table for determining the freezing point

depression of H2O-NaCl solutions. Geochimica et Cosmochimica Acta 57(3), 683-684.

120

de Meer, S., Spiers, C. J., Peach, C. J. & Watanabe, T. 2002b. Diffusive properties of fluid-filled grain boundaries measured electrically during active pressure solution. Earth and Planetary Science Letters 200, 147-157.

den Brok, B., Morel, J. & Zahid, M. 2002. In situ experimental study of roughness development at a stressed solid/fluid interface. In: Deformation mechanisms, rheology and tectonics; current status and future perspectives (edited by de Meer, S., Drury, M. R., de Bresser, J. H. P. & Pennock, G. M.). Geological Society Special Publications 200. Geological Society of London : London, UK, 73-83.

den Brok, S. W. J. 1998. Effect of microcracking on pressure-solution strain rate; the Gratz grain-boundary model. Geology 26(10), 915-918.

Dimanov, A., Dresen, G., Xiao, X. & Wirth, R. 1999. Grain boundary diffusion creep of synthetic anorthite aggregates: The effect of water. Journal of Geophysical Research 104(B5), 10,483-10,497.

Drury, M. R. & Urai, J. L. 1990. Deformation-related recrystallization processes. Tectonophysics 172(3-4), 235-253.

Durand, C. & Rosenberg, E. 1998. Fluid distribution in kaolinite- or illite-bearing cores: cryo-SEM observations versus bulk measurements. Journal of Petroleum Science and Engineering 19(1-2), 65-72.

Evans, B., Renner, J. & Hirth, G. 2001. A few remarks on the kinetics of static grain growth in rocks. International Journal of Earth Science (Geologische Rundschau) 90, 88-103.

Franssen, R. C. M. W. 1993. Rheology of Synthetic Rocksalt with Emphasis on the Influence of Deformation History and Geometry on the Flow Behaviour. PhD thesis. Universiteit Utrecht, 221p.

Gleiter, H. 1969. The mechanism of grain boundary migration. Acta Metallurgica 17(5), 565-573.

Gratz, A. J. 1991. Solution-transfer compaction of quartzites; progress toward a rate law. Geology 19(9), 901-904.

Griggs, D. 1974. A model of hydrolytic weakening in quartz. Journal of Geophysical Research 79(11), 1653-1661.

Guillopé, M. & Poirier, J. P. 1979. Dynamic recrystallization during creep of single-crystalline halite: An experimental study. Journal of Geophysical Research 84, 5557-5567.

Heard, H. C. & Ryerson, F. J. 1986. Effect of cation impurities on steady-state flow of salt. In: Mineral and rock deformation; laboratory studies; the Paterson volume. (edited by Hobbs, B. E. & Heard, H. C.). AGU Geophysical Monograph 36, 99-115.

Hickman, S. H. & Evans, B. 1991. Experimental pressure solution in halite; the effect of grain/interphase boundary structure. Journal of the Geological Society 148, 549-560.

Hirth, G. & Tullis, J. 1992. Dislocation creep regimes in quartz aggregates. Journal of Structural Geology 14(2), 145-159.

Jaoul, O., Tullis, J. & Kronenberg, A. 1984. The effect of varying water contents on the creep behavior of heavitree quartzite. Journal of Geophysical Research 89(B6), 4298-4312.

Jung, H. & Waff, H. S. 1998. Olivine crystallographic control and anisotropic melt distribution in ultramafic partial melts. Geophysical Research Letters 25(15), 2901-2904.

Kronenberg, A. K. & Tullis, J. 1984. Flow strengths of quartz aggregates: grain size and pressure effects due to hydrolytic weakening. Journal of Geophysical Research 89(B6), 4281-4297.

121

Lehner, F. K. 1990. Thermodynamics of rock deformation by pressure solution. In: Deformation processes in minerals, ceramics and rocks (edited by Barber, D. J. & Meredith, P. D.). Unwin Hyman, 296-333.

Mann, U., Neisel, J. D., Burchard, W. G., Heinen, V. & Welte, D. H. 1994. Fluid-rock interfaces as revealed by cryo-scanning electron microscopy. First Break 12(3), 131-136.

Martin, B., Roeller, K. & Stoeckhert, B. 1999. Low-stress pressure solution experiments on halite single-crystals. Tectonophysics 308, 299-310.

Mei, S. & Kohlstedt, D. L. 2000a. Influence of water on plastic deformation of olivine aggregates 1. Diffusion creep regime. Journal of Geophysical Research, B, Solid Earth and Planets 105(9), 21,457-21,469.

Mei, S. & Kohlstedt, D. L. 2000b. Influence of water on plastic deformation of olivine aggregates 2. Dislocation creep regime. Journal of Geophysical Research, B, Solid Earth and Planets 105(9), 21,471-21,481.

Monma, T., Kudo, M. & Masuko, T. 1997. Flow behaviors of smectite/water suspensions in terms of particle-coagulated structures. Nendo Kagaku 37(2), 47-57.

Peach, C. J. 1991. Influence of deformation on the fluid transport properties of salt rocks. PhD thesis. Universiteit Utrecht, 238p.

Peach, C. J., Spiers, C. J. & Trimby, P. W. 2001. Effect of confining pressure on dilatation, recrystallization, and flow of rock salt at 150°C. Journal of Geophysical Research 106(B7), 13,315-13,328.

Post, A. & Tullis, J. 1998. The rate of water penetration in experimentally deformed quartzite: implications for hydrolytic weakening. Tectonophysics 295(1-2), 117-137.

Prior, D. J., Wheeler, J., Peruzzo, L., Spiess, R. & Storey, C. 2002. Some garnet microstructures: an illustration of the potential of orientation maps and misorientation analysis in microstructural studies. Journal of Structural Geology 24(6-7), 999-1011.

Renard, F. & Ortoleva, P. 1997. Water films at grain-grain contacts: Debye-Huckel, osmotic model of stress, salinity, and mineralogy dependence. Geochimica et Cosmochimica Acta 61(10), 1963-1970.

Roedder, E. 1984. The fluids in salt. American Mineralogist 69(5-6), 413-439. Rutter, E. H. 1976. The kinetics of rock deformation by pressure solution.

