TY - JOUR
AU - Dresen, Georg
AB - Abstract We observe void growth and coalescence into cavity-bearing shear bands during deformation of wet synthetic anorthite aggregates containing <3 vol. % silica-enriched melt. Samples were deformed in the Newtonian creep regime to high strain during torsion experiments at 1100°C and 400 MPa confining pressure. Localized cavity-bearing shear bands show an S–C'-geometry: the bands (C') are oriented at about 30° to the compression direction of the imposed simple shear and the internal foliation (S) of the bands is rotated towards the horizontal external shear plane. Cavity-bearing shear bands started to nucleate in the sample periphery above a shear strain threshold of ≈2. Quartz crystallized from the water-saturated SiO2-rich melt within large cavities inside these bands, which requires that the melt is decompressed by >200 MPa during their formation. The dynamically evolving cavities are sites of locally reduced pressure that collect the melt distributed in the adjacent matrix. Therefore, cavitation damage under ductile conditions may result in the development of an efficient melt channelling system controlling SiO2-rich melt flow in the lower crust. Electron backscatter diffraction analysis shows that the quartz inside the cavity bands has a crystallographic preferred orientation (CPO). The development of the CPO is explained by the preferred dissolution of crystals oriented with the rhombohedra and trigonal dipyramids orthogonal to the compression direction and by preferential growth of crystals aligned with their <0001> axis in the extension direction of the externally applied simple shear deformation. INTRODUCTION Synthetic anorthite aggregates deformed to high strain during torsion experiments are excellent proxies for studying the development of melt and fluid channelling pathways through a feldspar-dominated lower crust. Rybacki et al. (2008, 2010) and Kohlstedt et al. (2010) have performed high-strain torsion experiments on anorthite aggregates in the diffusion creep regime at 950–1200°C and 100–400 MPa confining pressure. Some of the nominally melt-free aggregates used by Rybacki et al. (2008, 2010) contained small amounts (<3%) of silica-enriched residual glass, whereas all the aggregates, except one, used by Kohlstedt et al. (2010) contained 3–12% added mid-ocean ridge basalt (MORB)-type basaltic glass. Despite this difference, within both sets of deformed samples regularly spaced shear bands formed with an angle of about 15° to the shear plane and 30° to the principal stress direction σ1. After torsion tests performed to high strains (3–5), the samples of Rybacki et al. (2008, 2010) exhibited abundant cavities nucleating mostly at grain triple junctions and at grain boundaries. Cavitation growth has previously been observed in anorthite–diopside samples deformed in torsion by Dimanov et al. (2007). In those samples cavities coalesced to micro-fissures that linked to micro- and meso-cracks hundreds of micrometers in length. Instead, in the samples of Rybacki et al. (2008, 2010) at shear strains ≥2 an anastomosing network of regularly spaced, cavity-bearing shear bands formed. Because these shear bands contained the segregated residual glass phase, and because microstructural inspection suggests that growth and coalescence of the cavities is responsible for the formation of the shear bands, Rybacki et al. (2008, 2010) concluded that cavitation damage in high-temperature shear zones may control lower crustal melt and fluid flow. The recently published study of Kohlstedt et al. (2010) is a continuation of experimental work on the deformation of melt-bearing feldspar and olivine rocks going back to Bussod & Christie (1991), Hirth & Kohlstedt (1995a, 1995b), Kohlstedt & Zimmerman (1996), Daines & Kohlstedt (1997), Zimmerman et al. (1999), Holtzman et al. (2003a, 2003b, 2005), and Holtzman & Kohlstedt (2007), focusing on the understanding of melt segregation and melt channelling within partially molten rocks. Those researchers concluded that the regularly spaced shear bands result from compaction of the partially molten rocks under an applied differential stress. Orientation of the bands is the result of two factors: (1) preferential wetting of grain boundaries oriented at a low angle to the maximum compressive stress σ1 (Zimmerman et al., 1999; Hier-Majumder et al., 2004); (2) melt flow along a pressure gradient from bands oriented at high angles towards interconnected bands oriented at lower angles to σ1 (Holtzman & Kohlstedt, 2007). The experimental observations of Holtzman et al. (2003a, 2003b, 2005), Holtzman & Kohlstedt (2007) and Kohlstedt & Holtzman (2009) support the hypothesis of Stevenson (1989) that the spacing between melt-bearing shear bands and compaction length (McKenzie, 1984) are correlated. Compaction length, δc, depends on the permeability (k), melt viscosity (μ), bulk viscosity (λ) and shear viscosity (η) of the partially molten rock as expressed by the relation (McKenzie, 1984; Scott & Stevenson, 1984) (1) The compaction length defined in equation (1) decreases significantly with increasing melt viscosity. Kohlstedt et al. (2010) used mid-ocean ridge basalt (MORB)-type basaltic glass with a viscosity of 10 Pa s in their anorthite torsion experiments. Hydrous leucogranitic melts are at least two orders of magnitude more viscous (Hess & Dingwell, 1996) and dry rhyolitic melts are several orders of magnitude more viscous (Hui & Zhang, 2007), suggesting that the compaction length of samples with a silica-enriched melt is at least an order of magnitude smaller than in MORB-containing aggregates. Because the shear band spacing in the samples of Rybacki et al. (2008, 2010) is similar to that observed by Holtzman et al. (2003a, 2003b, 2005), Holtzman & Kohlstedt (2007), Kohlstedt & Holtzman (2009) and Kohlstedt et al. (2010), it is possible that in the presence of a highly viscous silica-rich melt its segregation is more efficiently achieved by opening of depressurized cavities in high-temperature shear zones than by differential stress-driven compaction. Here we present a detailed microstructural