Duncan, R. A., Backman, J., Peterson, L. C , et al., 1990Proceedings of the Ocean Drilling Program, Scientific Results, Vol. 115

10. MAGNETIC PROPERTIES, IRON-TITANIUM OXIDES,AND PETROLOGY OF FOUR LEG 115 BASALTS1

R. B. Hargraves2

ABSTRACT

A detailed study of the Fe-Ti oxides in four basalt samples—one from each of the four holes drilled into basementon Ocean Drilling Program Leg 115 (Sites 706, 707, 713, and 715)—has been performed. Ilmenite is present only insamples from Sites 706 and 715. In the sample from Site 715, Ti-magnetite intergrowths are characteristic of subaerial(?) high-temperature oxy-exsolution; Ti-magnetite in the other three samples has experienced pervasive low-temperatureoxidation to Ti-maghemite, as evidenced by the double-humped, irreversible, saturation magnetization vs. temperature(Js/T) curves.

The bulk susceptibility of these samples, which are similar in terms of major element chemistry, varies by a factor of— 20 and correlates semiquantitatively with the modal abundance of Fe-Ti spinel, as determined by image analysis withan electron microprobe. The variation in Fe-Ti oxide abundance correlates with average grain size: fine-grained samplescontain less Fe-Ti oxide. This prompts the speculation that the crystallization rate may also influence Fe-Ti oxide abun-dance.

INTRODUCTION

Considerable data on the magnetic properties of oceanic ba-salts has resulted from the study of dredged samples and base-ment cores acquired in various ocean drilling programs. Con-currently, exhaustive studies have been made of the petrochem-istry and petrogenesis of the materials. Iron-titanium oxides,the predominant carriers of the magnetic signature, are a ubiq-uitous phase in these basalts. Their composition, skeletal tex-tures, and susceptibility to low-temperature oxidation are nowwell documented and provide a sound basis for interpreting theamplitude, as a function of age, of the total-field magneticanomalies recorded in ocean surveys (Petersen et al., 1979; Bleiland Petersen, 1983).

During Ocean Drilling Program (ODP) Leg 115, on-boardmeasurements were made of the low-field magnetic susceptibil-ity of the basalt cores collected for paleomagnetic study (Van-damme and Courtillot, this volume). Variations in this property(see Fig. 1) proved to be a valuable empirical aid to petrologistsin logging the discontinuous core segments recovered.

Susceptibility varied by a factor of ~20, but this propertycould not readily be correlated, either (as might be expected)with the apparent abundance and composition of the"opaques" (given the limited facilities aboard), or with the sub-tle variations determined in the bulk-rock chemical composi-tions.

An attempt to understand better the petrologic significanceof the large susceptibility variations, which are assumed to berelated in some way to the abundance of iron-titanium oxides,has been undertaken here.

SAMPLES STUDIEDI have made a detailed investigation of the Fe-Ti oxides in

four basalt samples, one from each of the sites drilled into base-ment: Sample 115-706C-4R-3, 10-13 cm, Sample 115-707C-25R-

* 400 -

tn

A

D

»

i*3

706C707C71 3A71 5A

i

*

AA

• A

f

i

t•

•

AA A

\

A

A

A

A •

i i i i

1 Duncan, R. A., Backman, J., Peterson, L. C , et al., 1990. Proc. ODP, Sci.Results, 115: College Station, TX (Ocean Drilling Program).

2 Department of Geological and Geophysical Sciences, Princeton University,Princeton, NJ 08544, U.S.A.

0 10 20 30 40 50 60 70 80

DEPTH BELOW BASEMENT SURFACE (m)

Figure 1. Magnetic susceptibility as a function of depth in the basementcore of all minicores drilled for paleomagnetic study (Vandamme andCourtillot, this volume). Symbols for different holes as indicated on fig-

2, 130-132 cm, Sample 115-713A-15R-4, 43-47 cm, and Sample115-715A-25R-6, 34-37 cm (referred to in this paper by their sitenumbers only; i.e., 706, 707, 713, or 715). The samples were se-lected after a reflected light study of all the polished sectionsmade on board; they represent the full range of susceptibilitiesmeasured and microtextures observed. The data acquired, orpertinent to this study, are summarized in Tables 1-4, and thetextures are illustrated in Plate 1. The analytical equipmentand/or techniques used are described in the text.

