Geochemistry, Geophysics, Geosystems

Remanence stability and magnetic fabric development in synthetic shear zones deformed at 500°C

Authors


Abstract

Shear experiments were performed on magnetite-bearing calcite aggregates to examine magnetic fabric development and remanence stability in a deforming system using elevated temperature and pressure to encourage deformation by crystal-plastic processes. Samples composed of 1 wt % pseudo-single-domain magnetite (1–2 μm) in a calcite matrix were created with either strong or weak initial fabrics and deformed in coaxial simple shear to strains up to γ = 1.5 at constant strain rates between 6 × 10−5 and 1 × 10−4 s−1 at 500°C and confining pressure of 300 MPa. Samples were given weak field thermal remanent magnetizations prior to deformation. Demagnetization of postdeformation remanence reveals that a primary remanent magnetization can withstand deformation at pressures and temperatures approximately equivalent to greenschist facies metamorphic conditions on laboratory time scales, but this stability is found to depend on the character of the predeformation fabric. The origin of secondary remanence components acquired during deformation is uncertain but is likely to partially result from thermal viscous remagnetization. Complete postdeformation remagnetization in initially anisotropic samples appears to involve a stress-softening or piezoremanent magnetization mechanism. Postdeformation anisotropy measurements show progressive changes in magnetic fabric strength with strain. In the absence of a strong initial magnetic anisotropy, magnetic fabric intensity increases linearly as a function of strain; however, deformation that overprinted an existing fabric results in an apparent decrease of the initial anisotropy at low strains followed by rapid increases in magnetic fabric strength with increasing strain. Our results underscore the important role that initial fabric can play in determining the character of deformation fabrics.

1. Introduction

Magnetic fabric measurements are a popular tool for rapidly determining qualitative information about petrofabric orientation and relative strain [Tarling and Hrouda, 1993; Borradaile and Jackson, 2004]. In particular, anisotropy of magnetic susceptibility (AMS) measurements are frequently used to map changes in rock textures over broad regions. However, it is difficult to consistently obtain useful quantitative data from magnetic fabrics such as strain, paleostresses, and other deformation history information. Interpretation of magnetic fabrics is complicated because bulk magnetic properties always average signals from different mineral phases, grain size fractions and subfabrics [Borradaile, 1988]. Additionally, results from magnetic fabric analysis and standard microstructure analysis reflect the collective strain history of a rock and thus may fail to provide straightforward information about any particular stage of deformation.

Complicating factors that prevent extraction of quantitative data from magnetic fabrics in rocks include strain partitioning and inhomogeneous strain between phases in a rock, lack of information about predeformational rock textures, difficulty in interpreting the effects of multiple deformation stages [Evans et al., 2003] and magnetochemical alteration. Despite the necessary caveats involved with detailed interpretation of magnetic fabrics, AMS in bulk samples is still a valuable tool as a rapidly obtained textural indicator. Magnetic anisotropy studies are commonly used in structural and tectonic research; therefore, a comprehensive quantitative understanding of how magnetic properties reflect deformation history is needed to maximize the information obtained.

Since the work of Graham [1966], who recognized a pattern between AMS directions and finite strain, a number of studies have succeeded in formulating correlations between magnitudes of magnetic anisotropy and strain for naturally deformed rocks where strain estimates are available [Kneen, 1976; Kligfield et al., 1983; Cogné and Perroud, 1988]. However, these have sometimes been invalidated by subsequent work [Nakamura and Borradaile, 2001] and most determined relationships are only valid for a particular locality and will not necessarily hold for different lithologies or deformation regimes. Previous experimental magnetic fabric studies have deformed rocks or rock analogs at room temperature [Borradaile and Alford, 1988; Jackson et al., 1993]. Such studies usually involve quasi-brittle deformation regimes or large changes in sample volume that are not necessarily representative of natural modes of deformation or AMS development. We present experimental observations of changes in magnetic anisotropy and remanence in material deformed under simulated metamorphic conditions. The goal of this study is to examine the effects of elevated temperatures and pressures that facilitate plastic deformation on AMS development while additionally examining the extent of postdeformation remanence stability for potential paleomagnetic application.

2. Sample Preparation

Synthetic samples were composed of 1.0 wt % synthetic Wright 3006 (W3006) magnetite dispersed in a calcite matrix. This magnetite has a nominal grain size of 2–3 μm, placing it in the pseudo-single-domain (PSD) grain size range for magnetite. Yu et al. [2002] determined an estimated mean grain size of 1.06 ± 0.71 μm and an average axial ratio of 1.62 for W3006, and Özdemir and Banerjee [1982] noted a grain size range from 0.2 to 3.8 μm. The magnetite powder used in all samples was annealed in a reducing CO-CO2 atmosphere at 700°C for 15 h to remove any preexisting partial oxidation before mixing. Magnetite stoichiometry was verified through saturation magnetization (Ms) values and Mössbauer spectroscopy. The studied calcite powder was obtained from single crystals of Iceland spar crushed to an average grain size of 50 μm. Samples were thoroughly mixed until it was visually determined that the magnetite was well distributed (to reduce magnetic interaction effects). Although hand mixing appears to be effective at dispersing most of the magnetite, we acknowledge the general inability to uniformly distribute magnetite grains without some degree of clustering. In the photomicrographs in Figure 1, the magnetite grains are generally well dispersed, although some clustering is evident, especially along boundaries of large calcite grains.

Figure 1.

Reflected light photomicrograph of the spatial distribution of magnetite grains in the calcite matrix for (a) a typical sample after HIP and (b) a deformed sample. The small dark spots in the image are pits in the sample surface.

Test samples of W3006 magnetite in a matrix of high-temperature cement were created to test the thermal stability of thermoremanent magnetization (TRM). A TRM was given to the test samples in a field of 100 μT at 600°C and was subsequently subjected to stepwise thermal demagnetization in air (Figure 2b). These measurements indicate that the W3006 magnetite has a median blocking temperature Tb of 530°C, a Curie temperature, Tc of 577°C (Figure 2a), and retains approximately 60% of the original TRM intensity at 500°C. This was considered sufficiently stable for the purpose of our experiments.

Figure 2.