study of one (Pl10_1) of two anorthite aggregates from the set of samples deformed by Rybacki et al. (2008, 2010) containing quartz within the cavities of the shear bands. We show that quartz crystallization from the silica-rich melt can occur only if the cavities were strongly depressurized. METHODS Experimental approach Sample material We analysed the microfabric and microtexture of a synthetic cylindrical anorthite sample (Pl10_1), 10 mm in diameter and 6·5 mm long, containing ≈1 vol. % silica-rich melt. The sample belongs to a suite of samples, which were deformed experimentally in the Newtonian creep regime (Rybacki et al., 2008, 2010). Starting material for sample fabrication was a fine-grained anorthite glass powder (Schott Glass-Werke) with a mean grain size of <60 μm. The powder composition was nearly pure anorthite (An98·8Or0·2Ab0·9, normalized to 8O) with a trace amount of impurities such as TiO2 (0·1%), MgO (0·007%), and Fe2O3 (0·01%). After cold pressing, the glass powder was hot isostatically pressed using a Paterson-type gas apparatus at 300 MPa confining pressure and at three temperatures. The powder was initially densified above the glass transition temperature (871°C, Dresen et al., 1996) for 1 h and then further annealed for 2 h at 1050°C, slightly above the nucleation temperature of anorthite (1024°C, Dresen et al., 1996). Finally, to promote grain growth, the sample was kept for 2 h at 1100°C. Grains are typically prismatic with an aspect ratio of about 2·5. Assuming that the grains are rectangular with a square base, a mean grain size of 3·7 ± 0·7 μm can be calculated from the mean intercept length measured directly on scanning electron micrographs and the obtained stereological correction factor of 1·9 (Underwood, 1970; Dimanov et al., 1998). The porosity of the sample after crystallization, determined by the Archimedes method, was ≈1% and the bulk water content was ≈0·17 wt %, estimated using Fourier-transform infrared spectroscopy (FTIR) and Beer–Lambert’s law with a molar extinction coefficient of 32 L(H2O) mol–1 (Beran, 1987). The calculated values are about half as high compared with calculations using the integrated absorbance (Johnson & Rossman, 2003). Analysis of the undeformed sample shows a SiO2-enriched residual glass content of ≈1 vol. % homogeneously dispersed in the sample and located at grain triple junctions. Sample deformation Sample Pl10_1 was deformed in torsion using a Paterson-type gas apparatus (GFZ-Potsdam) at 400 MPa confining pressure, 1100°C, and a constant twist rate of ≈2 × 10–5 s–1 (corresponding to a peripheral shear strain rate of 1·6 × 10–5 s–1), yielding a maximum shear stress of ≈11 MPa. The sample achieved a maximum shear strain of γ = 4·2 and underwent ≈10% lengthening, as evaluated from the change in the dimension of the sample before and after the experimental run. Before removing the stress at the end of the experiment, the sample was cooled below 600°C at a rate of 30–50°C min–1, and finally the confining pressure was released. Measured torque–twist data were converted to stress–strain data assuming power law creep. The stress–strain data indicate continuous moderate strain hardening (Rybacki et al., 2010). Several twist rate steppings were performed to determine the stress exponent n (Paterson & Olgaard, 2000; Rybacki et al., 2003) revealing a value of n ≈ 1, indicative of Newtonian creep. Analytical techniques used for microstructural analysis After deformation two thin sections were cut from sample Pl10_1, one parallel to the sample cylinder axis (longitudinal axial section) and one tangential to the sample surface (longitudinal tangential section). During torsion of solid cylindrical specimens strain and strain rate increase from ≈0 at the specimen central axis to a maximum at the sample periphery (Paterson & Olgaard, 2000). Thus in longitudinal axial sections the microstructural evolution with increasing strain and strain rate can be studied in a single sample. For backscatter electron image analysis thin sections were chemically etched with HF vapour for 20 s and studied with a scanning electron microscope (CamScan Mx2500 SEM) at the Department of Geosciences (Padova). The SEM is equipped with a LaB6 filament, the electron backscatter diffraction (EBSD) Channel 5.9 software package from HKL-technology (Oxford Instruments), and a semi-quantitative energy-dispersive spectrometry (EDX) system for microanalysis. For EBSD analysis the thin sections were chemically–mechanically polished (Syton-polished) to remove surface damage (Prior et al., 1999) and carbon-coated with a very thin film to prevent charging without degrading the pattern quality too much. Patterns were collected manually and indexing of the EBSD pattern was accepted if at least five Kikuchi bands were identified by the computer simulation and if the mean angular deviation (MAD) was lower than 1°. RESULTS Backscatter electron (BSE) imaging of the longitudinal axial section shows that close to the torsion axis anorthite crystals have no uniform orientation, and silica-rich melt is localized within triple junctions and occasionally also along grain boundaries (Fig. 1a). With increasing strain towards the sample periphery, anorthite crystals develop a shape-preferred orientation defining the foliation of the sample. With increasing strain the foliation is progressively inclined towards the shear direction. At a shear strain >2, a network of regularly spaced cavity-bearing shear bands is observed. Cavities within the bands are often filled with a silica-enriched melt (Fig. 1b, and crystalline quartz is observed within the bands at a shear strain >3 (Figs 1c and 2). Semi-quantitative EDX analysis of the silica-rich melt reveals an SiO2 content of 81 wt % together with Al2O3 and CaO contents of 12·5 and 6·5 wt % respectively. Fig. 1. Open in new tabDownload slide (a) BSE image showing that at low shear strain close to the torsion axis (γ < 0·5) anorthite does not have a uniform orientation and silica-rich melt is confined mostly to triple junctions. (b) BSE image showing detail of a cavity-bearing shear band at high shear strain (γ = 