Petrographical DescriptionThe basalts are typical, fine- to medium-grained (Plate 1),

clinopyroxene, plagioclase phyric, with 707 and 715 also con-taining 3%-5% olivine (now completely altered). The 707 sam-ple is most conspicuously microporphyritic. In all, the ground-mass, which shows feathery quench textures in reflected lightand presumably was never true glass, is now largely altered tobrownish opaque clay material (most of the white backgroundsin Plate 1A).

103

R. B. HARGRAVES

Table 1. Petrographic properties of basalt samples studied.

Core, section,interval (cm)

Texture and average grain size (/an)

115-706C-4R-3, 10-13

115-707C-25R-2, 130-132

115-713A-15R-4, 43-47

115-715A-25R-6, 34-37

Plagioclase

Laths,<400

Laths,<200

Laths,<400

Stubbylaths,<200-400

Pyroxene

Subophitic,interstitial,<100

Laths,<100

Anhedral,interstitial,<300

Ophitic,enclosingplagioclase<300

Ilmenite

Parallelgrowthlaths,<200

—

—

Laths,<100

Spinel

Cross arm,skeletal,<200

Cruciform,skeletal,10-80

Cross arm,skeletal,100-200

Subhedral,30-50

Modal abundance (%)

Oxidea Magnetiteb

4.2

2.0

5.2

1.0

2.0

2.0

5.2

0.5

Note: Ilmenite is not present in samples from Sites 707 and 713.a Measured; see text for technique.b Estimated; ilmenite vs. magnetite abundance could not be resolved quantitatively, and the magnetite contents of sam-

ples from Sites 706 and 715 are based on a visual estimate.

Table 2. Magnetic properties of basalts from samplesstudied.

Core, section,interval (cm) JN T°C

115-706C-4R-3, 10-13 1.5 3.1 1.3 2 260115-707C-25R-2, 130-132 2.0 2.7 0.8 1.6 340115-713A-15R-4, 43-47 3.5 11.7 0.9 1.6 305115-715A-25R-6, 34-37 0.25 1.1 0.3 0.5 570

Note: x = susceptibility x 10 ~ 4 emu/cm3Oe; J N = intensity ofnatural remanence x 10~ 4 emu/cm3; J s = saturation mag-netization, emu/gm; J i / Jo = r a t i o °f saturation magnetiza-tion after heating to the original; and T°C = Curie point.

Bulk ChemistryThe major and trace element chemistry (determined by X-ray

fluorescence analysis on board) of samples adjacent, or closest,to the specific samples studied is listed in Table 3. Although allare generally similar, the largest differences are between the 706and 715 samples. The former contains over 3% TiO2, the latterabout 1%. The Fe/Mg ratio of the 706 sample, with over 15%"Fe2O3," is 2.8, whereas that for the 715 sample (11.8% "Fe203")is close to 1. The FeO values, as determined since by A. N. Bax-ter, are included to indicate the extreme oxidation of all the sam-ples. The 706 sample has 2.95% combined alkalis, 715 has1.84%, and 706 has more conspicuous trace element anomalies.

Iron-Titanium OxidesOxide textures are illustrated in Plates IB and 1C. Despite

the pervasive low-temperature oxidation to titanomaghemite (seemagnetic properties), the grains appeared optically homoge-

Table 3. Chemical analyses of the basalt samples studied.

interval (cm)

115-706C-4R-3, 10-13115-707C-25R-1, 39-43115-707C-25R-3, 44-47115-713A-15R-4, 7-13115-713A-15R-41, 43-146115-715A-25R-6, 38-42

interval (cm)

115-706C-4R-3, 10-13115-707C-25R-1, 39-43115-707C-25R-3, 44-47115-713A-15R-4, 7-13115-713A-15R-4, 143-146115-715A-25R-6, 38-42

Depth(mbsf)

