the interaction of ﬂuid inclusions and migrating grain ... · ethanol, focusing on the ......

The interaction of fluid inclusions and migrating grain boundariesin a rock analogue: deformation and annealing of polycrystallinecamphor–ethanol mixtures

J . SCHMATZ AND J. L . URAIStructural Geology, Tectonics and Geomechanics, Geological Institute, RWTH Aachen University, Lochnerstrasse 4-20, 52056Aachen, Germany ([email protected])

ABSTRACT The results of deformation and annealing experiments using a rock analogue containing a liquid phaseas observed in transmitted light microscopy are presented. Samples were made of camphor–ethanolmixtures, with the polycrystalline solid being a rhombohedral phase and the liquid a saturated solutionof camphor in ethanol. Results show the in situ pore fluid morphology during grain boundary migrationrecrystallization. Samples were deformed at high homologous temperatures (Th � 0.7) and strain ratesof 1 · 10)6 to 9 · 10)4 s)1. For grain boundary migration rates ranging from 10)10 to 10)6 ms)1 Zenerpinning, not only the drag and drop of fluid inclusions by migrating grain boundaries but also thepassage of grain boundaries over fluid inclusions without noticeable interaction was observed. The drag-limiting migration rate is three times lower for post-kinematic annealing than for dynamicrecrystallization and shows a weak dependence on fluid inclusion size. A detailed description ispresented of fluid inclusion–grain boundary interaction with respect to fluid inclusion sizes and grainboundary migration rates. Fluid inclusions show up to 39% reduction in diameter after passage of agrain boundary, indicating fluid flow along the mobile grain boundary.

Key words: fluid inclusion morphology; fluid transport; grain boundary migration recrystallization; grainboundary wetting; see-through deformation experiments.

INTRODUCTION

In Earth Sciences fluid inclusions are studied toreconstruct temperature and pressure conditions dur-ing metamorphism or basin formation (Boullier, 1999).Composition and density of the fluid are used tounderstand the circulation of fluid in the Earth�s crust.Detailed understanding of the processes of formationand alteration of fluid inclusions or melt pockets canbe valuable in applied studies of ore deposition(Roedder, 1984; Kolb et al., 2000; Kolb & Meyer,2002) or diagenesis of sedimentary basins (McLimans,1987; Schoenherr et al., 2009). Grain boundary–fluidinteraction can have a significant effect on the evolu-tion of texture and mechanical properties (Evans et al.,2001; Renner et al., 2002; Herwegh & Berger, 2004;Mancktelow & Pennacchioni, 2004; Jing et al., 2005).The microstructural evolution of polycrystalline rockaggregates during metamorphism is highly dependenton pressure, temperature, composition and morpho-logy of the pore fluid. These parameters affect grainboundary structure during recrystallization (Renneret al., 2002) but a migrating grain boundary can inturn affect the volume and distribution of fluid inclu-sions (Olgaard & FitzGerald, 1993).

There are four main classes of fluid inclusions,formed by different processes and producing distinct

microstructures (Roedder, 1984) (Fig. 1). Primaryinclusions (i) form in growth bands during growth ofcrystals in a free fluid (Hilgers & Urai, 2002; Passchier& Trouw, 2005; Schleder & Urai, 2005; Warren, 2006;Schoenherr et al., 2007) (Fig. 1a). Secondary fluidinclusions can be (ii) healed microcracks (Renardet al., 2002; Schoenherr et al., 2007) (Fig. 1b), or (iii)inclusions dropped from a fluid-filled, moving grainboundary (Urai, 1983; Olgaard & FitzGerald, 1993;Schenk & Urai, 2005) which may migrate over fluidinclusion arrays in ghost grain boundaries, or (iv) beremnants of a grain boundary which contained a fluidfilm (Urai et al., 1986) (Fig. 1c). Healed microcracksare preserved under low temperature (Passchier &Trouw, 2005) whereas the other mechanisms in theformation of secondary fluid inclusions require mobilegrain boundaries (Urai et al., 1986; Hirth & Tullis,1992). Interpretation of these fluid inclusions is oftenproblematic (Drury & Urai, 1990). A small amount offluid can increase the mobility of grain boundaries(Urai, 1985; Spiers et al., 1986; Urai et al., 1986;Schenk & Urai, 2004), because the diffusivity in a fluidphase is dramatically higher than in a dry grainboundary (Peach et al., 2001).

In the absence of deformation, grain-scale fluid flowis dependent on the geometry and connectivity ofpores. This in turn is controlled by the wetting angle

J. metamorphic Geol., 2010, 28, 1–18 doi:10.1111/j.1525-1314.2009.00849.x

� 2009 Blackwell Publishing Ltd 1

(e.g. Smith, 1964; Holness & Lewis, 1997; Holness &Siklos, 2000; Walte et al., 2003). In theory, underequilibrium conditions a wetting angle greater than 60�leads to isolated fluid inclusions on grain boundariesand in triple junctions while a wetting angle smallerthan 60� forms interconnected fluid channels alongtriple junctions. Walte (2005) showed that surfaceenergy-driven grain boundary migration recrystalliza-tion leads to a dynamic wetting angle which is up to10% smaller than the equilibrium one.

Urai et al. (1986) proposed that the rate of grainboundary migration is dependent on the thickness ofthe fluid film and that the incorporation of fluidinclusions may lead to a local thickening of the grainboundary. This process may induce Zener drag, whichreduces the grain boundary mobility (Humphreys &Hatherly, 1996; Renner et al., 2002). Schenk & Urai(2005) observed that for different migration rates fluidinclusions can be (i) swept without incorporation, (ii)incorporated and distributed along the migratingboundary, (iii) dragged through the material by themigrating boundary and (iv) dropped or left behind bya migrating grain boundary. A similar situation ispresent in many ceramics where during sintering fluidinclusions or pores are dragged by migrating bound-aries, or the boundaries may be separated from theinclusions dependent on grain boundary velocity andfluid inclusion mobility (Hsueh et al., 1982; Svoboda &Riedel, 1992). For surface energy-driven grainboundary migration recrystallization the velocity of agrain boundary (vGB) is proportional to its curvaturewhereas the maximum fluid inclusion velocity (vFI) is,to a first approximation, inversely proportional to size(Olgaard & FitzGerald, 1993). For vGB = vFI = 0 thegrain boundary is said to be pinned, for vGB ‡ vFI > 0the fluid inclusion is said to be dragged, and forvGB >> vFI the fluid inclusion is said to be dropped(these definitions are used throughout this paper).Many studies assume surface diffusion to be ratecontrolling during drag (e.g. Hsueh et al., 1982;Svoboda & Riedel, 1992; Petrishcheva & Renner,2005), especially at high homologous temperatures(Chuang et al., 1979).

As the in situ deformation of wet rocks at the grainboundary scale cannot be studied easily under labo-ratory conditions, it is useful to study model mate-rials using transmitted light deformation (e.g. Urai &

Humphreys, 1981; Means, 1983; Urai, 1983; Schutj-ens, 1991; Bons, 1993; Streit & Cox, 2000; Walteet al., 2003; Schenk & Urai, 2005). In these modelsthe liquid phase is made up of saturated solutionswhich can be inorganic (e.g. Schenk & Urai, 2005),or organic solvents (e.g. Walte et al., 2003; Walte,2005) or non-solvents (e.g. Bauer et al., 2000; Walteet al., 2007). Quantitative scaling of the results ofsuch experiments to natural prototypes is possible forthe case of isomechanical groups (sodium nitrate –calcite: Tungatt & Humphreys, 1981; Urai & Jessell,2001), for other materials the comparison is lessdirect and mainly focused on qualitative study of theevolution of microstructural processes. In this studyan analogue model is used to characterize the inter-action of migrating grain boundaries with fluidinclusions of a saturated solution of camphor inethanol, focusing on the detailed observation of fluidinclusion morphology and grain boundary migrationrates.

METHODS

Materials and sample preparation

The model materials used were polycrystalline cam-phor (C10H16O) and ethanol (C2H5OH). A commer-cially available camphor (Roth 6155.1) was used with apurity of >97% and Tm = 177–179 �C. Camphor isrhombohedral at room temperature with a transitionto a cubic phase at 92 (± 7) �C (Urai & Humphreys,1981). It was mixed with ethanol (Roth PO 76.2,purity: 99.5%). To prepare the sample material amixture of 5 ml ethanol and a few mg of SiC (grainsize = 13 lm) was heated to 80 �C in a closed, stirredglass phial, c. 0.2 mol camphor added and stirred for30 min before cooling to room temperature. Thisproduced a camphor mush in saturated solution.Solubility was measured to be 0.8 mol camphor ⁄ 100 gethanol at 24 �C with a linear increase to1.6 mol ⁄ 100 g ethanol at 60 �C.