3·75). Large cavities in the upper left are quartz filled, whereas next to this quartz-rich domain smaller cavities and open grain boundaries contain silica-rich melt. (c) Reflected light image showing the distribution of cavity-bearing shear bands within sample Pl10_1 in the longitudinal axial section. Shear strain increases from left (γ ≈ 2·7) to right (γ = 4). Bright spots inside the cavity-bearing shear bands are quartz crystals; these appear for the first time at a shear strain ≈3. Fig. 1. Open in new tabDownload slide (a) BSE image showing that at low shear strain close to the torsion axis (γ < 0·5) anorthite does not have a uniform orientation and silica-rich melt is confined mostly to triple junctions. (b) BSE image showing detail of a cavity-bearing shear band at high shear strain (γ = 3·75). Large cavities in the upper left are quartz filled, whereas next to this quartz-rich domain smaller cavities and open grain boundaries contain silica-rich melt. (c) Reflected light image showing the distribution of cavity-bearing shear bands within sample Pl10_1 in the longitudinal axial section. Shear strain increases from left (γ ≈ 2·7) to right (γ = 4). Bright spots inside the cavity-bearing shear bands are quartz crystals; these appear for the first time at a shear strain ≈3. Fig. 2. Open in new tabDownload slide Details of a cavity-bearing shear band in a chemically etched, longitudinal, tangential section (γ = 3·75). (a) BSE image and interpretative sketch. Because chemical etching highlights not only compositional but also crystallographic contrasts, quartz crystals have been coloured in white. EDX and EBSD analyses were used to identify quartz. The white continuous lines trace the rotated external foliation. Quartz-rich domains are lens-shaped and exist next to quartz-free domains. Grain boundaries in the quartz-free domains are typically open. The inset shows the geometrical relationships between external foliation, internal foliation and cavity band orientation. The two horizontal lines confining the cavity-bearing shear band at the top and bottom in the sketch trace the orientation of the shear plane relative to that of the shear band. They do not refer to the width of the sample, which is about 150 times the width of the shear band. (b) Same BSE image as in (a) without overlays. Rose diagrams show anorthite orientation in the external foliation (upper left) and in the cavity-bearing shear band (lower right). Fig. 2. Open in new tabDownload slide Details of a cavity-bearing shear band in a chemically etched, longitudinal, tangential section (γ = 3·75). (a) BSE image and interpretative sketch. Because chemical etching highlights not only compositional but also crystallographic contrasts, quartz crystals have been coloured in white. EDX and EBSD analyses were used to identify quartz. The white continuous lines trace the rotated external foliation. Quartz-rich domains are lens-shaped and exist next to quartz-free domains. Grain boundaries in the quartz-free domains are typically open. The inset shows the geometrical relationships between external foliation, internal foliation and cavity band orientation. The two horizontal lines confining the cavity-bearing shear band at the top and bottom in the sketch trace the orientation of the shear plane relative to that of the shear band. They do not refer to the width of the sample, which is about 150 times the width of the shear band. (b) Same BSE image as in (a) without overlays. Rose diagrams show anorthite orientation in the external foliation (upper left) and in the cavity-bearing shear band (lower right). Cavity-bearing shear bands: quartz-rich and quartz-free domains Detailed inspection of the longitudinal axial thin section using transmitted plus reflected light microscopy, EDX and EBSD analysis reveals that quartz never appears outside the cavity-bearing shear bands. Crystallization of quartz occurs only within cavity-bearing shear bands at shear strains >3 (Fig. 1b and c), occupying lens-shaped domains arranged in an en echelon configuration that are separated from quartz-free domains (Fig. 2a). Within these two distinct domains, grain boundaries have different characteristics. Anorthite–anorthite grain boundaries in the quartz-free domains often contain melt (Fig. 1b). In contrast, within quartz-rich domains, anorthite–anorthite grain boundaries are mostly closed (Figs 1b and 2b). Figure 2a shows that the orientation of the Qtz-rich domains differs from that of the cavity-bearing shear bands. The bands form an angle of about 20° with the shear plane (see inset in 2b) and about 25° with the direction of the maximum principal stress, σ1. The single quartz-rich domains are oriented close to the shear plane. Image analysis reveals that the melt content within the quartz-free domains inside the cavity-bearing shear bands exceeds 3%, and that the quartz content within the quartz-rich domains is close to 10%. Along the cavity-bearing shear bands the foliation has been dragged and rotated (Fig. 2a and b). Anorthite grains inside the bands are rotated towards the main shear plane (lower right inset in Fig. 2b). The foliation is formed by the shape-preferred orientation of elongate anorthite grains, whose long axes align with an angle of ≈30° to the shear plane (upper left inset Fig. 2b). Anorthite crystals forming the external foliation are frequently twinned and transmission electron microscopy analysis reveals that the dislocation density is very low. The deflection of the main foliation along the cavity-bearing shear bands confirms that the sense of shear along the bands is synthetic with that imposed by the torsion experiments and the bands resemble S–C' type shear bands (Fig. 2a and b). Anorthite grains in the cavity-bearing shear bands usually have a lower aspect ratio (2·83 ± 0·15) than crystals aligned parallel to the external foliation (3·11 ± 0·12). Quartz forms single grains with rounded to straight grain boundaries (Fig. 3). The grain size and shape appears to be controlled by the space opened between the pulled-apart anorthite grains. The average grain size of quartz is 5–6 μm, but smaller (1 