morphology and microstructure of magnetite and ilmenite...

American Mineralogist, Volume 96, pages 1316–1324, 2011

0003-004X/11/0809–1316$05.00/DOI: 10.2138/am.2011.3760 1316

Morphology and microstructure of magnetite and ilmenite inclusions in plagioclase from Adirondack anorthositic gneiss

Hans-Rudolf Wenk,* kai CHen,† and RebeCCa smitH

Department of Earth and Planetary Science, University of California at Berkeley, Berkeley, California 94720, U.S.A.

abstRaCt

It is well known that oriented iron and titanium oxide inclusions occur in pyroxenes and plagioclase of anorthosites and granulites, and they are attributed to exsolution at subsolidus conditions. The oxides occur as needles or platelets. In this study, we determine the morphology of oxide needles as well as their orientation in plagioclase (An 30–35) in anorthosite gneiss from the Adirondack mountains (New York). The investigation was done with electron backscatter diffraction (EBSD) in a scanning electron microscope, as well as Laue diffraction with a microfocus synchrotron X-ray beam at the Advanced Light Source in Berkeley. It was observed that the needle direction is [110] in magnetite and [1010] in ilmenite. The needle direction is consistently parallel to [001] of plagioclase. Furthermore, (111) of magnetite and (0001) of ilmenite are sub-parallel to (120) and (120) of plagioclase. We note that for directions [110] in the magnetite structure and [1010] in ilmenite, O atoms are close-packed, and (111) and (0001) are close-packed planes, correspondingly. In plagioclase, [001] is a direction with open channels as well as approximate alignment of Si tetrahedral edges, thus providing nucleation sites with a coincidence lattice relationship. (120) and (120) in this triclinic mineral are planes with approximate tetrahedral sides so that the relationship is structurally plausible. From Laue diffraction, we can determine that the magnetite needle axis is subject to an extensional stress, most likely attained during cooling of the inclusions within the plagioclase host.

Keywords: Plagioclase, magnetite needles, ilmenite needles, orientation of inclusions, residual stress

intRoduCtion

Pyroxene and plagioclase crystals of mafic plutonic and high-grade metamorphic rocks commonly contain inclusions of iron-titanium oxides (Feinberg et al. 2004), ranging in size from several nanometers to hundreds of micrometers. They often appear as needles, thin plates, or blades (Fleet et al. 1980). Most important to the paleomagnetic community are titanium-poor magnetite inclusions, as their silicate hosts may protect the magnetic particles from alteration during weathering and meta-morphism, preserving their remanent magnetism (Hargraves and Young 1969; Davis 1981; Feinberg et al. 2004; Usui et al. 2006). The older the rock, the more likely that it has undergone some level of metamorphism, thus these exsolved particles allow us to see beyond secondary geologic events, to gather information about even the earliest geodynamo (Tarduno et al. 2006).

The oxide inclusions in pyroxenes are often found in great abundance and are oriented in two to three visible directions within a single grain. Iron is incorporated in the silicate crystal structure during primary crystallization and is then exsolved during cooling to grow magnetite along convenient directions in the silicate host, guided by an optimal fit between the two crystal lattices (Davis 1981). This optimal fit can be used to determine the temperature at which the oxide particles were formed. The exsolution temperature is essential when determining the origin of remanent magnetization (Feinberg et al. 2004). Magnetite

particles formed above the Curie temperature of pure magnetite (585 °C at atmospheric pressure) would acquire thermo-remanent magnetization, which is the ideal form of magnetization for pa-leomagnetic studies (Tarduno et al. 2006). Feinberg et al. (2005) documented a domain microstructure in magnetite inclusions in Precambrian anorthosites in which iron-rich platelets were sur-rounded by ulvöspinel lamellae on the micrometer scale. It was suggested that compositional boundaries prevented magnetic domains from moving, thus enhancing magnetic stability.

The relationship between the structures of pyroxene host and magnetite particles has been well established (Bown and Gay 1959; Fleet et al. 1980). These studies found that magnetite (Mag) inclusions in augite are arranged in two orientations known as X and Z, which are subparallel to the [100] and [001] crystal-lographic directions of clinopyroxene (Cpx), respectively (Renne et al. 2002). Inclusions subparallel to [100]Cpx have [110]Mag//[010]Cpx, (1 11)Mag//(101)Cpx, and [112]Mag//[101]Cpx. Inclu-sions subparallel to [001]Cpx have [110]Mag//[010]Cpx, (111)Mag//(100)Cpx, and [1 12]Mag//[001]Cpx (Feinberg et al. 2004). The relationship between feldspar and magnetite inclusions is complicated by the triclinic lattice structure of plagioclase, and little work has been done to characterize this relationship.

