High-temperature magnetic fabric development from plastically deformed magnetite in experimental...

Geophys. J. Int. (2012) 189, 229–239 doi: 10.1111/j.1365-246X.2011.05338.x

GJI

Geo

mag

netism

,ro

ckm

agne

tism

and

pala

eom

agne

tism

High-temperature magnetic fabric development from plasticallydeformed magnetite in experimental shear zones

J.L. Till, B.M. Moskowitz and M.J. JacksonInstitute for Rock Magnetism, Department of Earth Sciences, University of Minnesota, Minneapolis, MN, USA. E-mail: [email protected]

Accepted 2011 December 14. Received 2011 December 8; in original form 2011 September 23

S U M M A R YMagnetic fabric development has been studied in synthetic magnetite-silicate aggregates ina series of high-temperature simple shear deformation experiments. Samples composed ofmagnetite grains with a nominal size of 20–40 μm dispersed at 3 wt % in a matrix of plagioclasewere deformed between 1000 and 1200 ◦C with a confining pressure of 300 MPa, shear stressesin the range 10–130 MPa, and shear strains up to γ = 3. We determined that both magnetite andplagioclase were deformed plastically at these conditions. An oblate magnetic fabric ellipsoiddevelops subparallel to the shear plane and the degree of AMS rapidly increases with strainup to a value of 2.5. Theoretical strain response models of magnetic fabric development wereapplied to the data to obtain estimated magnetite strains. The shape-preferred orientation ofmagnetite grains after deformation, determined from reflected light image analysis, was usedto calculate independent magnetite strain estimates. These results were then compared withstrains estimated from theoretical magnetic anisotropy development. Both strain estimationmethods indicate strain partitioning between magnetite and the plagioclase matrix and theresults are used to determine approximate viscosity ratios between the two phases at theexperimental conditions.

Key words: Magnetic fabrics and anisotropy; Rock and mineral magnetism; Plasticity,diffusion, and creep; Creep and deformation.

1 I N T RO D U C T I O N

Magnetic anisotropy methods are a valuable microstructural guidein field studies of deformed rocks. Information about relative strainprovided by this technique is particularly useful in zones of highstrain or extensive metamorphism where physical strain markersare typically absent, including shear zones or zones with very sub-tle degrees of subsolidus deformation where no macroscopic fabricis visible. Experimental work to understand how magnetic min-erals respond to strain in the laboratory is essential in trying tounderstand how deformation histories may be reflected in magneticsignatures of deformed rocks in nature. If it is possible to determinelaboratory relationships between strain and magnetic properties,particularly magnetic anisotropy, our interpretation of tectonic con-ditions through the use of magnetic fabric analysis may be vastlyimproved. Because magnetite is abundant in the Earth’s crust andtends to dominate the magnetic anisotropy of rocks when present(Borradaile & Jackson 2010), better constraints on the deforma-tion behaviour and magnetic properties of deformed magnetite willaid analysis of magnetic fabrics in many high-grade metamorphicsettings.

Although models exist to predict the changes in magneticanisotropy that occur in response to plastic strain in magnetite(Owens 1974; Hrouda 1993), these models are seldom applied to

field studies of magnetic fabrics in high strain zones. More attentionhas been given to settings in which magnetite is assumed to deformas a rigid marker, with the focus of study placed on determining ac-tive versus passive rotation behaviour. In addition, as an accessorymineral magnetite exerts little control on bulk rock rheology andits general deformation behaviour in field settings is often difficultto determine. Nevertheless, deformed magnetite has been docu-mented in several crustal shear zones (Housen et al. 1995; Mamtaniet al. 2007; Barbosa & Lagoeiro 2010) and its occurrence in locallyhigh concentrations in the oceanic crust is strongly associated withstrain localization in oceanic core complexes (Agar & Lloyd 1997;Natland 2002). Because magnetite rheology under both laboratoryand natural conditions is somewhat poorly constrained it is diffi-cult to identify the thermotectonic settings in which magnetite islikely to deform in the ductile regime. Controlled laboratory exper-iments to characterize the magnetic anisotropy signal of plasticallydeformed magnetite are therefore necessary to interpret Fe-oxidemineral strain responses from magnetic fabric data in deformedmagnetite-bearing rocks.

We present magnetic fabric results from high-temperature shearexperiments on synthetic magnetite–silicate aggregates to study thechanges in magnetic anisotropy that result from deformation ofmagnetite under ductile conditions. This study takes unique ad-vantage of the existing models of theoretical magnetic anisotropy

C© 2012 The Authors 229Geophysical Journal International C© 2012 RAS

Geophysical Journal International

230 J.L. Till, B.M. Moskowitz and M.J. Jackson

development in deforming rocks and illustrates the potential forextracting semi-quantitative magnetite strain data from magneticfabrics under certain circumstances.

2 E X P E R I M E N TA L M E T H O D S

2.1 Sample synthesis

Sample materials consisted of 3 wt % synthetic Wright 41183magnetite in a matrix of natural labradorite from the Beaver Bayanorthosite with a plagioclase composition of approximately An70.The magnetite used is multidomain (MD) with a mean grain sizearound 20 μm and a mean aspect ratio of 1.6 (Yu et al. 2002).This is a typical degree of grain elongation for natural magnetitein igneous rocks (Stacey & Banerjee 1974). The plagioclase matrixmaterial was sieved to remove the coarse-grained fraction above64 μm and then washed in HCl to remove traces of metallic Feimpurities. After washing in acid, the starting plagioclase is nearlypurely diamagnetic in hysteresis measurements. Sample powderswere mixed thoroughly by hand and then magnetically character-ized by measuring hysteresis loops and low-temperature remanencecurves. Mixed sample powders were cold-pressed at 130 MPa in ahydraulic oil-medium isostatic press and fitted into pre-oxidized Nicanisters. In the final synthesis step, the sample pellets were iso-statically hot-pressed in an Ar gas-medium Paterson pressure vesselat 300 MPa and 1200 ◦C resulting in solid, dense sample material.This high-temperature annealing step is designed to promote graingrowth and does not impart any strain to the sample. Magnetic prop-erties were measured before and after hot isostatic pressing (HIP),as described later.

During the synthesis of several trial samples, a large extent of re-action occurred between the magnetite and the plagioclase matrix.This was primarily indicated by large drops in saturation magnetiza-tion (M s) and was confirmed by microscopic analysis. One partiallyreacted sample was analysed with Raman spectroscopy, which in-dicated fayalite as the primary phase in reaction rims surroundingmagnetite grains. The occurrence of this reaction was interpreted toresult from a failure to maintain the oxygen fugacity of the sampleabove the quartz–fayalite–magnetite buffer. This problem was ulti-mately addressed by oxidizing the Ni canisters prior to hot-pressingby heating in air at 1200 ◦C for several hours and by packing smallamounts of hematite (Fe2O3) powder in the ends of the capsule be-tween the sample pellet and the canister end-caps. The regions ofsample material near the canister ends were avoided when sampleswere made from the hot-pressed material.