66.70404.90407.80131.90132.63237.00

Rb

3.51.50.82.7

23.64.7

SiO2 TiO2

48.04 3.1749.60 1.3849.70 1.4549.36 1.3149.07 1.5247.93 0.98

Ba Nb Co

139 23.0 50.030 5.3 6.928 6.4 7.448 4.6 7.450 7.5 9.871 8.1 6.8

A12O3

13.6714.9015.1014.1313.5814.98

Sr

27298999296

105

Fe2O3a

15.6513.4913.3213.2714.7111.82

Major element

MnO

0.200.160.170.190.250.18

Trace elements

Zr

250798373

10354

Y

433437334122

Zn

13710311284

11178

MgO

5.657.107.158.585.6410.97

Cu

12324916918763

117

CaO

10.3511.9911.2411.3212.1111.64

Ni

4869678653

235

Na2C

2.612.152.232.182.291.58

Cr

338688

222135483

> K2O P2O5

0.34 0.350.06 0.100.09 0.110.15 0.080.77 0.130.26 0.11

V FeOb

394 0.65414 —436 0.64373 0.36418 0.41222 0.50

Total

100.03100.93100.56100.57100.07100.45

LOI

0.310.480.230.221.120

Notes: Analyses reported are of samples equivalent, or nearest, to the samples used in this study. Major element valuesgiven in wt<% oxide; all trace element values in parts per million (ppm). LOI = loss on ignition.

a Total iron reported as Fe2O3 is on-board X-ray fluorescence analyses.b FeO values determined by A. N. Baxter.

104

MAGNETIC PROPERTIES, IRON-TITANIUM OXIDES, AND PETROLOGY OF BASALTS

Table 4. Recalculated electron microprobe analyses of Fe-Tioxides in samples from Leg 115 basalts.

Sample

707707707706706706706706706713713713713715715

Sample

706706706706715715715715

Grainanalysis

123lalblcId21219a2blb12

Grainanalysis

91011A13911

Usp

0.600.570.580.520.550.600.650.580.680.570.600.480.420.140.08

Ilm

0.860.840.880.860.910.820.800.83

Mt

0.370.390.390.370.370.350.310.350.280.390.360.480.520.840.90

Spinelsa

Sp/mf

0.010.010.010.080.060.040.020.050.020.020.020.020.020.010.01

Ilmeniteb

Hem

0.110.110.070.10—

6.070.110.11

Gei

0.020.040.040.030.070.100.080.05

Her

0.010.020.01_—_—————__——

Py

0.010.010.010.010.020.010.010.01

Hau

0.010.010.010.010.010.010.020.020.020.020.020.020.02

<0.01<0.01

Chr

_—

0.020.01————

_———

Notes: Electron microprobe analyses were performed on the RutgersUniversity JEOL Superprobe; 15-kV accelerating voltage, 15-mAcurrent. All elements were measured on W.D.S. crystals, and countscontinued until 0.5 standard deviation (SD) was achieved or 40 s hadelapsed. Grains la, lb, lc, and Id from Site 706 samples are spotanalyses from core to rim of a single zoned grain, as illustrated inPlate IB, Fig. 1.

a Molecular fractions of Usp = ulvospinel, Mt = magnetite, Sp = spi-nel and mf = magnesioferrite, Her = hercynite, Hau = hausman-nite, and Chr = chromite.

b Molecular fractions of Ilm = ilmenite, Hem = hematite, Gei =geikielite, and Py = pyrophanite.

neous. Their compositions, as determined by electron micro-probe, are given in Table 4.

Ilmenite is present only in the 706 and 715 samples. In the706 sample, the ilmenite occurs in remarkable "parallel growth"laths (Plate IB, Fig. 1), sometimes partly overgrown by magne-tite (Plate 1C, Fig. 1). The conventional explanation for suchaggregates is that they represent individual skeletal crystalsjoined in the third dimension (MacKenzie et al., 1982), but noclear evidence of such connections was observed.