Experimental set-up

The see-through deformation apparatus follows thedesign of Urai et al. (1987) and Schenk & Urai (2005)(Fig. 2). It consists of a stainless steel vessel equipped

(a) (b) (c)

(i)

(ii)

Fig. 1. Sketch showing types of fluid inclu-sion arrays: (a) fluid inclusions formed dur-ing crystal growth from a supersaturatedsolution, (b) plane with fluid inclusions in ahealed microcrack, (c) (i) fluid inclusionsdropped from a migrating grain boundary,(ii) fluid inclusion array in a equilibratedghost grain boundary. Arrow indicatesmigration direction of the grain boundary.

2 J . S CH M A TZ & J . L . U R A I

� 2009 Blackwell Publishing Ltd

with high-strength see-through windows. The first stepin setting up an experiment was to assemble the lowerspacer, the r1-piston, the piston guide and the lowerglass plate in the deformation rig. All grooves atmetal–glass contacts were carefully sealed with siliconepaste (Bayer-Silicone, Baysilone-paste). The next stepwas to place the moving r1-piston and the sampleholder (100 lm thick steel) into the cell. A 1-mm-widesilicone viscous jacket was formed with a rectangularcentral space, and this was filled with the camphor–ethanol mixture (�200 lm thick) (Fig. 3). The assem-bly was closed with the upper glass plate and the upperpiston guide before inserting the upper distance pieceand the high precision ball bearing. The precision nutwas tightened to compress the sample into a wafer of c.0.1 · 24 · 24 mm.

The assembly was heated with coils outside thepressure vessel inside a chamber of insulation material.Temperature was controlled by a thermocouple and aPID controller. The temperature range was between 25and 200 �C. The piston was connected to a constantspeed step motor and thus served as a r1-piston(cf. Schenk & Urai, 2005). An optical invertoscopewith long-distance objectives was used to allow in situobservations of the experiment. The invertoscope isequipped with a high-resolution digital camera thatrecords 2048 · 1536 px images at specific time inter-vals (15–900 s) with a pixel size of 1 px = 0.9 lm. Theimages were converted to movies.

Measurements

Strain measurements were carried out using markerparticles which do not affect the grain boundaries(Means et al., 1980; Jessell, 1986; Bons et al., 1993). Ifthe positions of three marker particles are knownbefore and after deformation, the position gradienttensor can be calculated assuming that the deformation

was homogeneous in the triangle (Jessell, 1986; Bonset al., 1993). The method has been used to determinethe approximate bulk strain and strain rate.

10 mm

s1 -piston Precision nut

High precisionball bearing

Notched glass plates

Heating block

Piston guide

Moving s1 -piston

Distance piece

"Sandwiched" sample

Distance piece Silicone jacket

Strain ring

Precision nut

Moving s1-piston

s 1-piston

10 mm

Fig. 2. Construction drawing and photo-graphs of the deformation apparatus. Upperpart-top view, lower part-side view. See textfor set-up details.

1

2

3

4

5

7

8

s 1

6

Fig. 3. Sketch showing set-up of deformation cell interior, with(1) lower piston guide, (2) lower glass plate, (3) sample holder,(4) moving r1-piston, (5) silicone jacket, (6) sample, (7) upperglass plate and (8) upper piston guide. Note that a distance pieceof the same thickness can be placed opposite to the movingr1-piston to increase finite strain. See text for details.

F L U I D I N C L U S I O NS AN D M O B I L E G R A I N B O U N D AR I E S 3


The sample thickness of 100 lm was chosen to allowobservation of large grain boundary areas with fluidinclusions up to 30 lm in radius inside the sample. Allmeasurements on grain boundary velocities were madealong orthogonal trajectories in the plane of the coverglass (see also Urai, 1983; Schenk & Urai, 2005). Thesize of the fluid inclusions was measured in plane offocus, focusing approximately at the centre of theinclusion. Measurements were made only in grainboundaries inclined more than 80� to the planeof section (see Appendix A1 for more details onmeasurements).

Experimental protocol

In each experimental set-up, the sample was firstheated to 105 �C for 60 min, and it was verified thatthe material was of cubic crystallography. In this sin-tering period, grains grew from 100 to 300 lm, andpart of the ethanol was removed from the sample(presumably by diffusion through the silicone). Inaddition, the process of grain growth moved part ofthe liquid to the samples� edge. At the end of this stagethe liquid fraction was predominantly distributed ongrain boundaries. The temperature was then reducedslowly (�0.5 �C min)1) to the desired temperature of50 �C. The phase transition to rhombohedral crystalsproduced elongated, up to 3000 lm size grains (cf.Urai & Humphreys, 1981) and fluid inclusions wereincorporated into the rhombohedral grains. After 120more minutes of annealing a grain size of 700–1000 lmwas reached and fluid inclusions were located on thegrain boundaries and inside the grains. The radius ofthe fluid inclusions varied from 0.5 to 30 lm. The endof this stage was the start of annealing or deformationexperiments. The liquid fraction was �10%, it wasreduced progressively to �2% at the end of theexperiment (see Appendix A2 for analysis of thisprocess).

Wetting angle measurement

Fluid inclusions inside grains were approximatelyspherical and without facets under equilibrium condi-tions, while they adopted an elliptical shape duringdeformation. The shape of fluid inclusions that wereattached to grain boundaries was controlled by thewetting angle. It is a common technique to measure theequilibrium wetting angle w of a solid–liquid interfacein triple junctions (Fig. 4a,c) (e.g. Walte et al., 2003;

ψ

200 µm

(a)

2

100 µm

(b)

Leading surface

Trailing surface

Equilibrium Dynamic

(c)

ψ

2ψ

2ψ

2ψ 2ψ

ψ

Fig. 4. Photographs showing wetting angles observed in theexperiments with (a) equilibrium wetting angle and (b) dynamicwetting angle in the camphor ethanol system. In (b) the directionof the migrating grain boundary is indicated with a double-arrow (Movie S1, experiment ce01 08, Table 1). (c) Sketchshowing measuring method for equilibrium and dynamic wettingangles. Note a low wetting angle at the leading surface of thefluid inclusion and a high wetting angle at the trailing surface.



Wark et al., 2003). Neglecting surface energy anisot-ropy the wetting angle can be described by the relation

w ¼ 2 arccoscss2csl

� �

with css-solid–solid and csl-solid–liquid being the inter-facial energies. The equilibrium wetting angle of24 ± 10� (n = 22) was measured for the camphor–ethanol system at a homologous temperatureTh = T ⁄Tm of 0.72. For fluid inclusions being draggedby a grain boundary, wetting angles were measured asshown in Fig. 4b,c. The wetting angle at the leadingsurface is usually smaller than the wetting angle at thetrailing surface (Fig. 4b,c & Movie S1). The dynamicwetting angle that was determined in the camphor–ethanol systemwas 27 ± 9� (n = 163) atTh of 0.72 and0.76. As expected for this wetting angle the pore fluidappeared as isolated inclusions at the grain boundaries.

Rates and temperatures

Ideally the development of grain boundary structuresis studied in a coarse-grained, polycrystalline materialas it provides long, slightly curved grain boundaries.Consequently relatively high temperatures (Th = 0.72and 0.76) were chosen to generate such a texture. Aftersintering and cooling the samples were first annealed at50 �C (stage 1, Table 1) or 70 �C (stage 2, Table 1),respectively, and then deformed at a constant tem-perature (50 or 70 �C) with strain rates between1 · 10)6 (low) and 9 · 10)4 s)1 (high) (stage 3,Table 1) and then post-kinematically annealed at 50 or70 �C respectively (stage 4, Table 1). All stages weredocumented and analysed with respect to grain size,grain boundary structure, grain boundary migrationrates, fluid inclusion size, distribution, behaviour andmicrostructural processes.