μm) and larger (10 μm) grains also exist. Occasionally, clusters of several grains occur, in which very small grains with rounded grain boundaries coexist next to larger ones (Fig. 3). The large majority of quartz grains are free of internal deformation microstructures. At anorthite contacts variable intensity of quartz impingement is evident, particularly along grain boundaries oriented orthogonally to the compression direction (Fig. 3). Fig. 3. Open in new tabDownload slide BSE images of chemically etched, longitudinal, tangential, thin section showing impingement and dissolution microstructures (thick white lines) in quartz (a). All quartz grains in the micrograph are contoured by a white dotted line. Because BSE imaging of chemically etched samples also reveals crystallographic orientation contrasts (see twinned anorthite crystals), darker grey shades are not exclusive to quartz. The upper and lower edges of the photomicrograph are parallel to the shear plane. Compression direction is shown by white arrows. Dissolution of quartz occurs preferentially along boundaries oriented at a high angle to the compression direction. Although quartz usually forms single grains, occasionally small grains with rounded grain boundaries coexist next to larger ones (lower left in image). (b) Same BSE image as in (a), but without contouring of quartz. Fig. 3. Open in new tabDownload slide BSE images of chemically etched, longitudinal, tangential, thin section showing impingement and dissolution microstructures (thick white lines) in quartz (a). All quartz grains in the micrograph are contoured by a white dotted line. Because BSE imaging of chemically etched samples also reveals crystallographic orientation contrasts (see twinned anorthite crystals), darker grey shades are not exclusive to quartz. The upper and lower edges of the photomicrograph are parallel to the shear plane. Compression direction is shown by white arrows. Dissolution of quartz occurs preferentially along boundaries oriented at a high angle to the compression direction. Although quartz usually forms single grains, occasionally small grains with rounded grain boundaries coexist next to larger ones (lower left in image). (b) Same BSE image as in (a), but without contouring of quartz. Quartz textures Analysis with EBSD reveals that the quartz crystals display a weak, but clearly perceptible, crystallographic preferred orientation (CPO; Fig. 4). The c [0001] axes preferentially align in a girdle oriented orthogonally to the compression direction and show a point maximum in the extension direction. Alignment of c-axes parallel to the compression direction is subordinate. Poles to a (11–20) and m (10–10) prisms form weak maxima about the compression direction. One of the three positive rhombohedra r (10–11) is parallel to the shear plane, whereas the other two are orthogonal. The negative rhombohedra z (01–11) are dispersed in a small girdle centred on an axis oriented closely orthogonally to the shearing direction. Overall, this distribution shows that rhombohedral faces are never orthogonal to the compression direction of the external stress field. The poles to the trigonal dipyramids ε (2–1–1–2) concentrate in a shear plane parallel girdle and in a maximum orthogonal to the shear plane. Fig. 4. Open in new tabDownload slide Quartz crystallographic orientation data plotted in pole figures. c-axes plot preferentially in a girdle orthogonal to the compression direction and form a discrete maximum in the extensional direction of the external stress field. Rhombohedra (r, z) and trigonal dipyramids (ε) are never orthogonal to the compression direction of the external stress reference frame. Fig. 4. Open in new tabDownload slide Quartz crystallographic orientation data plotted in pole figures. c-axes plot preferentially in a girdle orthogonal to the compression direction and form a discrete maximum in the extensional direction of the external stress field. Rhombohedra (r, z) and trigonal dipyramids (ε) are never orthogonal to the compression direction of the external stress reference frame. DISCUSSION Origin of Si-enriched melt Rybacki et al. (2006) showed that the intracrystalline water solubility for anorthite at the experimental conditions is about 0·04 wt %. However, the overall water content estimated for the undeformed sample Pl10 using FTIR was 0·17 ± 0·04 wt %. The presence of a strongly silica-enriched melt in sample Pl10_1 from which quartz crystallized suggests that most of the water was dissolved in the intercrystalline residual glass phase. The melt became silica-enriched during anorthite crystallization, either because the starting An-glass was locally non-stoichiometric or because of a Si4+ substitution process within anorthite (e.g. Si4+ ↔ 4H+, Rybacki et al., 2006). The solubility of H2O in a calcium–aluminosilicate melt is about 8 wt % at 400 MPa and 1170°C (McMillan et al., 1986), and depends only slightly on composition. Assuming that sample Pl10_1 has about 1 vol. % melt and about 0·13 ± 0·04 wt % free H2O, dissolution of most of it in the residual glass phase would result in a water-saturated melt. Quartz crystallization from a water-saturated melt during decompression Understanding the crystallization of quartz in the anorthite–quartz–H2O system (Stewart, 1967) helps to elucidate the driving forces that have controlled the redistribution of melt from the matrix to the cavity-bearing shear bands within sample Pl10_1. Figure 5 shows that at the experimental conditions of 1100°C and 400 MPa any water-saturated glass phase with a quartz content higher than 42 wt % will melt. Only melts with at least 58 wt % Qtz (composition of the eutectic at 400 MPa) would have crystallized quartz if the system were cooled below the eutectic temperature during deformation. Because the temperature during deformation was held constant at 1100°C, even from a water-saturated pure SiO2 melt no quartz would have crystallized, because at 400 MPa the liquidus temperature is below 1100°C. Fig. 5. Open in new tabDownload slide Projection of liquidi and solidi for water-saturated melt within the An–Qtz–H2O system at