In Sample 715, the titaniferous magnetite has undergonepervasive high-temperature oxidation with oxy-exsolution of il-menite (Plate 1C, Fig. 4). Discrete laths of ilmenite also occur inthis sample. The high-temperature oxidation of titaniferous mag-netite has come to be interpreted as almost a signature of sub-aerial eruption (Kono, 1980; Kono et al., 1980). Such a historyhad independently been inferred for the Site 715 basalts on thebasis of their common reddish alteration and their intercalationwith sediment.

Of the four samples studied, skeletal textured spinels arefound only in the three submarine flows (706, 707, and 713; PlateIB and Plate 1C, Figs. 1-3). Although comparably fine grained,the spinels in subaerial sample 715 are subhedral (Plate 1C, Fig.4). The finer grained 707 sample contains the most distinctlycruciform-type spinels, those in the slightly coarser grained 706

and 713 samples being more of the cross-arm type (Haggerty,1976, p. HglO6).

Modal AbundanceQuantitative microscopic determination of the modal abun-

dance of such minor, fine-grained opaque phases is essentiallyimpossible. Petersen et al. (1979) compared abundance estimatesderived from conventional point counting with results deter-mined by image analysis. They found the former to overestimatethe abundance by a factor of 1.5-2.5.

The technique employed here involved image analysis of thevariations in gray scale of back-scattered electron images usinga JEOL superprobe at Rutgers University (15-kV acceleratingvoltage, 20-mA current). The "Vista" image analysis programprovides the ability to optimize contrast, brightness, and otherfactors, easing the separation of phases in terms of gray scaleand quantitative modal analysis.

Qualitatively, ilmenite could be distinguished from magnetitein the images, but quantitative determination of each has notyet been achieved. A rough estimate of the relative abundancewas made when both phases were present, and the total opaquecontent was subdivided accordingly in Table 1.

Magnetic PropertiesSaturation magnetization (Js) was determined by measure-

ment in fields up to 7500 Oe. The samples were saturated by~ 2000 Oe and, to minimize the paramagnetic contribution, thisfield was used when measuring (in air) the temperature depen-dence of magnetization (Js/T).

As illustrated in Figure 2, the double-humped J/T spectralsignature of low-temperature oxidation of titaniferous magne-tite to titanomaghemite, characteristic of oceanic basalts (Ozimaand Ozima, 1971), with a substantial increase in magnetizationon cooling, was recorded in the 706, 707, and 713 samples. Thesample from Site 715, in which the titaniferous magnetite hadalready experienced high-temperature (subaerial?) oxidation,showed a magnetite Curie temperature, with a slight drop inmagnetization on cooling. This could be due either to the inver-sion of original maghemite or to oxidation during the heatingcycle.

The susceptibilities (Bartington Model MS.2) characteristicof the samples, intensities of remanence (measured on board),and saturation magnetization are listed in Table 2.

Fe-Ti Oxides and Basalt CrystallizationThe precipitation of an Fe-Ti phase (spinel or rhombohedral)

during crystallization of basalt magma appears to depend uponthe ferric-ferrous iron ratio and the titanium content. The fer-ric-ferrous ratio is, in turn, a function of the ambient oxygen fu-gacity, the temperature of the magma at the time of crystalliza-tion, and the alkali content (Carmichael et al., 1974). The Fe3+

is essentially excluded from the main high-temperature silicatephases that crystallize from basalt, and a high activity of thision promotes early crystallization of magnetite (Osborn, 1957).

Other things being equal, a high TiO2 concentration will bemanifest in the early crystallization of ilmenite and in the possi-ble partition of TiO2 into clinopyroxene. In addition to ilmenite,the clinopyroxene of the high-TiO2 sample from Site 706 doesindeed show the purplish color of titanaugite.