Magnetization properties: (a) comparison of Ms(T) curves with Curie temperatures for samples after HIP and after deformation, (b) stepwise thermal demagnetization of a 0.1 mT TRM given to a test sample of post-HIP material, (c) low-temperature warming curves of an SIRM given after field cooling, and (d) AF demagnetization spectra of predeformation TRM or postdeformation “NRM” for various samples.

Two methods of sample preparation were used during the first synthesis step. A hydraulic oil medium isostatic pressure vessel was used to cold press pellets of sample powders under a hydrostatic pressure of 135 MPa to create approximately isotropic samples with weak fabrics that were then shaped and fitted into Ni canisters. These are hereafter referred to as isostatically prepared material (IPM) samples. To create samples containing strong initial fabrics, the magnetite-calcite powders were cold pressed directly into a Ni canister under an axial load of approximately 140 MPa, referred to as uniaxially prepared material (UPM). Cold-pressed sample canisters were placed in a hot isostatic press (HIP) in a gas medium pressure vessel at 500°C for 5 h at a confining pressure of 300 MPa. HIP and deformation temperatures were constrained to 500°C to avoid the formation of siderite from magnetite due to decarbonation of calcite which was observed during trial HIP runs at higher temperatures. The color of the starting powders was a pale gray, which darkened to a medium gray after HIP, reflecting densification of the material. During HIP and deformation experiments, oxygen fugacity was maintained within the stability field of magnetite near the Ni-NiO buffer by the Ni canister [Frost, 1991].

No significant changes in critical transition temperatures occurred after hot pressing, although moderate changes in hysteresis parameters were observed. These values are included in Table 1 for comparison with postdeformation values. The rise in coercivity, Hc, and small drop in Ms and the slight rise in Tc after HIP may indicate a small degree of oxidation, however, Borradaile and Jackson [1993] demonstrated that room temperature hydrostatic compression of magnetite-calcite-cement aggregates at 200 MPa resulted in permanent increases in both coercivity and ratio of remanent magnetization, Mr to Ms, as well as permanent changes to the shape of coercivity spectra. Therefore, it is unclear whether changes in hysteresis parameters seen after HIP represent chemical changes to the magnetic mineralogy. Additionally, the Verwey transition temperature (Tv) is remarkably constant before and after HIP as is the sharpness of the transition (Figure 2c). This transition temperature has been shown to be sensitive to small changes in magnetite oxidation state and impurity content [Aragòn et al., 1985; Özdemir et al., 1993], and is therefore a good indication that changes in magnetite chemistry during HIP were minor. The Tv transition temperature at or near 108 K measured for our samples is lower than the Tv = 121 K value for ideal, stoichiometric magnetite, but is identical to the value measured for W3006 magnetite by Kosterov [2003]. The lower Tv value for W3006 is possibly due to minor impurity levels in the original material that affects Tv but not Tc. There is no evidence for growth of a secondary magnetic phase during HIP.

Table 1. Changes in Magnetic Properties After HIP, TRM Acquisition, and Deformation
 TV (K)TC (°C)Ms (Am2/kg)Mr (Am2/kg)Mr/MsHc (mT)
Mixed sample powders1085770.8120.0480.0596.9
HIP material1085850.7710.0650.0849.6
Post-TRM acquisition108-0.7520.0720.09511.0
Deformed IPM samples1075760.6770.0800.11813.7
Deformed UPM samples1085820.6800.0860.12314.6

Microscopic examination of the hot-pressed material revealed dispersed magnetite grains and the presence of abundant, randomly oriented calcite twins that likely originated during crushing (Figure 1). No shape-preferred orientation was observed for calcite grains after HIP. Although no overall change in calcite grain size was detected, it is likely that some grain growth occurs in the fine-grained calcite fraction during HIP, as demonstrated by Schmidt et al. [2008]. Several samples were cut from each batch of HIP material in the shape of elliptical discs with major and minor axes of 9 mm and 6 mm, respectively. Sample thickness was kept to approximately 1 mm in order to minimize the pure shear component of deformation.

3. Experimental Methods

After initial characterization, samples were given a weak field TRM perpendicular to the disc plane (the shear plane in subsequent deformation experiments) by heating to 600°C and cooling in an applied field of 0.1 mT in an atmosphere of flowing Ar. One sample was left unmagnetized before deformation. Test samples that were heated to 600°C showed minor changes in saturation magnetization and low-temperature remanence (Table 1 and Figure 2c), which indicates that thermal alteration of magnetite during TRM acquisition results in slight oxidation.

Sample discs were placed between pistons of alumina rod cut at 45° angles and then fitted into a nickel sheath and encased in an Fe jacket between larger ceramic pistons. Experiments were performed at 500°C with a confining pressure of 300 MPa in a gas medium pressure vessel designed for triaxial deformation [Paterson, 1990]. Figure 3 is a simplified diagram of the sample orientation in the deformation vessel. More details about the apparatus and a complete schematic diagram of a similar sample assembly are given by Hustoft and Kohlstedt [2006]. Simple shear experiments were performed at constant piston displacement rate, corresponding to constant shear strain rates between 6 × 10−5 s−1 and 1 × 10−4 s−1. The applied load on the pistons was increased as necessary to maintain the strain rate. Differential stress was therefore not constant throughout the experiment and the maximum differential stresses varied according to the total strain (Table 2). Experiment times ranged from 1.5 to 6 h. At the end of each experiment, differential loads on the samples were removed and samples were quenched to room temperature at an average rate of about 50°/min.

Figure 3.

Schematic diagram of sample coordinate scheme in relation to shear geometry and field direction from furnace heating coils during experiments.