pressures between 0·1 and 500 MPa (modified from Stewart, 1967). At 1100°C (dotted line) and 400 MPa (experimental conditions) any water-saturated glass with more than 42 wt % Qtz falls in the liquidus field (light grey). Melt with more than 77 wt % Qtz can crystallize quartz if the system becomes sufficiently decompressed to enter the Qtz + Lq + H2O field (dark grey). The amount of decompression needed is a function of composition, and is >200 MPa for any melt having between 77 and 90 wt % Qtz and >100 MPa for any melt having 90–100 wt % Qtz. The black arrow shows the minimum decompression for a melt with 83 wt % Qtz (i.e. the reference melt composition). E0 to E5 are eutectics between 0·1 and 500 MPa; curved continuous lines are liquidi, and straight continuous lines are solidi between 0·1 and 500 MPa. Fig. 5. Open in new tabDownload slide Projection of liquidi and solidi for water-saturated melt within the An–Qtz–H2O system at pressures between 0·1 and 500 MPa (modified from Stewart, 1967). At 1100°C (dotted line) and 400 MPa (experimental conditions) any water-saturated glass with more than 42 wt % Qtz falls in the liquidus field (light grey). Melt with more than 77 wt % Qtz can crystallize quartz if the system becomes sufficiently decompressed to enter the Qtz + Lq + H2O field (dark grey). The amount of decompression needed is a function of composition, and is >200 MPa for any melt having between 77 and 90 wt % Qtz and >100 MPa for any melt having 90–100 wt % Qtz. The black arrow shows the minimum decompression for a melt with 83 wt % Qtz (i.e. the reference melt composition). E0 to E5 are eutectics between 0·1 and 500 MPa; curved continuous lines are liquidi, and straight continuous lines are solidi between 0·1 and 500 MPa. Within the An–Qtz–H2O system crystallization of quartz from a water-saturated melt at 1100°C and 400 MPa confining pressure requires dehydration of the melt during decompression. The exact amount of decompression is a function of melt composition, and varies from 300 MPa for a melt with a minimum quartz content of 77 wt % to >100 MPa for an extremely quartz-rich melt (Fig. 5). The analysed silica-rich residual melt inside the shear bands corresponds compositionally almost perfectly to a molten mixture of 34 wt % An and 66 wt % Qtz (Fig. 5). This is a highly differentiated melt, and we ignore the exact melt composition prior to quartz crystallization. The quantitative quartz–melt relationship inside the shear bands (Fig. 1b and c) suggests that the melt’s initial Qtz content must have been significantly higher than 66 wt %. EDX analysis of melt contained within the quartz-free An-aggregates deformed by Rybacki et al. (2008, 2010) revealed an SiO2 content of 90 wt % as well as Al2O3 and CaO contents of 6 and 4 wt % respectively, corresponding to a molten mixture of about 17 wt % An and 83 wt % Qtz (black arrow in Fig. 5). If we take this as reference composition for the undifferentiated water-saturated melt in sample Pl10_1, then >200 MPa decompression is required to crystallize Qtz from it. Implications from quartz crystallization for the melt segregation process What are the implications from the above outlined quartz-crystallization process for the understanding of the interplay between melt segregation and cavity formation? The observed cavities are unlikely to result from hydro-fracturing owing to melt overpressure (Rosenberg & Handy, 2000), as quartz crystallization requires significantly reduced fluid or melt pressure compared with the experimental conditions. Instead, our observations suggest that depressurized cavities form in response to grain boundary sliding (GBS), and that melt is drained by the pressure difference from the matrix into the opening cavities. A certain amount of GBS is required to produce sufficiently large under-pressure for quartz crystallization within the cavity-bearing shear bands. Apparently, this under-pressure is not achieved at shear strains between two and three. Cavity-bearing shear bands nucleate at a shear strain >2 in the periphery. At this shear strain the relative displacement of anorthite grains is limited and only small voids open within discrete shear bands. With increasing twist the strain threshold for cavity nucleation propagate inwards and new cavities open next to the already existing bands in the more peripheral highly strained zones, resulting in larger relative displacement of anorthite crystals and bigger, more depressurized cavities. A critical shear strain >3 is necessary to depressurize the cavities to the point that quartz can crystallize from the melt sucked from the neighbouring matrix and from the smaller voids inside the cavity-bearing shear bands themselves. As a consequence, quartz-rich domains arrange next to quartz-free domains inside the cavity-bearing shear bands at shear strains >3 (Fig. 2), whereas at shear strains below this critical value shear bands are quartz-free. Two main aspects arise from the above model and need to be explained. The first aspect can be explained by the rates of the processes operating inside the sample. At the experimental strain rate of 2 × 10–5 s–1, and the given stress exponent of n ≈ 1, closure of the cavities by deformation requires orders of magnitude more time than filling them with melt, even if stress concentration around the cavities allowed anorthite to deform locally by dislocation creep. Therefore, quartz had enough time to crystallize within the large cavities. How can big depressurized cavities sustain >200 MPa under-pressure without collapsing, when anorthite deforms at a flow stress of about 11 MPa during the experiment? Why does quartz not dissolve back into the melt when the pressure in the cavities is turning back to 400 MPa? The second aspect implies that the melt’s partial water pressure inside the cavities remained at about 100 MPa, even when the pressure inside the cavities turned back to 400 MPa. Two different processes are envisaged to drive this. The pressure drop in the cavities must result in phase separation inside the melt: water exsolves to a supercritical fluid and quartz crystallizes. Differentiated