RESULTS

Microstructural processes

During sintering and annealing (stages 1 & 2) thedominant microstructural process was grain growth.Large grains grew at the expense of smaller grains whilereducing grain boundary curvature. At both tempera-tures, deformation (stage 3) led to dynamic grainboundary migration recrystallization accompanied bydynamic grain growth for low strain rates and grain sizereduction for high strain rates. During dynamicrecrystallization the grain boundaries were lobate andthe grain size was highly variable. The dominantmicrostructural process during post-kinematic anneal-ing (stage 4) was recovery (we observed mobile subgrainboundaries) accompanied by grain growth. Note thatthese experiments were performed at much lower strainrates than those by Urai et al. (1980) and Urai &Humphreys (1981), and grains grew instead of sizereduction during dynamic recrystallization.

Fluid inclusion–grain boundary interaction

Evolution of morphology

In all experiments, fluid inclusions were observed to bepassed by grain boundaries without any noticeableinteraction. This tended to be the case for high-velocitygrain boundaries; a quantitative analysis will be pre-sented in the Discussion. However, in all experiments,grain boundaries (especially slower-moving ones) wereobserved to interactwithfluid inclusions. Inwhat follows,a number of examples of this interaction are described.Fluid inclusions were left behind by grain boundaries inall experiments, in some cases from parts of a grainboundarywhich contained a visible fluid inclusion, but inother cases the source of the fluid was not readily visible.The process was very similar during annealing, dynamicrecrystallization and post-kinematic annealing, so thatthe descriptions below are quite general.

Mobile, fluid-filled triple junctions could drop large,flat fluid inclusions which later necked down toisolated, spherical inclusions (Fig. 5a; experiment ce0112, Table 1 & Movie S2). In short grain boundarysegments between triple junctions, fluid inclusions wereleft behind by the boundaries of a shrinking grain. Inthese cases, all boundaries of the shrinking grain con-tained fluids but only the fastest boundary dropped thefluid inclusions (Fig. 5b; experiment ce01 16, Table 1 &Movie S3). When a grain boundary containing anarray of elongated fluid inclusions moved away fromthese inclusions, these formed a ghost grain boundarywhich then broke up into an array of spherical fluidinclusions (Fig. 5c; experiment ce01 15, Table 1 &Movie S4). However, grain boundaries far from triplejunctions were also frequently observed to leave behindarrays of elongated fluid inclusions, even when noapparent source of the fluid was visible in the grainboundary (although the concave shape of the grainboundary suggests pinning). These fluid inclusions thencontracted into a spherical shape (Fig. 5d; experimentce03 01, Table 1 &Movie S5; cf. Schenk & Urai, 2005).

The evolution of fluid inclusion morphology duringinteraction with a grain boundary was also dependenton boundary velocity. An example is given in Fig. 6 forthree different fluid inclusions with the same radius(c. 25 lm), interacting with a migrating grain bound-ary which had a velocity that varies over three ordersof magnitude (a: 5 · 10)7, b: 9 · 10)8 and c: 2 ·10)10 m s)1). The fluid inclusion was not affected bythe fast migrating grain boundary (Fig. 6a &Movie S6); it was elongated by the slower boundary,re-equilibrating after interaction (Fig. 6b &Movie S7);and it necked down during interaction with the slowlymigrating grain boundary (Fig. 6c & Movie S8).

Overview of grain boundary–fluid inclusion interaction

All observations are summarized in Figs 7 & 8. Twoprincipal modes of interaction (Fig. 7) were observed.



The first mode concerns the change in shape of thefluid inclusion (change or no change), and the secondmode concerns the motion of the fluid inclusion andthe grain boundary [pass without moving the inclusion,pass with moving the inclusion (drag) and stopped by

the inclusion (pinning; Fig. 8)]. In this paper the term�no interaction� is used for the case that neitherposition nor shape changes when a grain boundarypasses a fluid interaction. However, it is noted that thedetection of both change in shape and motion of a fluid

Table 1. Experimental conditions.

Experiment Image

sequence

T (�C) Piston

velocity

(m s)1)

Strain

rate (s)1)

Duration

(min)

Strain

(%)

Finite

strain (%)

Modus Process Figure ⁄movie

ce01 Stage 2 01 50–70 0 0 120 0 0 Annealing Annealing

Stage 3 02 70 1.4 · 10)06 8.7 · 10)04 40 15 15 Dynamic GBM Rx

03 1.4 · 10)06 4.6 · 10)04 27 5 20 GBM Rx

04 1.4 · 10)06 6.8 · 10)04 36 11 31 GBM Rx

05 1.4 · 10)06 1.6 · 10)04 17 1 32 GBM Rx

06 2.7 · 10)07 6.4 · 10)05 58 2 34 GBM Rx

07 2.7 · 10)07 4.5 · 10)04 68 13 47 GBM Rx

08 2.7 · 10)07 4.3 · 10)06 68 0 47 GBM Rx ⁄ dyn GG 4b ⁄ S109 2.7 · 10)07 2.9 · 10)04 92 11 58 GBM Rx ⁄ dyn GG

10 2.7 · 10)07 3.2 · 10)06 144 0 58 GBM Rx ⁄ dyn GG

11 0 0 90 0 58 Post-kin. annealing GG

12 1.4 · 10)06 1.6 · 10)05 328 2 60 Dynamic GBM RX 5a ⁄ S213 4.0 · 10)02 2.3 · 10)04 4 0 61 Bulging Rx

Stage 4 14 70 0 0 1 0 61 Post-kin. annealing Recovery

15 4 Recovery 5c ⁄ S416 50 GG 5d ⁄ S517 47 GG

18 116 GG

19 802 GG

20 1318 GG

ce02 Stage 2 01 50–70 0 0 66 0 0 Annealing GG 6a, 6b, 10a ⁄S6, S7, S9

02 70 67 GG


04 2.7 · 10)07 4.1 · 10)04 70 12 12 GBM Rx

05 2.7 · 10)07 3.8 · 10)06 504 1 5 GBM Rx

ce03 Stage 3 01 50 2.7 · 10)07 5.0 · 10)05 1700 36 36 Dynamic GBM Rx 5b ⁄ S302 50 2.7 · 10)07 1.6 · 10)06 1082 1 37 GBM Rx

Stage 4 03 50 0 0 46 0 37 Post-kin. annealing GG

04 150 GG

05 70 57 GG

06 86 GG

07 100 GG

08 86 GG

Stage 3 (continued) 09 70 1.4 · 10)07 2.6 · 10)06 33 0 37 Dynamic

10 1.4 · 10)07 2.6 · 10)06 883 1 38

GBM Rx ⁄ dyn GG 14 ⁄ S10ce05 Stage 1 01 50 0 0 5480 0 38 Annealing Recovery ⁄GG

02 110 GG

03 1320 GG


05 2.7 · 10)07 1.0 · 10)05 157 1 1 GBM Rx

06 2.7 · 10)07 1.2 · 10)06 193 0 0 GBM Rx

07 2.7 · 10)07 3.9 · 10)05 999 17 55 GBM Rx

Stage 4 08 70 0 0 51 0 55 Post-kin. annealing Recovery ⁄GG

09 42 GG

10 126 GG

11 666 GG

12 5540 GG

13 275 GG

14 24 GG

15 84 GG

16 5000 GG

17 920 GG

18 296 GG

ce08 Stage 1 01 50 0 0 750 0 0 Annealing GG

02 1500 GG

03 1440 GG

04 2790 GG 6c ⁄ S805 8655 GG 6c ⁄ S806 3825 GG 6c ⁄ S807 9440 GG 6c ⁄ S808 16490 GG

GMB Rx, dynamic grain boundary migration recrystallization; dyn GG, dynamic grain growth; bulging Rx, bulging recrystallization; GG, grain growth.



inclusion is dependent on the resolution of the micro-scope and the digital images. Therefore, a more accu-rate definition of no interaction is the absence ofchange in shape or in position at the resolution of ourimaging system, which is around 1 lm. It is also clearthat the interaction is mutual: migrating grainboundaries affect fluid inclusions and grain boundaries

can be affected by fluid inclusions. In Fig. 8 this issimplified and the grain boundaries are shown asstraight lines.

Change in shape occurs in two regimes. In the first(regime 1) the shape of the fluid inclusions was notaffected by the migrating grain boundary; whereasin regime 2 there was a clear change in shape. Both

t1 t2

t4

t3

100 µmt5

Dynamicrecrystallization

Δt = 180 sec.