Philosophical Transactions of the Royal Society of London A 283, 203-219. Samson, I. M. & Walker, R. T. 2000. Cryogenic Raman spectroscopic studies in the

system NaCl-CaCl2-H2O and implications for low-temperature phase behavior in aqueous fluid inclusions. Canadian Mineralogist 38, 35-43.

Schenk, O. & Urai, J. L. 2004. Microstructural evolution and grain boundary structure during static recrystallization in synthetic polycrystals of Sodium Chloride containing saturated brine. Contributions to Mineralogy and Petrology 146(6), 671-682.

Schutjens, P. 1991. Intergranular pressure solution in halite aggregates and quartz sands: an experimental investigation. PhD thesis. Universiteit Utrecht, 233p.

Spiers, C. J., Urai, J., Lister, G. S., Boland, J. N. & Zwart, H. J. 1986. The influence of fluid-rock interaction on the rheology of salt rock and on ionic transport in the salt. University of Utrecht, 132p.

Spiers, C. J. & Schutjens, P. 1990. Densification of crystalline aggregates by fluid-phase diffusional creep. In: Deformation processes in minerals, ceramics and rocks (edited by Barber, D. J. & Meredith, P. D.). Unwin Hyman, 334-353.

122

Spiers, C. J., Schutjens, P. M. T. M., Brzesowsky, R. H., Peach, C. J., Liezenberg, J. L. & Zwart, H. J. 1990. Experimental determination of constitutive parameters governing creep of rocksalt by pressure solution. In: Deformation mechanisms, rheology and tectonics (edited by Knipe, R. J. & Rutter, E. H.). Geological Society Special Publications 54. Geological Society of London, United Kingdom, 215-227.

Timofeeff, M. N., Lowenstein, T. K., Brennan, S. T., Demicco, R. V., Zimmermann, H., Horita, J. & von Borstel, L. E. 2001. Evaluating seawater chemistry from fluid inclusions in halite: examples from modern marine and nonmarine environments. Geochimica et Cosmochimica Acta 65(14), 2293-2300.

Tullis, J. & Yund, R. A. 1982. Grain growth kinetics of quartz and calcite aggregates. Journal of Geology 90, 301-318.

Tullis, J., Yund, R. A. & Farver, J. 1996. Deformation-enhanced fluid distribution in feldspar aggregates and implications for ductile shear zones. Geology 24(1), 63-66.

Urai, J. L. 1983. Water assisted dynamic recrystallization and weakening in polycrystalline bischofite. Tectonophysics 96(1-2), 125-157.

Urai, J. L. 1985. Water-enhanced dynamic recrystallization and solution transfer in experimentally deformed carnallite. Tectonophysics 120(3-4), 285-317.

Urai, J. L., Means, W. D. & Lister, G. S. 1986a. Dynamic recrystallization of minerals. In: Mineral and rock deformation; laboratory studies; the Paterson volume. (edited by Hobbs, B. E. & Heard, H. C.). AGU Geophysical Monograph 36, 161-199.

Urai, J. L., Spiers, C. J., Zwart, H. J. & Lister, G. S. 1986b. Weakening of rock salt by water during long-term creep. Nature (London) 324(6097), 554-557.

Urai, J. L. 1987. Development of microstructure during deformation of carnallite and bischofite in transmitted light. Tectonophysics 135(1-3), 251-263.

Visser, H. J. M. 1999. Mass transfer processes in crystalline aggregates containing a fluid phase. PhD thesis. Universiteit Utrecht, 244p.

Walte, N. P., Bons, P. D., Passchier, C. W. & Koehn, D. 2003. Disequilibrium melt distribution during static recrystallization. Geology 31(11), 1009-1012.

Watanabe, T. & Peach, C. J. 2002. Electrical impedance measurement of plastically deforming halite rocks at 125°C and 50 MPa. Journal of Geophysical Research 107(B1), ECV 2-1 - 2-12.

Wheeler, J., Prior, D. J., Jiang, Z., Spiess, R. & Trimby, P. W. 2001. The petrological significance of misorientation between grains. Contributions to Mineralogy and Petrology 141, 109-124.

123

Chapter 5:

The migration of fluid-filled grain boundaries in recrystallizing synthetic bischofite – first results of in-situ HPHT deformation experiments in transmitted light5

5.0 Abstract The effect of fluids on the recrystallization behavior is well known, however their

detailed microscale distribution in grain boundaries and influence on grain boundary

migration is still unresolved. In this study we therefore carried out in-situ deformation

experiments in transmitted light microscopy as they allow continuous and direct

observation of the whole range of processes involved in fluid assisted grain boundary

migration. A new see-through deformation apparatus was developed enabling the

control of the fluid pressure.

We deformed bischofite containing small amounts of aqueous fluid at temperatures

between 50 and 90 °C, with the fluid pressure being 0.5 to 1 MPa. The strain rates

ranged from 5 · 10-6 to 1 · 10-4 s-1. The rates of grain boundary migration were

measured and assigned to the different temperatures and strain rates.

Detailed observations during and after deformation illustrate the evolution of the

migrating fluid-filled grain boundaries and show that the incorporation of fluids from

inclusions as well as their pinch-off is dependent on the grain boundary velocity, the

thickness of the grain boundary and the size and shape of the inclusions. We present

5 Schenk, O. & Urai, J.L., submitted. The migration of fluid-filled grain boundaries in recrystallizing synthetic bischofite – first results of in-situ HPHT deformation experiments in transmitted light. Journal of Metamorphic Geology.

124

direct evidence of the contraction of the grain boundary fluids into isolated inclusions

equilibrium conditions are attained.

5.1 Introduction Deformation experiments in transmitted light have been used in glaciology since the

late nineteen fifties by deforming ice. This technique was further used in geosciences

to get detailed insight into the complex dynamics of the microstructural evolution

during creep (Means, 1989 and references therein) and is now a well established and

useful tool in structural geology.

