Pdf/7/1/40/3341343/40.Pdf by Guest on 26 September 2021 High P, T Tomography on Deformed Composites

In situ high-pressure and high-temperature X-ray microtomographic imaging during large deformation: A new technique for studying mechanical behavior of multiphase composites

Yanbin Wang1, Charles Lesher2, Guillaume Fiquet3, Mark L. Rivers1, Norimasa Nishiyama1,*, Julien Siebert3, Jeffery Roberts4, Guillaume Morard3, Sarah Gaudio2, Alisha Clark2, Heather Watson4, Nicolas Menguy3, and Francois Guyot3 1Center for Advanced Radiation Sources, University of Chicago, 5640 S. Ellis Avenue, Chicago, Illinois 60637, USA 2University of California at Davis, Department of Geology, One Shields Avenue, Davis, California 95616, USA 3Institut de Minéralogie et de Physique des Milieux Condensés, Institut de Physique du Globe de Paris, 140 rue de Lourmel, 75015 Paris, France 4Lawrence Livermore National Laboratory, P.O. Box 808, L-184, Livermore, California 94551, USA

ABSTRACT static stress, migrated into silicate grain static stress, whereas weaker materials are more bound aries, and propagated in a manner susceptible to fl ow. Although researchers have We have examined the microstructural similar to melt inclusions in a deforming solid investigated the effects of LPO on fl ow proper- evolution of a two-phase composite (olivine matrix. The grain size of the silicate matrix ties and seismic anisotropy for several mantle + Fe-Ni-S) during large shear deformation, was signifi cantly reduced under large strain minerals (e.g., Bascou et al., 2002; Mainprice using a newly developed high-pressure X-ray deformation. The strong shape-preferred et al., 2004; Bystricky et al., 2006; Long et al., tomography microscope. Two samples were orientation thus developed can profoundly 2006; Warren et al., 2008), the effects of SPO examined: a load-bearing framework–type influence a composite’s bulk elastic and have received little attention, especially for texture, where the alloy phase (Fe-Ni-S) was rheological properties. High-pressure–high deep Earth dynamic processes. present as isolated spherical inclusions, and temperature tomography not only provides In studying the fl ow laws of multiphase com- an interconnected network–type texture, quantitative observations on textural evolu- posites, Handy (1994a) showed that two end- where the alloy phase was concentrated tion, but also can be compared with simula- member texture types can be defi ned for the along the silicate grain boundaries and tion results to derive more rigorous models plastic regime: (1) a load-bearing framework tended to form an interconnected network. of the mechanical properties of composite (LBF) texture, where the stronger phase sur- The samples, both containing ~10 vol% alloy materials relevant to Earth’s deep mantle. rounds isolated pockets of the weaker phase, inclusions, were compressed to 6 GPa, fol- and (2) interconnected layers of the weaker lowed by shear deformation at temperatures INTRODUCTION phase (IWL) separating boudins and clasts of up to 800 K. Shear strains were introduced the stronger phase. Under large deformation, by twisting the samples at high pressure and The entire rocky interior of Earth is com- bulk composite materials can undergo a tran- high temperature. At each imposed shear posed of multiphase composites. Volumetric sition from the initial LBF texture to IWL by strain, samples were cooled to ambient tem- fractions of the constituents and the spatial developing SPO (e.g., Handy, 1994a; Holyoke perature and tomographic images collected. relation of one phase relative to another have and Tullis, 2006; Takeda and Griera, 2006). The The three-dimensional tomographic images profound effects on the physical properties of materials science, mechanical engineering, and were analyzed for textural evolution. We the bulk constituents. It is well known that for structural geology literature provides a wealth found that in both samples, Fe-Ni-S, which a composite undergoing large deformation, two of information on models of the mechanical is the weaker phase in the composite, under- types of preferred orientations may develop. behaviors of composite materials (e.g., Guoan went signifi cant deformation. The resulting In cases where the constituents are deformed and Castañeda, 1993; Lee and Paul, 2005; lens-shaped alloy phase is subparallel to within the dislocation creep regime, a lattice- Tandon and Weng, 1986; Fletcher, 2004; Handy, the shear plane and has a laminated, highly preferred orientation (LPO) can develop (e.g., 1994a; Kanagawa, 1993; Treagus, 2002). anisotropic interconnected weak layer tex- see Karato, 1998; Wenk, 2006). Since deform- Experimental efforts have also yielded valua- ture. Scanning electron microscopy showed ing a weaker phase is energetically more ble information on the mechanical behavior of that many alloy inclusions became fi lm-like, favorable than deforming a stronger one, a crustal rocks and rock analogs (e.g., Jordan, with thicknesses <1 μm, suggesting that Fe- shape-preferred orientation (SPO) is also inevi- 1987; Shea and Kronenberg, 1993; Holyoke and Ni-S was highly mobile under nonhydro- table (Handy, 1994a, 1994b). Here we use the Tullis, 2006), by relating stress-strain curves terms “strong” and “weak” to indicate relative with textural information. *Present address: Geodynamics Research Center, rheological property contrast: stronger materi- Because of the technical challenges, virtu- Ehime University, Matsuyama, Ehime, Japan. als are more resistant to fl ow under nonhydro- ally no rheological data are available under

Geosphere; February 2011; v. 7; no. 1; p. 40–53; doi: 10.1130/GES00560.1; 12 ﬁ gures; 1 table; 3 animations.

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 High P, T tomography on deformed composites

deep-mantle conditions for rock composites cor- copy, to examine microstructures beyond the with homogeneous composition. Complex den- responding to the mineralogies in the transition current resolving power of the HPXTM. In sity contrast was observed in the darker silicate zone and the lower mantle. Theoretical bound- this report we present results on composites regions in backscattered electron micrographs ing analyses (e.g., Handy, 1994a; Takeda, 1998) consisting of Fe-Ni-S and silicates and dis- (labeled m in Fig. 1; also see inset). EPMA provide limits on the composite fl ow stress. cuss the potential implications for the rheol- showed that the compositions in these areas These studies indicate that the lower bound ogy and dynamic processes of Earth’s mantle. varied from stoichiometric olivine to more Mg corresponding to a homogeneous strain rate The current HPXTM setup does not allow us rich [i.e., with the (Mg + Fe)/Si ratio >2]. We approaches the strength of the composite with to measure stress and strain directly. Hence, we interpret these regions as quenched melt. Many an ideal IWL texture, whereas the upper bound concentrated on characterizing texture during fi ne Fe-Ni-S inclusions were present in these associated with homogeneous stress approaches large strain deformation. In the future we plan regions (a few are indicated by the arrows). that of an ideal LBF texture. to combine high-pressure tomography with dif- Most important, essentially all the Fe-Ni-S Quantitative analyses of the effects of texture fraction analyses using a multielement detector inclusions were spherical in shape, with no con- on rheological properties are diffi cult, if not so that we can relate textural evolution directly nection to the nearby neighbors (Fig. 1). The impossible, to determine using analytical solu- to stress-strain measurements. nature of this complex texture is still not well tions. One approach is to conduct numerical understood. It was probably due to an unex- simulations on a mixture of phases with known EXPERIMENTAL PROCEDURE pected power surge during sample synthesis, physical properties and a prescribed spatial and causing the temperature to fl uctuate and olivine shape distribution. So far, numerical models Sample Preparation and Characterization to melt incongruently to enstatite and liquid, as of large shear deformation have been limited has been reported at high pressure by Ohtani to two dimensions (e.g., Takeda and Griera, For this study we chose a mixture of olivine et al. (1998). An excursion into the melting 2006). The applicability of these results to more and Fe-Ni-S as the model composite for three region is further supported by the spherical realis tic three-dimensional (3D) situations may reasons: (1) silicate and iron-nickel sulfi de shape of Fe-Ni-S inclusions dispersed through- be limited because the effects of geometry and do not have a strong chemical interaction, so out the sample. While melting was not intended, mechanical interactions cannot be accounted for we can focus on the physical aspects of large this serendipitous event created an LBF texture, in the third dimension. Madi et al. (2005) con- deformation; (2) the two phases have a large the initial mechanical response of which should ducted a 3D fi nite-element study to model the X-ray absorption contrast, allowing us to read- be dominated by the strong phase (in this case behavior of the two-phase lower mantle miner- ily separate domains within the tomographic the silicate matrix). alogical assemblage during deformation. These images; and (3) the strength contrast between Sample B was synthesized in a graphite cap- simulations were limited to strains of a few olivine and Fe-Ni-S, expressed as a ratio of sule at a more stable temperature, where only percent because modeling large deformation shear stresses, ranges from 500:1 to 1000:1 (see the Fe-Ni-S alloy was molten. The resulting is time consuming and costly. Although their Discussion herein), similar to that between the texture consisted of an olivine matrix separated preliminary high strain calculations suggested a predicted strengths of silicate perovskite and by interconnected channels of quenched Fe- trend toward a transition from LBF-like texture ferropericlase (Yamazaki and Karato, 2001), Ni-S melt (Fig. 2). This texture approaches the to IWL, no quantitative results were given. two important components of Earth’s mantle. idealized IWL texture where the interconnected Another approach is to conduct large strain Two specimens, similar to those reported by phase is the weak phase. deformation experiments on analog composites Roberts et al. (2007), were prepared in a pis- In both samples, the metallic inclusions were to obtain a quantitative statistical description ton cylinder apparatus at Lawrence Livermore composed of quench crystals rich in FeNi or (SPO and LPO) of the microstructure. Effec- National Laboratory. Both samples A and B FeS (Figs. 1 and 2). Due to the small amount tive rheological properties of the composite may were made by mixing San Carlos olivine pow- of Ni in the starting material, FeS-rich quench