2.2 Deformation and characterization procedures

After hot pressing, all metallic material from the Ni canister wasremoved from the sample by cutting and grinding and the sam-ple material was cut into discs of 1 mm thickness. Rock mag-netic analysis was performed on synthesized samples along withquantitative compositional analysis using an electron microprobe.Rock magnetic properties measured before and after deformation onall samples include anisotropy of magnetic susceptibility (AMS),anisotropy of anhysteretic remanence (AARM), hysteresis loops,backfield remanence curves, Curie temperature (T c) measurementsand low-temperature (T < 0 ◦C) saturation remanence curves. AMSand Tc measurements were made on a KLY-2 KappaBridge suscepti-bility metre. Hysteresis loops and backfield remanence curves weremeasured on a Princeton Measurements MicroMag vibrating sam-ple magnetometer using a saturating field of 1.0 T. Low-temperature

Table 1. Deformation conditions.

T γ τmax τ avg Average viscositySample (◦C) γ (s−1) (MPa) (MPa) (Pa s)

011-02 1100 1 8.49 × 10−05 56 47.2 5.56 × 1011

011-03 1100 2 1.13 × 10−04 68.5 61.7 5.45 × 1011

011-04 1100 3 1.56 × 10−04 44 50.5 3.24 × 1011

011-05 1000 1 8.49 × 10−05 127 87.5 1.03 × 1012

011-06 1200 1 1.13 × 10−04 19 10.1 8.9 × 1010

011-08 1200 1.72 1.41 × 10−04 23 13.4 9.46 × 1010

measurements were made on a Quantum Designs Magnetic Prop-erties Measurement System and included field-cooled and zero-field-cooled saturation isothermal remanent magnetization (SIRM)warming curves and room-temperature SIRM cooling and warmingcurves using a saturating field of 2.5 T. AARM was measured ona 2G superconducting rock magnetometer. ARMs for remanenceanisotropy were given in a maximum alternating field of 100 mTwith a biasing field of 0.1 mT. The shape parameters of magneticlineation L (L = Kmax/K int) and magnetic foliation F (F = K int/Kmin)along with the shape parameter T (T = [ln (F) − ln (L)]/[ln (F) +ln (L)]) were used to describe the shape of the magnetic fabric el-lipsoid and the degree of anisotropy P (P = Kmax/Kmin), whichdescribes the aspect ratio of the anisotropy ellipsoid, was used toindicate the intensity of magnetic fabrics.

Deformation experiment procedures followed those of Till et al.(2010). Sample discs were shaped to fit the elliptical cross-sectionof a piston formed from alumina rod of diameter 5 mm cut at a45 ◦ angle. The angled surfaces of the small alumina pistons de-fine the shear zone boundaries and the shear plane lies in the planeof the sample disc parallel to these surfaces. The alumina pistonsand sample discs were enclosed in an oxidized Ni jacket. The NiOlayer on the interior surface of this jacket, produced by heatingthe jackets in air for several hours, maintains the oxygen fugac-ity near the Ni–NiO buffer during deformation. The full sampleassembly includes larger alumina and zirconia pistons on eitherside of the small angled pistons, all of which were enclosed in athin Fe jacket. See Hustoft & Kohlstedt (2006) for a schematicdiagram of a similar sample assembly. Deformation experimentswere performed in a triaxial gas-medium pressure vessel (Paterson1990) using Ar as a pressure medium at temperatures of 1000, 1100and 1200 ◦C. Samples were deformed under confining pressure of300 MPa and at constant piston displacement rate, approximatingconstant shear strain rates around 1 × 10−4 s−1 to shear strainsin the range γ = 1–3. The applied stresses needed to maintaina constant strain rate varied according to temperature, but meanshear stresses during experiments ranged between 10 and 130 MPa.Conditions for individual deformation experiments are shown inTable 1.

3 C H A N G E S I N RO C K M A G N E T I CP RO P E RT I E S

3.1 Initial sample characteristics

Significant changes in rock magnetic properties occurred in sam-ples after hot pressing due to chemical exchange between the ma-trix plagioclase and magnetite grains resulting in large changes inmagnetite composition. Electron microprobe analysis of the synthe-sized samples showed that opaque grains are Fe-oxides containingapproximately 5 wt % Al (Fig. 1). Microprobe analysis also re-vealed that the oxide grains are surrounded by rims composed of

C© 2012 The Authors, GJI, 189, 229–239

Geophysical Journal International C© 2012 RAS

Magnetic fabric from magnetite deformation 231

Figure 1. Elemental distribution maps from electron microprobe analysis around oxide grains in hot-pressed sample material.

Al–Ca–Na–silicate which are depleted in Al and Ca but enriched inSi and Na relative to the feldspar matrix. These rims are not visi-ble in reflected light microscopy and may be plagioclase of a morealbite-rich composition (Table 2). Electron microprobe analysis onoxide grains in selected samples was repeated after deformation andshowed that further compositional changes were minor in most sam-ples, although the Al content of oxide grains deformed at 1200 ◦Cis slightly higher than after lower-T experiments.

The post-synthesis changes in magnetic properties are summa-rized in Table 3. The most prominent change is a large drop inM s after hot pressing from a starting value of 2.35 Am2 kg−1 forthe cold-pressed sample pellet to 1.09 Am2 kg−1 after hot pressing.Thermomagnetic susceptibility curves indicate a Curie tempera-ture (T c) near 500 ◦C and low-temperature saturation remanencecurves showed no Verwey transition on warming, suppression ofwhich has been previously noted as an effect of cation-substitutionin magnetite (e.g. Dunlop & Ozdemir 1997,Ch. 3.5). Bulk sus-ceptibility decreased by approximately 50 per cent after HIP. Thecation-substituted Fe-oxide lies in the magnetite-hercynite solid so-

lution with a composition around x = 0.4 according to the formulaAlxFe3−xO4 where x = 0 for pure magnetite and x = 2 for hercynite.This composition is close to the value of x = 0.35 predicted for aCurie temperature of 500 ◦C based on magnetic property measure-ments of Al-bearing magnetite by Pickart & Turnock (1959). Thedecrease in M s is consistent with that predicted for a compositionin the range x = 0.4–0.5 for Al3+ replacement of Fe3+ on B-sites(Pickart & Turnock 1959). However, this oxide phase will be looselyreferred to as magnetite throughout the paper for simplicity.