This study focuses on a sample of anorthosite gneiss from the Adirondack Mountains in New York. The sample was found in the Frank Turner research collection at Berkeley (57-I-23) and was probably collected by Harry Hess near the border of the anorthosite mass in connection with his investigation of optical properties of clinopyroxenes (Hess 1949). The petrology and metamorphism of Adirondack rocks have been studied in detail (e.g., Buddington

* E-mail: [email protected]† Present address: Center for Advancing Materials Performance from the Nanoscale, Jiatong University, Xi’an 710049, China.

WENK ET AL.: MAGNETITE AND ILMENITE INCLUSIONS IN PLAGIOCLASE 1317

1939, 1950; Valley and O’Neil 1984; McLelland et al. 1996; Alcock and Muller 1999). Although there is some controversy (Alcock and Muller 2001), it has been established that the anorthosite was emplaced around 1100 Ma, resulting in anatexis and granulite facies metamorphism of adjacent rocks. The sample studied here is close proximity to the anorthosite contact, with large crystals of plagio-clase, garnet, clinopyroxene, ilmenite, and magnetite.

Phase relations of Fe-Ti oxide minerals are complex, de-pending on temperature, pressure, and oxygen fugacity (e.g., Ghiorso 1997; Lindsley 1962, 1963; O’Neill and Navrotsky 1984; Wechsler et al. 1984). In order of increasing oxidation, Fe-Ti pairs in plutonic rocks are ulvöspinel-rich magnetite + ilmenite, magnetite + hematite and hematite + rutile (Buddington and Lindsley 1964). Particularly for Adirondack anorthosites and granulite gneisses, Buddington and Balsley (1961) attribute

ilmenite-magnetite intergrowths to exsolution at subsolidus con-ditions. Buddington and Lindsley (1964) estimate temperatures from magnetite-ilmenite equilibria at 750–800 °C and oxygen fugacities log10 fO2(atm) at –13 to –18.

The sample chosen for this study exhibits not only large crystals of ilmenite and magnetite, but also small iron/titanium oxide needles within crystals of plagioclase and clinopyroxene. Here we investigate the morphology of the oxide inclusions, as well as the orientation and crystallographic relationships between plagioclase hosts and oxide inclusions, with electron backscatter diffraction (EBSD) and synchrotron X-ray micro-focus techniques.

metHodsA thin section of the anorthosite sample was prepared and subsequently pol-

ished. First the thin section was subjected to a 3 µm diamond polish for roughly 2 h, then for a half an hour using a 1/4 µm diamond polish. Finally, the sample was polished for 5 min by hand in colloidal silica. The final thickness was slightly greater than 30 µm so that the plagioclase crystals displayed first-order yellow and orange interference colors. No coating was applied to the sample.

The thin section was investigated with a petrographic microscope to determine the mineralogical composition and to identify regions in which oxide inclusions were abundant. The anorthosite gneiss contains symplectic aggregates of garnet, ilmenite, and clinopyroxene (Fig. 1a). Small, elongate oxide inclusions appear as irregular swarms in the interior of plagioclase crystals. Depending on their orien-tation, they appear as rods (Fig. 1b) or as dots, suggesting a needle morphology. Where they appear as rods they are mostly parallel, with a few exceptions where orientations are random. If Albite twins are present in plagioclase, the needles are parallel to the (010) twin planes (Fig. 1b).

Selected regions were studied with a Zeiss EVO scanning electron microscope (SEM) under the following settings: 25 kV, 100 µA beam current, 5 nA I Probe current, around 20 Pa variable pressure vacuum, and a working distance of 12–15 mm. The sample surface was tilted 70° relative to the horizontal. TSL-OIM Data Collection software was used to collect and index backscatter diffraction patterns of the plagioclase crystals and the iron oxide inclusions using both spot and map scanning modes. Inclusions in 10 differently oriented plagioclase grains were investigated.

Prior to EBSD analysis, energy-dispersive X-ray spectroscopy (EDS) was used to verify the chemical composition of the mineral components as well as the oxide inclusions. The EDAX Genesis software was used to create a semi-quantitative chemical map of a region of the sample. Map scans were created with 1 µm steps. Scans for Fe and Ti show that both elements are present in most needles, although with different concentrations (Fig. 2). No external standards were used and, par-ticularly for the small inclusions, the beam spread, and penetration always included plagioclase. Furthermore, compositionally different oxide phases in the inclusions cannot be separated quantitatively.

figuRe 1. (a) Micrograph of the anorthosite thin section at 10× magnification in plane polarized light, showing a large plagioclase crystal with a cloud of oxide inclusions. On the right side are globular garnet (Grt), ilmenite (Ilm), and clinopyroxene (Cpx). (b) Enlarged region with crossed polarized light showing Albite twins and oxide needles parallel to the twin boundaries.

figuRe 2. EDS maps of (a) Fe and (b) Ti of oxide needles in twinned plagioclase crystal. Note presence of Ti in the large needle, but in varying concentrations.