melt, supercritical fluid and quartz cannot be accommodated inside the same cavity, because the volume occupied by the supercritical water expands by a factor of four when the pressure is reduced from 400 to 100 MPa at 1100°C (Burnham et al., 1969; Helgeson & Kirkham, 1974) and cannot be compensated by the higher density of quartz. The observation that quartz occupies the big cavities implies that the fluid must have migrated into empty pore space opening in the sample with progressive deformation and associated fabric change. The amount of pore space required depends critically on the quantity of melt that decompressed and on its partial water pressure. If we hypothesize that the melt was redistributed over the full length of the shear bands, but quartz crystallized only at shear strain >3, then a maximum of half of the melt in the sample dehydrated down to a partial water pressure of about 100 MPa. According to McMillan et al. (1986), decompression of a water-saturated silica-rich melt from 400 to 100 MPa exsolves about 4·3 wt % H2O. This results in 0·02 wt % H2O that must have been redistributed in the sample’s open pore space, or less if the melt was water-undersaturated. Quartz CPO Quartz grains display a weak, but clearly perceptible, CPO (Fig. 4) defined by the alignment of c-axes normal to the compression direction of the applied external stress field and by the absence of rhombohedral and trapezohedral planes orthogonal to the compression direction. Formation of a CPO during deformation is commonly associated with dislocation creep, although a weak CPO in feldspar was found to develop during linear viscous creep (Gómez Barreiro et al., 2007). Activation of dislocation creep in quartz probably requires an order of magnitude higher differential stress than that applied in our experiments (Hirth & Tullis, 1994). Therefore, we consider it unlikely that the observed CPO results from dislocation creep of quartz. Instead, the observed quartz CPO is best explained by a mechanism implying melt-assisted dissolution–precipitation creep. This CPO formed under the control of the externally applied stress field once the quartz crystals formed a load-bearing framework inside the melt-filled voids. Tullis (1989) suggested that dissolution–precipitation creep will result in the formation of a CPO if crystallographic anisotropies control the growth and dissolution rates of mineral surfaces. A CPO can result either from selective dissolution of crystallographically unfavourably oriented minerals within a differential stress field (Hippertt, 1994) or potentially from rotation rate differences of crystals during deformation as a function of orientation and aspect ratio, the latter being controlled by the dissolution and growth kinetics of distinct crystal surfaces (Bons & den Brok, 2000). In our sample, dissolution of quartz is supported by impingement microstructures (Fig. 3). Comparison of our EBSD data with CPO data for quartz deformed in the dissolution–precipitation creep regime (Becker, 1995) supports the hypothesis that the measured CPO was formed during melt-assisted dissolution–precipitation creep controlled by the externally applied stress field. Becker (1995) has shown that during dissolution–precipitation creep quartz grains oriented with their c-axis at 45° to the compression direction are dissolved by pressure solution, whereas grains oriented with their c-axis either parallel or orthogonal to the compression direction are not. He interpreted this as evidence for rhombohedra (r {10–11}, z {01–1–1}) and trigonal dipyramids ε {2–1–12} being the most easily dissolvable crystal faces of quartz when they are oriented orthogonally to the compression direction. In our sample quartz rhombohedra or trigonal dipyramids are never oriented orthogonally to the compression direction of the externally applied stress field (Fig. 4). It is conceivable that crystals with these respective orientations were dissolved preferentially. In our sample, c-axes are preferentially oriented in a girdle orthogonal to the compression direction with a point maximum in the extension direction. Therefore, we believe that dissolved SiO2 has preferentially accreted crystals with these orientations. Because the EBSD data for the quartz crystals seem to reflect only the influence of the external stress state and not of the internal stress field, owing to shearing of the cavity bands themselves, we expect that the local strain is partitioned along the boundaries of these compact lenses. CONCLUSIONS The geometrical orientation and the spatial distribution of the cavity bands in our experimentally deformed anorthite aggregates are strikingly similar to those of the melt bands in the experimentally deformed samples of Holtzman et al. (2003a, 2003b, 2005) and Kohlstedt et al. (2010). Compaction length theory cannot explain their formation, nor can the preferred wetting of grain boundaries oriented at a low angle to σ1 (Zimmerman et al., 1999; Hier-Majumder et al., 2004) or the coalescence of favourably oriented melt pockets explain the formation of the cavity bands within our sample. It is likely that the slower flow rate of more highly viscous silica-rich melt in our samples inhibited the relaxation of local stress concentrations at the grain scale, resulting eventually in cavity nucleation, growth and coalescence. These differences may explain the discrepancies in the formation mechanisms of melt-rich bands in the samples of Holtzman et al. (2003a, 2003b, 2005) and Kohlstedt et al. (2010), and in our samples. In our sample, melt channelling bands in an S–C' geometry clearly form under Newtonian conditions at high shear strain. The melt channelling bands form in response to cavitation damage driven by co-operative grain boundary sliding not fully accommodated by diffusive mass transfer (Dimanov et al., 2007; Rybacki et al., 2008, 2010). The cavity-bearing shear bands in our sample are preferential conduits for melt transport through the deformed aggregate. There is unambiguous evidence that the opening of depressurized voids has driven melt to move from triple junctions and partially wetted grain