(a)

Post-kinematicannealing

Δt = 10 min.

t1 t2

t4

t3

125 µmt5

(b)

Fig. 5. The formation of isolated fluid inclusions in the bulk crystal. (a) Behind a fast migrating triple junction a large, flat fluidinclusion is left behind and shrinks to isolated bubbles, Dt = 180 s (Movie S2, experiment ce01 12, Table 1), (b) behind a slowmigrating triple junction isolated fluid inclusions are left behind, Dt = 10 min (Movie S3, experiment ce01 16, Table 1), (c) from aghost grain boundary, elongated fluid inclusions are left behind and shrink into isolated bubbles, Dt = 80 s (Movie S4, experimentce01 15, Table 1) and (d) behind a migrating fluid-filled grain boundary, isolated fluid inclusions are left behind, Dt = 20 min(Movie S5, experiment ce03 01, Table 1).



regimes 1 and 2 occurred in each experiment,depending on the fluid inclusion size and the velocity ofthe grain boundary (see Discussion – drag-limitinggrain boundary velocity). In regimes 1 and 2, amigrating grain boundary can (a) pass the fluid inclu-sion, or (b) drag and drop the fluid inclusion, or can (c)be pinned by a fluid inclusion, or (d) incorporate thefluid inclusion into the migrating grain boundary [(d)does not occur in regime 1]. In more detail, the pro-cesses are as follows. In regime 1 the inclusions did notchange in shape for any of the processes named above,they stayed spherical while being passed (a) or dragged

(b) by a migrating grain boundary and they also kepttheir spherical shape when a grain boundary waspinned (c). In regime 2 (a) the fluid inclusions wereelongated when being passed by migrating grainboundaries; they re-equilibrated to spheres after thegrain boundary passed (e.g. Fig. 6b & Movie S7). Inregime 2 (b) the elongated inclusion necked down, andpart of the fluid remained at the migrating grainboundary (see also Fig. 6c & Movie S8). In regime 2(c) the inclusions changed shape and wetting angle (seealso Fig. 4) or contracted to an array of isolatedinclusions (cf. Schenk & Urai, 2005). A rarely observed

t1 t2

t4

t3

100 µmt5

Post-kinematicannealing

Δt = 80 sec.

(c)

t1 t2

t1

t3

100 µm

t5

Dynamicrecrystallization

Δt = 20 min.

(d)

Fig. 5. (Continued).



case was the incorporation of fluid inclusions into themigrating grain boundary [regime 2 (d)] (Schenk &Urai, 2005). Processes were similar but more complexin the case of fluid inclusion arrays interacting withgrain boundaries (Fig. 9). As a variation to regime 2a,in (a) the elongated fluid inclusion is dropped as anarray of fluid inclusions (see also Fig. 5a & Movie S2).Dropping an array of isolated bubbles from a fluidreservoir at the migrating grain boundary could gen-erate a fluid inclusion trail as shown in (b) (see alsoFig 5b,d & Movies S3, S5). In (c) a grain boundarycontaining an array of fluid inclusions started to moveand dropped the fluid inclusions. Note, this is equiv-alent to regimes 1c and 2c. Finally in (d) we show how

a fluid film contracted into isolated fluid inclusions(see also Fig. 5c & Movie S4).

Mobility control and development of drag angle

In Figs 8 & 9, for simplification, grain boundaries aredrawn as straight lines. In reality the drag angle h (theangle under which the migrating grain boundaryexceeds a drag force on the fluid inclusion) (Hsuehet al., 1982; Svoboda & Riedel, 1992) is an importantparameter. According to theory (Hsueh et al., 1982;Svoboda & Riedel, 1992), under a critical drag anglehn the grain boundary will release the fluid inclusion.In our observations, hn was highly variable and nosystematic relationship to size and velocity wasapparent.

A number of observations were made where onesingle grain boundary swept fluid inclusions of differ-ent sizes and shapes (Fig. 10a & Movie S9). In thisexample six major features were apparent (Fig. 10b).(i) When the fluid inclusion came in contact with thegrain boundary it could change its shape producing adynamic wetting angle w. (ii) Small inclusions weredragged longer distances through the material thanlarge ones at the same grain boundary velocity. (iii)The critical drag angle hn under which the grainboundary was released from the fluid inclusions ap-pears to be independent of fluid inclusion size andgrain boundary velocity. (iv) The grain boundaryvelocity appears to decrease when being in contact withmany fluid inclusions. (v) For the same fluid inclusionsize a high-velocity grain boundary has less ability todrag a fluid inclusion than a slow one. (vi) A decreasein size of the fluid inclusions was observed when beingswept by a migrating grain boundary (see also Fig. 13).

DISCUSSION

Drag-limiting grain boundary velocity

Besides the wetting angle, the two main variablesexpected to control which of the interactions fromFigs 7 & 8 occurs are grain boundary velocity and fluidinclusion size (Hsueh et al., 1982; Svoboda & Riedel,

100 µm 100 µm 100 µm t1 t2 t1 t2t1 t2

(a) (b) (c)

Fig. 6. Photographs showing examples for interactions of fluid inclusions with migrating grain boundaries. (a) A fluid inclusion ispassed by a migrating grain boundary without being affected (Movie S6, experiment ce02 01, Table 1), (b) a fluid inclusion getselongated while being passed by a migrating grain boundary and equilibrates in a later time step (Movie S7, experiment ce02 01,Table 1) and (c) drag and drop, the fluid inclusion shape is affected, fluid inclusion necks down while being attached to grain boundary,fluid reservoir remains on migrating grain boundary, equilibration of left behind fluid inclusion (Movie S8, experiment ce08 04-07,Table 1). The area reduction for the isolated inclusion was 28% in this case. See text for details.

Shape not affected Shape affected

No

drag

Drag

No FI-GB interaction FI-GB interaction

FI-GB interaction FI-GB interaction

Fig. 8regime 1a

Fig. 8regime 2a

Fig. 8regime 1b

Fig. 8regime 2b

Fig. 7. Matrix summarizing two principal modes of grainboundary–fluid inclusion interactions. The first mode(columns) concerns the change in shape of the fluid inclusion(change or no change), and the second mode (rows) concerns themotion of the fluid inclusion and the grain boundary [passwithout moving the inclusion, pass with moving the inclusion(drag)]. The field where no interaction of fluid inclusions andgrain boundaries occurs is shaded in grey. See Fig. 8 for legend.



1992; Olgaard & FitzGerald, 1993; Petrishcheva &Renner, 2005). Figure 11 (experiment ce03) shows thatfor low grain boundary velocities (�10)6 ms)1) anddifferent fluid inclusion radius, the medium-sized fluidinclusions (6 > r > 3.5 lm) were elongated whenswept by migrating grain boundaries (cf. Fig. 6b &Movie S7). The largest fluid inclusions (r > 6 lm)showed necking followed by separation from themoving grain boundary (cf. Fig. 6c & Movie S8). Verysmall fluid inclusions (<3.5 lm) stayed spherical whilebeing dragged. This is in qualitative agreement withpredictions by Hsueh et al. (1982), Svoboda & Riedel(1992) and Petrishcheva & Renner (2005). The reasonfor this is that small inclusions allow mass transferfrom the leading to the trailing surface; hence they aredragged until they separate from the grain boundaryunder a critical drag angle. For large inclusions thematerial will precipitate and form a neck beforereaching the trailing surface.According to these models the occurrence of drag

and drop is also dependent on the fluid inclusion sizeand therefore mobility. For the same grain boundaryvelocity a small inclusion can be dragged over largerdistances than a large one. This is consistent with theobservations shown in Fig. 10 (Movie S9). Theunsystematic development of the critical drag angle ispresumably related to the number and distribution offluid inclusions (Humphreys & Hatherly, 1996) whichwas not further investigated in this study. Turning tothe presence or absence of interaction (Figs 7 & 8) wenow investigate the effect of grain boundary velocityand fluid inclusion radius. Here the data from fourdifferent experiments are plotted in Fig. 12, indicat-ing the presence or absence of interaction for (i)annealing, (ii) dynamic grain boundary migrationrecrystallization and (iii) post-kinematic annealing.All three diagrams show that the boundary betweenthe two fields (presence or absence of interaction)occurs at a critical grain boundary velocity, which is

Regime 2 Shape affectedRegime 1 Shape not affected

Pass(no drag)

Drag

Pin

Incorporate

(a)

(b)

(c)

(d)

t0 t1 t2 tn t0 t1 t2 tn

(a)

(b)

(c)

Migrating Grain Boundary

Stationary GB

GB in a previoustime step

Fluid filmon GB

Fluid inclusions FI-array or -trail

Legend:

Fig. 8. Matrices summarizing the evolutionof fluid inclusion morphology for fluidinclusions not being changed in shape bypassing grain boundaries (regime 1) and fluidinclusions being changed in shape by passinggrain boundaries (regime 2). Passing withoutdragging (a), the drag and drop of fluidinclusions (b) and pinning (c) of fluid inclu-sions was observed in both regimes. Theincorporation of fluid inclusions to themigrating grain boundary (d) does onlyoccur in regime 2. In (b) necking down of afluid inclusion is illustrated as an end-member of drag and drop accompanied byshape change. In this field fluid inclusionscan also become elongated without necking.The field where no interaction of fluidinclusions and grain boundaries occurs isshaded in grey.

t2 tn(a)

(b)

(c)

(d)

Fig. 9. Sketch illustrating the formation of fluid inclusion arraysbehind migrating grain boundaries. (a) The break-up of anelongated inclusion into an array of smaller inclusions, (b) amigrating grain boundary drops an array of inclusions, (c)reactivated grain boundary leaves a ghost grain boundarybehind and (d) a fluid film shrinks into an array of isolatedinclusions. See Fig. 8 for legend.