WENK ET AL.: MAGNETITE AND ILMENITE INCLUSIONS IN PLAGIOCLASE1318

BEARTEX software (Wenk et al. 1998) was used to calculate the orientation distribution (OD) from the Bunge-Euler angles provided by the TSL-OIM data collection software. Pole figures of spot scans were then calculated from the OD. The actual uncertainty of angle measurements is less than 1°. However, for representation, orientations were smoothed and appear therefore as larger regions in the pole figure illustrations. Euler angles collected during the map scan were checked for consistency and analyzed using Maptex, a function of BEARTEX.

The same samples that were investigated with EBSD were also studied with synchrotron X-ray Laue microdiffraction. The experiments were carried out at beamline 12.3.2 of the Advanced Light Source (ALS) at Lawrence Berkeley Na-tional Laboratory, and for details of the method we refer to Kunz et al. (2009). A polychromatic X-ray beam with an energy range of 4–22 keV was focused to ∼1 × 1 µm2. The sample was mounted on a high-resolution x-y scan stage and positioned at the focal point of the X-rays by laser triangulation. The diffraction experiment was conducted in reflection mode, with the sample tilted 45° to the incident X-ray. A two-dimensional X-ray MAR CCD detector with 1024 × 1024 pixel resolution, serving as a diffraction detector, was mounted with its center at 2θ of 90°, and a Si-drift detector that served as a fluorescence detector was also placed sideways. The diffraction geometries, including the distance from the CCD to the sample (∼8 cm), the center position of the CCD, and the CCD tilt angles, were calibrated by indexing a diffraction pattern of a strain-free Si sample.

X-ray microdiffraction measurements were taken from four different plagio-clase crystals. First, the regions of interests were raster scanned with the microfocus X-ray beam, and at each position a fluorescence spectrum was collected to map chemical composition. The X-ray fluorescence map for the Kα-line of iron is shown in Figure 3a; the grayscale in the map indicates the relative concentration of iron. Magnetite and ilmenite needles are clearly visible in the map, shown as light gray and white, respectively. Laue diffraction patterns were taken from the plagioclase matrix (black arrow on white background in Fig. 3a) and iron oxide needles (white arrow in Fig. 3a) with short and long exposure time, respectively, due to the much smaller volume of magnetite compared to plagioclase. A typical Laue diffraction image for the spot marked by the white arrow is shown in Figure 3b. Some reflections are indexed. Circles refer to plagioclase and squares to magnetite.

Results

General featuresSemiquantitative EDS chemical analysis revealed that the

anorthosite sample is mostly comprised of plagioclase, pyroxene, garnet, and Fe-Ti oxides. Approximate chemical formulas were normalized. Plagioclase has an average formula of (Na0.60Ca0.35

K0.03Fe0.015Ti0.007)(Al1.32Si2.68)O8 corresponding to andesine. The pyroxene is identified as clinopyroxene with an approximate formula of (Ca0.75Fe0.51Mg0.44 Al0.24Na0.07)Si2O6, correspond-ing to augite. The garnet is almandine with a composition (Al2.39Fe1.68Ca0.61Mg0.32)Si3O12. Large magnetite grains (not the needles) have an average formula of Fe2.967Ti0.033O4 and large ilmenite is, on average, Fe1.06Ti0.94O3. The small size of the needles and broadening of the electron beam after it entered the material (several micrometers) made it difficult to obtain quantitative chemical analyses of the oxide needles, because a plagioclase volume was always included. However, it was obvious that there are Fe-rich and Ti rich regions (Fig. 2), which were later identi-fied with diffraction to correspond to magnetite and ilmenite, respectively. Titanium is also present in Fe-rich needles and the possibility that titano-magnetite had exsolved to form more pure magnetite and ulvöspinel cannot be excluded. This could not be resolved with the techniques we employed.

Optical examination of iron oxide particles reveals a needle shape with an average aspect ratio of roughly 15:1, and an average length of 44 µm and width of 3 µm, averaged over 50 inclusions. We note that both the aspect ratio and length are a minimum because most needles are slightly tilted and were therefore truncated during sample preparation. Needles were not uniform throughout a plagioclase crystal, but were instead clustered in “clouds” (Fig. 1).

As indicated above, most needles are strictly parallel but there are exceptions. There are large globular grains of ilmenite (Fig. 1a), and within needle swarms there are small crystals of more irregular shape, probably nucleated on defects (Fig. 1b). Gener-ally these irregular crystals do not form long needles.

EBSD analysisThree phases were selected for indexing: C1 plagioclase

(Fig. 4a), Fd3m magnetite (Fig. 4b), and R3 ilmenite (Fig. 4c).

figuRe 3. X‑ray fluorescence map based on FeKα emission line (a) and Laue diffraction pattern (b) indexed for plagioclase (circles) and for magnetite (squares). The black arrow in a points to plagioclase and the white arrow to magnetite. The white square was scanned in detail for residual strain (cf. Fig. 9).