boundaries to newly forming, cavity-bearing shear bands. Our study suggests that segregation and transport of silica-rich melt is efficient in rocks undergoing cavity damage during ductile, high-strain deformation. Because in the crust there is a positive feedback relationship between partial melting and strain localization (e.g. Hobbson et al., 1998; Marchildon & Brown, 2001; Brown, 2004; Zavada et al., 2007; Jamieson et al., 2011), we envisage that cavitation damage within shear zones should be an important control for melt flow through the feldspar-dominated ductile crust. FUNDING This work was supported by the Ministero dell’Istruzione dell’Università e della Ricerca (grant number PRIN 2007BWMWM8_001). ACKNOWLEDGEMENTS We thank Ben Holtzman, David Kohlstedt and an anonymous reviewer for their very helpful and constructive reviews. REFERENCES Becker A . Quartz pressure solution: influence of crystallographic orientation , Journal of Structural Geology , 1995 , vol. 17 (pg. 1395 - 1405 ) Google Scholar Crossref Search ADS WorldCat Beran A . OH groups in nominally anhydrous framework structures: An infrared spectroscopic investigation of danburite and labradorite , Physics and Chemistry of Minerals , 1987 , vol. 14 (pg. 441 - 445 ) Google Scholar Crossref Search ADS WorldCat Bons P J , den Brok B . Crystallographic preferred orientation development by dissolution–precipitation creep , Journal of Structural Geology , 2000 , vol. 22 (pg. 1713 - 1722 ) Google Scholar Crossref Search ADS WorldCat Brown M . The mechanisms of melt extraction from lower continental crust of orogens , Transactions of the Royal Society of Edinburgh: Earth Sciences , 2004 , vol. 95 (pg. 35 - 48 ) Google Scholar Crossref Search ADS WorldCat Burnham C W , Holloway J R , Davis N F . The specific volume of water in the range 1000 to 8900 bars, 20° to 900°C , American Journal of Science , 1969 , vol. 267-A (pg. 70 - 95 ) Google Scholar OpenURL Placeholder Text WorldCat Bussod G Y , Christie J M . Textural development and melt topology in spinel lherzolite experimentally deformed at hypersolidus conditions , Journal of Petrology, Special Lherzolite Issue , 1991 (pg. 17 - 39 ) Google Scholar OpenURL Placeholder Text WorldCat Daines M J , Kohlstedt D L . Influence of deformation on melt topology in peridotites , Journal of Geophysical Research , 1997 , vol. 102 (pg. 10257 - 10271 ) Google Scholar Crossref Search ADS WorldCat Dimanov A , Dresen G , Wirth R . High-temperature creep of partially molten plagioclase aggregates , Journal of Geophysical Research , 1998 , vol. 103 B5 (pg. 9651 - 9664 ) Google Scholar Crossref Search ADS WorldCat Dimanov A , Rybacki E , Wirth R , Dresen G . Creep and strain-dependent microstructures of synthetic anorthite–diopside aggregates , Journal of Structural Geology , 2007 , vol. 29 (pg. 1049 - 1069 ) Google Scholar Crossref Search ADS WorldCat Dresen G , Wang Z C , Bai Q . Kinetics of grain growth in anorthite , Tectonophysics , 1996 , vol. 258 (pg. 251 - 262 ) Google Scholar Crossref Search ADS WorldCat Gómez Barreiro J , Lonardelli I , Wenk R H , Dresen G , Rybacki E , Ren Y , Tomé C N . Preferred orientation of anorthite deformed experimentally in Newtonian creep , Earth and Planetary Science Letters , 2007 , vol. 264 (pg. 188 - 207 ) Google Scholar Crossref Search ADS WorldCat Helgeson H C , Kirkham D H . Theoretical prediction of the thermodynamic behavior of aqueous electrolyte at high pressures and temperatures: I. Summary of the thermodynamic/electrostatic properties of the solvent , American Journal of Science , 1974 , vol. 274 (pg. 1089 - 1198 ) Google Scholar Crossref Search ADS WorldCat Hess K U , Dingwell D B . Viscosities of hydrous leucogranitic melts: A non-Arrhenian model , American Mineralogist , 1996 , vol. 81 (pg. 1297 - 1300 ) Google Scholar OpenURL Placeholder Text WorldCat Hier-Majumder S , Leo P H , Kohlstedt D L . On grain boundary wetting during deformation , Acta Materiala , 2004 , vol. 52 (pg. 3425 - 3433 ) Google Scholar Crossref Search ADS WorldCat Hippertt J F M . Microstructures and c-axis fabrics indicative of quartz dissolution in sheared quartzites and phyllonites , Tectonophysics , 1994 , vol. 229 (pg. 141 - 163 ) Google Scholar Crossref Search ADS WorldCat Hirth G , Kohlstedt D L . Experimental constraints on the dynamics of the partially molten upper mantle 2. Deformation in the dislocation creep regime , Journal of Geophysical Research , 1995 , vol. 100 (pg. 15441 - 15449 ) Google Scholar Crossref Search ADS WorldCat Hirth G , Kohlstedt D L . Experimental constraints on the dynamics of the partially molten upper mantle: Deformation in the diffusion creep regime , Journal of Geophysical Research , 1995 , vol. 100 (pg. 1981 - 2001 ) Google Scholar Crossref Search ADS WorldCat Hirth G , Tullis J . The brittle–plastic transition in experimentally deformed quartz aggregates , Journal of Geophysical Research , 1994 , vol. 99 (pg. 1131 - 11747 ) Google Scholar Crossref Search ADS WorldCat Hobbson A , Bussy F , Hernandez J . Shallow-level migmatitization of gabbros in a metamorphic contact aureole, Fuerteventura Basal Complex, Canary Islands , Journal of Petrology , 1998 , vol. 39 5 (pg. 1025 - 1037 ) Google Scholar Crossref Search ADS WorldCat Holtzman B K , Kohlstedt D L . Stress-driven melt segregation and strain partitioning in partially molten rocks: effects of stress and strain , Journal of Petrology , 2007 , vol. 48 12 (pg. 2379 - 2406 ) Google Scholar Crossref Search ADS WorldCat Holtzman B K , Kohlstedt D L , Zimmerman M E , Heidelbach F , Hiraga T , Hustoft J . Melt segregation and strain partitioning: implications for seismic anisotropy and mantle flow , Science , 2003 , vol. 301 (pg. 1227 - 1230 ) Google Scholar Crossref Search ADS PubMed WorldCat Holtzman B K , Groebner N J , Zimmerman M E , Ginsberg S B , Kohlstedt D L . Stress-driven melt segregation in partially molten rocks , Geochemistry, Geophysics, Geosystems , 2003 , vol. 4 pg. 8607 doi:10.1029/2001GC000258 Google Scholar Crossref Search ADS WorldCat Holtzman B K , Kohlstedt D L , Phipps Morgan J . Viscous energy dissipation and strain partitioning in