Table 2. Conditions for Individual Deformation Runs
SampleShear StrainStrain Rate (s−1)Estimated Maximum Stress (MPa)Experiment Length (hours)
IPM Samples
Iso07-110.36E–05681.17
Iso07-120.68E–05792
Iso07-0418E–05813.25
Iso07-0518E–05883.4
Iso07-101.51E–041505
 
UPM Samples
An07-010.36E–05711.17
An07-050.68E–051032
An07-0918E–05983.25
An07-071.31E–04964.25
An07-081.51E–041185.5

Rutter [1974] noted that brittle behavior is suppressed by a confining pressure of 150 MPa in experimental deformation of calcite rocks at 300°C and above. The larger confining pressure and temperature used in our experiments both strongly favor calcite deformation by crystal-plastic processes. Furthermore, several previous studies have documented plastic deformation of calcite at identical or similar conditions [Turner et al., 1954; Rutter, 1974; Walker et al., 1990; de Bresser and Spiers, 1993; Barnhoorn et al., 2004]. Although magnetite is not expected to deform plastically at our chosen conditions [Atkinson, 1977], the bulk sample deformation behavior is controlled by calcite rheology and is likely to be deforming in the ductile regime.

To correct for the strength of the Ni sheath in our experiments, a deformation run was performed on a Ni sheath containing only a thin Ni foil between angled ceramic pistons under the same conditions used in the experiments. We attempted to subtract the resulting stress-strain curve from that of each sample to estimate the maximum applied stress on each sample. However, the Ni strength was found to account for a significant proportion of the total applied load (about 60% for most experiments) and both the Ni and the sample experiments exhibited strain-hardening behavior. Therefore, there is large uncertainty associated with the Ni correction, and the maximum stresses shown in Table 2 should only be taken as upper limits.

It is important to note that the furnace heating coils in the deformation vessel produce a significant magnetic field when the vessel is at temperature. This field was calculated to be approximately 4 mT in the region of the samples using the known coil geometry and average current loads for operation at 500°C from furnace calibrations. The current is turned off during sample cooling and the field strength in the vessel at room temperature was measured to have a similar order of magnitude as the Earth's field with an upward inclination.

To distinguish remagnetization effects due to partial reheating in an applied field from possible effects of deformation and applied stress, acquisition of partial TRM (pTRM) was measured on undeformed samples from both IPM and UPM samples. First, pTRMs were given at 500°C in a 0.1 mT field at high angles (90° or 45°) to an original total TRM given in the same field. This was intended to determine the amount of remagnetization that can be attributed to simple reheating to the experimental temperatures. In IPM samples, the original total TRM was given in an upward direction, perpendicular to the sample plane. To test the effect of initial fabrics on pTRM acquisition, total TRMs were given to two UPM samples in directions parallel and perpendicular to Kmin. Additionally, one sample was given a total TRM and then reheated in the pressure vessel for 30 minutes at experimental conditions with no applied stress in order to observe pTRM acquisition resulting from the field of the furnace heating coils.

After deformation, samples were recovered by grinding away the metal jacket material. Samples were progressively demagnetized in an alternating field (AF) demagnetizer up to 200 mT to obtain demagnetization spectra and directional remanence components. AMS and anisotropy of anhysteretic remanent magnetization (AARM) measurements were used to characterize magnetic deformation fabrics, and anisotropy of isothermal remanent magnetization (AIRM) was also measured on deformed IPM samples. Hysteresis and low-temperature remanence measurements were made again after deformation to monitor changes in magnetic composition and coercivity. First-order reversal curve (FORC) measurements were also made on selected samples to observe changes in coercivity distributions and interaction field [Pike et al., 1999; Roberts et al., 2000]. FORC data were processed with the FORCinel software program [Harrison and Feinberg, 2008].

AMS measurements were performed in a KLY-2 Kappabridge susceptibility meter with a 15-position measurement scheme. Remanence anisotropy was determined by imparting an ARM in a maximum AF of 100 mT with a 0.1 mT bias field and a 9-position measurement scheme measured on a 2-G Enterprises superconducting rock magnetometer. Anisotropy of partial ARM (ApARM) was measured on IPM samples using a maximum AF of 100 mT with a bias field applied in AF windows of 100–10 mT (“hard” fabric) or 10–0 mT (“soft” fabric). AIRM was measured with a 36-position measurement scheme on a Princeton Measurements Corporation vibrating sample magnetometer (VSM) using a saturating field of 1.0 T. Magnetic fabric ellipsoid shapes are described in terms of magnetic lineations (L = Kmax/Kint), magnetic foliations (F = Kint/Kmin) and the shape parameter T of Jelinek [1981], where Kmax, Kint, and Kmin are the maximum, intermediate and minimum susceptibilities. The degree of anisotropy was determined by the parameter P (P = Kmax/Kmin) of Nagata [1961]. Hysteresis measurements were made on a VSM with a maximum field of 1.0 T. Low-temperature remanence measurements included field-cooled and zero-field-cooled warming curves of a 2.5 T saturation isothermal remanent magnetization (SIRM) that was imparted at 20 K along with room temperature SIRM cooling and warming curves from 20 to 300 K in a Magnetic Properties Measurement System (MPMS) by Quantum Designs. In Figures 37, 10, and 12, sample coordinates are chosen such that the shear plane contains the north and east directions, where east is updip and north is along strike, and the shear sense when looking at a cross section of the shear plane is top to the west, or downdip.

4. Results

4.1. Initial Fabrics

IPM samples were found to lack a distinct initial magnetic fabric, as indicated by P values of 2%–4% anisotropy from AMS measurements (Table 3). Principal directions of AMS ellipsoids among IPM samples lack a consistent orientation (Figure 4a). Additionally, AMS measurements for several samples were characterized by large angular uncertainties and therefore have no meaningful principal directions. Samples from uniaxially prepared material (UPM) were characterized by well-defined magnetic foliation fabrics (Figure 4b) perpendicular to the compression axis during synthesis, with P values ranging between 9% and 12% and comparatively small angular uncertainties about the Kmin axis (Table 3). UPM samples were cut and oriented so that the initial foliation was perpendicular to the shear plane in order to maximize the observable effects of fabric overprinting. AARM principal directions have small but measurable differences from AMS principal directions in both sets of samples.

Figure 4.

Stereographic plots of predeformation and postdeformation magnetic fabrics. Principal directions for AMS and AARM are shown in blue and red, respectively. Predeformation fabric for (a) IPM and (c) UPM samples. Open symbols represent directions that are not statistically significant. (b and d) Deformation fabrics for corresponding samples which are labeled with highest and lowest strains. All plots are lower hemisphere equal-area projections; sense of shear for Figures 4b and 4d is top to the west.