Initially, inclusions were indexed only using the phase magne-tite, and diffraction images of many crystals were successfully indexed, but others failed. We then tried indexing with other iron oxide phases and about one quarter of all inclusions could be indexed with hematite. EDS analysis revealed that the “hematite” regions were actually isostructural ilmenite. A local scan showed that magnetite and ilmenite are often intergrown, with ilmenite concentrated on the needle surface and on needle tips (Fig. 5).

A first goal was to quantify needle morphology. With op-tical microscopy, plagioclase grains were selected in which needles were roughly parallel to the thin section surface and others where they were perpendicular. Figure 6a shows a SEM image of magnetite needles that are approximately parallel to the surface. The corresponding crystallographic orientations of the host plagioclase matrix (Fig. 6b), magnetite (Fig. 6c), and ilmenite needle orientation (Fig. 6d) are represented as pole figures. Clearly, one <110> axis of magnetite is parallel to the needle direction (arrow in Fig. 6c). Figure 6d shows the orienta-tion relations for an ilmenite needle, which in this case is the tip of the magnetite needle shown in Figure 6c, and corresponds to the scan in Figure 5. Pole figures indicate that one <1010> axis is the needle direction (arrow in Fig. 6d). This morphology was verified for all crystals that were investigated. Furthermore, for all intergrown magnetite-ilmenite assemblages, one magnetite {111} plane is parallel to the basal plane (0001) of ilmenite (open arrows in Figs. 6c and 6d).

The next step was to investigate the crystallographic orienta-tion relationships of cubic magnetite and rhombohedral ilmenite needles in triclinic plagioclase. The lattice plane normal (102) of plagioclase is nearly parallel to the z-axis [001] of plagioclase (Fig. 6b). In all measured crystals, the needle axis [110] of mag-netite (arrow in Fig. 6c) is parallel to [001] of plagioclase (arrow in Fig. 6b). But there are other relationships that can be described in different ways, for example as (111)Mag sub-parallel to either (120)Plag or (120)Plag (arrows in Figs. 6b and 6c), or (552)Mag//(010)Plag (Figs. 7b and 7c). For ilmenite inclusions in plagioclase [1010] of ilmenite is parallel to [001] of plagioclase, and (0001) of ilmenite is sub-parallel to either (120)Plag or (120)Plag (arrows in Figs. 6b and 6d).

Plagioclase crystals frequently display lamellar twinning according to the Albite twin law (Fig. 1b). This has been veri-

fied by EBSD. Needles in host and twin are morphologically parallel, which is not surprising since the plagioclase direction [001], which is parallel to the needle direction, lies in the twin plane (010) of plagioclase (Fig. 6b).

X‑ray microdiffraction analysisThe software package XMAS (Tamura et al. 2002) was

employed to analyze the Laue patterns (LP). First, diffraction peaks in a plagioclase diffraction pattern were searched based on a user-defined peak-to-background threshold. The positions on the CCD (x-pixel and y-pixel), as well as width of each indi-vidual diffraction peak, were determined by fitting intensity of the two-dimensional peaks using a 2D Lorentzian function (Eq. 6.6, pp. 168, Noyan and Cohen 1987). Correlating the fitted peak positions with calculated positions, based on cell parameters of andesine (Fitz Gerald et al. 1986) allows one to establish the grain orientation. The positions of all the indexed plagioclase peaks were recorded and a list was prepared. Since all Laue dif-

figuRe 4. Automatically indexed EBSD diffraction pattern of plagioclase (a), magnetite (b), and ilmenite (c).

figuRe 5. Phase map based on EBSD patterns and plotted with Maptex. Light gray matrix is plagioclase, black is magnetite, and dark gray is ilmenite. Step size is 1 µm. White spots were not indexed due to low confidence index and a high density is observed in Plagioclase twin domains. The slight curvature of the pattern is due to drift.


figuRe 6. SEM image of needles (a) and corresponding pole figures based on EBSD for plagioclase (b), magnetite (c), and ilmenite (d). Equal area projection. Note that orientations have been artificially smoothed for better visualization. Indices refer to pole to lattice planes, except [001], which refers to a lattice direction.


fraction patterns taken from regions containing magnetite also contained signals from the plagioclase host, the peak positions previously identified as plagioclase reflections were substracted from the magnetite pattern. This reduced list was used for index-ing magnetite. Figure 3b shows a typical Laue diffraction pattern taken on the spot marked by the white arrow in Figure 3a, with Miller indices of selected peaks from both plagioclase (marked with circles) and magnetite (marked with squares).