be either measured directly or predicted based der with (Fe92.5Ni7.5)S (hereafter referred to as crystals dominate the alloy inclusions (volume on the observed microstructural details and the Fe-Ni-S alloy or the metallic phase). Samples fraction >95%). Thus, the mechanical behavior known properties of the constituent phases in A and B contained 12 vol% and 10 vol% of the alloy inclusions is dominated by FeS. the composite. The latter approach is widely metallic alloy, respectively. Both were sintered used in industrial engineering and rock physics for 24 h at 1 GPa and 1573 K. At these con- High-Pressure–High-Temperature (e.g., Garboczi et al., 1999; Arns et al., 2002; ditions the Fe-Ni-S was molten (Gaetani and Shear Deformation and Tomography Keehm, 2003). Grove, 1999). Data Collection In an effort to better constrain the relation- The samples were analyzed using a JEOL ship between composite material rheology and JSR-1000 scanning electron microscope (SEM) Experiments reported here were conducted textural evolution, we performed sample analy- at the Geodynamics Research Center, Ehime at the bending magnet beamline (13-BM-D) ses using the high-pressure X-ray tomography University, Japan, and a Zeiss Ultra 55 SEM of Sector 13 (GSECARS, GeoSoilEnviro- microscope (HPXTM). With HPXTM, we can at the Institut de Minéralogie et de Physique CARS) at the Advanced Photon Source, track the development of textures in three dimen- des Milieux Condensés, Paris. Electron probe Argonne National Laboratory. Details of the sions, detailing the transition from LBF to IWP microanalyses (EPMA) were conducted with a experimental setup can be found in Wang et al. texture at high pressure, high temperature, and Cameca SX-50 microprobe at Centre Camparis, (2005) and Lesher et al. (2009). Figure 3 shows large shear strains (Wang et al., 2005). We have Université Pierre et Marie Curie, Paris. the sample cell assembly that was placed in studied selected analog composite materials to Sample A, which was encapsulated in a a modiﬁ ed Drickamer anvil cell to generate examine their textural evolution under simple boron-nitride (BN) container, possessed a com- high pressure and high temperature (Uchida shear, with a spatial resolution of 4–5 µm, up plex texture for the silicates, indicative of par- et al., 2008). The tungsten carbide (WC) anvils to 10 GPa and 1300 K. The recovered samples tial melting. EPMA conﬁ rmed that the large (10 mm diameter) had a tapering angle of 20° were examined using scanning electron micros- grains in sample A (Fig. 1) were enstatite (En) and a truncation diameter of 4 mm. A mixture of

Geosphere, February 2011 41

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 Wang et al.

En step required ~30 min) during heating and then cooled to room temperature (but still under pressure) for the microtomography data collection m (see following). This procedure was repeated until the relative rotation between the two anvils reached 725°. We defi ne the maximum shear m strain as γ = (α r/h), where α is the true twist En angle, and r and h are the radius and height of the sample, respectively. Note that some slippage occurs between the rotating anvils and the En sample during tortional deformation. The apparent twist angle, i.e., the angle between the two rotating anvils, does not represent the true value of α. Careful analysis of the relative changes in the spatial distribution of metallic inclusions in the tomographic images (see following) was En used to estimate α. m After reaching a target shear strain level at high pressure and high temperature, the sample was rapidly cooled to room temperature for En tomography data collection. This process pre- vented sample creep during data collection and typically required ~30–120 min for a complete data set. To collect the tomography data, we rotated the two Drickamer anvils in unison with Figure 1. Backscattered electron image of sample A, as synthesized. Light gray areas radiographs collected every 0.5°, from 0° to are spherical Fe-Ni-S inclusions. The brighter contrast in these inclusions was caused by 179.5°. The incident white radiation was mono- quenching the near-eutectic composition during melting. Fine, brighter regions are almost chromatized by a Si (111) double-crystal mono- pure Fe-Ni, and darker regions are close to FeS in composition. The darker background con- chromator, with a photon energy of 35–40 keV tains two types of grains. Large euhedral grains with homogeneous appearance are enstatite for imaging, during which the incident beam (En); the darkest, complex areas are intergrowths of quenched melt (m) with compositions was collimated to ~2 mm × 3 mm. A CoolSNAP close or slightly more Mg-enriched than olivine. Also note the tiny Fe-Ni-S spherical inclu- charge-coupled device (CCD) camera was used sions (three arrows) that appear only in the complex, olivine-like grains. Inset shows the for image recording. Procedures for tomogra- contrast between the two silicate regions. phy data collection were similar to conventional ambient-condition microtomography (Rivers et al., 1999; Gualda and Rivers, 2006). Flat-fi eld amorphous boron and epoxy (BE, with a boron/ was used; however, to maintain temperature images were taken from a dummy Ultem-1000 epoxy weight ratio of 4:1) was used as the pres- within 15% of the desired value, we applied a ring, which had dimensions identical to those of sure medium. Two sintered MgO disks placed temperature versus power consumption calibra- the containment ring used in the high-pressure between the sample and the anvils served as tion at 2 t, using a C-type thermocouple inserted cell. Only two fl at-fi eld images were taken for both a pressure standard and a mechanical cou- through the Ultem-1000 containment ring– each tomographic data set, at 0° and 179.5°. pling for shear deformation. A containment ring pressure medium, with the junction located at These fl at-fi eld images were collected by driv- made of Ultem-1000 (polyetherimide; http:// the center of the graphite heater. ing the entire apparatus horizontally, perpendic- k-mac-plastics.net/data%20sheets/ultem_1000 A conventional energy-dispersive X-ray ular to the incident beam. Images were binned _technical_property_data.htm) was placed diffraction (EDXD) technique was used to by 2 × 2 pixels for sample A, corresponding outside the anvils and the cell assembly. The determine pressure during experiments. Two to a pixel size of 3.67 μm, and each image Ultem-1000 was chosen because its toughness pairs of WC slits driven into the beam path required ~2 s for data collection. For sample B, and rigidity at high temperature and its low collimated the white beam to dimensions of images were unbinned, corresponding to a pixel X-ray absorption. The cell was loaded such 0.1 mm × 0.1 mm, and EDXD signal was col- size of 2.4 μm, and each image required 8 s of that its cylindrical axis was vertical. Each anvil lected from the MgO pressure standard using acquisition time. was supported by a low-profi le thrust bearing a Ge solid-state detector (SSD) with neces- One unique feature of the 13-BM-D setup so that it could be rotated under the hydraulic sary collimating optics on the diffraction side. is the ability to quickly switch between the load applied by the 250 t hydraulic press (Wang Details of the diffraction and imaging setup polychromatic and monochromatic modes et al., 2005). The horizontal X-ray beam passed were given in Lesher et al. (2009). Pressures for EDXD and tomography data collection, through the sample, which was constrained by were determined using the equation of state respectively. The two monochromator crys- the Ultem-1000 containment ring (for details of specifi ed by Speziale et al. (2001). tals are offset vertically by ~14 mm. Lower- the setup, see Wang et al., 2005). Shear strain was imposed by rotating the ing the entire monochromator assembly by A cylindrical graphite furnace with the cur- upper and lower anvils in opposite directions ~6 mm allowed the white beam to pass through rent passing through the upper and lower anvils at any given pressure and temperature. The the opening between the two crystals and provided resistive heating. No thermocouple sample was twisted at a step size of 60° (each enter the high-pressure tomography apparatus

42 Geosphere, February 2011

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 High P, T tomography on deformed composites

(Lesher et al., 2009). The offset between the yet, this setup has the potential for in situ stress Image Reconstruction and Processing white and monochromatic beams remained measurement by replacing the current single- constant throughout the experiment, and both element SSD with a multielement detector, Three-dimensional tomographic reconstruc- the sample vertical position and the beam col- using the technique developed for rotational tions were performed using the GSECARS stan- limating slits were adjusted accordingly by deformation apparatus (Nishihara et al., 2008; dard tomography software Tomo_display (Rivers motorized stages. Although not implemented Kawazoe et al., 2009). et al., 1999). A Hahn fi lter was used during image processing, and a ring smoothing parameter of 20–25 was adopted to minimize ring artifacts. Visual examination of individual slices indi- A cated that small features with 4–5 pixels could be recognized. Quantitative evaluation of spatial resolution is diffi cult. We examined the 3D reconstructed images using the software package Blob3D (Ketcham and Carlson, 2001) and a dis- playing software vol_animate provided by Guil- herme Gualda (Vanderbilt University, Nashville, Tennessee) (Gualda and Rivers, 2006). Figure 2. Backscattered elec- Statistic analysis of microstructures was tron images of sample B, as carried out with Quant3D (Ketcham, 2005a, synthesized. Bright areas are 2005b). Quant3D extracts statistical fabric Fe-Ni-S quenched melts, which information from the tomographic image appear along grain boundaries reconstruction of a composite using a series of the silicate (pure olivine in of so-called “star” methods: star volume dis- this case). Some melts appear tribution (SVD) (Cruz-Orive et al., 1992), star to form channels (arrow in B). B length distribution (Odgaard et al., 1997), and In B, many olivine grain bound- mean intercept length (Cowin, 1986). In these aries show tubular channels, methods, points are placed within a material presumably caused by losing of interest, and lines are measured from these Fe-Ni-S during polishing. points in various directions until they encoun- ter an internal boundary (e.g., an interface between the phases present in the material). For SVD, these lines are considered infi nitesi- mal cones, with their vertex (the analysis direction vertex) at the origin point and subtending a solid angle as they approach the material inter- υ* face. The star volume component V for direction ω is defi ned as:

π n υω* ()= 3 ()ω Vi∑ L , (1) 3n i=1 WC anvil

TEL = 4.00 where n is the number of points used, and Li is the length of the line passing through point i with orientation ω that stays entirely within the material of interest. The star volume from a particular 0.65 MgO point within an object is the star volume com- BN ponents summed over all orientations. By this deﬁ nition, the star volume of a convex object is B + epoxy 1.20 1.20 2.50 equal to its standard volume, whereas for a more 4 : 1 by weight irregular object, the star volume is the volume that can be “seen” from the test point. Among the three measures, SVD is the most sensitive to tex- 0.65 MgO ture and therefore was adopted in our analyses. To summarize the SVD measurements, 2.50 we used a 3D version of a rose diagram (see Graphite Ketcham , 2005a, ﬁ g. 6 therein), created by pro- jecting each analysis direction vertex from the Figure 3. A schematic illustration of the cell assembly used in the high-pressure–high- unit sphere inward or outward from the origin temperature tomography experiments. Units are in millimeters. BN—boron-nitride; WC— inside an inclusion. Vertex positions and colors tungsten carbide. were normalized by dividing by the maximum

Geosphere, February 2011 43

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 Wang et al.

measurement value. A normalized value of 1.0 to 700 K. The maximum pressure was 6 GPa. Shear Strain Determination is plotted in red at a distance from the origin Sample B (R0912) was compressed to 3.5 t and equal to the coordinate axis length. Lower val- heated to 800 K. The maximum pressure was After each tomographic reconstruction, Fe- ues are plotted in successively “cooler” rainbow 4 GPa. The difference in pressure and tempera- Ni-S inclusions were identifi ed by selecting the colors and proportionally closer to the origin. ture conditions does not affect in any signifi - image contrast threshold according to the initial This coloring convention allows the relative cant way the mechanical properties of the con- volume fraction of the metallic phase. The spa- measurement values to be easily visualized. For stituents in the samples. During simple shear tial distribution of the inclusions and changes example, dark blue indicates that the difference deformation, both samples were twisted to an in inclusion shape after compression, heating, between the minimum and maximum measure- apparent shear angle of 725°. The rotational and twisting were used to estimate the maxi- ments is roughly a factor of 10. Also plotted on speeds of the anvils corresponded to a maximum mum shear strain applied to the sample. Figure 4 these diagrams are the eigenvector directions, apparent shear strain rate of ~5 × 10–4 s–1 near the shows an example of the shear strain determined with axis lengths scaled by their associated outer diameter of the sample. This rate is con- for sample A. We examined Fe-Ni-S inclusions at eigenvalues. Details of the analysis technique sidered an upper bound, given slippage between both ends of the sample and used them as strain were given in Ketcham (2005a, 2005b) and the sample, the pressure media, and the anvils. markers (Fig. 4A). For a Drickamer-type oppos- Ketcham and Carlson (2001). Initially, signifi cant slippage occurred between ing anvil assembly, the shear strain increases lin- the anvils and the sample, so that anvil twisting early from the center to the outer diameter. By RESULTS had little effect on shear in the sample. How- locating strain markers at both ends of the sample ever, with increasing anvil rotation, mechanical and knowing the distance of the markers from the Results are presented for two deformation coupling between the sample and the anvils was sample axis, we can thus calculate shear strain. experiments (samples A and B). Sample A gradually established, and shear strains of ~15 Figures 4B through 4H present a series of (run R0875) was compressed to 5 t and heated were achieved by the end of the experiments. tomographic image pairs of the top (left) and

Figure 4. Strain markers and determination of maximum A A B E shear strain during deforma- 175°-65°-2.51 tion at high pressure and high temperature. (A) Illustration of the locations of the markers examined: a pair of Fe-Ni-S inclusions, one at the top (viewed from above) and the other at the bottom (viewed from below) of the sample, is B F located near the cylindrical 0°-0°-0 325°-168°-6.81 surface, and then their relative rotation is determined. From this angle, the radius and the total height of the sample (also imaged) and the maximum shear strains are determined. (B–H) The locations of the marker pair (left: top; right: C bottom) at various twist angles. 50°-2°-0.07 G 525°-274°-11.30 In C, the white arrows indicate the direction of shear. Samples’ diameters are ~1 mm.

D 100°-30°-1.06 H 725°-353°-14.56

44 Geosphere, February 2011

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 High P, T tomography on deformed composites

bottom (right) ends, with the olivine matrix digi- bottom anvils were rotated by 50° relative to each introduced signifi cant fl attening of the origi- tally removed. On each end, a marker is identi- other, whereas the sample was actually twisted by nally spherical inclusions (e.g., cf. A1 and B1 fi ed (tied by the red line to the sample axis) and 2°, corresponding to a shear strain of 0.07. in Fig. 5). In later stages during shear deforma- tracked throughout the deformation process. The Table 1 summarizes observations for sample tion, axial strain continued to increase at a much three-number index for each image pair desig- A. Signifi cant axial shortening (~21%) also slower rate. nates, respectively, the relative rotation of the top occurred during cold compression. As the sam- Alloy inclusions in sample B were smaller and bottom anvils, the actual twist angle between ple was heated to 700 K and cooled to room and more complicated in shape than those in the top and bottom ends, and the calculated shear temperature, axial strain increased to 54% (row sample A, making quantitative analysis chal- strain. Thus 50°-2°-0.07 indicates that the top and under 5/640/0 in Table 1). Such large axial strain lenging. A 3D animation showing the spatial

TABLE 1. SHEAR STRAIN IN SAMPLE A 0/300/01 5/300/0 5/640/0 5/640/50 5/640/100 5/640/175 5/640/325 5/640/525 5/640/725 Sample height, h (mm) 1.40(8) 1.10(8) 0.64(8) 0.60(8) 0.60(8) 0.57(8) 0.56(8) 0.55(8) 0.55(8)

Sample radius, r (mm) 0.80(8) 0.92(8) 1.20(8) 1.22(8) 1.22(8) 1.26(8) 1.30(8) 1.30(8) 1.30(8)

Axial strain (%) 0 21(2) 54(2) 57(2) 57(2) 59(2) 60(2) 61(2) 61(2) Actual twist angle, α (°)2 0 0 0 2(3) 30(3) 65(3) 168(3) 274(3) 353(3) Shear strain 3 0 0 0 0.07(3) 1.06(3) 2.51(3) 6.81(3) 11.30(3) 14.56(3) Inclusion mean aspect ratio4 1.6(5) 1.8(6) 2.7(7) 3.0(7) 2.9(8) 2.9(9) 2.8(9) 2.8(10) 2.8(10) Inclusion mean inclination angle (°)4 19(14) 18(11) 13(9) 11(8) 13(6) 12(6) 12(5) 12(5) 13(5) Notes: 1Triple numerical combinations indicate load (t), temperature (K), and total twisting angle (°). For example, 5/640/100 means 5-ton load, after twisting at 640 K for a total of 100°. Tomography data were collected after cooling to room temperature. 2Rotation angles measured on inclusion markers near top and bottom of the sample. 3Shear strains (= r/h) estimated based on the true twist angle (1). 4Mean aspect ratios (max/min principal axes) and inclination angles (relative to the horizontal plane) for all melt inclusions (ellipsoidal ﬁ ts) with volumes greater than 2 × 10–5 (smaller volumes have larger uncertainties due to spatial resolution and are affected by silicate grains). Averaged over more than 250 melt inclusions. Numbers in parentheses correspond to statistics for inclusion volumes between 2 × 10–5 and 1.5 × 10–3 mm3.

A1: ambient B1: 3 GPa, rm T C1: 6GPa, 700K D1: 6GPa, 700K, 540°

A2 B2 C2 D2

0.1 mm

Figure 5. Fe-Ni-S inclusions extracted from a series of tomographic images using Blob3D (see text). Top row shows complete images of the sample when the silicate matrix is removed; bottom row shows examples of the inclusions extracted. (A) Ambient sample before compression. (B) Sample loaded to 3 GPa, at room temperature. (C) Sample loaded to 6 GPa after heating to 700 K. (D) Sample twisted by 540° at 6 GPa and 700 K. Note that because of resolution limitations and because Blob3D treats each inclusion as an ellipsoid, very small inclusions cannot be imaged properly. For each pair, the upper image (e.g., A1) is the three-dimensional view of the entire sample with one plane high- lighted, and the lower image (e.g., A2) is the selected image of individual inclusions extracted.