As most rock-forming minerals can not be studied in the ductile regime at laboratory

conditions with this experimental setup, materials such as magnesium (e.g. Burrows et

al., 1979), (nor-) camphor (e.g. Urai et al., 1980, 1981; Herwegh et al., 1997; Bauer et

al., 2000), sodium chlorate (e.g. den Brok et al., 1998) or octachloropropane (OCP)

(e.g. Jessell, 1986; Ree, 1994; Ree & Park, 1997) are used as analogue for quartz, or

sodium nitrate for calcite (e.g. Tungatt & Humphreys, 1981, 1984).

Some ionic salts as bischofite or carnallite are also suitable for the see-through

experiments (Urai, 1987) as they are transparent, optically anisotropic and as they

deform in the ductile regime at conditions of T < 300 °C and p < tens of MPa.

In this study we focus on in-situ observations of migrating grain boundaries that are

filled with saturated solution as the micro-scale distribution of fluids strongly affects

the transport and mechanical properties of rocks. In ionic salts the effect of fluids on

recrystallization was shown by observations from nature and experiments during both

solution transfer creep (e.g. Spiers et al. 1990; Hickman & Evans, 1991) and fluid

assisted grain boundary migration (e.g. Urai et al. 1986a, b; Peach et al., 2001).

However the structure of the fluid-filled grain boundaries is still under debate. For

pressure solution one model favors the presence of a non-mobile thin, structured fluid

film separating two grains (Rutter, 1976; Hickman & Evans, 1991), through which the

dissolved material is proposed to diffuse.

Experiments on wet halite in the non-dilatant field between 25 °C and 150 °C show

that grain boundary mobility is strongly enhanced (e.g. Urai et al., 1986b; Watanabe

& Peach, 2002; Schenk & Urai, 2004). This again is an indication for grain

boundaries to be filled with fluids, as diffusion through a fluid phase is dramatically

higher than in solid-solid contacts. The migration of such fluid-filled grain boundary

125

is interpreted as a process of i) dissolution of the deformed grains, ii) diffusion

through the brine film and iii) precipitation on recrystallized grains.

Urai et al. (1986b) deformed natural halite in the presence of brine: a sample broken

one month after deformation showed smooth grain boundaries, whereas a sample

observed after one year was characterized by isolated fluid inclusions on its grain

boundaries. This was interpreted as evidence for necking down of the fluid film into

its equilibrium configuration after grain boundary migration has stopped.

A method that offers direct evidence of fluid-filled grain boundaries is the application

of the cryo-SEM (Schenk et al., submitted): statically recrystallizing wet,

polycrystalline sodium chloride samples were shock frozen and observed at

temperatures of -194 °C; the migrating grain boundaries are characterized by the

segregation patterns that represent the frozen fluid phase, composed of the phases

hydrohalite and evaporated ice.

In a recent paper on in-situ experiments under transmitted light, Walte et al., (2003)

used the system norcamphor + ethanol to study the development of solid-liquid

systems. The experiments show that disequilibrium features such as completely

wetted grain boundaries and large melt patches can form during fluid-enhanced static

recrystallization when small grains are consumed. Although these features have the

tendency to evolve back toward equilibrium geometry, continuous static grain

coarsening results in the omnipresence of fully wetted grain boundaries and large melt

patches.

This study builds on the in-situ experiments of Urai (1987) on bischofite containing

saturated aqueous solution in order to obtain a fundamental understanding of the

processes that control the fluid geometry during grain boundary migration. We

critically examine the microstructural evolution during recrystallization with a special

focus on migrating fluid-filled grain boundaries up to their break-up into arrays of

isolated inclusions after migration has stopped. We used bischofite in our experiments

due to its suitable properties as transparency, optical anisotropy, plastic deformation

and recrystallization at relatively low temperatures.

5.2 Experimental techniques The newly developed see-though deformation apparatus (Fig. 5.1) follows the design

of the Urai-rig (1987), but includes a controlled pore fluid system that allows fluid

pressures up to 30 MPa. The design of the apparatus is described in the appendix.

126

Figure 5.1: The see-through deformation apparatus. See appendix for a detailed description.

The material used in the experiments is synthetic, polycrystalline bischofite

(MgCl2 · 6 H2O) (Roth; Art.-Nr. A.537.1), because of its high solubility and very

rapid recrystallization at low temperature.

Due to the hygroscopic nature of bischofite the sample was prepared as far as possible

before sample insertion. The first set of o-ring, cover glass, and notched glass plate

was inserted and covered with the sample holder plates (thickness: 300 µm) consisting

of the circular notched plate and the respective rectangular pin guide.

The samples were prepared by quickly pressing molten bischofite between two ingots

separated by space holders with a thickness of 300 µm. It was cut into the desired size

(14 · 8.5 mm) and immediately inserted into the cell, with a few droplets of saturated

bischofite solution (resulting in a fluid content of ~ 2 to 5 %). The outlets were sealed

with silicon oil (M5) to prevent dissolution of the sample due to capillary forces of the

brine layer (Fig. 5.2). Subsequently, the second set of notched glass plate, o-ring and

cover glass plate was inserted and the cell was closed with the precision nut to keep

the seals in place. Then the cell was placed onto the stage and connected with the

loading rams for the moving and pressure regulating pistons, heating coil,

thermocouple and motor controller. The stage was mounted onto the inverse optical

microscope equipped with a digital video camera for continuous documentation of the

observations.

127

The experiments were performed at temperatures of 50, 70 and 90 °C with the fluid

pressure maintained between 0.5 to 1 MPa. The strain rates ranged between 5 · 10-6 to

1 · 10-4 s-1. The microstructural evolution was studied during and after deformation.

Grain boundary migration rates were measured by tracing grain boundaries along

selected orthogonal trajectories.

Figure 5.2: Schematic illustration of the sample setup; a) view from top and b) view in profile. The wet bischofite sample is enclosed by saturated brine, while the outlets of the deformation cell are sealed with silicon oil. The immiscible fluid-fluid-boundary prevents dissolution of the sample and allows experiments in a closed system.

The experiments were performed at temperatures of 50, 70 and 90 °C with the fluid

pressure maintained between 0.5 to 1 MPa. The strain rates ranged between 5 · 10-6 to

1 · 10-4 s-1. The microstructural evolution was studied during and after deformation.

Grain boundary migration rates were measured by tracing grain boundaries along

selected orthogonal trajectories.