DISCUSSIONAs would be expected, there is clearly a first-order correla-

tion between the susceptibility and modal abundance of titanif-erous magnetite or maghemite in these basalts (cf. Tables 1 and2). In detail, the magnetic susceptibility is further modulated bygrain size and texture (fine grain and skeletal texture tending to

105

R. B. HARGRAVES

1.5-

.5-

\1 1

706

s

\

\

\\

\\

V\

1 1

\

\

1 ^ - " • 1

2 -

1.5-

1-

5-

707

100 200 300 400

Temperature °C

500 600 100 200 300 400

Temperature °C

500 600

713 1.5-

1 -

.5 -

715

100 200 300 400

Temperature °C

500 600 100 200 300 400

Temperature °C

500 600

Figure 2. Normalized (to initial intensity, Jt) magnetization (J) vs. temperature (T) curves for samples studied: solid, on heating (8°C/min); broken,on cooling. Note double-humped heating curves for samples from Sites 706, 707, and 713, which is characteristic of low-temperature oxidation in oce-anic basalts (Ozima and Ozima, 1971). The extra hump on the heating curve for the sample from Site 707 is considered to be an artifact of the heatingrate employed (Housden et al., 1988).

decrease the initial susceptibility) and the extent of the low-tem-perature alteration (Stacey and Banerjee, 1974). The latter, in-ferred from the Curie temperatures, which increase with oxida-tion, also correlates with increasing age of the basalts, as hasbeen reported.

The influence of grain size and texture of the magnetite grainson the susceptibility is difficult to establish quantitatively. Cer-tainly, in the samples studied here, those with higher susceptibil-ity and modal magnetite are coarser grained in general, and themagnetite grains themselves are larger (cf. the samples fromSites 706 and 713 with those from 707 and 715 in Plate 1).

There is no clear indication in the bulk-rock chemistry (Table3) as to why the abundance of the Fe-Ti spinel phase shouldvary as it does. The high TiO2 content of the 706 sample is con-sistent with the presence of ilmenite, but this phase is also presentas discrete grains in the 715 sample, with the lowest TiO2 con-tent.

The ferric-ferrous ratios (see Table 3) are very high, but sam-ples have clearly undergone extensive low-temperature alterationand oxidation. The present value of this parameter, therefore,has no relevance to the primary abundance of the Fe-Ti oxides.

SUMMARY AND SPECULATIONThere is no obvious clue in these data to the petrologic rea-

son for the variation in Fe-Ti oxide abundance. Major elementvariations appear minor, and the significance of the trace ele-ments is unknown. Intrinsic differences in the magmatic oxygenfugacity could well be the primary cause, but evidence of thesehas been obscured by the pervasive low-temperature alteration.

An additional factor that may conceivably contribute to themodal variation is the rate of crystallization: That these mag-mas crystallized rapidly is likely, given that they were extruded.There is (or was) no glass in the samples studied, but the origi-

106

MAGNETIC PROPERTIES, IRON-TITANIUM OXIDES, AND PETROLOGY OF BASALTS

nal silicate phases in the cryptocrystalline groundmass—nowlargely altered to brownish "opaque" clay—may well have servedas a sink for components that otherwise would have crystallizedas Fe-Ti oxides. There is, however, no apparent variation in thegroundmass:crystal proportion commensurate with the varia-tion in oxide mode. Although all lavas cooled rapidly, some ofthe samples came from the interior of flows, or from thickerunits, and presumably crystallized somewhat less rapidly, as evi-denced by the variation from fine to medium grain.

The position of spinels (and ilmenites) in the paragenesis ofthese basalts is not obvious. As they are neither micropheno-crysts enclosed by silicate grains nor late-stage products of alter-ation or devitrification, they appear to have crystallized concur-rently with the bulk of the silicate phases.

In general, the grain sizes of the oxides vary directly withthat of the whole rock in which they occur. It is this grain-sizevariation that correlates best with the modal abundance varia-tion and prompts the following speculation. Rapid nucleationand growth of all crystals has occurred. Under such conditions,the major phases (i.e., silicates) may incorporate impurity ionsmore readily, and ions that, under slower, more ideal crystalliza-tion conditions would be partitioned into oxide crystals, may in-stead tend to be incorporated into silicates, thereby minimizingthe eventual oxide mode!

With the advent of a fairly easy technique for determiningthe oxide mode quantitatively (as described above), a straight-forward means of testing the hypothesis is at hand: To measurethe modal oxide abundance in samples from an unequivocallysingle, chemically homogeneous unit that shows a variation fromquenched border to medium-grained interior. If the results arepositive, this could be followed by experimental studies (e.g.,Lofgren, 1980) to constrain further the intensive variables in-volved.