Geosphere, February 2011 45

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 Wang et al.

distribution of the large alloy inclusions in sample B before and after shear deformation is provided as supplemental information (see Ani- mations 1 and 2, respectively). It was difﬁ cult to identify with conﬁ dence major inclusions for such an analysis; therefore, no statistical data are presented here. Nevertheless, SEM analysis indicates that sample B underwent a larger shear deformation than sample A.

Tomographic Microstructural Analysis

Fe-Ni-S inclusions were extracted using Blob3D with a prescribed matrix-alloy interface threshold based on the technique described in Lesher et al. (2009). The utilities in Blob3D Animation 1. Three-dimensional tomo- Animation 2. Three-dimensional tomo- treated each inclusion as an ellipsoid, with graphic reconstruction of sample B at 1 GPa graphic image of sample B at 3.5 GPa, 800 K, lengths of the three axes, volume, and surface and room temperature (before heating and after twisting the Drickamer anvils by 540°. all recorded. Figure 5 provides a few examples. shear deformation). The starting view is Similar view as in Animation 1. Animation 2 As the samples were compressed, the original along the sample axis. The view is about one- is based on the data collected in situ in the sphere-like and tube-like inclusions became quarter of the circular cross section. Anima- HPXTM (high-pressure X-ray tomography fl attened along the directions perpendicular to tion 1 is based on the data collected in situ in microscope). If you are viewing the PDF of the loading axis. With increasing shear defor- the HPXTM (high-pressure X-ray tomogra- this paper or reading it offl ine, please visit mation, inclusions became increasingly mutu- phy microscope). If you are viewing the PDF http://dx.doi.org/10.1130/GES00560.S2 or the ally connected (e.g., B2, C2, D2 in Fig. 5). of this paper or reading it offl ine, please visit full-text article on www.gsapubs.org to view However, at larger deformations, some alloy http://dx.doi.org/10.1130/GES00560.S1 or the Animation 2. inclusions were severely deformed, becoming full-text article on www.gsapubs.org to view smears with very thin, fi lm-like features. SEM Animation 1. analyses indicated that the shortest dimension of ple during deformation. As a result, the large these thin layers is often below the current reso- inclusions near the sample’s outer boundary did lution of the high-pressure tomography instru- properly imaged with our current spatial resolu- not undergo as much shear deformation as the ment. Imaging these features may be possible tion (see further discussion on the minor effects inclusions well inside. Inclusions near the sam- in recovered samples using higher resolution of imaging resolution in the following). ple’s outer surface tended to migrate outward X-ray computed microtomography. and remained less deformed than those well We applied Quant3D (Ketcham, 2005a; Virtual Serial Sections and SEM Analyses inside the sample. This phenomenon, unique Ketcham and Ryan, 2004) to obtain statistics to sample A, can be more clearly seen in the of the inclusions in response to axial and shear We generated virtual thin sections from the virtual thin sections in Figure 7. For a higher strains. Figure 6 shows a series of rose diagrams tomographic reconstructions and visually exam- resolution tomographic image (a 3D anima- based on SVD analysis for sample A. Initially, ined the textural evolution. These observations tion) of this sample, after it was recovered from the eigenvalues of the Fe-Ni-S fabric can be were then compared with the fi nal, post mortem the high-pressure deformation experiment, see represented by an ellipsoid, and the average analysis using SEM, the spatial resolution Animation 3. ratio between the maximum (red) and minimum of which is well below 1 μm. Figure 7 shows Figure 8B is a backscattered electron image (green to light blue) axis lengths is ~2 (Fig. 6A). selected virtual serial sections of sample A at of sample B, showing texture essentially identi- There appears to be some preferred orientation various stages of the deformation process. All of cal to that in sample A at comparable scales. The in the distribution of the inclusions: the long these sections are oriented parallel to the rota- slight differences are attributed to the greater axes tend to align with the radial direction, prob- tion axis. Figure 7 shows that inclusions were axial deformation given the higher temperature. ably indicating that an axial stress component fl attened during the initial loading and heating Figure 9 provides higher magnifi cation SEM was introduced when the sample was synthe- and then smeared along planes nearly perpen- images taken of sample A, showing Fe-Ni-S sized in the piston cylinder device. Loading to dicular to the rotational axis, forming very thin veins at scales of 10 μm (Fig. 9A) and 1 μm 6 GPa at room temperature (axial strain of 21%) layers that are barely detectable with our tomog- (Fig. 9B). These veins are preferentially aligned signifi cantly fl attened the inclusions; the eigen- raphy technique. in directions approximately perpendicular to the value representation became disk-like rather We confi rmed that the bright thin features loading axis and parallel to the shear direction than rod-like, and the elongation in the horizon- observed in the virtual sections are not imaging (Fig. 9A). At a fi ner scale, extremely thin Fe- tal directions increased, with the ratio of maxi- artifacts by directly comparing them with SEM Ni-S fi lms appear to penetrate the silicate grains mum (red) to minimum (intermediate blue) axis images made of actual serial sections through and connect to the adjacent metallic inclusions lengths of ~5. Large shear deformation appears the recovered samples (Fig. 8). The SEM image (Fig. 9B). These fi ne veins were nearly invisi- to have mostly increased this ratio to ~10. In the of sample A (Fig. 8A may be directly compared ble in the tomographic images because of the end, the statistical shape of the inclusions is an with the last virtual thin section in Fig. 7). Note technique’s resolution limitations. Sample A’s almost perfectly round disk. This shape, how- that the BN sleeve (see Fig. 3) surrounding microstructure resembles melt distribution fab- ever, applies to only the inclusions that could be sample A created a weak coupling to the sam- rics in partially molten rocks and rock analogs

46 Geosphere, February 2011

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 High P, T tomography on deformed composites

AB CD

Ambient 6 GPa, 300K 6 GPa, 700K, 0° 6 GPa, 700K, 353° Figure 6. Rose diagram pairs for the star volume distribution analyses. Top row shows the rose diagrams in sample radial directions. In the bottom row, Up and Down indicate the sample axial direction. Note the gradual disappearance of the horizontal shape-preferred orientation as the sample was sheared (top row, from left to right) and the signifi cant fl attening of the inclusions (bottom row, from left to right). The diagrams at the end of shear indicate that, at ~0.005 mm spatial resolution, the average inclusion shape is a signifi cantly fl attened disk, with an aspect ratio (thickness to diameter ratio) <0.1.

A E

(30°, 1.06)

Figure 7. Virtual thin sections from F tomographic images for sample A. (A) Ambient condition, before compression. (B) 6 GPa, room tem- ambient (65°, 2.51) perature (rm T), before heating. (C–I) After heating to 700 K and B twisting at various angles. Values G in parentheses are twisting angle and shear strain. Bright contrasts are Fe-Ni-S inclusions imaged. (168°, 6.81) Note the development of linear features as shear strain increases. 6 GPa, rm T H Rotation/compression axis is vertical in all the images. The initial C horizontal dimension (A) is 1 mm. (274°, 11.30) (0°, 0) I D

(2°, 0.07) (353°, 14.56)

Geosphere, February 2011 47

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 Wang et al.

0.5 mm

0.2 mm B

Animation 3. Three-dimensional tomographic image of sample A after recovery from the high-pressure–high-temperature shear deformation experiments. The sample became a thin disk, ~1.5 mm in diameter and 0.3 mm in thickness. Images collected at Advanced Light Source. If you are viewing the PDF of this paper or reading it offl ine, please visit http://dx.doi.org/10.1130/ Figure 8. Backscattered electron images of sample A (A) and sample B (B), recovered after GES00560.S3 or the full-text article on the high-pressure–high-temperature tomography experiment. Sample A can be directly www.gsapubs.org to view Animation 3. compared with the last virtual thin section in Figure 7. High-pressure–high-temperature tomography has captured most of the features. The rotating axis is vertical for torsional shear generation. ~2–4 μm). Some of the alloy fi lms (e.g., Figs. 8–10) are well below these voxel dimensions and therefore are not resolved by this technique. Several improvements to spatial resolution are (e.g., Rosenberg and Handy, 2000; Rosenberg These observations document the develop- being implemented, including more accurate and Berger, 2001; Hustoft and Kohlstedt, 2006). ment of SPO in multiphase composites as a alignment of the rotational axis, software cor- This fi nding indicates that the softer metallic result of large shear deformation. We show that rections for off-axis errors in the mechanical phase behaved in an almost fl uid-like manner SPO caused the texture of sample A to change setup, as well as more sensitive detectors and during deformation and propagated along sili- from LBF to IWL. Furthermore, although the more responsive phosphor screens. cate grain boundaries in a manner analogous to initial microstructures differed in the two sam- One technical challenge is to maintain deformation-assisted melt migration mecha- ples, their fi nal textures were very similar after sample height during high-temperature defor- nisms (e.g., Phipps Morgan and Holtzman, large deformation. mation, which may be mitigated by adopting 2005; Stevenson, 1989; Kohlstedt and Holtz- harder gasket materials. These improvements man, 2009). The eutectic melting point in the DISCUSSION to imaging quality will be the focus of future Fe-S system is well above 1000 K under our technical development. However, in general, experimental conditions (Fei et al., 1997). In A comparison of the virtual tomographic axial compression fl attens weak phase inclu- sample B, such thin fi lms are more pronounced serial sections collected at the fi nal deforma- sions in the directions perpendicular to load- (Fig. 10), most likely because sample A initially tion stage (e.g., Fig. 7I) and the SEM images ing axis and does not signifi cantly affect the had an IWL-type texture. of the recovered sample (Figs. 8A and 9) shows later shear-induced texture. Further quantifi ca- The silicate matrix underwent dramatic grain- that the current HPXTM setup captures general tion of microstructure based on tomographic size reduction in both samples, from an original features of the true microstructure. Of course, images, however, will require more sophisti- 30–40 μm to ~1 μm (cf. Figs. 9 and 10 to Figs. the utility of any given imaging techniques is cated image analysis software. For example, an 1 and 2). Some remnants of large olivine grains limited by the spatial resolution. Gualda and ability to describe inclusions in complex shapes remain and often contain numerous straight Rivers (2006) analyzed rock inclusions using and to correlate inclusions at various deforma- grain boundaries (e.g., Fig. 10B), indicative of X-ray tomography and concluded that 3D tion stages based on spatial distribution statistics severe brittle deformation. The enstatite grains objects can be properly imaged when the linear will help researchers extract detailed informa- in sample A, however, show a lesser degree of dimensions are at least fi ve times the voxel size. tion on deformation and strain partitioning from grain-size reduction, perhaps due to its greater This conclusion is in general agreement with the images. Although the experimental tech- strength (e.g., Carter, 1976; Mackwell, 1991), our analysis: inclusions with linear dimensions nique is still being developed and the results are and the presence of the surrounding fi ner- >15–20 μm could be imaged with suffi cient preliminary, several interesting observations are grained quench silicates. confi dence (recall that our voxel sizes were worthy of discussion.