Table 3. Predeformation Magnetic Fabric Intensity, Shape, and Orientation for Synthesized Samples
SampleLFTP0Dec. KmaxInc. KmaxDec. KminInc. Kmine12ae23e31χ × 10−6 m3/kg
  • a

    The eXX values are angular uncertainties for orientations of ellipsoid principal axes.

AMS
Iso07-041.0181.0190.0271.03721622586649.6848.6530.055.80
Iso07-051.0161.004−0.5611.040244441371663.8982.2357.935.60
Iso07-101.0071.003−0.4021.01126267152810.6523.907.525.38
Iso07-111.0051.0150.5241.02076331872914.034.513.435.30
Iso07-121.0111.007−0.1821.01831177157118.9912.995.365.58
An07-011.0071.0950.8651.1023187687819.441.491.395.35
An07-051.0121.0910.7611.103357187410.031.451.275.31
An07-071.0131.0990.7581.113185727019.911.461.275.33
An07-081.0121.0780.7251.091173226399.541.691.435.31
An07-091.0121.1070.7971.119186496112.031.431.285.23
 
AARM
Iso07-041.0191.0350.3061.055282121794570.5757.1245.02-
Iso07-051.0611.017−0.5451.0791517582141.0371.9634.14-
Iso07-101.0481.019−0.4131.0681715981039.5564.0530.50-
Iso07-111.0321.010−0.5271.04272321782459.0979.7152.03-
Iso07-121.0531.030−0.2731.0842682131938.0855.0026.84-
An07-011.0761.1540.3241.24115361259948.8132.9822.48-
An07-051.0731.1670.3721.2523546087235.1419.8013.40-
An07-071.0471.2440.6531.3033534786355.1818.9615.50-
An07-081.0371.2890.7501.337345272066.5520.8118.07-
An07-091.0011.2030.9841.20534047932087.9413.7413.63-

4.2. Acquisition of pTRM

For both IPM and UPM samples, pTRM acquisition experiments (Figures 5a and 5b) resulted in acquisition of soft components that are well aligned with the direction of the cooling field. The newly acquired component is generally demagnetized in AFs around 12 mT. The direction of the sample fabric with respect to pTRM acquisition had a small, but observable, effect (Figures 5a and 5b). The sample that was given a total TRM in the plane of the magnetic foliation (TRM⊥Kmin) acquired a slightly larger pTRM in the direction perpendicular to the foliation (pTRM∥Kmin) than the sample that was magnetized with a total TRM out of the foliation plane, with an in-plane pTRM. The reheating experiment performed in the pressure vessel without deformation of the sample (Figure 5c) was designed to measure in situ pTRM acquisition resulting from the field in the furnace. AF demagnetization revealed multiple overlapping components of remanence, including a stable component oriented parallel to the original total TRM. The soft components acquired during this reheating experiment are removed by demagnetizing fields of 25 mT.

Figure 5.

Orthogonal vector component [Zijderveld, 1967] diagrams with results of pTRM acquisition and demagnetization experiments for (a) UPM test samples, (b) IPM test samples, and (c) remagnetization of a test sample by reheating in the deformation vessel. Open symbols are projections onto the vertical plane, solid symbols are projections onto the horizontal plane, and initial total TRM vectors are shown in red. Points are labeled with peak AF field steps in mT.

4.3. Demagnetization of Postdeformation NRM

Orthogonal vector component plots [Zijderveld, 1967] of AF demagnetization results for all samples after deformation are displayed in Figure 6. Two distinct patterns of remagnetization are observed. IPM samples (Figure 6a) typically have multiple remanence components, including a nearly vertical, high-coercivity component (removed in fields >20 mT) oriented parallel or subparallel to the predeformation TRM. The low-coercivity components (removed in fields <20 mT) in IPM samples do not have a systematic orientation, and they have variable inclinations, with directions that are not uniformly consistent with a vector rotation relative to the sense of shearing. One IPM sample, Iso07–04, was not magnetized prior to deformation and contained a single, low-coercivity component with shallow inclination. The SuperIAPD software was used to perform principal component analysis on the demagnetization data (Figure 7a); these results are listed in Table 4. Although we have only identified soft and hard components of remanence for IPM samples, there is curvature from overlapping coercivity spectra that makes it difficult to rule out the existence of additional components. The proportion of remanence represented by weak field pTRMs in test samples is small compared to the proportion of remanence represented by the remagnetized portion in deformed IPM samples. Similarly, the coercivities of remagnetized components observed in IPM samples are larger on average than the coercivity of the remagnetized components in weak field pTRM tests, but are nearly identical to that of the pTRM acquired during reheating in the pressure vessel.

Figure 6.

Orthogonal vector component diagrams of remanence behavior during AF demagnetization for (a) deformed IPM and (b) deformed UPM samples. Symbols are the same as in Figure 5. Orientations of predeformation TRM vectors are shown as red squares and have been normalized to the postdeformation intensity values; therefore, relative predeformation remanence intensity is not reflected. Please refer to Figure 8c for relative remanence intensity data.

Figure 7.

(a) Upper hemisphere equal-area stereographic projection of orientations of postdeformation “NRM” components determined by principal component analysis (see Table 4). The red diamond indicates the orientation of the ambient field during deformation, which is coaxial with the maximum compressive stress direction. (b) Comparison of vertical rotation of AMS minimum axes in deformed UPM samples with vertical deflection of remanence directions.

Table 4. Postdeformation Remanence Component Orientations Determined by Principal Component Analysis
SampleStrainSoftHard
Dec.Inc.Dec.Inc.
Iso07-04145−14  
Iso07-051105−63252−88
Iso07-101.538−246−72
Iso07-110.374−54277−80
Iso07-120.6311−5536−78
SampleStrainMain Component 
Dec.Inc.
An07-010.356−79  
An07-041238−46  
An07-050.6248−77  
An07-071.3156−57  
An07-081.5162−38  
An07-091178−80  

Demagnetization of several UPM samples revealed a single remanence component that is not aligned with the original TRM, and that is distinctly different than the pattern seen in IPM samples. However, samples An07-09 and An07-08 (Figure 6b) have more complex “NRM” patterns. Additionally, one UPM deformation experiment (sample An07-04) failed to provide a reliable strain measurement and was considered unsuitable for magnetic fabric analysis; however, demagnetization of this sample also revealed a single remanence component so it is included in Figure 6 for comparison. Nearly all postdeformation “NRM” intensities are higher than the predeformation TRM (Figure 8c).