3.2 Rock magnetic properties after deformation

Changes in Curie temperature, hysteresis properties and bulk sus-ceptibility after deformation are varied among samples. Some ofthese changes reflect varying degrees of further cation-exchangereaction between the magnetite and the matrix in each experiment,which slightly obscure the effects of stress and deformation oncertain magnetic properties. Post-deformation Curie temperatures

Table 2. Quantitative composition of synthesized HIP samples from electron microprobe measure-ments in wt % oxides.

Oxide Oxide after 1000◦C Oxide after 1200◦Cafter HIP deformation deformation Silicate rim Silicate matrix

TiO2 0.109 0.061 0.112 a a

Na2O 0.044 a a 2.059 1.898Al2O3 6.193 5.785 7.138 27.29 32.90FeOb 84.41 84.82 82.87 1.224 0.434NiO 0.037 0.021 0.662 0.018 0.067CaO 0.088 0.090 0.098 12.80 16.08SiO2 0.264 0.327 0.246 55.99 48.28MgO 1.088 0.928 1.038 0.166 0.126

Total 92.23 92.46 92.03 99.55 99.78aNot analyzed.bTotal iron as FeO.

C© 2012 The Authors, GJI, 189, 229–239



Table 3. Changes in rock magnetic properties before and after synthesis and deformation. Median destructivefields (MDF) of remanence components identified from coercivity distributions (Supporting Information) arelisted as a proxy for the coercivity of remanence.

M r M s Hc Tc χ MDF1 MDF2Sample M r/M s (Am2 kg−1) (Am2 kg−1) (mT) (◦C) 10−6 (m3 kg−1) (mT) (mT)

Mixed sample powders011-00a 0.047 0.111 2.346 5.433 572 1.616Hot-pressed011-0xa 0.041 0.045 1.089 4.038 499 0.831 33 135

±0.004 ±0.026 ±0.189 ± 0.144Deformed011-02 0.094 0.148 1.577 9.00 458 0.468 23 170011-03 0.114 0.168 1.478 8.57 489 0.772 23011-04 0.086 0.076 0.880 5.58 405 0.439 17 151011-05 0.110 0.164 1.487 10.32 478 0.525 30011-06 0.084 0.058 0.692 7.58 470 0.287 15 148011-08 0.057 0.060 1.070 4.86 474 0.490 14 138aValues shown are means of predeformation values for all samples with standard deviations indicated.

decreased slightly in all samples relative to the average value near500 ◦C in HIPed material, but the magnitude of the decrease variedsignificantly between experiments (Table 3). Measurements of low-field susceptibility versus temperature up to 600 ◦C indicate smallamounts of a second phase with a Tc near 355 ◦C after deformationin some samples (Fig. 2). This corresponds to the Tc of metallicNi and results from traces of the Ni jackets that were incompletelyremoved during sample recovery. However, in thermomagnetic mea-surements of saturation magnetization, M s, the small magnitude ofthe drop in M s at 355 ◦C indicates that Ni contributes no more than5 per cent of the room temperature M s and thus the Ni contributionto remanence properties is minor.

Sample hysteresis properties after deformation are significantlyvaried and complex. M s is approximately 50 per cent higher in halfof deformed samples and is lowered in the rest. Decreases in M s

after deformation in samples 011–06 and 011–08 are thought torepresent further cation-exchange between the magnetite and the

Figure 2. Changes in the shape of normalized thermomagnetic susceptibil-ity measurements on heating indicating the typical change in Curie temper-ature between the mixed sample powder, a HIPed sample and a deformedsample.

matrix at 1200 ◦. The lower M s in sample 011-04 may reflect partialreaction of the type observed in several trial synthesis runs due tofailure to maintain oxygen fugacity. Because this experiment had alonger duration than the others, it is possible that the buffer layerwas consumed and conditions in the deformation vessel became tooreducing for magnetite stability. If Al content in magnetite in HIPedsamples represents a quasi-equilibrium concentration at 1200 ◦C,increases in M s after deformation at 1000 and 1100 ◦C could indi-cate lower miscibility of Al in magnetite and reduced Al substitutionat these temperatures.

Bulk coercivity is generally higher after deformation and thisincrease is highest in sample 011-05, which was subjected to thehighest differential stress during deformation and has a coercivityover double that of the undeformed HIP material. Samples 011-02and 011-03, deformed at 1100 ◦C, show a significant increase inHc over the HIP value, while samples deformed at 1200 ◦C showa more modest increase. Complicated patterns of change in coer-civity distributions were also seen after HIP and deformation andprobably reflect complex changes in chemistry and stress state (Sup-porting Information). Bulk susceptibility, χ 0, is lower in all samplesafter deformation in accordance with findings from previous exper-imental deformation studies involving magnetite (Borradaile 1996;Jackson et al. 1993; Till et al. 2010). Changes in χ 0 are likely afunction of changes in both M s and Hc.

4 M A G N E T I C FA B R I C S O F D E F O R M E DS A M P L E S

Initial magnetic fabric intensities for individual HIPed samples haveaverage AMS P values of 1.072 and the fabric ellipsoids are slightlyoblate in shape. Magnetic fabric parameters for all samples beforeand after deformation are listed in Table 4. AARM fabrics show asignificantly stronger degree of anisotropy with P values near 40 percent. The stronger AARM fabrics reflect the fact that remanentmagnetization is inherently more anisotropic than induced magne-tization (e.g. Borradaile & Jackson 2004, eq. 16), and our resultsare in accordance with previous experimental and theoretical com-parisons of remanence and susceptibility anisotropy (Stephensonet al. 1986; Hrouda et al. 2000). Initial fabric ellipsoid principaldirections are highly scattered (Fig. 3a), indicating that the start-ing material does not have a clear preferred direction of magnetic

C© 2012 The Authors, GJI, 189, 229–239



Table 4. Pre- and post-deformation magnetic fabric parameters and orientations. D and I refer todeclination and inclination, respectively, of principal anisotropy orientations.