Should this relationship be confirmed, the results of mag-netic susceptibility logging, both downhole and of individualsamples on board, may contribute significantly to the determi-nation of the basic stratigraphy of the oceanic crust, which is es-sential to a better understanding of its origin and evolution. Inaddition, the demonstration that the kinetics of nucleation andgrowth may affect significantly the modal abundance of (at least)accessory phases in a closed-system crystallization of magmawill be of more general petrologic interest.

ACKNOWLEDGMENTS

A. N. Baxter made the whole-rock ferrous iron determina-tion, S. Swapp performed the electron microprobe analyses ofthe oxides, and D. L. Thompson provided the photomicrographsand modal analyses. Their essential contributions are gratefullyacknowledged.

REFERENCES

Bleil, U, and Petersen, N., 1983. Variations in magnetization intensityand low-temperature titanomagnetite oxidation of ocean floor ba-salts. Nature, 301:384-388.

Carmichael, I.S.E., Turner, F. J., and Verhoogen, J., 1974. Igneous Pe-trology: New York (McGraw-Hill).

Haggerty, S. E., 1976. Opaque mineral oxides in terrestrial igneous rocks.In Rumble, D., Ill, Oxide Minerals. Mineral. Soc. Am., Short CourseNotes, 3:HglOl-Hg3OO.

Housden, J., DeSa, A., and O'Reilly, W., 1988. The magnetic balanceand its application to studying the magnetic mineralogy of igneousrocks. J. Geomagn. Geoelectr., 40:63-75.

Kono, M., 1980. Magnetic properties of DSDP Leg 55 basalts. In Jack-son, E. D., Koizumi, I., et al., Ink. Repts. DSDP, 55: Washington(U.S. Govt. Printing Office), 723-752.

Kono, M., Clague, D., and Larson, E. E., 1980. Fe-Ti oxide mineralogyof DSDP Leg 55 basalts. In Jackson, E. D., Koizumi, I., et al., Init.Repts. DSDP, 55: Washington (U.S. Govt. Printing Office), 639-652.

Lofgren, G., 1980. Experimental studies on the dynamic crystallizationof silicate melts. In Hargraves, R. B. (Ed.), Physics of MagmaticProcesses: Princeton, NJ (Princeton Univ. Press).

MacKenzie, W. S., Donaldson, C. H., and Guilford, C , 1982. Atlas ofIgneous Rocks and Their Textures: New York (Wiley).

Osborn, E. F , 1957. Role of oxygen pressure in the crystallization anddifferentiation of basaltic magma. Am. J. Sci., 259:609-647.

Ozima, M., and Ozima, M., 1971. Characteristic thermomagneticcurves in submarine basalts. J. Geophys. Res., 76:2051-2056.

Petersen, N., Eisenach, P., and Bleil, U., 1979. Low-temperature altera-tion of the magnetic minerals in ocean floor basalts. Am. Geophys.Union, Maurice Ewing Sen, 2:169-209.

Stacey, F, and Banerjee, S. K., 1974. The Physical Principles of RockMagnetism: Amsterdam (Elsevier).

Date of initial receipt: 1 March 1989Date of acceptance: 13 November 1989Ms 115B-127

107

R. B. HARGRAVES

Plate 1. Photomicrographs illustrating variation in texture and grain size of basalt samples studied. Figures numbered 1-4 are of samples from Sites706, 707, 713, and 715, respectively. Plate 1A shows transmitted-light negative images of thin sections (scale bar = 500 fim). Plates IB and 1C arephotographs of electron backscatter images taken on screen of electron probe. Scale bars in Plate IB are 200 jum. Scale bar lengths in Plate 1C areindicated on the images.

108

MAGNETIC PROPERTIES, IRON-TITANIUM OXIDES, AND PETROLOGY OF BASALTS

B

Plate 1 (continued).

109

R. B. HARGRAVES

Plate 1 (continued).

110