5.3 Observations In the presence of saturated solution, bischofite recrystallizes readily at temperatures

below 100 °C.

Several interesting features can be seen, such as the incorporation of fluids into

migrating grain boundaries, large changes in grain boundary migration rate and the

development of twinning and surface grooves.

An overview of the experiments is given in Table 5.1.

128

Table 5.1: Conditions of the experiments described in this paper (abbreviations: rx = recrystallization; dyn. = dynamic; metadyn. = metadynamic; stat. = static; gbm = grain boundary migration).

syndeformational postdeformational

tem-

perature strain rate prior strain rate sample

main observations

[°C] [s-1] [s-1]

sg surface grooves 80 fast

db inclusions during heating 70-80

3d-2 dyn rx; nucleation of grain 50 2·10-04

3d-1 dyn rx; consumed grains; gbm rates 50 4·10-06

3s metadyn rx; stat rx; gbm rates 50 4·10-05

2d-2 dyn rx; separation of inclusion 70 4·10-05

2d-1 dyn rx; contraction of fluid films; gbm rates 70 4·10-05

2s metadyn rx; stat rx; fluid films; gbm rates 70 4·10-05

1d twinning; dyn rx; gbm rates 90 2·10-04

1s metadyn rx; stat rx; gbm rates 90 2·10-04

5.3.1 Surface grooves

For one experimental setup the deformation cell was filled with a mixture of synthetic

bischofite grains and saturated solution and heated up to 95 °C, resulting in complete

dissolution of the grains. With subsequent cooling crystals nucleated and grew

parallel to 110 as elongated bischofite (hexahydrate) needles. The crystallization of

tetrahydrate grains was not observed.

At the temperature of 80 °C the piston moved into the deformation chamber. Most

needles were oriented with their long axes perpendicular to the shortening direction.

During shortening at fast rates (~1 · 10-4 s-1) surface grooves developed on surfaces of

some bischofite needles perpendicular to the 110 plane (Fig. 5.3). The grooves

occurred in a more or less regular distance of ~ 10 µm with various depth of up to

5 µm. Five minutes after the deformation had started most of the needles were rotated

and new grooves developed with different orientation with respect to the 110 plane

(Figs. 5.3b & c). After cessation of deformation the roughened surface evolved into a

flattened one again (Fig. 5.3d).

129

Figure 5.3: Stress induced grooves on surfaces of elongated bischofite grains at a

temperature of 90 °C. After deformation the surfaces smoothen again (image 022).

Discussion on surface grooves

The motion of the piston is interpreted to have caused bending of some bischofite

needles. The initially flat surface of the elastically strained crystal is morphologically

unstable (see den Brok et al., 2002). The interaction of surface energy and elastic

strain energy drives this surface to roughen by development of valley-ridge structures.

The amplitude of such surface grooves is directly related to the applied stress

(Srolovitz, 1989). Rotation of some needles resulted in different stress conditions on

the bent needles, which are accommodated by different orientations of the grooves.

After cessation of deformation the stress-related grooves disappear as the surface

energy exceeds again and drives the grain’s surface to smoothen again.

5.3.2 Nature of fluid inclusions

The dominant aim of the experiments was the investigation of the nature of fluids

during grain boundary migration. Even if the fluid pressure inside the cell suppresses

air bubbles, very minor amounts of air-filled inclusions are present inside the sample.

Due to the sealing of the outlets with silicon oil, the occurrence of additional small oil

130

droplets inside the sample cannot be excluded. These phases are characterized by

different refractive indices that result in different intensities of the solid-liquid

surfaces. The process of recrystallization affects the shape of the inclusions ranging

from circular droplets (spheres) to straightened, tubular or even flattened occurrences.

Thus, to ensure that the investigated inclusions and grain boundaries are filled with

saturated bischofite solution, an inclusion-rich bischofite grain resting in saturated

bischofite solution was heated (Fig. 5.4): if focused the elongated inclusions are

characterized by a sharp solid-liquid boundary, similar to the grain’s transition to the

surrounding solution. With successive dissolution the grain surface reaches the

inclusions and their content flows into the surroundings liquid. The fact that there are

no disturbances or “clouds” around the mouths indicates that the inclusions are filled

with the same phase as the liquid i.e. saturated bischofite solution.

Figure 5.4: Image sequence showing the behavior of fluid inclusions during heating

of bischofite grains (T = 70 to 80 °C). Note the different intensity of the fluid-solid surfaces of the disappearing fluid inclusions. See text for details.

131

An example of the pulling back of fluid inclusions is illustrated in Figure 5.5: Directly

after the elongated inclusion is separated the lower right bubble contracts towards a

spherical shape.

Figure 5.5: Image sequence showing an elongated tubular fluid inclusion that is

separated during deformation. Note that the smaller inclusion evolves towards a sphere to minimize its surface energy (T=70 °C; strain rate: 4·10-05 s-1).

5.3.3 Microstructural development

Deformation of the bischofite samples causes intracrystalline defects indicated by

undulose extinction and twinning of some grains.

Predominantly at high angle boundaries new grains nucleate and grow into grains

with presumably higher dislocation density (Fig. 5.6). Due to ongoing deformation

these grains are successively strained and the rate of grain boundaries migration

decreases. Such grain boundaries again are preferred sites for nucleation of new

dynamically recrystallizing grains. By this process the grain size is reduced

significantly, however due to the thickness of our samples (300 µm) and grain sizes

smaller than 20 µm, the grains overlap and make a statistically valuable grain size

analysis difficult.

132

Figure 5.6: Image sequence showing the nucleation of grains at old high angle grain

boundaries. Subsequently the strain-free grain grows with its boundaries migrating into the deformed microstructure (T = 50 °C; strain rate: 2·10-4s-1).

During dynamic recrystallization the grain boundary migration rate is observed to be

strongly dependent on the strain rate. In sample 1d deformed with a strain rate of ~ 2 ·

10-4 s-1 at 90 °C migration rates of up to 4000 nm/s are measured, whereas in

experiments performed with strain rates of ~ 6 · 10-5 s-1 at 70 °C (sample 2d-1) and ~

4 · 10-6 s-1 at 50 °C (sample 3d-1) grain boundary migration rates are observed to

reach values of up to 800 nm/s and 70 nm/s, respectively (Fig. 5.7a). Sudden changes

in grain boundary migration rates are common in all deformation experiments,

however more or less constant migration rates are typical (Fig. 5.8a).