Sample Strain L F P T Dmax Imax Dmin Imin

Pre-deformationAMS

‘Site mean’ 0 1.017 1.018 1.036 0.019 33.1 57.4 302.8 0.2Sample average 0 1.025 1.046 1.072 0.318 – – – –

AARM‘Site mean’ 0 1.068 1.304 1.393 0.602 26.9 31.4 289.6 11.8Sample average 0 1.066 1.313 1.399 0.470 – – – –

Post-deformationAMS

011-02 1 1.054 1.254 1.322 0.622 98.2 16.7 272.4 73.2011-03 2 1.069 1.614 1.725 0.757 87.3 11.4 276.3 78.5011-04 3 1.515 1.653 2.504 0.095 97.8 0.10 354.6 89.7011-05 1 1.047 1.197 1.254 0.591 77.8 17.9 287.8 69.5011-06 1 1.054 1.308 1.378 0.672 86.1 13.8 244.3 75.2011-08 1.72 1.075 1.515 1.629 0.703 97.3 10.7 267.9 79.2

AARM011-02 1 1.133 1.623 1.839 0.591 107.2 17.0 285.0 73.0011-03 2 1.162 2.305 2.679 0.695 85.1 13.8 277.1 75.9011-04 3 2.020 2.131 4.305 0.037 100.6 3.60 310.5 85.9011-05 1 1.187 1.415 1.679 0.340 97.7 17.4 285.6 72.5011-06 1 1.287 1.489 1.916 0.224 110.9 10.6 298.8 79.3011-08 1.72 1.142 1.462 1.669 0.481 88.1 18.6 300.6 68.2

Figure 3. Magnetic fabric results for deformed samples. (a) Lower-hemisphere equal-area stereoplots showing orientation of principal magnetic fabricdirections for hot-pressed samples before deformation (left hand side) and deformed samples (right hand side). The shear direction is top to the west, with theshear plane lying in the circumference of the stereonet. (b) Shape parameters of magnetic fabrics in deformed samples. Solid (open) symbols represent AMS(AARM) data.

C© 2012 The Authors, GJI, 189, 229–239



Figure 4. Variation in degree of AMS with bulk shear strain. The AMSresults from the deformation experiments of Till et al. (2010) are shown forcomparison.

anisotropy. Because all samples were cut from the HIPed materialin the same orientation, a ‘site’ mean AMS value can be taken fromthe averaged directions of all HIPed samples. The ‘site’ mean yieldsan anisotropy intensity of less than 4 per cent, compared with themean AMS magnitude in individual samples (calculated by simplyaveraging P values) around 7 per cent, indicating some slight het-erogeneity among samples. AMS and AARM principal directionsare in relatively good agreement in most of the starting material, butsome samples show a slight directional discrepancy (∼15–20 ◦).

Post-deformation magnetic fabric maximum directions clusternear the shear plane in the direction of shearing, and magnetic min-imum directions are subperpendicular to the shear plane (Fig. 3a).The fabric orientation does not vary with either shear strain ordeformation temperature. Magnetic fabric ellipsoids of deformedsamples are generally oblate in shape, although sample 011-04 istriaxial. Magnetic lineations, foliations and shape parameters areshown in Fig. 3(b). AMS fabric intensity increases monotonicallywith shear strain, up to P values of 2.5 (Fig. 4). Fig. 4 also containsmagnetic fabric data from deformed synthetic magnetite–carbonateaggregates in experiments by Till et al. (2010), which used the samesynthesis and deformation procedures as in this study, but were per-formed at a lower temperature (500 ◦C). AMS and AARM principaldirections and intensities are both in excellent agreement in alldeformed samples, although AARM intensities are systematicallyhigher in all samples for the reasons explained earlier.

Anisotropy development in the samples of Till et al. (2010) re-sulted from the alignment of rigidly rotating magnetite particles ina ductile matrix. The AMS intensities for deformed samples in thisstudy exceed those of Till et al. (2010) for equivalent shear strainsby a factor of approximately 3. The more rapid development ofanisotropy with strain in the current experiments strongly suggeststhat magnetite deforms plastically during deformation of the bulksamples and that the high degrees of AMS and AARM are due toincreasing magnetite grain shape anisotropy with increasing bulkshear strain. Although the magnetic particles used in experiments ofTill et al. (2010) were finer-grained (1–2 μm) and purer in compo-sition, we estimate that the intrinsic susceptibilities of the magneticphases used in these two sets of experiments are similar and there-fore the differences in anisotropy development primarily reflect thedifferent responses of magnetic particles to deformation (crystal

Figure 5. Comparison of AMS P values for samples 011-05, 011-02 and011-06, each deformed to a shear strain of γ = 1, illustrating temperature-dependence of the degree of anisotropy after deformation.

plastic vs. rigid behaviour). In addition, microstructural examina-tion of magnetite grains after deformation revealed the presence oflow-angle subgrain boundaries in several samples, confirming thatplastic deformation mechanisms are active in magnetite during theexperiments (Till et al. in preparation).

4.1 Temperature dependence of anisotropy

Fig. 5 is a comparison of AMS intensity for three samples deformedto shear strains of γ = 1 at different temperatures in which a sig-nificant dependence of magnetic fabric strength on temperature isevident. The degree of AMS in sample 011-06, deformed at 1200 ◦C,is 1.378 (38 per cent anisotropy) which is approximately 49 per centmore anisotropic than sample 011-05, deformed at 1000 ◦C to thesame bulk shear strain, with a degree of AMS of 1.254 (25 per centanisotropy). Sample 011-02, deformed at 1100 ◦C, has a degree ofAMS that is intermediate between the previous two samples. Thedegree of remanence anisotropy measured by AARM is 33 per centmore intense in sample 011-06 (P = 1.927) than in sample 011-05(P = 1.695). Differences in experimental strain rates between thesesamples are small and are unlikely to account for the difference inanisotropy strength. However, the difference in applied stresses ateach temperature to maintain similar strain rates reflects the chang-ing viscosity of the plagioclase matrix with temperature. Magnetitedeforming plastically as a passive marker, that is, as particles withno viscosity contrast with the matrix, would be expected to producethe same degree of anisotropy in samples deformed to the same bulkshear strain, independent of deformation temperature. The observedtemperature dependence of magnetic anisotropy is only possible ifmagnetite deforms plastically, with higher magnetite strain rates athigher temperatures, but at overall lower strain rates than that of theplagioclase. The magnetite in our samples therefore does not be-have as a perfect passive marker during deformation and undergoesstrain partitioning with the matrix.

4.2 Comparison to field studies

To determine whether the magnetic fabric intensities produced inour experiments are representative of anisotropy values in deformedrocks in nature, we compared our experimental fabrics with previ-ously reported AMS measurements for two high-grade shear zones

C© 2012 The Authors, GJI, 189, 229–239



Figure 6. AMS measurements from samples deformed at 1100 ◦C com-pared with AMS data for a natural anorthosite mylonite from Parry Sound,Ontario (Housen et al. 1995), and a granitic mylonite from the Santa Catalinametamorphic core complex, Arizona (Ruf et al. 1988).

for which strain estimates were made. AMS P values for samplesfrom the 1100 ◦C experiments are shown in comparison with fieldstudy data in Fig. 6. Housen et al. (1995) documented deformedmagnetite in highly strained anorthosites in the Parry Sound ShearZone (PSSZ), Ontario, and established a correlation between strainand magnetic fabric strength, noting that fabric strength progres-sively increased from protolith to mylonite to ultramylonite. Thisstudy determined that magnetite deformed plastically in the my-lonitic zones. Although the main components of the anorthosites inthe PSSZ are similar in composition to our synthetic samples, thefabric development is much more rapid in the experimentally de-formed samples. This discrepancy may be partly due to a higher vis-cosity contrast and different strain partitioning between magnetiteand silicates at the lower temperatures and geologic strain ratesof the PSSZ compared with the experimental conditions. Alterna-tively, differences in grain size and composition of the magnetite,and thus intrinsic susceptibility, may also control the different ratesof anisotropy development between the different sets of samples.