1 0 J . S C H M A T Z & J . L . U R A I


t1

t8t7t6t5

t4t3t2

(a)

300 µm

t1 t7t6t5t4t3t2 t8

qq

i)

ii)

iii)

iv)

v)

vi)

(b)

Fig. 10. (a) Image sequence showing camphor with ethanol during annealing (temperature 50–70 �C). A migrating grain boundarysweeps over an array of differently shaped and sized fluid inclusions (Movie S9, experiment ce02 01, Table 1). (b) Sketch showing anoverlay of the eight time steps. Grey shading of the inclusions indicates the successive time steps, Dt = 120 s. See text for details.

F L U I D I N CL US I O N S A N D M O B I L E G R A I N B O UN D A R I E S 1 1


dependent on the fluid inclusion size (in Fig. 12c thedata are not sufficient to test the presence of an effectdue to fluid inclusion size). For annealing and dy-namic recrystallization the drag limit was mapped tobe in the range of 1 · 10)7 ms)1 for fluid inclusionradius below10 lm; for post-kinematic annealing thisis slightly slower, at 2 · 10)8 ms)1. Above the limitsshown in Fig. 12, interaction cannot be excluded butmust have been below the limit of resolution of ourimaging system. The slightly higher limit for anneal-ing (stages 1 & 2) and dynamic recrystallization(stage 3) in comparison with post-kinematic anneal-ing (stage 4) could be explained by the higher dislo-cation density in the polycrystal, which increases thefluid inclusion mobility (e.g. Tullis et al., 1996). Priorto sintering the samples were pre-compacted whichincreases dislocation density (Gottstein, 2004).Recovery processes during post-kinematic annealingdecrease the dislocation density (e.g. Humphreys &Hatherly, 1996).

Fluid inclusion re-equilibration

When a fluid inclusion was dropped by a grainboundary after interaction (Fig. 5) it was frequentlyelongated (cigar shaped) with a gradient of curvaturealong the surface. During annealing these inclusionseither contracted into a sphere (indicating a very lowsurface energy anisotropy) or broke up into an array offluid inclusions.

The gradient in curvature produces chemicalpotential gradients which drive diffusive materialtransport (Chuang & Rice, 1973; Reuschle et al., 1988).The final, stable structure is an array of fluid inclusionsas the transformation from an elongated or tube-shapedinclusion to an array of smaller spherical inclusionsresults in a reduction in total interfacial energy of thesystem (Nichols & Mullins, 1965). The formation andbreak-up of cylindrical tubes is also observed for crack-healing experiments. The phenomenon is explained

with rate-limited diffusion transport processes (Smith &Evans, 1984; Hickman & Evans, 1987; Leroy &Heidug,1994) and surface tension instability.

Preferential leaking of fluid inclusions along grainboundaries

Figure 13 shows the size of fluid inclusion beforeand after being passed (with and without interaction)by migrating grain boundaries. An analysis ofmeasurement uncertainty of fluid inclusion radius isgiven in Appendix A1. It is clear from this diagramthat after passage of a grain boundary, fluid inclu-sions tend to be smaller. In 12.5% of the casesmeasured, there was no significant change in radius,while in 12.5% the fluid inclusion became signifi-cantly larger and in 75% of the cases the inclusionreduced in diameter, up to 39% (39% decrease inradius equals 77% decrease in volume assuming aspherical inclusion).This change in size indicates a preferential leaking

of fluid inclusions along grain boundaries, and sug-gests that in our experiments grain boundaries wererelatively easy pathways for the fluid, in channels orfilms below the resolution of our imaging system. Wefound no clear effect of grain boundary orientation orgrain boundary velocity in this data set. A reductionin fluid inclusion size was observed for all modes andregimes of fluid inclusions–grain boundary interac-tion. High angle grain boundaries are defined aslattice defects (e.g. Gottstein, 2004) and are known asfast diffusion paths. Schenk & Urai (2005) showedthat at low grain boundary migration rates the con-tent of the fluid inclusion is incorporated into theboundary and distributed laterally. In our study theincorporation of the entire inclusion was an excep-tion. However, our measurements indicated that therewas preferential leakage along grain boundaries,which may lead to local wetting of grain boundaries.Many studies on natural rocks, such as quartz

7.5

–7.0

–6.5

(i) (ii) (iii)

1.25 5.0 10.0Fluid inclusion radius [µm]

Drag(regime 1b)

Drag with elongation(regime 2b)

Drag with necking(regime 2b)

log

GB

vel

ocity

[ms–1

]

Fig. 11. Diagram showing grain boundaryvelocity v. fluid inclusion size. Dependent onthe size a fluid inclusion can be (i) draggedkeeping its spherical shape, (ii) becomeelongated while being dragged or (iii) neckdown being in contact with a migrating grainboundary. As an example the morphologychange of an attached fluid inclusion withincreasing fluid inclusion size was sketched.See text for details.

1 2 J . S C H M A T Z & J . L . U R A I


mylonites (Mancktelow & Pennacchioni, 2004) orhalite (Ter Heege et al., 2005), have shown that wet-ting of grain boundaries can have great effects on therheological and seismic properties of rocks (Hirth &Kohlstedt, 1996).

Implications for metamorphic studies

In agreement with a number of earlier studies (e.g.Urai, 1983; Drury & Urai, 1990; Schenk & Urai,2005), this study has shown that in crystalline mate-rials at high homologous temperatures, mobile grainboundaries and fluid inclusions can have a range ofvery complex interactions. The most prominentdiagnostic microstructure for these fluid inclusions isthat they are arranged in linear arrays in the grains(Hansen et al., 1984; Beurlen et al., 2001). Jessellet al. (2003) showed that subsequent to grain growth,unswept cores of grains are preserved that recordimportant information on initial grain size and ori-entation that can be used to reconstruct the meta-morphic conditions of formation. Preserved arrays offluid inclusions provide additional information.Looking at snapshots of our experiments at a certaintime such trails are relatively rare. They may havebeen overlooked in metamorphic studies. In addition,we have shown that fluid inclusions can experiencea number of interactions with migrating grainboundaries, thereby possibly changing the chemicalcomposition of the fluid. An example is shown inFig. 14 (Movie S10) (see figure caption for description).Although the processes are complex, our study couldprovide a basis for improved interpretation of fluidinclusions in metamorphic tectonites.

Many studies show that the presence of a secondphase may have a strong effect on the recrystallized

ce 01 no interaction

ce 03 no interactionce 01 interaction

ce 03 interactionce 03 interactiondataset Fig. 11

Fluid inclusion radius [µm]



log

grai

n bo

unda

ry v

eloc

ity [m

s–1]

log

grai

n bo

unda

ry v

eloc

ity [m

s–1]

log

grai

n bo

unda

ry v

eloc

ity [m

s–1]

–9

–8

–7

–6

0 5 10 15 20 25 30

–9

–8

–7

–6

0 5 10 15 20 25 30

–9

–8

–7

0 5 10 15 20 25 30


ce 03 no interactionce 02 interaction

ce 03 interaction


ce 03 interactionce 01 interaction


a(a)

(b)

(c)

b

c

Fig. 12. Diagrams showing results for grain boundary migrationrates as a function of fluid inclusions size for (a) annealing, (b)dynamic grain boundary migration recrystallization and (c) post-kinematic annealing. Measurements allow mapping of a draglimit (dashed line) above which the fluid inclusions are notaffected by a passing grain boundary. The data set of Fig. 11 ismarked in (b). Please note error bars in the lower left and upperright corner. See text for details.