After cessation of deformation the grain boundaries continue to move through the

microstructure. Independent on temperature or prior strain rate a general decrease of

migration rates with time is observed in all experiments decreasing rapidly direct after

deformation and slowing down subsequently (Fig. 5.8b). Sudden changes in migration

133

rates are also present but only to a minor extent, predominantly occurring in the

metadynamic recrystallization stage, i.e. shortly after deformation stopped.

Figure 5.7: Diagram showing the dependence of strain rate on the grain boundary

migration rate. a) during deformation and b) after deformation. The range of grain boundary migration rates is indicated by the arrows. See Table 5.1 for sample description.

The grain boundary migration rates in the post-kinematically recrystallized samples

(deformed with a strain rate of ~ 4 · 10-5 s-1 at 50 °C (sample 3s) and 70 °C

(sample 2s)) are significantly slower than during deformation (Fig. 5.7b). However

inside the sample that was deformed with the high strain rate of 2 · 10-4 s-1 the

migration rates during syn- and post-kinematic recrystallization (samples 1d and 1s,

respectively) are of the same order, before it significantly decreases two minutes after

deformation stopped.

134

Figure 5.8: Time-displacement curves of migrating grain boundaries for different

experiments. See Table 5.1 for sample description. a) during dynamic recrystallization; note the sudden changes in grain boundary migration rates. b) during post-kinematic recrystallization; during the first minutes the graphs represent metadynamic recrystallization, whereas during static recrystallization the migration rates decrease.

135

Subsequently the microstructural evolution is dominated by the rearrangement of

grains. Grain boundaries are straightened (Fig. 5.9, see solid arrows) and triple

junctions evolve towards equilibrium values of 120° (Fig. 5.9, dashed arrow).

Figure 5.9: Image sequence illustrating the post-kinematic microstructural evolution.

Directly after cessation of deformation a recrystallizing grain grows at a high rate (upper row). With time and decreasing driving force the migration slows down (lower row). The surface energy becomes more important and results in straightened grain boundaries (solid arrow) and triple junctions with 120° angles (dashed arrow) (T = 90 °C; prior strain rate: 2·10-4 s-1).

Fluids in grain boundaries

Inside the samples saturated bischofite solution is present in fluid inclusions and fluid-

filled grain boundaries. Due to deformation, temperature and subsequent

recrystallization grain boundaries start to migrate through the microstructure. When

encountering inclusions they often incorporate the inclusions’ content which then is

distributed along the grain boundary (Fig. 5.10). Behind the moving boundary an

inclusion-free region is common. This image sequence shows that fluid inclusions on

both sides of the boundary are able to flow into the boundary. The thickness of the

fluid-filled grain boundaries is less than 1 µm (see arrow in Fig. 5.10).

136

Figure 5.10: Image sequence showing migration of a fluid-filled grain boundary that

incorporates fluids from inclusions leaving behind a zone that is free of inclusions. Note however that the fluid-richer (thicker) part of the boundary is dragged (T= 70 °C; prior strain rate: 4·10-5 s-1).

Situations also exist, in which fluids are left behind, forming cigar-shaped tubular

inclusions which subsequently contract into spherical bubbles. In Figures 5.11 and

5.12 we focus on the spherical inclusions a and b, which are similar in size and

refraction index. Inclusion a is encountered by a grain boundary and its shape changes

while it is swept (images 002-004 in Fig. 11). Subsequently this inclusion is overrun

by a faster migrating grain boundary (images 005-007; see arrow), again without

taking up the fluids. The same grain boundary continues to migrate, but at a

significantly slower rate; now it is able to incorporate inclusion b (Fig. 5.12; images

027-030). It is interesting to mention that inclusion b is moving towards the grain

boundary before both get in contact to each other. The inclusions c are isolated

remnants of a fluid-filled grain boundary of a shrinking grain (Fig. 5.11; images 007-

009).

137

Figure 5.11: Image sequence showing a migrating grain boundary after cessation of

deformation. The initially fast migrating grain boundary slows down with time. The effect of the different migration velocity on the inclusions a and b is explained in detail in the text; note that Figure 5.12 represents the same area at a later annealing stage. Note also inclusions c, which are the isolated remnants of a shrinking grain (T = 70 °C; prior strain rate: 4·10-5s-1).

138

Figure 5.12: Image sequence showing the same migrating grain boundary as in

Figure 5.11. Here the grain boundary migrates slowly and is now able to incorporate the content of inclusion b. It leaves behind a zone that is free of inclusions. See text for discussion (T = 70 °C; prior strain rate: 4·10-5s-1).

139

The experiments also document the formation of isolated inclusions when

straightened fluid-filled grain boundaries stop to migrate and pull back into isolated

bubbles (Fig. 5.13).

Figure 5.13: Image sequence showing the contraction of a fluid film into isolated

bubbles after grain boundary has stopped (last image). The experiment was stopped at this stage, and the isolated inclusions are interpreted to become spherical after additional annealing (T = 70 °C; prior strain rate: 4·10-5s-1).

5.4 Discussion The migration of fluid-filled high angle grain boundaries is the dominant process in

all experiments. In the following we briefly discuss their appearance in transmitted

light and the microstructural evolution observed during and after deformation, before

we focus onto the role of fluids during grain boundary migration.

5.4.1 Appearance of fluid-filled grain boundaries and fluid inclusions

The appearance of fluid-filled grain boundaries and fluid inclusions is the result of the

angle light rays make with the surface between media of different optical density

(Snell’s law). The transition of light rays from the optically denser material bischofite

(refraction index n1 = ~ 1.5) into the optically less dense saturated bischofite solution

140

(n2 = ~ 1.3) is characterized by a critical angle above which the light is completely

reflected (Fig. 5.14). For this system it follows that fluid-filled grain boundaries

appear dark if the angle of incidence α is larger than ~ 62 ° (critical angle), while the

parts of the grain boundary are transparent for α smaller the critical angle. Another

possibility of transparency is given, if the fluid-filled grain boundary is oriented

parallel to the path of rays (see right hand part of Figure 5.14b). This optical effect is

also responsible for the appearance of fluid inclusions in a focused plane. While the

center on which the light rays reach the surface under α < ~ 62 ° are transparent, the

outer regions with α being steeper than the critical angle are visible as dark rims.