Ruf et al. (1988) also reported AMS intensities associated withestimated strains for a granitoid mylonite in the Santa CatalinaCore Complex (SCCC) mylonite zone. Fig. 6 shows that the degreeof AMS for experimentally deformed samples overlaps with thestrongest fabrics from the SCCC. However, Ruf et al. (1988) didnot conclusively determine that magnetite was plastically deformedin the mylonite and this may account for the scatter in their mea-sured fabric strengths as a function of strain and the overall loweranisotropy. On the basis of this brief comparison with the abovefield studies, the magnetic anisotropies produced by experimentaldeformation broadly appear to be reasonable values for high-strainzones in nature.

5 T H E O R E T I C A L M O D E L S O F A M SD E V E L O P M E N T

Theoretical calculations of AMS development as a function of strainare generally categorized into four types of strain response. Thepassive model as defined by Owens (1974) describes deforma-tion of magnetic grains that have no viscosity contrast with thematrix, whereby magnetic grains undergo internal strain at the

same rate as the matrix. This strain response produces the mostrapid anisotropy development with respect to strain and predictsthe strongest anisotropy intensities at high strains. Two additionalmodels describing the rotation of rigid magnetic grains that do notundergo internal strain, the line/plane model and the viscous model,are also defined in Owens (1974) (confusingly, these are sometimesreferred to as either passive rotation or active rotation models). Thetwo types of rigid rotation models predict slower anisotropy devel-opment with strain and maximum anisotropy intensities that dependon the initial degree of particle elongation. A fourth model intro-duced by Hrouda & Lanza (1989) was termed the ductile model,describing the deformation of magnetic particles undergoing inter-nal strain at a reduced rate with respect to the deforming matrix, dueto higher viscosity of the magnetic particles. The rate of anisotropydevelopment in the ductile model varies as a function of the viscos-ity ratio between the magnetic particle and the matrix, with morerapid increases in anisotropy at lower viscosity ratios.

Because the strong magnetic fabrics in our samples developquickly with respect to both the lower-temperature experiments ofTill et al. (2010), where anisotropies correlate with the rigid par-ticle viscous model, and to natural shear zones where plasticallydeformed magnetite has been documented (Fig. 6), it is unlikelythat the strain-response of magnetite in the current experiments isone of rigid rotation. Calculations by Hrouda (1993) of anisotropydevelopment with strain in rigid particle rotation models indicatethat alignment of magnetite particles with an initial aspect ratio of1.5 begins to plateau near shear strains over γ = 7, at P valuesaround 1.35 (Hrouda 1993, Fig. 6a). Because the mean aspect ratioof magnetite in the synthetic samples is close to 1.5 before defor-mation (Yu et al. 2002), it is reasonable to assume that increases inthe degree of anisotropy beyond 1.35 must be caused by a change inmagnetite grain shape, that is, plastic deformation of the magnetite.

The strain-partitioning indicated by the temperature-dependenceof AMS strength also rules out the passive model as an appropri-ate description of the strain-response of magnetite in the samples.Because magnetite appears to have been deformed at a lower strainrate than the plagioclase, a version of the ductile model should beable to adequately describe the relationship between the observedanisotropy and magnetite strain in the experiments. The absence ofpassive marker behaviour in magnetite in our experiments supportsthe interpretation by Hrouda (1993) that the calculated magneticanisotropy arising from passive marker deformation of magnetite isunrealistically high when compared to the range of P values actuallymeasured in typical metamorphic rocks. This observation may beinterpreted as suggesting that typical metamorphic conditions arenot favourable for high strain rates in magnetite relative to strain inother rock-forming minerals. If the passive marker model representsa relatively uncommon strain response, then strain partitioning be-tween oxides and silicates and therefore AMS development accord-ing to the ductile model for magnetite may be the dominant modeof magnetite deformation in high-grade metamorphic settings.

5.1 Intrinsic susceptibility

It is apparent from the calculations of AMS development in Hrouda(1993) that the modelled outcome of AMS intensity is highly sensi-tive to the value of intrinsic susceptibility, χ i, chosen for magnetite,with lower values of χ i resulting in slower anisotropy development(Fig. 7a). The implication of this is that in inverting the problem,that is, attempting to estimate a strain value from a particular AMSintensity, strains will be overestimated if the chosen value of χ i

C© 2012 The Authors, GJI, 189, 229–239



Figure 7. (a) Degree of AMS for samples deformed at 1100 ◦C (solidsymbols) compared with theoretical models of AMS development accordingto a rigid rotation and alignment model (dashed line), and a passive markermodel for grains with χ i = 6 (solid grey line) and χ i = 17 (solid blackline). (b) The passive model does not describe experimental AMS valueswhen plotted against bulk sample strain (open symbols), but can be used toestimate the effective mean strain in magnetite grains (solid symbols).

is too low. Intrinsic susceptibility is related to the measured bulksusceptibility, χ 0, by

χ0 = χi

(1 + Nχi ), (1)

where N is the demagnetizing factor (Uyeda et al. 1963). Thispresents a problem in our attempt to apply models of AMS develop-ment to our samples, since very few methods exist for measuring χ i

in a bulk specimen. The method of Hodych (1986) to determine χ i

using low-temperature hysteresis properties yielded unsatisfactoryresults because coercivity increases rather than decreases at lowtemperatures for the aluminomagnetite in our samples.

Intrinsic susceptibility in magnetic particles is a function of grainsize (Stacey & Banerjee 1974), composition (O′Reilly 1984) andprobably also depends on the stress state of the grains, because highdefect concentrations will lower domain wall mobility and willinhibit the magnetization response to an applied field. In addition,an empirical inverse relationship between Hc and χ i has been noted(Stacey & Banerjee 1974).

Hrouda (1993) reported χ i values from published estimates forMD magnetite between 6 and 21 [SI], while Jackson et al. (1998)calculated values as high as 61 for large MD grains of nearly puremagnetite. Several studies have suggested that for magnetite, χ i

should be inversely proportional to Hc in the form χ i ≈ k/Hc wherek is a constant (Stacey & Banerjee 1974; Dunlop 1984; Hodych1986). Stacey & Banerjee (1974) suggested a proportionality con-stant of 56.25 [SI; 45 (c.g.s.)] if Hc is given in mT. A coercivitynear 4 mT for the HIPed samples before deformation gives a χ i near14. Hodych (1986) argues that k may vary substantially from onerock to another and determined k values between 32.5 and 89.75[SI] for a variety of rocks containing MD magnetite, which for ourHIPed samples indicates χ i approximately in the range 8–22.5. Wechose a χ i value of 17 for the purpose of applying theoretical AMSmodels. However, in practice χ i is not necessarily constant dur-ing deformation, especially if stress-induced changes in coercivityoccur (Jackson et al. 1993) or as is also the case in this study, ifchanges in composition occur.