Pole figures (Figs. 7a–7c) confirm that the magnetite needle is extended parallel to the [110] direction (solid square in Fig. 7a), which is parallel to the (102) = [001] direction of plagioclase (Fig. 7c), and (552)Mag (solid square in Fig. 7b) is parallel to (010)Plag (Fig. 7c). Twenty-three oxide needles provided reliable indexing, and displayed the same relationships, consistent with results from the EBSD analysis.

In addition to orientation information, the Laue diffraction technique also provides detailed data about lattice distortions. In cubic magnetite, symmetrically equivalent {100} directions may not be equivalent because of residual strain. From the indexed Laue diffraction pattern, a local strain tensor given by a 3 × 3 symmetric matrix with six variants can be calculated (Chung and Ice 1999; Pavese 2005). For example, the deviatoric strain tensor at the position where the white arrow is pointing in Figure 3a is expressed by the matrix

ε =

–1.09 1.13 0.04

1.13 1.84 2.60

0.04 2.60 –0.75

× −10

3

in laboratory coordinates, i.e. x horizontal and perpendicular to the incident X-ray beam, y parallel to the projection of the incident beam on the sample plane, z perpendicular to the sample plane. Figure 7d displays the strain tensor in a polar coordinate system and spherical projection (Chen et al. 2011a). The highest tensile strain, +3.66 × 10−3, is found close to the needle direc-tion, and compressive deviatoric strain is perpendicular to this direction. The principal strains of this tensor are +3.66 × 10−3, –2.59 × 10−3, and –1.06 × 10−3 (in microstrains) and the principal vectors in laboratory coordinates are [0.20, 0.84, 0.50], [0.38, –0.54, 0.75], and [0.90, 0.04, –0.43], respectively. Thus, the highest compressive strain is loaded close to the plane normal to (111). The strain tensors of all magnetite needles that were tested have very similar shapes to the one shown in Figure 7d.

Because strains are calculated with respect to a set of strain-free lattice parameters, which highly depend on chemical compositions and ordering state for triclinic plagioclase, it is not meaningful to calculate the deviatoric strains in the plagioclase matrix.

disCussion

In the sample that was investigated, oxide inclusions in plagioclase occur as needles, rather than platelets as has been described for inclusions in pyroxene (Feinberg et al. 2005). The needles are parallel to one of the <110> directions of cubic magnetite and to one of the <1010> directions of rhombohedral ilmenite. Both magnetite and ilmenite structures have close-packed arrangements of oxygen atoms, which are aligned along these two directions. Also, where the two oxides are intergrown,

figuRe 7. Pole figures of (a) {110} and (b) {552} planes of magnetite, and (c) (102) = [001] and (010) planes of plagioclase determined from Laue diffraction patterns. Solid squares indicate planes of magnetite with obvious orientation relationship to plagioclase (c). (d) The strain tensor in a polar coordinate system measured at the square outlined and marked by the white arrow in Figure 3a. Units of contours are micro-strains.


a {111}Mag plane is parallel to {0001}Ilm, corresponding to planes with close-packed O atoms.

How do these simple oxide structures fit into triclinic plagioclase? This is not immediately obvious from conven-tional plagioclase structure projections in textbooks, such as the “double crankshaft,” or the “Y-projection” with four-membered tetrahedral rings used to explain Al-Si order (e.g., Smith and Brown 1988). Rotating a crystal structure model reveals that a [001] projection is amazingly regular and displays different channels that are parallel to the z-axis. Figure 8 displays an ideal-ized [001] projection. Of course plagioclase is triclinic and all tetrahedra are irregular and tilted in different directions. What appears here as rings, displays tetrahedra at different levels. But the most conspicuous features are large six-membered “ring”

channels that can accommodate octahedra, one of which is il-lustrated. The [110] direction of magnetite, corresponding to an edge of the Fe-O coordination polyhedron is parallel to [001] of the plagioclase structure and (111) faces of the coordination octahedron are more or less parallel to (120) and (120) planes of plagioclase. Thus the orientation relationship—while not corresponding to a true coincidence lattice—is plausible. The channels provide excellent nucleation sites for Fe- and Ti-oxide octahedra to accommodate these ions that are no longer in solid solution as the plagioclase cools. It is natural that the needles preferentially grow along the channel direction but obviously also replace plagioclase material around the channels, though at a much slower rate.

The presence of Ti in the magnetite needles is suggestive of ilmenite (FeTiO3) and possibly ulvöspinel (Fe2TiO4) lamellae. We have documented ilmenite at tips and surfaces of magnetite needles, but with the techniques employed, we cannot char-acterize finer structures or exclude the presence of ulvöspinel domains. Multidomain-sized magnetite inclusions in pyroxene that contain ulvöspinel have been found to exhibit single-domain states, increasing the coercivity of the magnetization (Feinberg et al. 2005). These microstructures could only be confirmed using magnetic force microscopy or transmission electron microscopy, which is not the subject of this study.