Figure 8.

(a–d) Changes in hysteresis and remanence properties after deformation relative to predeformation values. Solid (open) symbols represent IPM (UPM) samples. (e) Changes in bulk susceptibility after deformation. (f) Normalized IRM acquisition curves (thin lines) and gradients of acquisition curves (heavy lines) from FORC measurements.

4.4. Changes in Rock Magnetic Properties

No correlation is observed between median destructive field (MDF) of “NRM” and strain in either set of deformed samples (Figures 2d and 8d) although it is generally lower than before deformation. MDF values of around 30 mT for both the HIP material and starting magnetite powders are similar to MDFs measured by AF demagnetization of ARM in W3006 magnetite by Yu et al. [2002] and Özdemir and Banerjee [1982]. The remanence intensity (JNRM) of our samples typically increased after deformation with respect to the predeformation TRM intensity (J0) but did not vary consistently with strain (Figure 8c). Although Borradaile [1994] found that remanence intensity was reduced at small strains in triaxial deformation, our observed increase likely originates from the presence of a stronger field during experiments compared to that applied during TRM acquisition. Likewise, variation in the JNRM/J0 ratio among deformed samples may result from differences in vertical position of the sample within the furnace between experiments or from small field variations due to differences in the amount of current applied to the furnace heating coils to maintain experimental temperatures.

Mr/Ms ratios vary slightly among deformed samples and are elevated relative to the predeformation ratio. Table 1 indicates a drop in Ms for deformed samples, however, considering the small volume of our samples we note that these values are susceptible to mass normalization errors due to small amounts of mass loss during repeated sample handling. The consistency of diagnostic transition temperatures in Table 1 suggests that lowered postdeformation Ms may not reflect a real change in magnetite stoichiometry. Nonetheless, if limited magnetite oxidation does occur during deformation, its effect on magnetic anisotropy development will be minimal and it is unlikely to have a significant impact on remanence stability. Bulk Hc increased significantly after deformation for all experiments. Published flow laws for magnetite [Atkinson, 1977] indicate that magnetite would not accumulate an appreciable amount of plastic strain at the experimental conditions, but the rise in coercivity indicates that the deformation process has increased the internal stress state in the magnetite grains [Heider et al., 1987], as was determined for room temperature deformation by Jackson et al. [1993]. Similarly, gradients of IRM acquisition curves from FORC measurements show a single coercivity component that shifts from low to moderate coercivities after HIP and deformation (Figure 8f). FORC diagrams (Figure 9) have closed contour structures with no difference in interaction fields between the starting HIP material and deformed samples, and are comparable to FORCs measured on similarly sized synthetic PSD Wright magnetites by Muxworthy and Dunlop [2002]. The small sample size necessitated the use of large smoothing factors during FORC processing. Although some degree of magnetostatic interaction is indicated by the FORC diagrams, it is not possible to determine how much of the interaction field is attributable to domain interactions within single grains or to interactions between grains, or how much the distribution is smeared by the high smoothing factor (SF) values [cf. Roberts et al., 2000].

Figure 9.

Comparison of FORC measurements for (a) starting material, (b and d) HIP material, and (c and e) deformed samples. Smoothing factor (SF) = 10 for Figure 9c, and SF = 8 for Figures 9a, 9b, 9d, and 9e.

Bulk susceptibility, κ, was calculated as the mean of AMS measurements and decreased in all samples after deformation by 10%35% (Figure 8e). This observed drop in mean susceptibility is somewhat larger than the 10% decrease in κ observed in previous room temperature studies [Jackson et al., 1993; Borradaile, 1996] for samples exposed to comparable stresses (∼100 MPa). Jackson et al. [1993] identified two mechanisms by which a drop in bulk susceptibility may result after deformation. An increase in Hc resulting from higher defect densities that inhibit domain wall motion can lower the intrinsic susceptibility, or an increase in the demagnetizing factor, N, can result in lower observed κ. Although we see no evidence for changes in magnetic grain shape anisotropy after the experiments, N has been found to depend on the domain structure of grains [Merrill, 1977] and domain wall displacement from minimum energy positions can produce large changes in N for small MD grains [Dunlop, 1983]. Therefore, domain wall pinning by increased defect concentrations could potentially raise N and lower κ by enough to account for the observed changes in our samples. The larger magnitude of susceptibility decrease in our study compared to earlier studies may be explained by the smaller magnetic grain size used in our experiments (1 μm compared with 40 μm) because Dunlop [1983] calculated that the largest potential for increase in N by domain wall pinning occurs at the small end of the MD grain size spectrum.

4.5. Changes in Magnetic Fabric Character

Magnetic fabrics of deformed samples are largely characterized by a flattened anisotropy ellipsoid, or a magnetic foliation. Table 5 contains a summary of magnetic fabric measurements for all deformed samples. The degree of anisotropy in IPM samples increases in a progressive linear pattern with increasing shear strain for both AMS and AARM measurements (Figure 10a). The shape parameter T (Figure 11a) indicates that the anisotropy ellipsoid becomes oblate at low strains and maintains this shape with increasing degree of anisotropy at higher strains. Although the degree of anisotropy for remanence fabrics measured by both AIRM and AARM is slightly higher than that measured by AMS, both remanence fabrics exhibit the same trend with changing strain intensity and also generally show the same changes in fabric shape displayed by AMS measurements. ApARM for IPM samples reveals distinct remanence fabrics carried by high- and low-coercivity grain subpopulations (Figure 12). Kmin inclinations for the lower-coercivity fractions are generally shallower than those for total AARM measurements but are clustered around the same orientations. The higher-coercivity ApARM axes are more scattered, possibly reflecting a different rotational response to deformation by different grain size fractions.

Figure 10.