5.2 Application of AMS-strain models

The degree of AMS in a single grain is given by Uyeda et al. (1963)

P = 1 + χi Nb

1 + χi Na, (2)

where Na and Nb are the demagnetizing factors along the maximumand minimum grain dimensions, respectively, which give rise to thestrong grain shape-dependence of P for high-χ i minerals. Equationsfor N values along the principal directions of ellipsoids of any shapeare given by Osborn (1945). The magnetic susceptibility anisotropyof an assemblage of homogeneously strained grains is characterizedby the mean magnetic anisotropy of the assemblage which in turn isequivalent to the AMS of a single grain with the same dimensionsas the mean grain shape of the assemblage. Eq. (2) was used tocalculate the change in degree of AMS for a single grain withan intrinsic susceptibility of 17 and increasing shape aspect ratioto represent deformation according to the passive model (Fig. 7).P values were calculated assuming an initially spherical grain sothat the theoretical strain is defined by the grain shape. BecauseAMS fabrics in the deformed samples are predominantly oblate, Nvalues for oblate spheroidal grains were used.

The AMS strength for particles with intrinsic susceptibilities of6 and 17 [SI] corresponding to the passive model are plotted asa function of strain ratio, S, in Fig. 7. S is defined as the ratio ofmaximum to minimum principal strains and is related to shear strainby the equation

S =√√√√0.5(γ 2 + 2 + γ

√γ 2 + 4)

0.5(γ 2 + 2 − γ√

γ 2 + 4)(3)

(Ramsay & Huber 1983, eq. 2.7). It is evident from the temperature-dependence of magnetic fabric strength that magnetite grains donot behave as passive (low viscosity) markers during the deforma-tion experiments, and indeed when AMS values from deformedsamples are plotted as a function of bulk sample strain in Fig. 7,it is clear that the anisotropy development in the experiments isslower than predicted by the passive model, although more rapidthan a rigid rotation model. The ductile model presented by Hrouda(1993) calculates the rate of AMS development for an assemblageof magnetic particles deforming in a matrix with various viscositycontrasts. However, a simpler and more direct approach to deter-mining approximate magnetite strains is to use the passive modelto predict the effective magnetite strain that should produce thesame degree of AMS as that measured in the deformed samples.The magnetite grain strains that correspond to the measured degreeof anisotropy according to the passive model for grains with an

C© 2012 The Authors, GJI, 189, 229–239



Figure 8. (a) Strain estimate results from SPO analysis and AMS-strain models as a function of deformation temperature (right hand side) and strain (lefthand side). (b) Representative reflected light images of samples deformed at 1100 ◦C showing the development of shape-preferred orientation of magnetite asa function of shear strain. The bright phase in all images is magnetite and the grey phase is the plagioclase matrix. Dark grey portions of the images are pits orscratches in the sample surface.

intrinsic susceptibility of 17 are shown in Fig. 8. The AMS val-ues of samples deformed at 1100 ◦C correspond to effective mag-netite strains equal to approximately 30 per cent of the bulk samplestrains.

Because the grain shape of the starting magnetite is slightlyelongated, part of the total AMS intensity probably results fromalignment and rotation of magnetite particles during deformation,in addition to the contribution from plastic strain and changinggrain shape anisotropy (Gay 1968). The AMS strength resultingfrom alignment of rigid particles with an aspect ratio of 1.5 andan initially random distribution of orientations according to theline/plane model (Hrouda 1993) is shown by the dashed line inFig. 7(a). This type of strain response produces weaker anisotropiesthan the measured AMS in our samples and it alone does not ac-count for the observed anisotropy. To approximate the effect of rigidrotation on strain estimates made with AMS models, the theoret-ical degree of anisotropy from the line/plane model for the bulkshear strains in each experiment was subtracted from the measuredAMS P values of deformed samples before determining effectivemagnetite strains from the passive model. The strains indicated bythese rotation-corrected anisotropies are shown for comparison inFig. 8(a). Although the rate of grain rotation depends on particleaspect ratios which change during deformation, we make the sim-plifying assumption that most of the grain alignment occurs earlyin the deformation experiment and only consider the rotation andalignment contribution for grains with a constant aspect ratio of1.5.

6 I M A G E A NA LY S I S

Reflected light image analysis was used to obtain direct estimatesof magnetite strain by determining the change in mean magnetite

grain shape and the shape-preferred orientation (SPO). These resultswere then compared with the magnetite strains estimated from AMSmodels. Reflected light images were obtained from sample surfacesparallel to the shear direction and perpendicular to the shear planeto observe the magnetite in the shear zone cross-section, becausethe AMS fabrics indicate that this surface contains the directionof maximum grain elongation. Representative photomicrographs atdifferent strains are shown in Fig. 8(b). To compensate for the smallsample size and low concentration of magnetite grains, the cross-sectional surfaces were subjected to several iterations of grinding,polishing and image acquisition to provide an adequate number ofmagnetite grains for image analysis (at least 200 grains per sample).

The program SPO (Launeau & Robin 1996) was used to calculatea mean grain orientation and a mean grain shape (aspect ratio) whichwas taken as an estimate of the magnetite strain ratio, S. Because themagnetite grains are well dispersed in the samples and it is there-fore easy to isolate individual grain shapes in reflected light images,we used the inertia tensor calculation in the SPO program to de-termine the average magnetite grain shape and orientation for alldeformed samples and one undeformed sample. The inertia tensormethod has the advantage of weighting individual grains by area.For this reason, Launeau & Cruden (1998) found that the grainshape ellipsoid of the inertia tensor for magnetite grains gave bet-ter agreement with AMS data than ellipsoids determined by theintercept method.

As with the degree of anisotropy, SPO intensity increasesmonotonically as a function of bulk shear strain (Fig. 8a). Thetemperature-dependence of AMS intensity is also mirrored in thepattern of SPO versus deformation temperature for samples de-formed to identical strains (Fig. 8a). As with the AMS measure-ments, the SPO of undeformed HIP material has a weak but mea-surable initial fabric. In all samples, the SPO results give magnetite

C© 2012 The Authors, GJI, 189, 229–239



grain shape anisotropies that are significantly less than the strainratios of the bulk samples. This is consistent with the interpretationof strain partitioning determined above from analysis of AMS data.