As the true inclination of the grain boundaries with respect to the path of light rays

can not be resolved, it is unsuitable to draw conclusions on the thickness of the fluid

layer. In addition reflected dark regions of unfocused planes of the sample mask the

true thickness.

Figure 5.14: Sketches to illustrate the appearance of fluid-filled grain boundaries and fluid inclusions as an effect of reflection and refraction at surfaces separating media with different optical density. a) schematic diagram showing the relationship between the angle of incidence (α) and the reflective coefficient in a situation when light rays travel from an optically denser material into a less dense one, i.e. n1 > n2 (with n1 = ~ 1.5 for bischofite and n2 = ~ 1.3 estimated for the fluid); for α > ~ 62 ° the light rays are completely reflected; b) lower part: sketch to show the traveling of light rays through bischofite grains separated by a fluid-filled grain boundary; for 0 < α < 62 °, the light rays are refracted at both solid-fluid interfaces; however, in the steeper inclined part of the grain boundary (α > 62 °) the light rays are fully reflected; the upper part sketches the appearance of this situation in transmitted light microscopy in a focused plane.

141

5.4.2 Microstructural evolution

Due to continuous deformation the dislocation density inside the aggregates increases.

This is e.g. documented by the mechanical twins from deformations experiments at

high strain rates. After reaching a critical strain dynamic recrystallization initiates by

nucleation of new grains, preferentially at pre-existing high angle grain boundaries

(Fig. 5.6). The grain boundaries of such strain-free grains migrate into the deformed

microstructure. This is driven by the stored energy of deformation due to elimination

of large amounts of dislocations and results in growth of the recrystallizing grains.

However due to continuing deformation new dislocations are generated and

accumulated also in these grains. The difference in dislocation density decreases and

so does the driving force for further growth, eventually until its cessation. The growth

can also be limited by the nucleation of further grains at the migrating grain

boundaries, which then grow into both deformed old and deformed recrystallized

grains. With continuous deformation at constant strain rate and temperature,

recrystallized grains are characterized by a specific grain size, which is estimated to

be approximately two to three times smaller than the initial grain size; however due to

the thickness of the sample we are not able to give valuable data on both initial and

recrystallized grain size.

With cessation of deformation the nucleation of new grains is stopped. Now, the

dynamically recrystallized microstructure consists of a) small, strain-free

recrystallized grains, b) larger recrystallized grains with a moderate dislocation

density (as they were deformed during growth) and c) eventually unrecrystallized

grains with high dislocation densities.

During the early stages of post-deformation the strain-free grains (a) continue to grow

into the heterogeneously deformed microstructure by the mechanism of metadynamic

recrystallziation (Fig. 5.9; images 063-071). The rate of grain boundary migration

decreases with time as these grains grow progressively either into less deformed

material or into recovering recrystallized (b) or unrecrystallized (c) grains.

This early stage of post-deformation is followed by static recrystallization until the

microstructure is completely recrystallized. During this stage the influence of grain

boundary energy increases. This is indicated by decreasing grain boundary migration

142

rates (Fig. 5.8b), successively straightened grain boundaries and triple junctions

evolving towards equilibrium angles of 120° (Fig. 5.9).

After recrystallization is completed normal grain growth takes place. The driving

force is the reduction of grain boundary area, while the triple junctions try to maintain

120° angles. Grains with more than six sides (in two dimensions) grow at the expense

of grains with less than six sides. As the grain boundary energy is significantly

smaller than the stored energy of deformation for recrystallization, grain boundary

velocities at this stage are very slow.

5.4.3 Grain boundary migration rates

The grain boundary migration rates are characterized by the temperature and

deformation conditions at the specific phases of recrystallization (i.e. dynamic,

metadynamic and static recrystallization) and grain growth.

In the metallurgical literature (e.g. Humphreys & Hatherly, 1996; Doherty et al.,

1997) dynamic recrystallization is – after the critical strain is reached – described to

be strongly dependent on strain rate and less sensitive to temperature and strain. This

is also observed in our samples that show the fastest migration rates in experiments

deformed with high strain rates.

Also during metadynamic recrystallization the prior strain rates are directly related to

fast migration rates. In the experiments deformed at a high strain rate (samples

1d and 1s) syn- and initial post-kinematic migration are characterized by the same

rates. However the subsequent static recrystallization is suggested not to be influenced

by prior strain rate, but to be more sensitive to temperature.

The grain boundary migration rates underlie sudden changes (Figs. 5.8a & b) during

both dynamic and metadynamic recrystallization which was also reported by Guillopé

& Poirier (1979) and Urai (1987). These mobility changes can occur when a

migrating grain boundary crosses a high angle boundary or if it sweeps through a

single grain with regional differences in dislocation density. They can be caused by

heterogeneous distribution of dislocations and/or by recovery processes (dynamic and

static) when dislocations start to accumulate in subgrain boundaries.

Besides the effect of stored energy on migration rates, the effect of fluids inside the

grain boundaries has to be considered as their presence raises the grain boundary

mobility and thus also enhance the influence of the orientation-dependence on grain

boundary migration rate (e.g. Urai, 1987).

143

Our experiments show that during grain boundary migration inclusions are

incorporated, left behind or even overrun. In the following we describe a simple

model of the evolution of a migrating fluid-filled grain boundary.