partially molten rocks , Journal of Petrology , 2005 , vol. 46 (pg. 2569 - 2592 ) Google Scholar Crossref Search ADS WorldCat Hui H , Zhang Y . Toward a general viscosity equation for natural anhydrous and hydrous silicate melts , Geochimica et Cosmochimica Acta , 2007 , vol. 71 2 (pg. 403 - 416 ) doi:10.1016/j.gca·2006·09·003 Google Scholar Crossref Search ADS WorldCat Jamieson R A , Unsworth M J , Harris N B W , Rosenberg C L , Schulmann K . Crustal melting and the flow of mountains , Elements , 2011 , vol. 7 (pg. 253 - 260 ) Google Scholar Crossref Search ADS WorldCat Johnson E A , Rossman G R . The concentration and speciation of hydrogen in feldspars using FTIR and H MAS NMR spectroscopy , American Mineralogist , 2003 , vol. 88 (pg. 901 - 911 ) Google Scholar Crossref Search ADS WorldCat Kohlstedt D L , Holtzman B K . Shearing melt out of the Earth: An experimentalist’s perspective on the influence of deformation on melt extraction , Annual Review of Earth and Planetary Sciences , 2009 , vol. 37 (pg. 16.1 - 16.33 ) doi:10.1146/annurev.earth.031208.100104 Google Scholar Crossref Search ADS WorldCat Kohlstedt D L , Zimmerman M E . Rheology of partially molten mantle rocks , Annual Review of Earth and Planetary Sciences , 1996 , vol. 24 (pg. 41 - 62 ) Google Scholar Crossref Search ADS WorldCat Kohlstedt D L , Zimmerman M E , Mackwell S J . Stress-driven melt segregation in partially molten feldspathic rocks , Journal of Petrology , 2010 , vol. 51 (pg. 9 - 19 ) doi:10.1093/petrology/egp043 Google Scholar Crossref Search ADS WorldCat Marchildon N , Brown M . Melt segregation in late syn-tectonic anatectic migmatites: an example from the Onawa contact aureole, Maine, USA , Physics and Chemistry of the Earth (A) , 2001 , vol. 26 (pg. 225 - 229 ) Google Scholar Crossref Search ADS WorldCat McKenzie D . The generation and compaction of partially molten rock , Journal of Petrology , 1984 , vol. 25 (pg. 713 - 765 ) Google Scholar Crossref Search ADS WorldCat McMillan P , Peraudeau G , Holloway J , Coutures J P . Water solubility in a calcium aluminosilicate melt , Contributions to Mineralogy and Petrology , 1986 , vol. 94 (pg. 178 - 182 ) Google Scholar Crossref Search ADS WorldCat Paterson M S , Olgaard D L . Rock deformation tests to large shear strains in torsion , Journal of Structural Geology , 2000 , vol. 22 (pg. 1341 - 1358 ) Google Scholar Crossref Search ADS WorldCat Prior D J , Boyle A P , Brenker F , Cheadle M C , Day A , Lopez G , Peruzzo L , Potts G J , Reddy S , Spiess R , Timms N E , Trimby P , Wheeler J , Zetterström L . The application of electron backscatter diffraction and orientation contrast imaging in the SEM to textural problems in rocks , American Mineralogist , 1999 , vol. 84 (pg. 1741 - 1759 ) Google Scholar Crossref Search ADS WorldCat Rosenberg C L , Handy M R . Syntectonic melt pathways during simple shearing of a partially molten rock analogue (Norcamphor–Benzamide) , Journal of Geophysical Research , 2000 , vol. 105 (pg. 3135 - 3149 ) Google Scholar Crossref Search ADS WorldCat Rybacki E , Paterson M S , Wirth R , Dresen G . Rheology of calcite–quartz aggregates deformed to large strain in torsion , Journal of Geophysical Research , 2003 , vol. 108 (pg. 1 - 24 ) Google Scholar OpenURL Placeholder Text WorldCat Rybacki E , Gottschalk M , Wirth R , Dresen G . Influence of water fugacity and activation volume on the flow properties of fine-grained anorthite aggregates , Journal of Geophysical Research , 2006 , vol. 111 B3 pg. B03203 Google Scholar Crossref Search ADS WorldCat Rybacki E , Wirth R , Dresen G . High-strain creep of feldspar rocks: Implications for cavitation and ductile failure in the lower crust , Geophysical Research Letters , 2008 , vol. 35 pg. L04304 doi:10.1029/2007GL032478 Google Scholar Crossref Search ADS WorldCat Rybacki E , Wirth R , Dresen G . Superplasticity and ductile fracture of synthetic feldspar deformed to large strain , Journal of Geophysical Research , 2010 , vol. 115 pg. B08209 doi:10.1029/2009JB007203 Google Scholar Crossref Search ADS WorldCat Scott D R , Stevenson D J . Magma solitons , Geophysical Research Letters , 1984 , vol. 11 (pg. 1161 - 1164 ) Google Scholar Crossref Search ADS WorldCat Stevenson D J . Spontaneous small-scale melt segregation in partial melts undergoing deformation , Geophysical Research Letters , 1989 , vol. 16 (pg. 1067 - 1070 ) Google Scholar Crossref Search ADS WorldCat Stewart D B . Four-phase curve in the system CaAl2Si2O8–SiO2–H2O between 1 and 10 kilobars , Schweizerische Mineralogische und Petrographische Mitteilungen , 1967 , vol. 47 (pg. 35 - 59 ) Google Scholar OpenURL Placeholder Text WorldCat Tullis T E . Development of preferred orientation due to anisotropic dissolution/growth rates during solution-transfer creep , EOS Transactions, American Geophysical Union , 1989 , vol. 70 (pg. 457 - 458 ) Google Scholar OpenURL Placeholder Text WorldCat Underwood E E . , Quantitative Stereology , 1970 Boston, MA Addison Wesley pg. 274 Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zavada P , Schulmann K , Konopasek J , Ulrich S , Lexa O . Extreme ductility of feldspar aggregates—melt-enhanced grain boundary sliding and creep failure: Rheological implications for felsic lower crust , Journal of Geophysical Research , 2007 , vol. 112 pg. B10210 doi:10.1029/2006JB004820 Google Scholar Crossref Search ADS WorldCat Zimmerman M E , Zhang S , Kohlstedt D L , Karat S . Melt distribution in mantle rocks deformed in shear , Geophysical Research Letters , 1999 , vol. 26 (pg. 1505 - 1508 ) Google Scholar Crossref Search ADS WorldCat © The Author 2012. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com © The Author 2012. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com
TI - Depressurized Cavities within High-strain Shear Zones: their Role in the Segregation and Flow of SiO2-rich Melt in Feldspar-dominated Rocks
JF - Journal of Petrology
DO - 10.1093/petrology/egs032
DA - 2012-09-01
UR - https://www.deepdyve.com/lp/oxford-university-press/depressurized-cavities-within-high-strain-shear-zones-their-role-in-f0sTGGqgv0
SP - 1767
EP - 1776
VL - 53
IS - 9
DP - DeepDyve
ER -