Changes in (a) degree of anisotropy and (b) orientation of poles to magnetic foliation. Data points are averages over several measurements, and error bars were calculated from standard deviations. Solid (open) symbols represent data from IPM (UPM) samples and circles (squares) represent AMS (AARM) data. The dashed line in Figure 10b represents the rotation of the minimum stretching axis of the theoretical strain ellipsoid.

Figure 11.

Changes in magnetic fabric shape parameters. (a) Shape parameter T versus degree of anisotropy P. (b) Modified Flinn plot of shape change in magnetic fabric ellipsoid. Solid (open) symbols represent data from IPM (UPM) samples. Points are labeled in increments of shear strain.

Figure 12.

Lower hemisphere stereographic projection of maximum and minimum directions of anisotropy of partial ARM imparted to high-coercivity (Hc >10 mT) and low-coercivity (Hc <10 mT) magnetic fractions compared with total ARM fabrics of deformed IPM samples.

Table 5. Magnetic Fabric Parameters, Changes in Anisotropy, and Orientations of Ellipsoid Principal Axes for Deformed Samples
SampleStrainFTP0PfΔPTDec. KmaxInc. KmaxDec. KminInc. Kmin
AMS
Iso07-110.31.0291.0061.021.0350.0150.67455627083
Iso07-120.61.0511.0211.0181.0730.0550.39636824880
Iso07-0411.0771.0131.0371.0910.0540.6791421628872
Iso07-0511.0461.021.041.0670.0260.387753125859
Iso07-101.51.1361.0321.0111.1720.1620.604841926970
An07-010.31.0511.0361.1021.093−0.0090.1219821017
An07-050.61.0291.0411.1031.071−0.032−0.16317132616
An07-0911.0371.0481.1191.086−0.033−0.124212721
An07-071.31.1141.0221.1131.1390.0260.6611801227834
An07-081.51.1881.061.0911.260.1680.49290−8255
 
AARM
Iso07-110.31.0281.0531.0551.0820.028−0.271232720260
Iso07-120.61.1681.0271.1671.1990.0320.68461013673
Iso07-0411.3271.0421.0691.3830.3140.75892426366
Iso07-0511.1591.0581.0421.2260.1850.4481062424460
Iso07-101.51.3211.2751.0841.6840.60.063902226768
An07-010.31.1991.0771.2411.2910.050.391825911
An07-050.61.1391.1241.2521.2750.0230.1091811027729
An07-0911.1011.0671.2051.175−0.03−0.1971612127142
An07-071.31.3751.0931.3031.4970.1940.572412627545
An07-081.51.9131.268-2.425-0.4646127256
 
AIRM
Iso07-110.31.0141.038-1.052-−0.451762823058
Iso07-120.61.0711.031-1.104-0.386302155475
Iso07-0511.0931.041-1.138-0.372843026460
Iso07-0411.0081.101-1.11-−0.851833226958
Iso07-101.51.1291.129-1.275-−0.002862226068

In UPM samples deformed to small shear strains, the deformation fabrics resemble the initial fabric but are slightly weaker and less oblate. At higher strains, the predeformation fabric is overprinted and the shape of the anisotropy ellipsoid changes sharply. As the initial foliation is overprinted, a magnetic lineation develops at the intersection of the shear plane and the initial foliation plane (Figure 11b), and the fabric shape returns to a foliation around a shear strain of γ = 1 as the deformation fabric becomes more pronounced. This results in an initial apparent decrease in the degree of anisotropy at small strains followed by a rapid increase after the initial fabric becomes overprinted. As in IPM samples, the remanence anisotropy in UPM samples follows the same pattern of changes in intensity and fabric shape seen with AMS.

4.6. Fabric Orientation

The magnetic foliation of deformed IPM samples is subparallel to the shear plane and dips shallowly for all strains (Figure 4c). Changes in orientation of the fabric ellipsoid are best defined by Kmin axes, or poles to magnetic foliation. The inclination of Kmin undergoes only small changes from low to high strain, and does not consistently rotate in the direction of shearing (Figure 10b). AARM fabric orientations tend to be distinct from those of the AMS ellipsoid, with the largest directional differences seen at low strains (γ <1). Neither the AMS nor AARM fabric is well aligned with the minimum stretching axis of the bulk strain ellipsoid, but rather they remain subparallel to the orientation of the strain ellipsoid. The rate of rotation for both remanence and AMS minimum axes is slower than that of the theoretical minimum stretching axis. AIRM fabric orientations were nearly identical to those of the AARM fabric in all samples. Borradaile and Alford [1987] also reported rotation of AMS principal axes that does not match the rate expected for lines rotating in homogeneous strain in their deformation experiments.

UPM samples deformed to small strains have a foliation oriented perpendicular to the shear plane, which mimics the predeformation fabric (Figure 4d). With progressive strain, Kmin axes remain in or near the shear plane up to strains of γ = 1, and rotate rapidly to become steeply inclined with the shear plane at high strains (Figure 10b). After the initial fabric is effectively overprinted, Kmin axes rotate much more rapidly than the minimum stretching axis of the strain ellipsoid over the higher-strain interval. These samples were also characterized by significant differences between AMS and AARM orientations. Rotation of AARM minimum axes is relatively smooth and progressive and is also more rapid than that of the minimum stretching axis. The net rotations of magnetic fabric ellipsoids are consistent with shear direction in UPM samples. Also, the deflection of remanence vectors toward the shear plane in UPM samples closely mirrors the rotation of UPM minimum AMS directions toward the axis of maximum compression but is only loosely correlated with the rotation of AARM minimum directions (Figure 7b).