48 Geosphere, February 2011

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 High P, T tomography on deformed composites

Examining dominant deformation mechanisms in the constituent silicate phases and their effects on texture development provides a context for further analysis and discussion. The silicates in sample A appear to have deformed in a ductile or semibrittle regime, as suggested by the irregular and elongated enstatite grains Figure 9. Backscattered electron (Fig. 9A). The quenched silicate melt behaves images of sample A at higher En differently, showing signifi cant particle-size magnifi cations. (A) Alloy lenses reduction that may be partly due to the com- appear in the silicate matrix. positional inhomogeneity in the quench glass. Shear occurred perpendicular Microprobe analyses showed a varied Mg/Si to the rotational axis (white ol ratio across the quenched melt, for which the arrow). Note that the “olivine” A defi nition of “particle” is somewhat ambiguous. (ol) matrix has undergone a sig- There may be numerous domains (quenched nifi cant grain-size reduction (cf. crystals) with varied composition and mechani- Fig. 1). En—enstatite. (B) Thin cal properties. Under large shear strain, these alloy fi lm passing though a sili- domains may be separated, resulting in an cate grain boundary connecting apparent particle-size reduction. For sample B, two alloy lenses. Extremely thin the silicate matrix is essentially homogeneous alloy fi lms (arrows) are barely olivine, which was most likely deformed in the visi ble. Block arrows indicate the brittle regime (Fig. 10), given the low pressures rota tional axis for shear generation. and temperatures during deformation. As the shape of alloy inclusions becomes more lens-like, stress concentration increases near the tips of the lenses, causing the silicates B to fracture. The alloy material can then propa- gate through the fractured interface in a manner similar to that of fl uid (Fig. 9B). Previous experiments showed that the texture transition from LBF to IWL occurs in numerous polyphase rocks and rock analogs, including gneiss (quartz-plagioclase-biotite; Holyoke and Tullis, 2006), aplite (quartz-feldspar; Dell’Angelo and Tullis, 1996), mica-quartz- feldspar (Shea and Kronenberg, 1993; Tullis and Wenk, 1994), muscovite/kaolinite-halite (Bos and Spiers, 2001), halite-calcite (Jordan, 1987), and camphor-octachloropropane (Bons Figure 10. Backscattered electron and Cox, 1994). These experiments also found images of sample B. (A) Thin that the LBF-IWL texture transition can occur lamination is well developed when the stronger phase (matrix) is either brittle in planes nearly perpendicular or ductile. When matrix deformation is predom- A to the loading axis (parallel to inantly brittle, the LBF-IWL transition occurs the shear plane). (B) The olivine in localized shear zones, weakening the entire matrix also underwent signif- rock sample by concentrating deformation in icant grain-size reduction. In these shear zones. When deformation is ductile, places, the original large grain the IWL texture occurs more or less uniformly can still be recognized. Adjacent throughout the sample. small grains have close crystal- One of the most important factors in control- lographic orientations. Block ling the LBF-IWL texture transition is the phase arrows indicate the rotational strength contrast (PSC), i.e., the strength ratio axis for shear generation. between the strong and the weak constituents (e.g., Handy, 1994a, 1994b; Holyoke and Tullis, 2006). High PSC generally results in greater stress concentration in the strong phase adjacent B to the weaker phase. In the gneiss samples studied by Holyoke and Tullis (2006), local semibrittle fl ow of quartz and plagioclase allowed biotite grains (13 vol%) to become interconnected by

Geosphere, February 2011 49

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 Wang et al.

shearing at ~1000 K. Holyoke and Tullis (2006) rate, than the strength alone. Using experi- 1998; Yamazaki and Karato, 2001). Thus, once estimated the PSC ratio to be ~30:1 for feldspar- mentally measured diffusion coeffi cients and the transition from LBF to IWL occurs, bulk biotite and ~50:1 for quartz-biotite. homologous temperature scaling, Yamazaki and viscosity will suddenly decrease from the upper For our samples, the PSC ratio is diffi cult Karato (2001) calculated that the viscosity of Pv curve to the lower one. to determine precisely because we do not have is about three orders of magnitude higher than To further explore this scenario, we con- strength data for the alloy. Near stoichiometric that of Fp. sider a lower mantle dominated by upwelling FeS has widely varying strengths depending Figure 11 shows one attempt to estimate the and downwelling fl ows separated by relatively on crystal structure and slip systems. At 673 K rheological properties of the lower mantle and stationary regions (Karato [1998] called these and atmospheric pressure, hexagonal pyrrhotite the effects of texture on bulk viscosity. Based on regions “stagnant cores” of convection cells;

(Fe1–xS, with x = 0–0.2) single crystals slip along the rheological properties for Pv and Fp given Fig. 12A). We examine a cross section from the basal plane with shear strengths as low as by Yamazaki and Karato (2001, their Table 1), the edge of the subducting slab (point A) to the 5 MPa (Kübler, 1985). Assuming that the Fe-Ni- viscosities are estimated for depths of 1000 and edge of the upwelling conduit (point B), at a S phase in our samples has a similar strength, the 2000 km using the approach outlined by Takeda constant depth of, e.g., 1000 km. Points A′ and ratio of shear strength contrast between olivine (1998). Here we adopt a lower mantle geotherm B′ are the demarcation points of fl ow-induced (e.g., Carter, 1976) and the alloy is ~500:1 to with a temperature gradient of 0.3 K/km, with shear zones; regions AA′ and BB′ are under 1000:1. Because of the weak strength of the Fe- homologous temperature scaling for the vis- pronounced shear deformation due to the drag Ni-S phase, the presence of quenched silicate cosities of the two constituents (Yamazaki and of viscous fl ows. These shear zones are also melts in sample A is unlikely to have signifi cant Karato, 2001). Across the currently accepted thermal boundary layers: temperatures adja- effect on the observed fabric evolution. Bruhn mineralogical models (70–100 vol% Pv; shaded cent to the downwelling and upwelling conduits et al. (2000) deformed an olivine sample with 4 area in Fig. 11), the viscosities of the two-phase may be a few hundred Kelvin cooler and hot- vol% gold at 0.3 GPa and ~1500 K; i.e., under mineralogical assemblage vary by roughly three ter, respectively, than the ambient mantle (Fig. conditions where the strong phase (olivine) was orders in magnitude. For each depth, the upper 12B). Figure 12C gives the likely deformation deforming in the dislocation creep regime. In line represents viscosities for the ideal LBF tex- mechanisms of different regions along the cross that experiment, PSC may be even higher given ture, whereas the lower curve corresponds to section. In zone AA′, where temperatures are that gold was in the liquid state. Under such large viscosities for the ideal IWL texture (Takeda, lower and stresses higher than the surrounding PSC, IWL texture develops readily. The similar- ity between the shear-induced IWL morphology (i.e., shape and spatial distribution of the weaker 23 phase) reported here and that reported by Bruhn et al. (2000) indicates that in the stronger phase, the specifi c attributes of the deformation mechanisms are less important. Therefore, the LBF- IWL transition predominantly depends on PSC 22 and the amount of strain. Note, however, that in the work of Bruhn et al. (2000) and other studies dealing with deformation of partially molten systems by tortional shear (e.g., Kohlstedt and Holtzman, 2009), the weak phase is generally 21 LBF, 1000km oriented at 25°–30° with respect to the shear