From the strain tensor, which is determined with the Laue X-ray microdiffraction technique (Fig. 7d), and knowing single-crystal elastic properties, residual strain can be converted to residual stress by applying Hooke’s law σ'i = Cijε'j, where σ'i, Cij, and ε'j are deviatoric stress tensor (second rank), stiffness tensor (fourth rank), and deviatoric strain tensor (second rank), respectively. It is worth emphasizing that this equation is only applicable to cubic crystals, otherwise deviatoric stress is not directly linked to deviatoric strain (for details see Chen et al. 2011b). Figure 9 shows a map of the diagonal components of the stress tensor, σ'x'x', σ'y'y', and σ'z'z' in the region marked by a white dashed square in Figure 3a. We employed the stiffness constants of magnetite reported by Reichmann and Jacobsen (2004). The stiffness tensor of magnetite is almost isotropic, ranging from 260 to 271 GPa, contrary to plagioclase, which is highly anisotropic ranging from about 80 to 275 GPa for andesine

figuRe 8. [001] projection of the plagioclase structure, idealized. The rings of tetrahedra in this representation do not actually connect in a plane. Open circles represent large cations. An octahedron is inserted into one of the large channels.

figuRe 9. Map of the diagonal stress tensor components (in MPa) relative to a coordinate system where y' is parallel to the needle axis, x' is perpendicular to y' and roughly in the plane of the thin section, z' is almost perpendicular to the thin section. Step size is 2 µm. The region corresponds to the square outlined in Figure 3a.


(Ryzhova 1964). The coordinate system in Figure 9 is defined by y' parallel to the needle direction, which is roughly lying in the sample plane in this case, x' perpendicular to the (552)Mag plane and (010)Plag plane, which is perpendicular to the needle direction but approximately parallel to the sample plane in this case, and z' perpendicular to the x'-y' plane, which is about the sample plane. We note that σ'y'y' is larger (positive) than σ'x'x' and much larger than σ'z'z' (negative), thus a ∼350 MPa tensile stress exists parallel to the needle direction, which is less confined by the plagioclase structure.

To understand the stress distribution in magnetite, we explore a simple model to simulate the interactions between plagioclase matrix and magnetite needles. If magnetite needles precipitated in plagioclase channels along [001]Plag at elevated temperatures with no stress, differences in coefficients of thermal expansion (CTE) may impose stresses during cooling. In the model, we as-sume that CTE values of magnetite and plagioclase are constant over the temperature range 800–300 K. For isotropic magnetite, the linear CTE is 0.90 × 10−5 K−1 (Run 2 of Levy et al. 2004). For plagioclase, the CTE is highly anisotropic (e.g., Saucier and Saplevitch 1962; Tribaudino et al. 2010). Tribaudino et al. (2010) show that the three principal components of the CTE tensor are –0.32 × 10−5 K−1, 0.70 × 10−5 K−1, and 1.80 × 10−5 K−1 for plagioclase An35Ab65 at 550–935 K (their Fig. 6), all of which are inclined to crystallographic directions (their Fig. 7). In Figure 10, we plot the CTE ellipsoid relative to crystal coordinates in a spherical projection. We observe that along [001]Plag (center of plot) the CTE of plagioclase is lower than for magnetite, while perpendicular to [001]Plag the CTE of plagioclase is overall higher. Thus, it is plausible that the residual stress was attained during cooling. Alternatively, the stress could have been produced dur-ing growth of magnetite needles in the rigid plagioclase channels (e.g., Scherer 2004).

In this study, EBSD techniques and microfocus X-ray dif-

fraction were used to characterize the epitaxial relationships of plagioclase with magnetite and ilmenite needles. The anorthositic gneiss from Adirondack shows a very strong and consistent orien-tation control that seems to be related to the crystal structures. In the future, it needs to be established if these relationships apply to other samples with different compositions and histories as well.

A simple relationship between plagioclase, magnetite, and ilmenite has been established: [110]Mag//[1010]Ilm//[001]Plag. But oxide needles do not have rotational freedom around this axis. The (111)Mag and (0001)Ilm planes are sub-parallel to either or both (120)Plag and (120)Plag. Also this relationship should be explored further before generalizations are made. Once a definite orienta-tion relationship is established, optimal phase boundary theory could then be applied following the methods used by Fleet et al. (1980) who were able to determine the nucleation temperatures of magnetite inclusions within crystals of the pyroxene augite. The nucleation temperature could then be used to determine the type of remanent magnetization that the needles had acquired.