5. Discussion

5.1. Remagnetization During Deformation

Postdeformation remanence patterns of IPM samples (Figure 6a) are in most cases essentially indistinguishable from a pTRM acquired by reheating in the deformation vessel (Figure 5c), which would result in partial overprinting of the original TRM. Neither the univectorial UPM magnetizations nor the low-coercivity IPM components are consistently oriented away from the maximum stress direction, so there is insufficient evidence to suggest that strain-induced grain rotation is the primary mechanism that determines postdeformation remanence directions. Vector rotation produced by shearing would be expected to result in remanence directions upward out of plane and in the dip direction of the shear zone since the shear sense is top moving downdip, and should also result in similar remanence intensities before and after deformation. Any remanence produced by an axial field in the deformation vessel should be updip and upward out of plane in sample coordinates, reflecting the 45° updip inclination of the furnace field to the shear plane (Figure 3). Neither of these expected direction patterns is observed consistently in either set of samples, although Borradaile and Mothersill [1989, 1991] also observed nonsystematic and unexpected rotation of NRM directions during room temperature experimental deformation. On the basis of pTRM experiments on test samples and generally higher postdeformation remanence intensities, it is likely that the remagnetization in IPM samples is a high-field pTRM; however, it is difficult to determine what factor is controlling the direction of the remagnetized components. The high-coercivity ChRM components are interpreted simply as those portions of the original total TRM that have withstood deformation.

Most UPM samples lack a stable component that could be identified as part of a predeformation TRM, which suggests that most samples have been completely remagnetized. Also, the different demagnetization patterns in IPM and UPM samples suggest that the presence of a strong predeformation fabric has somehow facilitated remagnetization of UPM samples. This is possible if a piezoremanent magnetization (PRM) mechanism contributes to remagnetization during experiments. Initial fabrics in UPM samples are oriented such that the average direction of maximum remanence susceptibility is near that of the maximum compressive stress direction. This alignment of magnetite grains results in a higher potential for stress demagnetization than the nearly random grain orientation distributions in IPM samples. It has been noted in previous studies of synstrain PRM acquisition [e.g., Borradaile, 1994, 1996] that magnetic softening from differential stresses has a stronger effect than strain-induced grain rotation, and that the direction of the PRM acquired is typically independent of the stress direction and is controlled by the direction of the ambient field. Although we do not observe systematic field control over the remanence directions, a PRM remagnetization mechanism in the presence of a field, or equivalently, a high-field pTRM facilitated by stress softening are possible explanations for different remagnetization patterns between IPM and UPM samples.

Like the IPM samples, the “NRM” directions for UPM samples do not all lie in the quadrant expected if the furnace field controlled the remanence direction, so again the main factor controlling the direction of postdeformation remanence cannot be determined. However, when the experiment geometry is considered, where the compressive stress axis is coaxial with the field during experiments, this may not be a surprising result. The net effect of the applied stress is to reorient the remanence away from the axis of compression, while the field exerts a force to reorient the remanence toward the same axis. If the two competing forces largely cancel, the resulting remanence directions may be predictably nonsystematic. Also important to note is that although we have tried to show progressive changes by using different strain increments for different samples, repeated experiments on a single sample greatly reduce the effects of heterogeneity between samples when possible.

Large uncertainties associated with our estimated experimental stresses make it impossible to discuss the extent or character of remagnetization in terms of differential stress. In addition, these results cannot be directly compared to natural stress-assisted remagnetization processes in part because fields of the same order of magnitude used in our experiments are exceedingly rare in nature. However, viscous remagnetization effects on geologic time scales could effectively induce remagnetization comparable to that observed in experiments. Although a viscous remagnetization mechanism in our samples is possible given the relative proximity of our experiment temperatures to the sample blocking temperatures, it would be difficult to isolate its effect from other remagnetization processes. If a viscous process was dominant, one would expect it to operate identically in both sets of samples and any effect should be dependent on experiment duration, which is not clearly the case for our samples. Further analysis would benefit from development of a theoretical treatment incorporating the combined effects of stress reorientation, temperature, field, and time to calculate the threshold conditions for complete remagnetization in PSD magnetite.

5.2. Fabric Character

The changes in principal directions of the magnetic fabric in IPM samples are not large enough to be reliably used as shear sense indicators. The net change in orientation of the AMS ellipsoid from small to large strain indicates a shear sense opposite to that imposed (Figure 4b). Rathore and Becke [1980] also noted conflicting senses of shear movement indicated by magnetic fabrics along different sections of a natural shear zone, suggesting that interpretation of shear sense from AMS directions is not always straightforward. Net rotation of the AARM ellipsoid in IPM samples provides a better, and correct, indication of shear sense (Figure 10b). However, the systematic and progressive increase in degree of anisotropy with shear strain means that both magnetic fabric measurement methods provide an excellent representation of relative strain in IPM samples, even though determination of principal stress directions from magnetic ellipsoid axes is less straightforward in this case.

The behavior of UPM magnetic fabrics can be easily understood in terms of overprinting an initial fabric with a deformation fabric. Previous studies have noted that when simple shear strains are superimposed on pure shear strains, the strain ellipsoid may change shape in an undulating fashion, and the two strains may partially cancel each other, returning the strain ellipsoid to a more spherical shape before becoming ellipsoidal once again [Ramberg, 1975; Means et al., 1980; Ramsay and Huber, 1983]. Similarly, Ruf et al. [1988] observed that magnetic fabrics in a mylonitic zone became weaker with increasing strain compared to the well-developed anisotropy in the granite protolith and Borradaile and Alford [1987] noted decreasing eccentricity of AMS fabric in early stages of deformation experiments for samples with initial fabrics at high angles to the shortening direction. Deformed UPM magnetic fabrics mimic this type of behavior at small strains with an apparent decrease in anisotropy as the predeformation fabric is overprinted, which is observed as a single undulation of the fabric ellipsoid shape (Figure 11a). A similar pattern of magnetic fabric shape change has been observed previously in weakly deformed mudrocks [e.g., Kligfield et al., 1981; Parès, 2004], where a deformation-induced cleavage overprints a primary bedding fabric, producing fabric shape changes from oblate to prolate and back to oblate. The direction of the net rotation of fabric axes in UPM samples is consistent with the direction of shearing. AARM ellipsoids change more progressively in orientation than AMS ellipsoids, and therefore provide a clearer indication of shear sense. However, the apparent decrease in anisotropy at low strains means that relative deformation intensity is not accurately indicated by comparing fabric intensities among UPM samples.