plane. In our study, the weak Fe-Ni-S alloy is Pa-s) (viscosity,

oriented almost parallel to the shear plane. This 10 LBF, 2000km

difference is most likely related to the substan- log 20 tial amount of axial compressive stress that the IWL, 2000km samples undergo before and during deformation in simple shear. IWL, 1000km Let us consider effects of PSC and shear- induced texture development on the rheological 19 properties of the lower mantle. For simpliﬁ ca- 0 20406080100 tion, we assume the mantle consists of silicate perovskite (Pv) and ferropericlase (Fp) only. We Pv vol% estimate the volume fraction of Pv to range from ~70‰ to 90%, depending on the mineralogical Figure 11. Bounds of viscosity for the lower mantle mineralogy assemblage (Pv + Fp, models preferred (e.g., Stixrude et al., 1992; perovskite and ferropericlase), based on rheological properties proposed by Yamazaki and Ringwood, 1991). A number of observations Karato (2001) and the theory proposed by Takeda (1998) at depths of 1000 (black curves) suggest that Fp is the weaker phase (Weaver, and 2000 km (gray curves). These bounds are presented as a function of Pv vol% (hori- 1967; Hulse et al., 1963; Paterson and Weaver, zontal axis) and texture. The upper bound corresponds to the idealized LBF (load-bearing 1970; Uchida et al., 2004; Chen et al., 2002). framework) fabric and the lower bounds to the idealized IWL (interconnected layers of the When discussing mantle dynamics, it is more weaker phase) fabric. The light gray area to the right indicates the current range of min- convenient to use viscosity, which, for solids, eralogical models for the lower mantle in Pv vol%. An LBF-IWL transition could reduce is proportional to the ratio of strength to strain viscosity by two orders of magnitude.

50 Geosphere, February 2011

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 High P, T tomography on deformed composites

A T

Hot (d1) ? Def. Mech. Diff. creep Disl. creep ′ Superplastic Diff. creep B B D IWL IWL? LBF?

′ Texture A E μ Pv A (d2) ?

Viscosity μ Fp Cold AA′ B′ B Figure 12. Postulated variation in mechanical properties in the lower mantle due to shear-induced shape-preferred orientation. (A) Dia- gram showing an assumed profi le along AB, where A′ and B′ mark the boundaries between the normal mantle and the shear zones caused by the downwelling and the upwelling fl ow, respectively. (B) Schematic temperature profi le. (C) Possible deformation mechanisms across AB. Diff—diffusion; Disl—dislocation. (D) Fabric of the Pv + Fp (perovskite and ferropericlase) composite. For AA′, an IWL (interconnected layers of the weaker phase) fabric (illustrated in inset d1) with lamination direction perpendicular to the fl ow direction is the most prob- able. Because of the many competing deformation mechanisms for zones A′B′ and B′B, the fabric may be complicated. Inset d2 shows an LBF (load-bearing framework) type texture (Yamazaki et al., 2009). (E) Schematic viscosity profi le for the Pv + Fp composite. In zones AA′ and B′B, the effective viscosity along the fl ow direction is dominated by properties of Fp, whereas A′B′ may adopt a wide range of effective viscosity ranging from that of Fp to Pv. See text for discussion.

mantle, the dominant deformation mechanism observations show that at such shear strains, mantle introduces large deformation in the sur- is likely dislocation creep. In the normal mantle a transition from LBF to IWL had already rounding mantle, especially around the tip of (A′B′), both diffusion creep (Karato and Li, occurred (Fig. 12, inset d1). As a result, the the slab. The mantle material undergoes a tex- 1992; Yamazaki and Karato, 2001) and super- viscosity along the ﬂ ow direction is dominated tural transition to adapt an IWL fabric, and the plasticity (Karato et al., 1995) have been pro- by that of Fp (Fig. 12E). In the normal mantle, effective viscosity decreases. The magnitude of posed. In the hotter zone (B′B), diffusion creep the viscosity may be controlled by that of Pv if this decrease in effective viscosity may range is likely to dominate. With these competing this phase forms a LBF (Fig. 12, inset d2), or from 100 to 1000, resulting in an avalanching deformation mechanisms, Figure 12D summa- by superplasticity if both phases are ﬁ ne grained enhancement for the downwelling process. The rizes the expected fabric of the lower mantle and deform by grain-boundary sliding. Hence, weakened zone (AA′) may be only a few tens of phase assemblage Pv + Fp. Large shear defor- dynamic processes are likely controlled by the kilometers in thickness, but it could play a criti- mation is likely to occur in the boundary layer rheological properties of Fp. cal role in mantle dynamics. Measurements of (AA′): for a slab motion speed of ~10 mm/yr, Texture-induced rheological weakening may the rheological properties of Pv and a full under- shear strains of ~5 will be produced in ~5 m.y., have profound effects on the dynamics of a sub- standing of dominant deformation mechanisms for a thickness of the shear zone of ~10 km. Our ducting slab. A slab penetrating into the lower under various pressure, temperature, grain-size,

Geosphere, February 2011 51

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 Wang et al.