FI r

adiu

s [µ

m] a

fter

cont

act w

ith G

BFI radius [µm] before contact with GB

0

5

10

15

20

25

0 5 10 15 20 25

No interactionInteraction

Fig. 13. Diagram showing the fluid inclusion size for fluidinclusions before v. after being passed by a migrating grainboundary. Fluid inclusions show up to 39% reduction indiameter. The reduction was observed for both cases, the inter-action of fluid inclusions and grain boundaries and for nointeraction of fluid inclusions and grain boundaries.



grain size and therefore the mechanical properties of apolycrystalline material (e.g. Couturier et al., 2005).Zener pinning may delay or inhibit grain growth(Olgaard & Evans, 1988; Humphreys & Hatherly,1996) dependent on the number, size and distributionof the second phase particles, or in our case fluidinclusions (see also Evans et al., 2001). The pinning ofgrain boundaries is mainly caused by fluid inclusionsin which the host crystal is weakly soluble, or solidinclusions (Mancktelow et al., 1998). However, in thisstudy a strong solvent (camphor in ethanol) was usedand the results show that for sufficiently high grainboundary velocities fluid inclusions show no interac-tion with the grain boundary. Enduring pinningeffects have rarely been observed and it can beassumed that the liquid phase will have a minoreffect on the recrystallized grain size, with respectto the grain boundary–fluid inclusion interaction.Moreover, studies of Ter Heege et al. (2005) showedthat a material deformed together with a solvent(halite in water) adjusted to a larger grain size com-pared to a �dry� sample. Mancktelow et al. (1998)point to the higher-order effect of enhanced grainboundary mobility due to the presence of water insamples of quartz and feldspar towards pinningeffects of insoluble second phases on the recrystallizedmicrostructure.

CONCLUSIONS

The experiments we report show that fluid inclusionsmay or may not be affected by a migrating grainboundary, depending on the grain boundary velocity.The critical grain boundary velocity for the interactionof migrating grain boundaries with fluid inclusions isdependent on the fluid inclusion size. For the samegrain boundary velocity a large inclusion can bemodified while a small inclusion is passed withoutbeing affected. In addition, the interaction of a fluidinclusion with a migrating grain boundary maystrongly affect the fluid inclusion morphology. It isdemonstrated that the contact of a fluid inclusion witha grain boundary allows leaking of fluid and fluidtransport along the grain boundary. However, thepreserved microstructure is only a snapshot andprovides no full information on the formation andalteration of fluid inclusions in recrystallized rocks.

ACKNOWLEDGEMENTS

We thank F. -D. Scherberich for constructing thedeformation cell. We are very grateful to P. D. Bons,S. Hickman and O. Schenk for inspiring discussionson our experiments. M. Jessell and an anonymousreviewer are acknowledged for their constructive and

t1 t2

FI FI

GB1 GB1

t3

FI

GB2200 µm t4

FIGB3

Fig. 14. Image sequence showing the inter-action of fluid with fluid-filled, migratinggrain boundaries, Dt = 100 min(Movie S10, experiment ce03 10, Table 1).The example illustrates that overprinting bya fluid-filled grain boundary may also causean alteration of isolated fluid inclusions byfluid-filled grain boundaries. The processrecorded in this image sequence wasdynamic grain boundary migration recrys-tallization. In the three time increments, oneand the same fluid inclusion came in contactwith grain boundaries of three differentgrains. The first passing grain boundaryshowed a microscopically visible fluidcontent, for the other ones the fluid contentcould be assumed to be apparent in adifferent scale. In the first time increment thefluid inclusion has not only been in contactwith the fluid-filled grain boundary, but ithas also been merged with a neighbouringfluid inclusion. The same is valid for thesecond time increment and after the thirdcontact with a grain boundary in t4, the fluidinclusion again is located inside the bulkcrystal at a considerable distance from anygrain boundary and there are no indicatorsleft that hint the grain boundary–fluidinclusion interaction.

1 4 J . S C H M A T Z & J . L . U R A I


thorough reviews and especially M. Brown for hiseditorial help. The project was funded by the GermanScience Foundation (DFG, UR 64 ⁄ 8-1) and was partof the European Science Foundation (ESF)-fundedcollaborative research project EuroMinSci.

REFERENCES

Bauer, P., Rosenberg, C. & Handy, M.R., 2000. See-throughdeformation experiments on brittle-viscous norcamphorat controlled temperature, strain rate and appliedconfining pressure. Journal of Structural Geology, 22, 281–289.

Beurlen, H., da Silva, M.R.R. & de Castro, C., 2001. Fluidinclusion microthermometry in Be-Ta-(Li-Sn)-bearingpegmatites from the Borborema Province, Northeast Brazil.Chemical Geology, 173, 107–123.

Bons, P.D., 1993. Experimental deformation of polyphase rockanalogues. Geologica Ultraiectina, PhD Thesis, Utrecht Uni-versity, 110, 207.

Bons, P.D., Jessell, M.W. & Passchier, C.W., 1993. The analysisof progressive deformation in rock analogues. Journal ofStructural Geology, 15, 403–411.

Boullier, A.-M., 1999. Fluid inclusions: tectonic indicators.Journal of Structural Geology, 21, 1229–1235.

Chuang, T.-J. & Rice, J.R., 1973. The shape of intergranularcreep cracks growing by surface diffusion. Acta Metallurgica,21, 1625–1628.

Chuang, T.-J., Kagawa, K.I., Rice, J.R. & Sills, L.B., 1979.Overview no. 2: non-equilibrium models for diffusive cavita-tion of grain interfaces. Acta Metallurgica, 27, 265–284.

Couturier, G., Doherty, R., Maurice, C. & Fortunier, R., 2005.3D finite element simulation of the inhibition of normal graingrowth by particles. Acta Materialia, 53, 977–989.

Drury, M.R. & Urai, J.L., 1990. Deformation-related recrys-tallization processes. Tectonophysics, 172, 235–253.

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

Gottstein, G., 2004. Physical Foundations of Materials Science.Springer, Berlin, 504 pp.

Hansen, E.C., Newton, R.C. & Janardhan, A.S., 1984. Fluidinclusions in rocks from the amphibolite-facies gneiss tocharnockite progression in southern Karnataka, India: directevidence concerning the fluids of granulite metamorphism.Journal of Metamorphic Geology, 2, 249–264.

Herwegh, M. & Berger, A., 2004. Deformation mechanisms insecond-phase affected microstructures and their energybalance. Journal of Structural Geology, 26, 1483–1498.

Hickman, S.H. & Evans, B., 1987. Influence of geometry uponcrack healing rate in calcite. Physics and Chemistry ofMinerals, 15, 91–102.

Hilgers, C. & Urai, J.L., 2002. Microstructural observations onnatural syntectonic fibrous veins: implications for the growthprocess. Tectonophysics, 352, 257–274.

Hind, R.K., McLaughlin, E. & Ubbelohde, A.R., 1960. Struc-ture and viscosity of liquids. Camphor+pyrene mixtures.Transactions of the Faraday Society, 56, 328–330.

Hirth, G. & Kohlstedt, D.L., 1996. Water in the upper ozeanicmantle: implications for rheology, melt extraction and evolu-tion of the lithosphere. Earth and Planetary Science Letters,144, 93–108.

Hirth, G. & Tullis, J., 1992. Dislocation creep regimes in quartzaggregates. Journal of Structural Geology, 14, 145–159.

Holness, M.B. & Lewis, S., 1997. The structure of the halite-brine interface inferred from pressure and temperature varia-tions of equilibrium dihedral angles in the halite–H2O–CO2

system. Geochimica et Cosmochimica Acta, 61, 795–804.Holness, M.B. & Siklos, S.T.C., 2000. The rates and extent of

textural equilibration in high-temperature fluid-bearing sys-tems. Chemical Geology, 162, 137–153.

Hsueh, C.H., Evans, A.G. & Coble, R.L., 1982. Microstructuredevelopment during final ⁄ intermediate stage sintering, I.Pore ⁄ grain boundary separation. Acta Metallurgica, 30, 1269–1279.

Humphreys, F.J. & Hatherly, M. 1996. Recrystallization andRelated Annealing Phenomena. Pergamon, Oxford, 497 pp.