It is clear that the close packing of oxygen atoms is what determines the orientation of the exsolved magnetite. The channels formed by the distributions of the silica and alumina tetrahedra restrict the morphology of magnetite to rods. The elongation of magnetite’s cubic structure along an axis does produce strain in the crystal lattice, which is still expressed in substantial residual stresses.

aCknoWledgmentsWe acknowledge support from DOE-BES (DE-FG02-05ER15637) and NSF

(EAR-0836402) as well as access to beamline 12.3.2 at ALS. ALS is supported by the Director, Office of Science, Office of Basic Energy Sciences, Materials Sci-ence Division, of the U.S. Department of Energy. The micro-diffraction program at 12.3.2 was made possible by NSF grant no. 0416243. We are appreciative to Joshua Feinberg who got us interested in magnetite exsolution in silicates and to Paul Renne for making us aware of this particular sample that we analyzed. Martin Kunz helped with microfocus measurements and made useful suggestions about the manuscript. We also acknowledge helpful comments by two reviewers, Nick Timms and Mario Tribaudino, that helped to improve the manuscript.

RefeRenCes CitedAlcock, J. and Muller, P.D. (1999) Very high-temperature, moderate pressure

metamorphism in the New Russia gneiss complex, northeastern Adirondack Highlands, metamorphic aureole to the Marcy anorthosite. Canadian Journal of Earth Sciences, 36, 1–13.

——— (2001) Very high temperature, moderate pressure metamorphism in the New Russia gneiss complex, northeastern Adirondack Highlands, metamorphic aureole to the Marcy anorthosite: Reply. Canadian Journal of Earth Sciences, 38, 471–475.

Bown, M.G. and Gay, P. (1959) The identification of oriented inclusions in pyroxene crystals. American Mineralogist, 44, 592–602.

Buddington, A.F. (1939) Adirondack igneous rocks and their metamorphism, vol. 7, 354 pp. Memoirs Geological Society of America, Boulder, Colorado.

——— (1950) Composition and genesis of pyroxene and garnet related to Adirondack anorthosite and anorthosite-marble contact zones. American Mineralogist, 35, 659–670.

Buddington, A.F. and Balsley, J.R. (1961) Micro-intergrowths and fabrics of iron–titanium oxide minerals in some Adirondack rocks. Mehaveden, 1, 1–16.

Buddington, A.F. and Lindsley, D.H. (1964) Iron-titanium oxide minerals and synthetic equivalents. Journal of Petrology, 5, 310–357.

Chen, K., Kunz, M., Tamura, N., and Wenk, H.R. (2011a) Evidence for high stress in quartz from the impact site of Vredefort, South Africa. European Journal of Mineralogy, 23, 169–178.

Chen, K., Kunz, M., Tamura, N., and Wenk, H.-R. (2011b) Deformation twinning and residual stress in calcite studied with synchrotron polychromatic X-ray microdiffraction. Physics and Chemistry of Minerals, 38, 491–500.

Chung, J.S. and Ice, G.E. (1999) Automated indexing for texture and strain measurement with broad-bandpass X-ray microbeams, Journal of Applied Physics, 86, 5249–5255.

Davis, K.E. (1981) Magnetite rods in plagioclase as the primary carrier of stable NRM in ocean floor gabbros. Earth and Planetary Science Letters,

figuRe 10. Coefficient of thermal expansion (CTE) tensor ellipsoid of plagioclase, based on values from Tribaudino et al. (2010), represented in a polar coordinate system. Crystallographic directions are indicated. Equal area projection.


55, 190–198.Feinberg, J.M., Wenk, H.-R., Renne, P.R., and Scott, G.R. (2004) Epitaxial relation-

ships of clinopyroxene-hosted magnetite determined using electron backscatter diffraction (EBSD) technique. American Mineralogist, 89, 462–466.

Feinberg, J.M., Scott, G.R., Renne, P.R., and Wenk, H.-R. (2005) Exsolved mag-netite inclusions in silicates: Features determining their remanence behavior. Geology, 33, 513–516.

Fitz Gerald, J.D., Parise, J.B., and Mackinnon, I.D.R. (1986) Average structure of an An48 plagioclase from the Hogarth ranges. American Mineralogist, 71, 1399–1408.

Fleet, M.E., Bilcox, G.A., and Barnett, R.L. (1980) Oriented magnetite inclusions in pyroxenes from the Grenville Province. Canadian Mineralogist, 18, 89–99.

Ghiorso, M.S. (1997) Thermodynamic analysis of the effect of magnetic ordering on miscibility gaps in the Fe-Ti cubic and rhombohedral oxide minerals and the FeTi oxide geothermometer. Physics and Chemistry of Minerals, 25, 28–38.

Hargraves, R.B. and Young, W.M. (1969) Source of stable magnetism in Lambert-ville diabase. American Journal of Science, 267, 1161–1177.

Hess, H. (1949) Chemical composition and optical properties of common clinopy-roxenes. American Mineralogist, 34, 621–666.

Noyan, I.C. and Cohen, J.B. (1987) Residual Stress—Measurement by Diffraction and Interpretation. Springer, Heidelberg.