7 D I S C U S S I O N

Magnetite strain estimates from SPO data are lower than those madeusing AMS-strain models assuming plastic deformation without ro-tation for most highly strained samples, but are comparable to theAMS estimates for the samples with γ = 1. Attempting to correctfor particle rotation in the AMS models results in magnetite strainestimates that are much lower than the other estimate methods forγ = 1, but slightly higher than the SPO estimates for higher strains.Because the development of a magnetite SPO will be affected byparticle rotation during strain in the same way as the magnetic fab-ric, it qualitatively appears that the rotation correction performedhere is not valid at low strains (γ = 1). More focused experimentalwork on the behaviour of magnetite during deformation involvingstrain partitioning with silicates is needed to understand the rota-tional contribution to shape-preferred orientation of highly viscousparticles.

In comparing the SPO results with the AMS data, a limitation ofthe image analysis is that the grain population sampled is restrictedto a single plane and samples a limited number of grains, while AMSintegrates the anisotropy signal from every grain in the sampleand provides information in three dimensions. Another cause ofdiscrepancy between the strain estimate methods may be that theestimate of 17 for χ i is too low and that χ i varies somewhat fromone sample to another or changes during deformation.

Rheological descriptions of strain partitioning (Handy 1990;Tullis et al. 1991; Vigneresse & Tikoff 1999) in two-phase materialstypically refer to the end-member behaviours of constant stress andconstant strain-rate (Voigt and Reuss bounds, respectively). Giventhe small volume fraction of magnetite in the samples, magnetitecannot support a greater load than the matrix. Therefore the as-sumption of constant stress between the two phases is reasonableand strain rate partitioning is the only quantity determining the vis-cosity contrast between the two phases. Using the assumption ofconstant stress, we also assume the measured sample viscosity iseffectively the viscosity of the matrix plagioclase. Estimated mag-netite strain ratios were converted into shear strains to calculateapproximate viscosity ratios with the plagioclase. Assuming ourvarious strain estimates are reasonable bounds for the true mag-netite strain and the magnetite deformed at an approximately con-stant rate, we estimate that magnetite viscosity is between 3.3 and7.7 times that of plagioclase at 1000 ◦C and is between 2.5 and 4.4times the viscosity of plagioclase at 1200 ◦C for the stresses usedin the experiments. Table 5 contains the results of viscosity ratioestimates made by each method.

Table 5. Magnetite/plagioclase viscosity ratios from magnetite strain esti-mates.

T Rotation-corrected Inertia tensorSample (◦C) γ AMS model AMS model SPO

Estimated magnetite/plagioclase viscosity ratios011-02 1100 1 2.67 4.44 3.13011-03 1100 2 2.55 3.51 4.76011-04 1100 3 2.27 2.84 4.41011-05 1000 1 3.33 7.69 3.57011-06 1200 1 2.27 3.45 2.67011-08 1200 1.72 2.50 3.66 4.42

In field studies, determining bulk rock strain is often of moreinterest than strain in Fe-oxides or other low-abundance phases.The limited strain analysis performed here for magnetite based onmagnetic fabric measurements may find application in field studiesfor rocks with certain compositions or special conditions. For ex-ample, if independent bulk strain markers are already present in arock, strain estimates for magnetite made using AMS could be usedto calculate viscosity contrasts that may provide constraints on thethermobarometric conditions present during metamorphism if therheological behaviour of the bulk rock is well known. Conversely, ifstrain partitioning behaviour between magnetite and common rock-forming silicates is well known over a range of geologic tempera-tures and stress conditions, magnetite strain estimates may be able togive some indication of bulk rock strain. However, direct estimatesof the magnetic anisotropy carried by magnetite will be complicatedif large quantities of paramagnetic minerals are present. In addition,the predictive accuracy achieved with theoretical models of AMSis limited by the ability to determine intrinsic susceptibility for aparticular magnetic component.

8 C O N C LU S I O N S

High-temperature shear deformation of magnetite–silicate aggre-gates produces strong magnetic anisotropy that rapidly intensifieswith progressive strain. Models of AMS development during strainshow that neither a passive marker model nor a rigid rotation modeldescribe the experimental fabric data. Rather, a model involvingstrain partitioning between magnetite and the plagioclase matrixbetter predicts the observed AMS. While magnetite grain rotationand alignment probably contributes to the total anisotropy, it is diffi-cult to evaluate the relative importance of this component. Changesin the strength of magnetite shape-preferred orientation obtainedwith image analysis qualitatively confirms that magnetite is de-forming plastically and that changes in magnetite shape anisotropyare responsible for the development of a strong magnetic fabric.Magnetite viscosities are estimated to be higher than that of pla-gioclase by at least a factor of two and as much as seven at thetemperatures and stresses of our experiments. Direct experimentsto study the rheological behaviour of magnetite and its strain parti-tioning with silicates will greatly aid the interpretation of magneticfabrics formed under high-grade conditions in both laboratory andnatural environments.

A C K N OW L E D G M E N T S

David Kohlstedt and Mark Zimmerman are thanked for assistancein planning and performing deformation experiments. Ellery Frahmis thanked for assistance with electron microprobe analysis. Wethank Frantisek Hrouda and an anonymous reviewer for helpfulcomments that improved this manuscript. This work was supportedby a National Science Foundation EAR grant. This is publication1107 of the Institute for Rock Magnetism, which is supported bygrants from the Instruments and Facilities Program, Division ofEarth Science, National Science Foundation.

R E F E R E N C E S

Agar, S. & Lloyd, G., 1997. Deformation of Fe-Ti oxides in gabbroic shearzones from the Mark area, in Proc. ODP, Sci. Results, 153, 123–141.

Barbosa, P.F. & Lagoeiro, L., 2010. Crystallographic texture of themagnetite-hematite transformation: evidence for topotactic relationships

C© 2012 The Authors, GJI, 189, 229–239



in natural samples from Quadrilatero Ferrifero, Brazil, Am. Mineral.,95(1), 118–125.

Borradaile, G. & Jackson, M., 2010. Structural geology, petrofabricsand magnetic fabrics (AMS, AARM, AIRM), J. Struct. Geol., 32(10),1519–1551.

Borradaile, G.J., 1996. Experimental stress remagnetization of magnetite,Tectonophysics, 261(4), 229–248.

Borradaile, G.J. & Jackson, M., 2004. Anisotropy of magnetic susceptibility(AMS): magnetic petrofabrics of deformed rocks, Geol. Soc. Lond. Spec.Publ., 238, 299–360, doi:10.1144/GSL.SP.2004.238.01.18.

Dunlop, D., 1984. A method of determining demagnetizing factor frommultidomain hysteresis, J. geophys. Res., 89(B1), 553–558.