5.3. Potential for Complex Fabrics

The stable (nonrotated) ChRM components of postdeformation remanence in IPM samples suggests the presence of a nonrotating fraction of grains during deformation, while the progressive increase in the degree of anisotropy with strain clearly suggests that a fraction of rotating grains is becoming aligned by the deformation process. Thus, we can infer that separate magnetic subfabrics are carried by these distinct grain fractions, and that IPM samples really carry a composite magnetic fabric. We also note that because the stable ChRM is carried by the higher-coercivity component, the nonrotating fraction is likely to be composed of smaller or higher aspect ratio grains that were aligned in stable or metastable positions prior to deformation (possibly in or near the shear plane) that inhibited rotation. Figure 13 is a conceptual model in which deformation of a theoretical rock containing two distinct subsets of rotating and nonrotating grains results in a composite fabric with a magnetic foliation that is more shallowly dipping than that of the strain ellipsoid. This model also produces a smaller net rotation of principal fabric axes than the strain ellipsoid. We propose that the deformation behavior of IPM samples is similar to this model, and is responsible for the small changes in orientation and the subparallel alignment of the magnetic fabric with the strain ellipsoid. Results from ApARM measurements support this idea since the fabric carried by the “soft” fraction undergoes a larger net rotation and has generally more steeply dipping foliation planes than total ARM fabric (Figure 12). However, the pARM fabrics are not clear enough to elucidate the behavior of all magnetic grains in the sample. There is no evidence from remanence measurements to suggest that the deformation fabrics in UPM samples are composite in nature, or that different grain fractions have exhibited different deformation responses as in IPM samples. However, orientation differences between AMS and AARM principal directions in both sets of samples suggest some degree of fabric complexity.

Figure 13.

Conceptual model of theorized grain rotation behavior in IPM samples, illustrating how a composite magnetic fabric may arise out of complex grain rotation behavior during deformation. The idealized steroplots at the top represent Kmax directions of individual grains.

5.4. Origin of AMS/AARM Discrepancies

5.4.1. Inverse Fabrics

The observations of [Özdemir and Banerjee, 1982] suggest that the synthetic magnetite used in our experiments has a broad grain size distribution, and there potentially exists a small fraction of single-domain (SD) or SD-like material in our samples. The measured deformation fabrics are clearly not pure inverse fabrics as expected for SD particles, however, a fraction of fine-grained magnetic material could produce an inverse fabric component that would result in small differences between AMS and AARM principal direction orientations [Rochette, 1988; Rochette et al., 1992]. Since the misorientation between AMS and AARM directions varies among samples, this explanation would require high variability in the quantity of fine-grained material among samples. This seems unlikely if one considers that the theoretical average number of magnetite grains per sample is on the order of 108 for the magnetite grain size and concentration used in our synthesis. Nevertheless, some heterogeneity among samples is possible. Although it would be possible to choose some configuration of SD-like grains for each sample that would reproduce our fabric orientation results when combined with a normal fabric component, there is not enough evidence to show that such inverse fabric components exist in our samples. The analysis is further complicated by the fact that rotation behavior during deformation can vary among grain size fractions as described above for IPM samples.

5.4.2. Interaction Effects

Alternatively, interaction effects could contribute to discrepancies between magnetic fabric measurement methods. Due to the difficult nature of dispersing magnetic grains, it is generally not possible to produce synthetic samples with uniformly distributed particles, especially with fine grained material, such as that used here. A possible effect of magnetic particle clumping or clustering during initial mixing of sample materials is the creation of distribution anisotropy [Hargraves et al., 1991]. Since it is known that interactions among fine-grained magnetic particles affect ARM acquisition, even at concentrations as low as 0.01% [Sugiura, 1979; Egli, 2006], it is conceivable that interactions caused by distribution anisotropy can give rise to discrepancies between AMS and AARM fabric directions. Interactions tend to have opposite effects on ARM and AMS acquisition [Muxworthy and Williams, 2004; Egli, 2006], therefore, certain distribution geometries may achieve an inverse fabric similar to that carried by SD particles. Although most work on interactions in magnetite refers only to SD particles, the relatively high concentration of PSD magnetite used in our experiments could produce nonnegligible interaction effects. Further work on magnetic interactions in grain sizes larger than SD could help to assess their significance in this study, but ultimately, without knowing the geometry of grain distribution in our samples it is difficult to predict how distribution anisotropy would affect our measured magnetic fabrics.

6. Conclusions

High-coercivity remanence components present after deformation in IPM samples are interpreted as ChRMs, and demonstrate the ability of a primary thermal magnetization to survive deformation at pressure and temperature conditions approximately equivalent to upper greenschist facies metamorphism on laboratory time scales. However, this remanence stability may be highly dependent on the presence (or absence) and character of a predeformational fabric, as preexisting anisotropy could either stabilize a remanence or facilitate overprinting. The postdeformational remanence displayed by UPM samples probably represents a complete remagnetization by a combination of stress, field, and temperature and appears to be independent of grain rotation effects. PRM acquisition and stress-softening mechanisms likely contribute to complete remagnetization during our deformation experiments but were not sufficient to remove a primary magnetization in IPM samples. We also note that viscous remagnetization processes are not considered in our experiments, and remanence stability in natural systems will depend on the duration of exposure to heat and stress associated with a metamorphic event.

Progressive changes in anisotropy intensity for material with weak initial fabric suggests that magnetic fabric strength is a good indicator of relative strain for these samples, but a relatively poor kinematic indicator. Conversely, in samples with strong initial fabric, magnetic fabric strength is a poor indicator of relative strain due to superposition of deformation fabrics with a preexisting anisotropy, but sense of shearing is more clearly demonstrated than in IPM samples. Distinct differences in orientations of AMS and AARM principal axes are attributed to a combination of fine-grained material and magnetic interaction effects. The presence of a predeformation fabric in our experiments has been shown to have significant effects on both remanence stability and magnetic fabric development in deformed samples.

Acknowledgments

We thank David Kohlstedt and Mark Zimmerman for assistance in planning and performing experiments and Jim Stout for helpful discussions. Reviews by Ann Hirt, Graham Borradaile, and Andrew Roberts improved this manuscript. Parts of this work were carried out in the Institute of Technology Characterization Facility, University of Minnesota, which receives partial support from NSF through the NNIN program. This is publication 1002 of the Institute for Rock Magnetism, which is supported by grants from the Instruments and Facilities Program, Division of Earth Science, National Science Foundation.