Kunz, M., Tamura, N., Chen, K., MacDowell, A.A., Celestre, R.S., Church, M.M., Fakra, S., Domning, E.E., Glossinger, J.M., Plate, D.W., and others. (2009) A dedicated superbend X-ray microdiffraction beamline for materials, geo-, and environmental sciences at the advanced light source. Review of Scientific Instruments, 80, 035108.

Levy, D., Artioli, G., and Dapiaggi, M. (2004) The effect of oxidation and reduction on thermal expansion of magnetite from 298 to 1173K at different vacuum conditions. Journal of Solid State Chemistry, 177, 1713–1716.

Lindsley, D.H. (1962) Investigations in the system FeO-Fe2O3-TiO2. Carnegie Institution of Washington Yearbook, 61, 100–106.

——— (1963) Equilibrium relations of coexisting pairs of Fe-Ti oxides. Carnegie Institution of Washington Yearbook, 62, 60–66.

McLelland, J.M., Daly, J.S., and McLelland, J.M. (1996) The Grenville Orogenic Cycle (ca. 1350–1000 Ma): an Adirondack perspective. Tectonophysics, 265, 1–28.

O’Neill, H.St.C. and Navrotsky, A. (1984) Cation distributions and thermody-namic properties of binary solid spinel soluttions. American Mineralogist, 69, 733–753.

Pavese, A. (2005) About the relations between finite strain in non-cubic crystals and the related phenomenological P-V equation of state. Physics and Chemistry of Minerals, 32, 269–276.

Reichmann, H.J. and Jacobsen, S.D. (2004) High-pressure elasticity of a natural magnetite crystal. American Mineralogist, 89, 1061–1066.

Renne, P.R., Scott, G.R., Glen, J.M.G., and Feinberg, J.M. (2002) Oriented

inclusions of magnetite in clinopyroxene: Source of stable magnetization in gabbros of the Messum Complex, Namibia. Geochemistry Geophysics Geosystems, 3, 1079.

Ryzhova, T.V. (1964) Elastic properties of plagioclase. Bulletin Academy of Sci-ences USSR Geophysical Series, 7, 633–635 (Engl. Transl.) (Or Izv/Akad Nauk SSSR, Ser. Geotiz., 1964, p. 1049–1051, in Russian).

Saucier, H. and Saplevitch, A. (1962) La dilatation thermique des feldspaths. Norsk Geologisk Tidsskrift, 42, 224–243.

Scherer, G.W. (2004) Stress from crystallization of salt. Cement and Concrete Research, 34, 1613–1624.

Smith, J.V. and Brown, W.L. (1988) Feldspar Minerals, vol. 1, 2nd ed., 828 pp. Springer Verlag.

Tamura, N., MacDowell, A.A., Celestre, R.S., Padmore, H.A., Valek, B., Bravman, J.C., Spolenak, R., Brown, W.L., Marieb, T., Fujimoto, H., Batterman, B.W., and Patel, J.R. (2002) High spatial resolution grain orientation and strain map-ping in thin films using polychromatic submicron X-ray diffraction. Applied Physics Letters, 80, 3724–3727.

Tarduno, J.A., Cottrell, R.D., and Smirnov, A.V. (2006) The paleomagnetism of single silicate crystals: Recording geomagnetic field strength during mixed polarity intervals, superchrons, and inner core growth. Reviews of Geophys-ics, 41, 1–31.

Tribaudino, M., Angel, R.J., Camara, F., Nestola, F., Pasqual, D., and Margiolaki, I. (2010) Thermal expansion of plagioclase feldspars. Contributions to Mineral-ogy and Petrology, 160, 899–908.

Usui, Y., Nakamura, N., and Yoshida, T. (2006) Magnetite microinclusions in silicate and magmatic flow fabric of the Goyozan granitoid (NE Japan): Significance of partial remanence anisotropy. Journal of Geophysical Research, 111, B11101, 19pp, doi:10.1029/2005JB004183.

Valley, J.W. and O’Neil, J.R. (1984) Fluid heterogeneity during granulite facies metamorphism in the Adirondacks: stable isotope evidence. Contributions to Mineralogy and Petrology, 85, 158–173.

Wechsler, B.A., Lindsley, D.H., and Prewitt, C.T. (1984) Crystal structure and cation distribution in titanomagnetites (Fe3-xTixO4). American Mineralogist, 69, 754–770.

Wenk, H.-R., Matthies, S., Donovan, J., and Chateigner, D. (1998) BEARTEX, a Windows-based program system for quantitative texture analysis. Journal of Applied Crystallography, 31, 262–269.

Manuscript received deceMber 9, 2010Manuscript accepted april 12, 2011Manuscript handled by Florian heidelbach

morphology and microstructure of magnetite and ilmenite...

Documents