and strain-rate conditions are required to evalu- between the constituent phases. In addition, the Bons, P.D., and Cox, S.J.D., 1994, Analogue experiments ate the effects quantitatively. observed texture transition should be a com- and numerical modelling on the relation between microgeometry and fl ow properties of polyphase mate- Development of an IWL fabric is also associ- mon phenomenon in rock assemblies with high rials: Materials Science and Engineering, v. A175, ated with seismic anisotropy. A simple estimate, phase-strength contrast. The texture transition p. 237–245. Bos, B., and Spiers, C.J., 2001, Experimental investigation based on an IWL-induced, weak, transversely from LBF to IWL will infl uence the dynamic into the microstructural and mechanical evolution of anisotropic composite (e.g., Brittan and Warner , process in several important ways. A rheologi- phyllosilicate-bearing fault rock under conditions 1995), yields ~0.5% shear-wave anisotropy, cal weakening is likely to occur in regions of favouring pressure solution: Journal of Structural Geol- ogy, v. 23, p. 1187–1202, doi: 10.1016/S0191-8141 with S wave speed in the laminating direction the lower mantle where the Pv + Fp assembly (00)00184-X. faster than that parallel to the layers. LPO asso- is undergoing large deformation. The regions Brittan, J., and Warner, M.P.G., 1995, Anisotropic parame ters ciated with the texture may further enhance the most likely to be affected are within the ther- of layered media in terms of composite elastic properties: Geophysics, v. 60, p. 1243–1248, doi: 10.1190/ degree of anisotropy. Future high-resolution mal boundary layers adjacent to subducting 1.1443854. seismic studies may be able to resolve the con- slabs, where a strong IWL texture is expected to Bruhn, D., Groebner, N., and Kolhstedt, D.L., 2000, An interconnected network of core-forming melts pro- tributions of anisotropy caused by this shear develop, thereby greatly reducing the bulk vis- duced by shear deformation: Nature, v. 403, p. 883– zone from that within the slab. cosity in these regions. This scenario should be 886, doi: 10.1038/35002558. considered in geodynamic models. Also, since Bystricky, M., Heidelbach, F., and Mackwell, S., 2006, Large-strain deformation and strain partitioning CONCLUSIONS lamination develops in planes subparallel to the in polyphase rocks: Dislocation creep of olivine- fl ow direction, SPO-induced elastic anisotropy magnesiowüstite aggregates: Tectonophysics, v. 427, We have developed a new experimental tech- will also develop in the composite. A simple p. 115–132, doi: 10.1016/j.tecto.2006.05.025. Carter, N.L., 1976, Steady state fl ow of rocks: Reviews of nique, called HPXTM (high-pressure X-ray model indicates that in these highly deformed Geophysics and Space Physics, v. 14, p. 301–360, doi: tomography microscope), to study the mechani- regions, SPO-induced anisotropy may be as 10.1029/RG014i003p00301. cal properties of multiphase materials. Stress much as 0.5% for the lower mantle. Seismic Chen, J., Weidner, D.J., and Vaughan, M.T., 2002, The strength of Mg0.9Fe0.1SiO3 perovskite at high pressure measurements were not performed at this stage studies on anisotropy may provide the clues and temperature: Nature, v. 419, p. 824–826, doi: of the development; however, by implementing researchers need to infer the degree of deforma- 10.1038/nature01130. Cowin, S.C., 1986, Wolff’s law of trabecular architecture a multi-element detector and conical slits (e.g., tion and, hence, the rheological weakening in at remodeling equilibrium: Journal of Biomechanical as described in Nishihara et al., 2008; Kawazoe the lower mantle. Engineering, v. 108, p. 83–88, doi: 10.1115/1.3138584. et al., 2009), we can conduct such measure- Cruz-Orive, L.M., Karlsson, L.M., Larsen, S.E., and ACKNOWLEDGMENTS Wainschtein, F., 1992, Characterizing anisotropy: A ments in future high-pressure tomography new concept: Micron and Microscopica Acta, v. 23, experiments. Further improvements in optical We thank Frank Westferro for the excellent engi- p. 75–76, doi: 10.1016/0739-6260(92)90076-P. components (such as scintillator materials and neering support during the high-pressure tomography Dell’Angelo, L.N., and Tullis, J., 1996, Textural and experiments at GSECARS (GeoSoilEnviroCARS, mechanical evolution with progressive strain in experi- cameras) and mechanical components will be Argonne National Laboratory) and Anne-Line mentally deformed aplite: Tectonophysics, v. 256, sought to increase spatial resolution. Auzende for assistance during transmission electron p. 57–82, doi: 10.1016/0040-1951(95)00166-2. Fei, Y., Bertka, C.M., and Finger, L.W., 1997, High-pressure Using HPXTM, we examined the evolution microscopy studies at IMPMC (Institut de Minéralo- iron-sulfur compound, Fe3S2, and melting relations in of strain and texture in an analog two-phase gie et de Physique des Milieux Condensés). We are the Fe-FeS system: Science, v. 275, p. 1621–1623, doi: grateful to G. Gualda, who provided the software rock composite (olivine + Fe-Ni-S). Two types 10.1126/science.275.5306.1621. vol_animate, which was very helpful in examining Fletcher, R.C., 2004, Anisotropic viscosity of a dispersion of of starting samples were studied: an LBF-type three-dimensional tomography microstructure, and aligned elliptical cylindrical clasts in viscous matrix: texture, where the alloy phase was present as to S. Karato for his valuable comments on early ver- Journal of Structural Geology, v. 26, p. 1977–1987, isolated spherical inclusions, and a near-IWL- sions of the manuscript. We also thank two anony- doi: 10.1016/j.jsg.2004.04.004. mous reviewers, whose thorough and constructive Gaetani, G., and Grove, T., 1999, Wetting of mantle olivine type texture, where the alloy phase was con- by sulfi de melt: Implications for Re/Os ratios in mantle reviews signifi cantly improved the manuscript. This centrated near grain boundaries and tended to peridotite and late-stage core formation: Earth and work was performed at GSECARS (Sector 13), Planetary Science Letters, v. 169, p. 147–163, doi: form an interconnected network. The 3D tomo- Advanced Photon Source, Argonne National Labora- 10.1016/S0012-821X(99)00062-X. graphic image analyses (with a spatial resolu- tory. GSECARS is supported by the National Science Garboczi, E., Bentz, D.P., and Martys, N.S., 1999, Digital tion of 4–5 μm) allowed us to track individual Foundation (NSF)–Earth Sciences (EAR-0622171) images and computer modeling, in Wong, P.-Z., ed., and Department of Energy–Geosciences (DE-FG02- Methods in the physics of porous media: San Diego, alloy inclusions at various shear deformation 94ER14466). Use of the Advanced Photon Source California, Academic Press, p. 1–41. stages. SEM images of the recovered samples, was supported by the U.S. Department of Energy, Gualda, G.A.R., and Rivers, M., 2006, Quantitative 3D petrography using X-ray tomography: Application to with a spatial resolution of <1 µm, provided Offi ce of Science, Offi ce of Basic Energy Sciences, Bishop Tuff pumice clasts: Journal of Volcanology and supporting observations and confi rmed that at under Contract DE-AC02-06CH11357. We gratefully Geothermal Research, v. 154, p. 48–62, doi: 10.1016/ the resolution level of the tomography setup, acknowledge fi nancial support from the NSF through j.jvolgeores.2005.09.019. grants EAR-0711057 (Wang) and EAR-0711599 Guoan, L., and Castañeda, P.P., 1993, The effect of par- HPXTM adequately captures the general fea- (Lesher). Work by J. Roberts was performed under the ticle shape and stiffness on the constitutive behavior tures of a composite’s texture. Results indicate auspices of the U.S. Department of Energy by Law- of metal-matrix composites: International Journal that regardless of the initial texture, large shear rence Livermore National Laboratory under Contract of Solids and Structures, v. 30, p. 3189–3209, doi: DE-AC52-07NA27344. 10.1016/0020-7683(93)90109-K. deformation causes a dramatic texture transi- Handy, M.R., 1994a, Flow laws for rocks containing two tion from LBF to IWL, the alloy phase forming REFERENCES CITED non-linear viscous phases: A phenomenological laminated layers subparallel to the shear plane. approach: Journal of Structural Geology, v. 16, p. 287– Arns, C., Knackstedt, M.A., Val Pinczewski, W., and 301, doi: 10.1016/0191-8141(94)90035-3. For such a laminated IWL texture, effective Garboczi, E.J., 2002, Computation of linear elastic prop- Handy, M.R., 1994b, The energetics of steady state hetero- rheological properties should be dominated by erties from microtomographic images: Methodology geneous shear in mylonitic rock: Materials Science and Engineering A, v. 175, p. 261–272, doi: 10.1016/ the weak phase. Our results indicate that the and agreement between theory and experiment: Geo- physics, v. 67, p. 1396–1405, doi: 10.1190/1.1512785. 0921-5093(94)91065-0. deformation mechanisms in the strong matrix Bascou, J., Tommasi, A., and Mainprice, D., 2002, Plastic Holyoke, C.W., III, and Tullis, J., 2006, Mechanisms of material (silicate in our study) play a less impor- deformation and development of clinopyroxene lat- weak phase interconnection and the effects of phase tice preferred orientations in eclogites: Journal of strength contrast on fabric development: Journal of tant role in the development of the IWL fabric Structural Geology, v. 24, p. 1357–1368, doi: 10.1016/ Structural Geology, v. 28, p. 621–640, doi: 10.1016/ than does the contrast in rheological properties S0191-8141(01)00137-7. j.jsg.2006.01.008.

52 Geosphere, February 2011

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021 High P, T tomography on deformed composites

Hulse, C.O., Copley, S.M., and Pask, J.A., 1963, Effect of Long, M.D., Xiao, X., Jiang, Z., Evans, B., and Karato, pressure-volume-temperature equation of state: Jour- crystal orientation on plastic deformation of magne- S.-I., 2006, Lattice preferred orientation in deformed nal of Geophysical Research, v. 106, p. 515–528, doi: sium oxide: Journal of the American Ceramic Soci- polycrystalline (Mg,Fe)O and implications for seis- 10.1029/2000JB900318. ety, v. 46, p. 317–323, doi: 10.1111/j.1151-2916.1963 mic anisotropy in D’’: Physics of the Earth and Stevenson, D.J., 1989, Spontaneous small-scale melt segre- .tb11738.x. Planetary Interiors, v. 156, p. 75–88, doi: 10.1016/ gation in partial melts undergoing deformation: Geo- Hustoft, J., and Kohlstedt, D., 2006, Metal-silicate seg- j.pepi.2006.02.006. physical Research Letters, v. 16, p. 1067–1069, doi: regation in deforming dunitic rocks: Geochemistry Mackwell, S.J., 1991, High-temperature rheology of enstatite: 10.1029/GL016i009p01067. Geophysics Geosystems, v. 7, Q02001, doi: 10.1029/ Implications for creep in the mantle: Geophysical Stixrude, L., Hemley, R.J., Fei, Y., and Mao, H.K., 1992, 2005GC001048. Research Letters, v. 18, p. 2027–2030, doi: 10.1029/ Thermo elasticity of silicate perovskite and magnesio- Jordan, P.G., 1987, The deformational behaviour of biminer- 91GL02492. wustite and stratiﬁ cation of the Earth’s mantle: Sci- alic limestone-halite aggregates: Tectonophysics, v. 135, Madi, K., Forest, S., Cordier, P., and Boussuge, M., 2005, ence, v. 257, p. 1099–1101, doi: 10.1126/science.257 p. 185–197, doi: 10.1016/0040-1951(87)90160-0. Numerical study of creep in two-phase aggregates with .5073.1099. Kanagawa, K., 1993, Competence contrasts in ductile defor- a large rheology contrast: Implications for the lower Takeda, Y.-T., 1998, Flow in rocks modelled as multiphase mation as illustrated from naturally deformed chert- mantle: Earth and Planetary Science Letters, v. 237, continua: Application to polymineralic rocks: Jour- mudstone layers: Journal of Structural Geology, v. 15, p. 223–238, doi: 10.1016/j.epsl.2005.06.027. nal of Structural Geology, v. 20, p. 1569–1578, doi: p. 865–885, doi: 10.1016/0191-8141(93)90181-9. Mainprice, D., Bascou, J., Cordier, P., and Tommasi, A., 10.1016/S0191-8141(98)00043-1. Karato, S., 1998, Seismic anisotropy in the deep mantle, 2004, Crystal preferred orientations of garnet: Compari- Takeda, Y.-T., and Griera, A., 2006, Rheological and kine- boundary layers and the geometry of mantle convec- son between numerical simulations and electron back- matical responses to ﬂ ow of two-phase rocks: Tec tono- tion: Pure and Applied Geophysics, v. 151, p. 565–587, scattered diffraction (EBSD) measurements in naturally physics, v. 427, p. 95–113, doi: 10.1016/j.tecto.2006 doi: 10.1007/s000240050130. deformed eclogites: Journal of Structural Geology, .03.050. Karato, S.-I., and Li, P., 1992, Diffusion creep in perovskite: v. 26, p. 2089–2102, doi: 10.1016/j.jsg.2004.04.008. Tandon, G.P., and Weng, G.J., 1986, Average stress in the Implications for the rheology of the lower mantle: Sci- Nishihara, Y., Tinker, D., Kawazoe, T., Xu, Y., Jing, Z., matrix and effective moduli of randomly oriented com- ence, v. 255, p. 1238–1240, doi: 10.1126/science.255 Matsukage, K., and Karato, S., 2008, Plastic deforma- posites: Composites Science and Technology, v. 27, .5049.1238. tion of wadsleyite and olivine at high-pressure and p. 111–132, doi: 10.1016/0266-3538(86)90067-9. Karato, S.-I., Zhang, S., and Wenk, H.-R., 1995, Superplas- high-temperature using a rotational Drickamer apparatus Treagus, S.H., 2002, Modelling the bulk viscosity of two- ticity in Earth’s lower mantle: Evidence from seismic (RDA): Physics of the Earth and Planetary Interiors, phase mixtures in terms of clast shape: Journal of anisotropy and rock physics: Science, v. 270, p. 458– v. 170, p. 156–169, doi: 10.1016/j.pepi.2008.03.003. Structural Geology, v. 24, p. 57–76, doi: 10.1016/ 461, doi: 10.1126/science.270.5235.458. Odgaard, A., Kabel, J., van Rietbergen, B., Dalstra, M., and S0191-8141(01)00049-9. Kawazoe, T., Karato, S., Otsuka, K., Jing, Z., and Huiskes, R., 1997, Fabric and elastic principal directions Tullis, J., and Wenk, H.R., 1994, Effect of muscovite on the Mookherjee, M., 2009, Shear deformation of dry poly- of cancellous bone are closely related: Journal of Bio- strength and lattice preferred orientations of experi- crystalline olivine under deep upper mantle conditions mechanics, v. 30, p. 487–495, doi: 10.1016/S0021-9290 mentally deformed quartz aggregates: Materials Sci- using a rotational Drickamer apparatus (RDA): Physics (96)00177-7. ence and Engineering, v. A175, p. 209–220. of the Earth and Planetary Interiors, v. 174, p. 128–137, Ohtani, E., Moriwaki, K., Kato, T., and Onuma, K., 1998, Uchida, T., Wang, Y., Rivers, M.L., and Sutton, S.R., 2004, doi: 10.1016/j.pepi.2008.06.027. Melting and crystal-liquid partitioning in the system Yield strength and strain hardening of MgO up to