5.4.4 Evolution of a migrating fluid-filled grain boundary

Consider a recrystallizing grain that nucleates at an old fluid-filled high angle

boundary and subsequently grows into the deformed microstructure under static

conditions. The grain boundary contains fluids that are derived from the old grain

boundary and migrates by dissolution of the deformed grain, diffusion through the

fluid layer and precipitation on the recrystallizing grain. As the migration is driven by

the difference in dislocation density, its rate is very high in the beginning as the new

grain is still free of dislocations. This often results in sweeping of the fluid-filled grain

boundary over an inclusion without incorporation (Fig. 5.15a-c). It is proposed that

the surface energy of the inclusion exceeds the dragging force of the fast migrating

boundary resulting in the maintenance of a spherical bubble that is now present inside

the recrystallizing grain. With continuing recrystallization the boundary migrates at a

successively decreased rate as recovery process are proposed to result in a decreasing

difference in dislocation density. It takes up the inclusion’s fluid content but only for a

short period (Fig. 5.15d-e). This is due to the still high migration rate that does not

allow the additional fluid to distribute laterally. This results in local thickening of the

fluid layer and hence slowing down of the diffusion-driven migration. The fluid-filled

bulge is dragged by the boundary (Zener drag) until it is cut off and left behind. While

the inclusion contracts immediately into its low energy configuration, the boundary is

released from the dragging force and continues to migrate at a higher rate again. Such

cigar-shaped left-behind inclusions were also reported by Urai (1987). With

subsequent grain boundary migration at slow rates, further incorporated fluids are able

to be distributed laterally and dragged bulges occur less frequently (Fig. 5.15f). This

may explain the often observed fluid-inclusion-free regions of recrystallized grains

behind slow migrating grain boundaries.

144

Figure 5.15: Schematic illustration of the effect of the grain boundary migration rate

on the behavior of encountered fluid inclusions (compare with Figs. 5.11 & 5.12): a) a deformed grain contains inclusions of similar size and shape, and is consumed by a strain-free grain at a successively decreasing grain boundary migration rate indicated by the size of the arrows (b-f); in c) the highly mobile fluid-filled grain boundary runs over the fluid inclusion, whereas at lower rates in d) the incorporated inclusion’s content drags the grain boundary until it is cut off and left behind; however in e) the grain boundary migrates at a rate that allows the fluids to redistribute laterally in to the grain boundary. The stippled circles represent the former position of the inclusions.

In the absence of a driving force (equilibrium conditions) grain boundary migration

ceases (Fig. 5.16). Subsequently the fluid layer in the grain boundary contracts into an

array of isolated inclusions.

Figure 5.16: Schematic illustration of the reorganization of fluids in a microstructure

that evolves towards equilibrium conditions. After the grain boundaries are straightened and hence further driving forces are absent, the fluid-filled grain boundaries contract into isolated inclusion that arranged along the fluid-free grain boundaries (compare with Fig. 5.13).

145

5.5 Conclusions • Our transmitted light experiments document the development and migration of

fluid-filled grain boundaries in recrystallizing wet bischofite during and after

deformation.

• Formation of the fluid-filled grain boundaries occurs only at existing high

angle grain boundaries during deformation.

• During grain boundary migration fluid inclusions are swept, incorporated

and/or left behind. This depends on grain boundary velocity, the thickness of

the fluid-filled grain boundary and the size and shape of the fluid inclusions.

• We illustrate direct evidence of the contraction of fluid-filled grain boundaries

into isolated inclusions after attaining equilibrium conditions (grain boundary

migration stops).

146

5.6 Appendix

5.6.1 Design of the deformation apparatus

The newly developed see-though deformation apparatus (Fig. 5.1) follows the design

of the Urai-rig (1987), but includes a controlled pore fluid system that allows fluid

pressures up to 30 MPa. The design of the apparatus is shown in Figure 5.A1.

Figure 5.A1: Construction drawing of the essential parts of the deformation

apparatus (lower part: side view; upper part: view from top).

It consists of a pressure vessel (Fig. A2) equipped with high-strength see-through

windows at the top and bottom (cover glass plates). The windows are sealed with o-

rings and the top window is held in place by a precision nut. The assembly is heated

with coils outside the pressure vessel and mantled with insulation material. The

temperature is controlled by a thermocouple.

147

Figure 5.A2: The deformation cell with the sandwiched sample assembly. The cell interior is a sandwich of glass plates, sample holder and sample (Fig. A3).

The inner glass plates have slots on one side which prevent buckling of the piston.

The sample assembly consists of a stainless steel holder and a pin guide that acts as a

forcing block, (both with a thickness of 300 µm). The piston is connected to a

constant-speed step motor and thus serves as moving σ1 piston. It is internally

compensated, i.e. that the loading ram moves into the cell without change of fluid

volume in the cell (Tullis & Tullis, 1986). The opposite piston finely regulates the

fluid pressure.

Although in this design the fluid pressure does not have an effect on the effective

stress, control of the fluid pressure is necessary because the dihedral angle α is a

function of fluid pressure (Holness & Lewis, 1997) and the formation of gas bubbles

in the pore fluid must be suppressed. Suppression of gas bubbles is important since

they pin the grain boundaries and thus affect the grain boundary mobility.

An optical invertoscope with long working distance objectives is used to allow

observations of the experiments. To guarantee continuous investigation of the same

location over days, the microscope is equipped with an image recording system

equipped with a digital camera that takes high resolution images at specific time

intervals.

148

Figure 5.A3: Schematic illustration of the sample assembly. The sample is enclosed

in the sample holder, that is sandwiched between pairs of notched and cover glass plates. The notches allow the loading ram to enter the cell interior.

Acknowledgements

The authors are very grateful to F.-D. Scherberich (Institute for Crystallography,

RWTH Aachen) for constructing the deformation cell and acknowledge C. Hilgers

and F.-D. Scherberich for the many discussions and their essential help in designing

the machine. This project is funded by the Deutsche Forschungsgemeinschaft (UR

64/4-1).

5.7 References Bauer, P., Palm, S. & Handy, M. R. 2000. Strain localization and fluid pathways in

mylonite; inferences from in situ deformation of a water-bearing quartz analogue (norcamphor). Tectonophysics 320(2), 141-165.

Burrows, S., Humphreys, J. & White, S. 1979. Dynamic recrystallization. A comparison between magnesium and quartz. Bulletin de minéralogie 102, 75-79.

den Brok, B., Zahid, M. & Passchier, C. 1998. Cataclastic solution creep of very soluble brittle salt as a rock analogue. Earth and Planetary Science Letters 163(1-4), 83-95.