Dunlop, D.J. & Ozdemir, O., 1997. Rock Magnetism: Fundamentals andFrontiers, Cambridge University Press, Cambridge.

Gay, N., 1968. Pure shear and simple shear deformation of inhomogeneousviscous fluids. 1. Theory, Tectonophysics, 5(3), 211–234.

Handy, M., 1990. The solid-state flow of polymineralic rocks, J. geophys.Res., 95(B6), 8647–8661.

Hodych, J., 1986. Determination of self-demagnetizing factor N for multido-main magnetite grains in rock, Phys. Earth planet. Inter., 41(4), 283–291.

Housen, B.A., Pluijm, B.A.V.D. & Essene, E.J., 1995. Plastic behavior ofmagnetite and high strains obtained from magnetic fabrics in the ParrySound shear zone, Ontario Grenville Province, J. Struct. Geol., 17(2),265–278.

Hrouda, F., 1993. Theoretical models of magnetic anisotropy to strain rela-tionship revisited, Phys. Earth planet. Inter., 77(3-4), 237–249.

Hrouda, F. & Lanza, R., 1989. Magnetic fabric in the Biella and Traversellastocks (Periadriatic Line): implications for the mode of emplacement,Phys. Earth planet. Inter., 56(3-4), 337–348.

Hrouda, F., Henry, B. & Borradaile, G., 2000. Limitations of tensor sub-traction in isolating diamagnetic fabrics by magnetic anisotropy, Tectono-physics, 322(3-4), 303–310.

Hustoft, J.W. & Kohlstedt, D.L., 2006. Metal-silicate segregationin deforming dunitic rocks, Geochem. Geophys. Geosyst., 7(2),doi:10.1029/2005GC00148.

Jackson, M., Borradaile, G.J., Hudleston, P. & Banerjee, S., 1993. Experi-mental deformation of synthetic magnetite-bearing calcite sandstones: ef-fects on remanence, bulk magnetic properties, and magnetic anisotropy.,J. geophys. Res., 98(B1), 383–401.

Jackson, M., Moskowitz, B., Rosenbaum, J. & Kissel, C., 1998. Field-dependence of AC susceptibility in titanomagnetites, Earth planet. Sci.Lett., 157, 129–139.

Launeau, P. & Cruden, A., 1998. Magmatic fabric acquisition mechanismsin a syenite: results of a combined anisotropy of magnetic susceptibilityand image analysis study, J. geophys. Res., 103(B3), 5067–5089.

Launeau, P. & Robin, P., 1996. Fabric analysis using the intercept method,Tectonophysics, 267(1-4), 91–119.

Mamtani, M., Mukherji, A. & Chaudhuri, A., 2007. Microstructuresin a banded iron formation (Gua mine, India), Geol. Mag., 144(2),271–287, doi:10.1017/S0016756806002998.

Natland, J., 2002. Magnetic susceptibility as an index of the lithology andcomposition of gabbros, ODP Leg 176, Hole 735B, Southwest IndianRidge, Proc. ODP, Sci. Results, 176, 1–69.

O′Reilly, W., 1984. Rock and Mineral Magnetism, Blackie, Glasgow.Osborn, J.A., 1945. Demagnetizing factors of the general ellipsoid, Phys.

Rev., 67(11-12), 351–357.

Owens, W., 1974. Mathematical model studies on factors affecting the mag-netic anisotropy of deformed rocks, Tectonophysics, 24(1-2), 115–131.

Paterson, M.S., 1990. Rock deformation experimentation, in The Brittle-Ductile Transition in Rocks, Geophys. Monograph Vol. 56, pp. 187–194,eds Duba, A.G., Durham, W.B., Handin, J.W. & Wang, H.F., AmericanGeophysical Union, Washington, DC.

Pickart, S. & Turnock, A., 1959. Magnetic properties of solid solution ofFe3O4 and Fe2AlO4, Phys. Chem. Solids, 10, 242–244.

Ramsay, J.G. & Huber, M.I., 1983. The Techniques of Modern StructuralGeology Vol. 1: Strain Analysis., Academic Press, London.

Ruf, A.S., Naruk, S.J., Butler, R.F. & Calderone, G.J., 1988. Strain and mag-netic fabric in the Santa Catalina and Pinaleno Mountains metamorphiccore complex mylonite zones, Arizona, Tectonics, 7(2), 235–248.

Stacey, F. & Banerjee, S., 1974. The Physical Principles of Rock Magnetism,Elsevier, Amsterdam.

Stephenson, A., Sadikun, S. & Potter, D., 1986. A theoretical and exper-imental comparison of the anisotropies of magnetic susceptibility andremanence in rocks and minerals, Geophys. J. Roy. astr. Soc., 84(1),185–200.

Till, J., Moskowitz, B. & Jackson, M., 2010. Remanence stabil-ity and magnetic fabric development in synthetic shear zones de-formed at 500 ◦C, Geochem. Geophys. Geosyst., 11(12), Q12Z21,doi:10.1029/2010GC003320.

Tullis, T., Horowitz, F. & Tullis, J., 1991. Flow laws of polyphase aggregatesfrom end-member flow laws, J. geophys. Res., 96(B5), 8081–8096.

Uyeda, S., Fuller, M.D., Belshe, J.C. & Girdler, R.W., 1963. Anisotropy ofmagnetic susceptibility of rocks and minerals, J. geophys. Res., 68(1),279–291.

Vigneresse, J. & Tikoff, B., 1999. Strain partitioning during partial meltingand crystallizing felsic magmas, Tectonophysics, 312(2-4), 117–132.

Yu, Y., Dunlop, D.J. & Ozdemir, O., 2002. Partial anhysteretic remanentmagnetization in magnetite 1. Additivity, J. geophys. Res., 107(B10),2244, doi:10.1029/2001JB001249.

S U P P O RT I N G I N F O R M AT I O N

Additional Supporting Information may be found in the online ver-sion of this article:

Figure S1. Results of coercivity unmixing analysis on curves ofisothermal remanence magnetization acquisition showing the shapeof coercivity distributions in starting material, HIPed material, anddeformed samples. The gradient of the acquisition curves is shownby the grey lines, and the unmixed components are shown by theblack lines. Large shifts in coercivities of each remanence compo-nent are evident as a result of Al substitution and smaller increasesafter deformation are proportional to the differential stress appliedduring deformation. Median destructive fields (MDF) of the rema-nence components for each sample are listed in Table 3.

Please note: Wiley-Blackwell are not responsible for the content orfunctionality of any supporting materials supplied by the authors.Any queries (other than missing material) should be directed to thecorresponding author for the article.

C© 2012 The Authors, GJI, 189, 229–239


High-temperature magnetic fabric development from plastically deformed magnetite in experimental...

Documents

Transcript of High-temperature magnetic fabric development from plastically deformed magnetite in experimental...