Keehm, Y., 2003, Computational rock physics: Transport Mg2SiO4-Fe2SiO4 to 25 GPa: Physics of the Earth and 8 GPa measured in the deformation-DIA with mono- properties in porous media and applications [Ph.D. Planetary Interiors, v. 107, p. 75–82, doi: 10.1016/ chromatic X-ray diffraction: Earth and Planetary thesis]: Palo Alto, California, Stanford University, S0031-9201(97)00125-8. Science Letters, v. 226, p. 117–126, doi: 10.1016/ 135 p. Paterson, M.S., and Weaver, C.W., 1970, Deformation of poly- j.epsl.2004.07.023. Ketcham, R.A., 2005a, Three-dimensional grain fabric crystalline MgO under pressure: Journal of the Ameri- Uchida, T., Wang, Y., Westferro, F., Rivers, M.L., Gebhardt, measurements using high-resolution X-ray computed can Ceramic Society, v. 53, p. 463–471, doi: 10.1111/ J., and Sutton, S.R., 2008, X-ray microtomography tomography: Journal of Structural Geology, v. 27, j.1151-2916.1970.tb12678.x. under high pressure, in Ohtani, E., ed., Advances p. 1217–1228, doi: 10.1016/j.jsg.2005.02.006. Phipps Morgan, J., and Holtzman, B., 2005, Vug waves: A in high-pressure mineralogy: Geological Society of Ketcham, R.A., 2005b, Computational methods for quan- mechanism for coupled rock deformation and fl uid America Special Paper 421, p. 227–237. titative analysis of three-dimensional features in geo- migration: Geochemistry Geophysics Geosystems, v. 6, Wang, Y., Uchida, T., Westferro, F., Rivers, M.L., Nishi- logical specimens: Geosphere, v. 1, p. 9, doi: 10.1130/ Q08002, doi: 10.1029/2004GC000818. yama, N., Gebhardt, J., Lesher, C.E., and Sutton, S.R., GES00001.1. Ringwood, A.E., 1991, Phase transformations and their 2005, High-pressure X-ray tomography microscope: Ketcham, R.A., and Carlson, W.D., 2001, Acquisition, bearing on the constitution and dynamics of the man- Synchrotron computed microtomography at high optimization and interpretation of X-ray computed tle: Geochimica et Cosmochimica Acta, v. 55, p. 2083– pressure and temperature: Review of Scientifi c Instru- tomographic imagery: Applications to the geosciences: 2110, doi: 10.1016/0016-7037(91)90090-R. ments, v. 76, 073709, 7 p., doi: 10.1063/1.1979477. Computers & Geosciences, v. 27, p. 381–400, doi: Rivers, M.L., Sutton, S.R., and Eng, P., 1999, Geoscience Warren, J.M., Hirth, G., and Kelemen, P.B., 2008, Evo- 10.1016/S0098-3004(00)00116-3. applications of x-ray computed microtomography: lution of olivine lattice preferred orientation dur- Ketcham, R.A., and Ryan, T.M., 2004, Quantifi cation and Society for Photo-Instrumentation Engineers Proceed- ing simple shear in the mantle: Earth and Planetary visualization of anisotropy in trabecular bone: Journal ings, v. 3772, p. 78–86. Science Letters, v. 272, p. 501–512, doi: 10.1016/ of Microscopy, v. 213, p. 158–171, doi: 10.1111/j.1365 Roberts, J.J., Kinney, J.H., Siebert, J., and Ryerson, F.J., 2007, j.epsl.2008.03.063. -2818.2004.01277.x. Fe-Ni-S melt permeability in olivine: Implications for Weaver, C.W., 1967, Kinking in MgO single crystals under Kohlstedt, D.L., and Holtzman, B.K., 2009, Shearing melt planetary core formation: Geophysical Research Let- high confi ning pressure: Philosophical Magazine, v. 15, out of the Earth: An experimentalist’s perspective on ters, v. 34, L14306, doi: 10.1029/2007GL030497. p. 177–186, doi: 10.1080/14786436708230356. the infl uence of deformation on melt extraction: Annual Rosenberg, C.L., and Berger, A., 2001, Syntectonic melt Wenk, H.-R., ed., 2006, Neutron diffraction texture analy- Review of Earth and Planetary Sciences, v. 37, p. 561– pathways in granitic gneisses, and melt-induced transi- sis: Reviews in Mineralogy and Geochemistry, v. 63, 593, doi: 10.1146/annurev.earth.031208.100104. tions in deformation mechanisms: Physics and Chemis- p. 399–426, doi: 10.2138/rmg.2006.63.15. Kübler, L., 1985, Deformation mechanisms in experimen- try of the Earth. Part A: Solid Earth and Geodesy, v. 26, Yamazaki, D., and Karato, S., 2001, Some mineral physics

tally deformed single crystals of pyrrhotite, Fe1–xS: p. 287–293, doi: 10.1016/S1464-1895(01)00058-8. constraints on the rheology and geothermal structure Physics and Chemistry of Minerals, v. 12, p. 353–362, Rosenberg, C.L., and Handy, M.R., 2000, Syntectonic melt of Earth’s lower mantle: American Mineralogist, v. 86, doi: 10.1007/BF00654346. pathways during simple shearing of a partially molten p. 385–391. Lee, K.Y., and Paul, D.R., 2005, A model for composites rock analogue (Norcamphor-Benzamide): Journal Yamazaki, D., Yoshino, T., Matsuzaki, T., Katsura, T., and

containing three-dimensional ellipsoidal inclusions: of Geophysical Research, v. 105, p. 3135–3149, doi: Yoneda, A., 2009, Texture of (Mg,Fe)SiO3 perovskite Polymer, v. 46, p. 9064–9080, doi: 10.1016/j.polymer 10.1029/1999JB900371. and ferro-periclase aggregate: Implications for rhe- .2005.06.113. Shea, W.T., Jr., and Kronenberg, A.K., 1993, Strength and ology of the lower mantle: Physics of the Earth and Lesher, C.E., Wang, Y., Gaudio, S., Clark, A., Nishiyama, anisotropy of foliated rocks with varied mica contents: Planetary Interiors, v. 174, p. 138–144, doi: 10.1016/ N., and Rivers, M., 2009, Volumetric properties of Journal of Structural Geology, v. 15, p. 1097–1121, j.pepi.2008.11.002. magnesium silicate glasses and supercooled liquid at doi: 10.1016/0191-8141(93)90158-7. high pressure by X-ray microtomography: Physics of Speziale, S., Zha, C.-S., Duffy, T.S., Hemley, R.J., and MANUSCRIPT RECEIVED 30 OCTOBER 2009 the Earth and Planetary Interiors, v. 174, p. 292–301, Mao, H.-K., 2001, Quasi-hydrostatic compression REVISED MANUSCRIPT RECEIVED 24 APRIL 2010 doi: 10.1016/j.pepi.2008.10.023. of magnesium oxide to 52 GPa: Implications for the MANUSCRIPT ACCEPTED 22 MAY 2010

Geosphere, February 2011 53

Downloaded from http://pubs.geoscienceworld.org/gsa/geosphere/article-pdf/7/1/40/3341343/40.pdf by guest on 26 September 2021