149

den Brok, B. & Morel, J. 2001. The effect of elastic strain on the microstructure of free surfaces of stressed minerals in contact with an aqueous solution. Geophysical Research Letters 28(4), 603-606.

den Brok, B., Morel, J. & Zahid, M. 2002. In situ experimental study of roughness development at a stressed solid/fluid interface. In: Geological Society Special Publications (edited by de Meer, S., Drury, M. R., de Bresser, J. H. P. & Pennock, G. M.) 200. Geological Society of London : London, United Kingdom, United Kingdom, 73-83.

Doherty, R. D., Hughes, D. A., Humphreys, F. J., Jonas, J. J., Juul Jensen, D., Kassner, M. E., King, W. E., McNelley, T. R., McQueen, H. J. & Rollett, A. D. 1997. Current issues in recrystallization: a review. Material Science and Engineering A 238, 219-274.

Guillopé, M. & Poirier, J. P. 1979. Dynamic recrystallization during creep of single-crystalline halite: An experimental study. Journal of Geophysical Research 84, 5557-5567.

Herwegh, M., Handy, M. R. & Heilbronner, R. 1997. Temperature- and strain-rate-dependent microfabric evolution in monomineralic mylonite; evidence from in situ deformation of norcamphor. Tectonophysics 280(1-2), 83-106.

Hickman, S. H. & Evans, B. 1991. Experimental pressure solution in halite; the effect of grain/interphase boundary structure. Journal of the Geological Society 148, 549-560.

Holness, M. B. & Lewis, S. 1997. The structure of the halite-brine interface inferred from pressure and temperature variations of equilibrium dihedral angles in the halite-H2O-CO2 system. Geochimica et Cosmochimica Acta 61(4), 795-804.

Humphreys, F. J. & Hatherly, M. 1996. Recrystallization and related annealing phenomena. Pergamon, 497pp.

Jessell, M. W. 1986. Grain boundary migration and fabric development in experimentally deformed octachloropropane. Journal of Structural Geology 8(5), 527-542.

Koehn, D., Dysthe, D. K. & Jamtveit, B. 2004. Transient dissolution patterns on stressed crystal surfaces. Geochimica et Cosmochimica Acta 68(16), 3317-3325.

Means, W. D. 1989. Synkinematic microscopy of transparent polycrystals. Journal of Structural Geology 11(1-2), 163-174.

Morel, J. 2000. Experimental investigation into the effect of stress on dissolution and growth of very soluble brittle salts in aqueous solution. PhD thesis, Johannes Gutenberg-Universität, Mainz, Germany, 117pp.

Peach, C. J., Spiers, C. J. & Trimby, P. W. 2001. Effect of confining pressure on dilatation, recrystallization, and flow of rock salt at 150°C. Journal of Geophysical Research 106(B7), 13,315-13,328.

Ree, J. H. & Park, Y. 1997. Static recovery and recrystallization microstructures in sheared octachloropropane. Journal of Structural Geology 19(12), 1521-1526.

Ree, J.-H. 1994. Grain boundary sliding and development of grain boundary openings in experimentally deformed octachloropropane. Journal of Structural Geology 16(3), 403-418.

Rutter, E. H. 1976. The kinetics of rock deformation by pressure solution. Philosophical Transactions of the Royal Society of London 283(1312), 203-217.

Schenk, O. & Urai, J. L. 2004. Microstructural evolution and grain boundary structure during static recrystallization in synthetic polycrystals of Sodium Chloride

150

containing saturated brine. Contributions to Mineralogy and Petrology 146(6), 671-682.

Schenk, O., Urai, J. L. & Piazolo, S. submitted. Structure of grain boundaries in wet, synthetic polycrystalline, statically recrystallizing halite – evidence from cryo-SEM observations. Geofluids.

Spiers, C. J., Schutjens, P. M. T. M., Brzesowsky, R. H., Peach, C. J., Liezenberg, J. L. & Zwart, H. J. 1990. Experimental determination of constitutive parameters governing creep of rocksalt by pressure solution. In: Deformation mechanisms, rheology and tectonics (edited by Knipe, R. J. & Rutter, E. H.) 54, Leeds, United Kingdom, 215-227.

Srolovitz, D. J. 1989. On the stability of surfaces of stressed solids. Acta Metallurgica 37(2), 621-625.

Tullis, T.E. and J. Tullis, 1986. Experimental rock deformation techniques. In: Mineral and rock deformation; laboratory studies; the Paterson volume. AGU Geophysical Monograph (edited by Hobbs, B. E. & Heard, H. C.) 36, 297-324.

Tungatt, P. D. & Humphreys, F. J. 1981. An in-situ optical investigation of the deformation behaviour of sodium nitrate - an analogue for calcite. Tectonophysics 78, 661-675.

Tungatt, P. D. & Humphreys, F. J. 1984. The plastic deformation and dynamic recrystallization of polycrystalline sodium nitrate. Acta metallurgica 32(10), 1625-1635.

Urai, J. L., Humphreys, F. J. & Burrows, S. E. 1980. In-situ studies of the deformation and dynamic recrystallization of rhombohedral camphor. Journal of Materials Science 15, 1231-1240.

Urai, J. L. & Humphreys, F. J. 1981. The development of shear zones in polycrystalline camphor. Tectonophysics 78(1-4), 677-685.

Urai, J. L., Means, W. D. & Lister, G. S. 1986a. Dynamic recrystallization of minerals. In: Mineral and rock deformation; laboratory studies; the Paterson volume. AGU Geophysical Monograph (edited by Hobbs, B. E. & Heard, H. C.) 36, 161-199.

Urai, J. L., Spiers, C. J., Zwart, H. J. & Lister, G. S. 1986b. Weakening of rock salt by water during long-term creep. Nature (London) 324(6097), 554-557.

Urai, J. L. 1987. Development of microstructure during deformation of carnallite and bischofite in transmitted light. Tectonophysics 135(1-3), 251-263.

Walte, N. P., Bons, P. D., Passchier, C. W. & Koehn, D. 2003. Disequilibrium melt distribution during static recrystallization. Geology 31(11), 1009-1012.

Watanabe, T. & Peach, C. J. 2002. Electrical impedance measurement of plastically deforming halite rocks at 125°C and 50 MPa. Journal of Geophysical Research 107(B1), ECV 2-1 - 2-12.