Jessell, M.W., 1986. Grain boundary migration and fabricdevelopment in experimentally deformed octachloropropane.Journal of Structural Geology, 8, 527–542.

Jessell, M.W., Kostenko, B. & Jamtveit, B., 2003. The preser-vation potential of microstructures during static grain growth.Journal of Metamorphic Geology, 21, 481–491.

Jing, X.N., Zhao, J.H., Subhash, G. & Gao, X.L., 2005.Anisotropic grain growth with pore drag under applied loads.Materials Science and Engineering: A, 412, 271–278.

Kolb, J. & Meyer, F.M., 2002. Fluid inclusion record of thehypozonal orogenic Renco gold deposit (Zimbabwe) duringthe retrograde P–T evolution. Contributions to Mineralogy andPetrology, 143, 495–509.

Kolb, J., Kisters, A.F.M., Hoernes, S. & Meyer, F.M., 2000. Theorigin of fluids and nature of fluid-rock interaction in mid-crustal auriferous mylonites of the Renco mine, southernZimbabwe. Mineralium Deposita, 35, 109–125.

Leroy, Y.M. & Heidug, W.K., 1994. Geometrical evolution ofstressed and curved solid-fluid phase boundaries; 2, Stabilityof cylindrical pores. Journal of Geophysical Research, B, SolidEarth and Planets, 99, 517–530.

Mancktelow, N.S. & Pennacchioni, G., 2004. The influence ofgrain boundary fluids on the microstructure of quartz-feldsparmylonites. Journal of Structural Geology, 26, 47–69.

Mancktelow, N.S., Grujic, D. & Johnson, E.L., 1998. A SEMstudy of porosity and grain boundary microstructure in quartzmylonites, Simplon Fault Zone, Central Alps. Contributions toMineralogy and Petrology, 131, 71–85.

McLimans, R.K., 1987. The application of fluid inclusions tomigration of oil and diagenesis in petroleum reservoirs.Applied Geochemistry, 2, 585–603.

Means, W.D., 1983. Microstructure and micromotion inrecrystallization flow of octachloropropane: a first look.Geologische Rundschau, 72, 511–528.

Means, W.D., Hobbs, B.E., Lister, G.S. & Williams, P.F., 1980.Vorticity and non-coaxiality in progressive deformations.Journal of Structural Geology, 2, 371–378.

Nichols, F.A. & Mullins, W.W., 1965. Surface (Interface) andvolume diffusion contributions to morphological changesdriven by capillarity. Transactions of the Metallurgy SocietyAIME, 233, 1840–1848.

Olgaard, D.L. & Evans, B., 1988. Grain growth in syntheticmarbles with added mica and water. Contributions to Miner-alogy and Petrology, 100, 246–260.

Olgaard, D.L. & FitzGerald, J.D., 1993. Evolution of poremicrostructures during healing of grain boundaries in syntheticcalcite rocks. Contributions to Mineralogy and Petrology, 115,138–154.

Passchier, C.W. & Trouw, R.A.J., 2005. Microtectonics.Springer, Berlin, 266 pp.

Peach, C.J., Spiers, C.J. & Trimby, P.W., 2001. Effect of con-fining pressure on dilatation, recrystallization, and flow ofrock salt at 150 �C. Journal of Geophysical Research, 106,13,315–13,328.

Petrishcheva, E. & Renner, J., 2005. Two-dimensional analysisof pore drag and drop. Acta Materialia, 53, 2793–2803.

Renard, F., Dysthe, D., Feder, J. & Jamtveit, B., 2002. Healingof fluid-filled microcracks. Poromechanics II (Proceedings ofthe second Biot Conference on poromechanics), 925–931.

Renner, J., Evans, B. & Hirth, G., 2002. Grain growth andinclusion formation in partially molten carbonate rocks.Contributions to Mineralogy and Petrology, 142, 501–514.

Reuschle, T., Trotignon, L. & Gueguen, Y., 1988. Pore shapeevolution by solution transfer: thermodynamics and mechan-ics. Geophysical Journal International, 95, 535–547.



Roedder, E., 1984. Fluid inclusions. Reviews in Mineralogy, 12,644.

Schenk, O. & Urai, J.L., 2004. Microstructural evolution andgrain boundary structure during static recrystallization insynthetic polycrystals of sodium chloride containing saturatedbrine. Contributions to Mineralogy and Petrology, 146, 671–682.

Schenk, O. & Urai, J.L., 2005. The migration of fluid-filled grainboundaries in recrystallizing synthetic bischofite: first resultsof in-situ high-pressure, high-temperature deformationexperiments in transmitted light. Journal of MetamorphicGeology, 23, 695–709.

Schleder, Z. & Urai, J., 2005. Microstructural evolution ofdeformation-modified primary halite from the Middle TriassicRot Formation at Hengelo, The Netherlands. InternationalJournal of Earth Science (Geologische Rundschau), 94, 941–955.

Schoenherr, J., Urai, J.L., Kukla, P.A. et al., 2007. Limits to thesealing capacity of rock salt; a case study of the infra-Cambrian Ara Salt from the South Oman salt basin. AAPGBulletin, 91, 1541–1557.

Schoenherr, J., Reuning, L., Kukla, P.A. et al., 2009. Halitecementation and carbonate diagenesis of intra-salt reservoirsfrom the Late Neoproterozoic to Early Cambrian Ara Group(South Oman Salt Basin). Sedimentology, 56, 567–589.

Schutjens, P. 1991. Intergranular Pressure Solution in HaliteAggregates and Quartz Sands: An Experimental Investigation.Unpublished PhD thesis, Universiteit Utrecht, Utrecht, 233pp.

Smith, C.S., 1964. Some elementary principles of polycrystallinemicrostructure. Metallurgical Reviews, 9, 1–48.

Smith, D.L. & Evans, B., 1984. Diffusional crack healing inquartz. Journal of Geophysical Research, 89, 4125–4135.

Spiers, C.J., Urai, J., Lister, G.S., Boland, J.N. & Zwart, H.J.1986. The Influence of Fluid–Rock Interaction on the Rheologyof Salt Rock and on Ionic Transport in the Salt. University ofUtrecht, The Netherlands, 132 pp.

Streit, J.E. & Cox, F.C., 2000. Asperity interactions during creepof simulated faults at hydrothermal conditions. Geology, 28,231–234.

Svoboda, J. & Riedel, H., 1992. Pore-boundary interactionsand evolution equations for the porosity and the grain sizeduring sintering. Acta Metallurgica et Materialia, 40, 2829–2840.

Ter Heege, J.H., De Bresser, J.H.P. & Spiers, C.J., 2005.Dynamic recrystallization of wet synthetic polycrystallinehalite: dependence of grain size distribution on flow stress,temperature and strain. Tectonophysics, 396, 35–57.

Tullis, J., Yund, R. & Farver, J., 1996. Deformation-enhancedfluid distribution in feldspar aggregates and implications forductile shear zones. Geology, 24, 63–64.

Tungatt, P.D. & Humphreys, F.J., 1981. An in situ opticalinvestigation of the deformation behaviour of sodium nitrate-an analogue for calcite. Tectonophysics, 78, 661–675.

Urai, J.L., 1983. Water assisted dynamic recrystallization andweakening in polycrystalline bischofite. Tectonophysics, 96,125–157.

Urai, J.L., 1985. Water-enhanced dynamic recrystallization andsolution transfer in experimentally deformed carnallite.Tectonophysics, 120, 285–317.

Urai, J.L. & Humphreys, F.J., 1981. The development of shearzones in polycrystalline camphor. Tectonophysics, 78,677–685.

Urai, J.L. & Jessell, M.W., 2001. Recrystallization and graingrowth in minerals: recent developments. In: Recrystallizationand Grain Growth, Proceedings of the First Joint InternationalConference 1 (eds Gottstein, G. & Molodov, D.), pp. 87–96.Springer Verlag, Berlin.

Urai, J.L., Humphreys, F.J. & Burrows, S.E., 1980. In-situstudies of the deformation and dynamic recrystallization ofrhombohedral camphor. Journal of Material Science, 15,1231–1240.

Urai, J.L., Means, W.D. & Lister, G.S., 1986. Dynamic recrys-tallization of minerals. In: Mineral and Rock Deformation;Laboratory Studies; The Paterson Volume. AGU GeophysicalMonograph (eds Hobbs, B.E. & Heard, H.C.), pp. 161–199.Washington, DC, USA.

