The interaction of fluid and rock during deformation plays an important role in numerous processes inside the Earth’s crust. The presence and composition of fluids affects rock properties, such as rheology and transport, on different scales and by a variety of processes. Understanding these processes is needed for extrapolation of rheology or paleo-conditions from experiments to nature, and to determine the reliability of, for example, temperature and paleostress estimates. This effect is important for free fluid as well as for fluid occurring in inclusions.

Since quartz is an important component of the Earth's continental crust and fluids are thought to occur abundantly at depth, it is important to understand the interaction of the fluid with the quartzite. It is well known that water affects the deformation of quartzites in nature and in experiments (e.g., Jaoul et al., 1984; Kronenberg et al., 1990; Post and Tullis, 1998), though the process of weakening is not yet understood in detail and there is still much debate about the exact role of fluids in deformation mechanisms (e.g., den Brok, 1992, Wang, 1992, Rutter and Brodie, 2004; Stipp et al., 2006). Early experimental work on understanding quartz deformation in the presence of fluids has been performed with pure water (e.g, Jaoul, 1984 and references therein). However, the fluids that occur in the crust are not pure H2O, and FTIR studies on natural quartz suggest more CO2-rich fluids at depth (e.g., Nakashima et al., 1995). Some work has been done on effects of the fluid composition (Post et al., 1996; Wang et al., 1993; Wang, 1992; Ord and Hobbs, 1986) and a recent study by Chernak et al. (2009) shows the effect of aqueous and carbonic fluids on the strength of quartz taking into account the chemical environment, such as the effect of reducing and oxidizing conditions. Post et al. (1996) varied the water fugacity in experiments on Heavitree quartzite by changing the confining pressure and found evidence that water fugacity affected the dislocation creep strength. They studied the effect of varying fluid composition (pure water or 90% CO2) on the rate of dislocation climb and recrystallization and observed higher rates of dislocation climb and recrystallization in the samples with pure water added.

Recrystallization is almost always strongly affected by the presence of a second phase. The effect depends on the type of second phase and several other parameters. The presence of a solid second phase will in general decrease the average grain boundary mobility, i.e., it will pin the grain boundary or drag may occur (Olgaard and Evans, 1988; Herwegh and Berger, 2004, Schmatz and Urai, 2010; Schmatz et al., 2011). When the driving force is high enough the boundary may break free of the particles (Drury and Urai, 1990). The interaction with a fluid phase is somewhat different and depends on the type of fluid. When the fluid is a solvent for the mineral, its presence may increase the grain boundary mobility because it enhances the across-boundary diffusion rate, e.g., in 1968 Hobbs showed that dry samples of quartz did not recrystallize under conditions where water-wet samples did (see also Tullis and Yund, 1982). A similar role of water was found for static grain growth in anorthite (Dresen et al., 1996). However, if the mineral is not soluble in the fluid phase the second phase will not increase the migration rate but may even pin the grain boundaries (e.g., calcite at high temperature, Tullis and Yund, 1982, or air: Olgaard and Evans, 1988). Schmatz and Urai (2011) show that oxidant bearing grain boundaries in quartz have the ability to transform solid graphite into CO2 in a naturally deformed and partly recrystallized quartz vein from the Hunsrück slate, Germany. They also show that various interactions of grain boundaries, moving at different velocities, with fluid inclusions are responsible for significant redistribution and modification of the fluid phase at different rates.

Fluid-enhanced recrystallization has also been attributed to the wetting characteristics of the fluid with the solid, i.e., if the fluid wets the solid the grain boundary migration rate would increase as more surface was exposed to a fast diffusion environment. It is therefore useful to consider what "wetting" means. Under equilibrium conditions a fluid will only fully wet a grain boundary (wetting angle = 0°) if the surface energy of the solid-liquid interface is equal to or smaller than half that of the solid-solid interface (γssfs = 2 cos (a/2), where γssis the solid-solid interfacial energy, γfs is the liquid-solid interface energy, and α the dihedral, or wetting angle; e.g., Watson et al., 1990; Laporte, 1994, Holness, 1992; 1993). However, a moving grain boundary or a boundary under load is not in thermodynamic equilibrium and thus the properties of the boundary may be different from that in equilibrium, and a continuous fluid film may exist on such a boundaries. This phenomenon has been suggested from several experimental studies (e.g., Urai, 1983; Urai et al., 1986; Jin et al. 1994; Drury and Fitz Gerald, 1996; Tullis et al., 1996; Bai et al., 1997; Rutter and Brodie, 2004; Schenk and Urai, 2005, Schmatz and Urai, 2010; Schmatz et al., 2011). When using the term wetting in this paper we refer to this non-equilibrium situation, not the thermodynamic equilibrium case.

The interaction between grain boundaries and fluid inclusions during recrystallization is dependent on the fluid composition. This is important since fluid inclusions give information about paleo-temperature in the rocks. Holness (1995) points to the effect of fluid-composition on wettability, which is a major parameter controlling the pore fluid distribution. Drury and Urai (1990) have proposed that H2O -rich fluids form continuous fluid films on moving grain boundaries, whereas CO2-rich fluids do not, based on observations in natural rocks. They suggested that a moving grain boundary may significantly change the fluid inclusion composition when CO2 is preferentially partitioned to inclusions, and H2O remains on the boundary. A release of H2O can be even promoted taking into account moderately reducing conditions (Chernak et al., 2009). Thus errors can occur when using fluid inclusions for paleo-thermometry. These issues were already addressed in studies dealing with the effect of decrepitation on the composition of fluid inclusions (e.g. Bodnar et al., 1989; Sterner and Bodnar, 1989), also with respect to the role of (pipe) diffusion (Bakker and Jansen 1991,1994; Hall and Sterner, 1993) and the effect of hydrogen diffusion through the quartz lattice (Mavrogenes and Bodnar; 1994; Morgan et al., 1993; Rutter and Brodie, 2004). There are several studies that discuss the effect of deformation on H2O-loss from fluid inclusions (e.g., Kerrich, 1976; Hollister, 1990; Johnson and Hollister, 1995; Audétat and Günther, 1999; Vityk et al., 2000; Tarantola et al., 2010) but the role of grain boundary migration (Drury and Urai, 1990; Schmatz and Urai, 2011) is not often considered. Grain boundary diffusion can be a much faster process than lattice diffusion (Atkinson, 1984), and is accordingly important to understand. However, the effect of grain boundary migration on the composition of fluid inclusions has not been given much attention.

With this study we present some preliminary results that show an effect of the ratio of aqueous and gaseous inclusions on rheology and recrystallization. We have deformed single crystals of quartz that contain fluid inclusions with different aqueous-gaseous inclusion ratios and observed strength and microstructure as a function of fluid composition. The samples recrystallized during deformation and during a post-kinematic annealing period and the effect of recrystallization on the distribution and composition of fluid inclusions was studied. We believe these results are important and representative for processes of wide relevance but more experiments and systematic observations are needed to fully understand the process.

N.B. This paper based on the results of the master thesis by I. Dijkstra (1990). All results are presented here as described coherent with that study. We were not able to include further analytics, as the samples were not available anymore.