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

Walte, N.P. 2005. The Grain-Scale Distribution and Behaviourof Melt and Fluid in Crystalline Analogue Systems. Unpub-lished PhD thesis, Johannes Gutenberg Universitat, Mainz.147 pp.

Walte, N.P., Bons, P.D., Passchier, C.W. & Koehn, D., 2003.Disequilibrium melt distribution during static recrystalliza-tion. Geology, 31, 1009–1012.

Walte, N.P., Becker, J.K., Bons, P.D., Rubie, D.C. & Frost,D.J., 2007. Liquid-distribution and attainment of texturalequilibrium in a partially-molten crystalline system with ahigh-dihedral-angle liquid phase. Earth and Planetary ScienceLetters, 262, 517–532.

Wark, D.A., Williams, C.A., Watson, E.B. & Price, J.D.,2003. Reassessment of pore shapes in microstructurallyequilibrated rocks, with implications for permeability of theupper mantle. Journal of Geophysical Research, 108, ECV12-1–ECV 12-16.

Warren, J.K., 2006. Evaporites: Sediments, Resources andHydrocarbons. Springer, Berlin, 1035 pp.

APPENDIX

A1. Measurements on fluid inclusions

The automated time-lapse image capture does notallow re-focusing on individual fluid inclusions duringan image sequence. Due to this, measurements onfluid inclusion size may contain an error if theinclusion moves up or down in the sample. Toquantify this we focused at different levels on twoisolated fluid inclusions. This is illustrated for oneinclusion in Fig. A1. (a) Level 1 lies in the centre ofthe inclusion, level 2 above the centre (+10 lm) andlevel 3 below ()10 lm) it. Focusing too high or toolow did not produce an image of a dark ring arounda bright area – such images were not used in ourexperiments (levels 4 & 5, >20 lm from the centre).Adobe Illustrator CS 13 was used to trace the outerlimits of the inclusion in the centre of the dark reflec-tion rim by defining about 10 anchor points around theboundary, on the darkest pixel in the image (b). ABezier curve was then created from the anchor points(setting curve precision = 70%) (c). The area of theobject was measured using Image SXM 1.86 (d, e) andthe radius was set equal to the area-identical sphericalradius. The largest area was measured in level 1, thecentre of the inclusion. Accordingly, measurements indifferent levels had a smaller value. The measurementswere repeated three times (Table A1). A standarddeviation of 0.5 lm was estimated for the inclusionwith a radius of 19.5 and 0.08 lm for the inclu-sion with a radius of 5.8 lm. An uncertainty in

1 6 J . S C H M A T Z & J . L . U R A I


(a)

(b)

(c)

(d)

(e)

Fig. A1. Image sequence and sketch showing the measuring procedure for uncertainty approximation of fluid inclusion measurements.See text for details.

Table A1. Fluid inclusion size measurements.

Inclusion ⁄measurement Level Area (lm) Radius (lm) Level (lm) SD (lm) Average (lm) Error (lm) Error (max) (lm) Error (max) (%)

A ⁄a 3 1196.93 19.52 )10 0.02 19.50 0.67 0.67 3.44

b 3 1191.76 19.48 )10c 3 1194.34 19.50 )10

A ⁄a 1 1300.34 20.34 0 0.17 20.17 0.00

b 1 1256.39 20.00 0

c 1 1277.07 20.16 0

A ⁄a 2 1140.06 19.05 10 0.53 19.62 0.55

b 2 1269.31 20.10 10

c 2 1220.2 19.71 10

B ⁄a 3 98.24 5.59 )2.5 0.00 5.59 0.17 0.24 4.35

b 3 98.24 5.59 )2.5c 3 98.24 5.59 )2.5

B ⁄a 1 105.99 5.81 0 0.08 5.76 0.00

b 1 105.99 5.81 0

c 1 100.82 5.66 0

B ⁄a 2 98.24 5.59 2.5 0.07 5.52 0.24

b 2 93.07 5.44 2.5

c 2 95.65 5.52 2.5

measurement of 4% for the radius of a fluid inclusionwas determined.

A2. Distribution of the fluid phase

In a triaxial gas apparatus, Bons (1993) determined apower law flow-law for pure camphor of logð_eÞ ¼�4877þ 3:3 logðrÞ at 28 �C. Hind et al. (1960) pro-

posed an Arrhenius type of equation for the viscosity ofcamphor with g ¼ 0:342 expð1:96� 104=RT (Fig. A2a).This suggests that the differential stress in our experi-ments was �0.05 MPa. Prior to deformation the MohrCircle lies in the origin (Fig. A2b). As soon as a stress isapplied differential stresses build up and simultaneouslythe sample is squeezed out at grooves in the sampleholder.



SUPPORTING INFORMATION

Additional Supporting Information may be found inthe online version of this article:

Movie S1. Experiment ce01 08, stage 3; deformationat 70 �C with a constant strain rate of 2.7 · 10)6, themicrostructural process is grain boundary migrationrecrystallization accompanied by dynamic graingrowth. Please note the dynamic wetting angle withvariations at leading and trailing surface. This experi-ment is also shown in Fig. 4b.

Movie S2. Experiment ce01 12, stage 3; deformationat 70 �C with a constant strain rate of 1.6 · 10)5, themicrostructural process is grain boundary migrationrecrystallization. A flat fluid inclusion left behind bymigrating grain boundary breaks up into an array ofsmaller inclusions. This experiment is also shown inFig. 5a.

Movie S3. Experiment ce01 16, stage 4; post-kine-matic annealing at 70 �C, the microstructural processis grain boundary migration recrystallization. Isolatedfluid inclusions are left behind a slowly migrating triplejunction; only the fastest boundary drops the fluid.This experiment is also shown in Fig. 5b.

Movie S4. Experiment ce01 15, stage 4; post-kine-matic annealing at 70 �C, the microstructural processis recovery followed by grain growth. A ghost grainboundary is formed behind a migrating grainboundary. The fluid film breaks up into an array ofinclusions. This experiment is also shown in Fig. 5c.

Movie S5. Experiment ce03 01, stage 3; deformationat 50 �C with a constant strain rate of 5.0 · 10)5, themicrostructural process is grain boundary migrationrecrystallization. Isolated fluid inclusions are leftbehind a migrating, fluid-filled grain boundary. Thisexperiment is also shown in Fig. 5d.

Movie S6. Experiment ce02 01, stage 1; annealing at70 �C, the microstructural process is grain growth. Afluid inclusion (radius = 25 lm) is passed by migra-ting grain boundary without being affected. Thisexperiment is also shown in Fig. 6a.Movie S7. Experiment ce02 01, stage 1; annealing at

70 �C, the microstructural process is grain growth. Thefluid inclusion (radius = 25 lm) gets elongated whilebeing passed by a migrating grain boundary. Thisexperiment is also shown in Fig. 6b.Movie S8. Experiment ce08 04-07, stage 1; annealing

at 50 �C, the microstructural process is grain growth.A fluid inclusion (radius = 25 lm) necks down whilebeing attached to a migrating grain boundary. Thisexperiment is also shown in Fig. 6c.Movie S9. Experiment ce02 01, stage 1; annealing at

70 �C, the microstructural process is grain growth. Amigrating grain boundary passes an array of differentshaped and sized fluid inclusions. This experiment isalso shown in Fig. 10a.Movie S10. Experiment ce03 10, stage 3; deforma-

tion at 70 �C with a constant strain rate of 2.6 · 10)6,the microstructural process is grain boundary migra-tion recrystallization. One and the same fluid inclusionis passed by three different, presumably fluid-filledgrain boundaries. This experiment is also shown inFig. 14.Please note: Wiley-Blackwell are not responsible for

the content or functionality of any supporting mate-rials supplied by the authors. Any queries (other thanmissing material) should be directed to the corre-sponding author for the article.

Received 9 February 2009; revision accepted 23 August 2009.

(a) (b)

Fig. A2. (a) Diagram showing the stress–strain rate relation for pure camphor at 28 �C as proposed by Bons (1993) and for a linear-viscous material behaviour at 28 and 70 �C deduced from the equation by Hind et al. (1960) (small diagram). See text for furtherexplanation. (b) Sketch showing sample and sample jacket prior (1) and during deformation (2). Mohr diagram illustrates stressconditions. See text for further explanation.

1 8 J . S C H M A T Z & J . L . U R A I


the interaction of ﬂuid inclusions and migrating grain ... · ethanol, focusing on the ......

Documents