American Mineralogist, Volume 60, pages 29-34, I97S
MagneticProperties
of Separated
MineralPhasesI n
Unoxidized
and OxidizedlcelandicBasalts
Ronpnr B. Hlncnevps
Department of Geological and Geophysical Sciences,
Princeton Uniuersity, Princeton, New lersey 08540
eNn J,qrunsM. Aos-Harr
Department ol Geology, Dalhousie (l niuersity,
Halifax, Noua Scotia, Cqnada
Abstract
Following microscopic study, concentratesof Fe-Ti oxide (predominantly titaniferous magnetite), ferromagnesian silicates (pyroxene plus subsidiary olivine) and plagioclase, were
separated from crushed samples of (1) unoxidized and (2) oxidized Icelandic basalt. Saturation and thermal (Snivr and TnIvt) remanent magnetism plus alternating field (Ar) demagnetization, magnetic moment (J) us magnetic field (H), and saturation magnetization (J") os
temperature (7) were measured for these concentrates and for chips of each original rock.
The results indicate that although minor amounts of ferrimagnetic oxides of distinctive composition are present in the silicates, the experimental (and presumably natural) Rrra in both
rocks is predorninantly associatedwith the primary oxide grains.
fntroduction
may be associatedwith small magnetiteparticles
silicate,
It is common experience in paleomagnetic studies producedby oxidation of a ferromagnesian
that the stability of the natural remanent magnetism especiallythe fayalite moleculein olivine. Because
(Nnrvr) of basalts with respect to alternating field stable remanencehas been found to be associated
(An) demagnetization, increases with increasing with small particlesin the silicatesin coarser-gtained
high-temperature oxidation (Watkins and Haggerty, basicrocks (EvansandMcElhinny,1969;Hargraves
1965; Zijderveld, 1967; Strangway, Larson, and and Young, 1969), it was consideredworthwhile
to compare the magneticproperties of individual
Goldstein, 1968; Ade-Hall, 1969). The increasein
and
stability with oxidation has been demonstrated mineral phase (Fe-Ti oxides,ferromagnesians,
experimentally, and has been interpreted as result- feldspars) presentin samplesof both unoxidizedand
ing from the sub-solidus exsolution of ilmenite highly oxidized basalt, the commonestmaterial in
studies.
lamellae following oxidation in primary ulvdspinel- usein palaeomagnetic
rich titanomagnetite. Presumably an original homoTechnique
geneous multidomain grain is partitioned into an
aggregateof discrete single-domain or pseudo-singleThe samplesstudied consistedof two 2.5 cm
domain magnetite particles (Strangway et al, 1968).
diameter cores, each 2.5 cm long, of Icelandic
This explanation is plausible for samples in inter- basalt: one, grey-black,unoxidized (U16-1a) and
mediate oxidation state, but it presumably cannot one reddish,and apparentlyhighly oxidized (Ulzbe applied to highly oxidized samples in which the
2a). Both samplesare from a sequenceof Pliocene
primary spinel oxides appear to be completely
flows in the Noxduradalur area of Eastern Iceland.
oxidized to an assemblageof pseudobrookite, hema- Locations with respect to flow boundaries are untite, and rutile. Whereas the hematite in such an known. Magnetic and mineralogicalstudiesindicate
aggregato may contribute stability to the remanent that this sequenceof flows has never been buried
magnetism, the intensity should be drastically reby more than 5@ m of younger material. These
duced. For this reason, Watkins and Haggerty particular samples were selected to represent the
(1965) suggestedthat the strong stable remanence extremesof deutericoxidationstatesfound in,basaltic
29
R. B. HARGRAVES. AND T. M. ADE-HALL
30
Tr.sls 1. Magnetic Properties of SeparatedMinerals and of Whole Rock Chips
Rock Chip
U ]6-1a
(unoxidixed)
rluudr
wcatsrr
Oxide
Components
PlagiocJ-ase
Pyroxene
+ Olivine
Rock Chip
tJ I2-2a
(oxidized )
Oxide
Conponents
Pynoxene
Plagioclase
+ Olivine
L
61
30
0.030
0.015
r.9
0.70
I
64
28
6.8
0.020
0.0025
1 1
(x fo-";'t'r
20
tq
0.3
20
uunfe rornl-^..
Heating
Cooling
5850
5800
5850
5800
5800
5800
6250
2800
. 1 .2 5 x
_LU
. AX
t, +)X
1.05x
2.43x
40
r20
l-'10
300
r70
r70
Per"cents
Satur"ation
Js:
Magnetization*
Tnnnoree
in
l aco
,I^f
MDFTT
SRM
TRM
50
170
Remanent Masnetic Momentt+i
4.9
J q p M ( xr o - 2 )
Jr*r(x I0
-)
Jsnu
,/ Jrnrq
J**n (x 10 3 )
JrR,l /
J,'rRo,r"f
4.3
I0.t
9
250
330
85
0.99
0,22
6. r+
39
1.4
0.64
3.6
21.8
a2
3.4
17.8
f.l
3.0
?q
1.I
n
200
290
5800
3s0
32s
6 3 0 -6 8 0 0
5800
3.6x
300
?EO
r.,65
0.30
79
r o
o 11
3 1 .8
8.7
250
27,L
*of saturabTe
(enu/gn)
ferriregnetic
component
**of
(y) enu/gm oe-(see
3a' b, c)
Figs.
parawgnetic
apparent
component
*** (Tc) on heating
and cooJ-ing in vacuum of 2 x l-0-)
totr
tafter
thermT
cgcle
in vacuum
needed to reduce
in oersteds
bg 502 the saturation
and
lluedian
field
demgnetizing
tield
or aTternating
reMnent
nagnetism
the thernai
(=JSnil
to 650oC and cooTing
in ajr
and bg heating
lilin
fiel.d
in a 9 kiTogauss
enu/gn. nlduced isothennaLLv
(=Jrnn)
in a L.O4 g field
*+The large
the
sample is at -least latgelg
for
the unoxidized
unit7
JfnU/JnnU
deviation
fron
of the tatio
theilnaT cgclinglaboratorg
resuLt
of changes induced
in the sample during
Frc. 1. Photomicrograph of highly oxidized specimen, Ul2-2a. (a) Deuteric oxidation class 4 grain with small rectilinear
islands of magnetite(medium gray) separatedby broad off-white lamellaeconsisting of a fine aggregateof rutile and titanohematite.
(b) Partly oxidized olivine grain containing and partly surrounded by thin (white) veins of hematite. Scale: 1| inches : 50 s.
MAGNETIC
PROPERTIES OF MINERALS IN BASALTS
3l
Ftc' 2. Photomicrograph of unoxidized specimen, UlGla. (a) I-arge primary ulvrispinel-rich titanomagnetite (light grain in
upper right) and many examplesof the small-sizedform of magnetite (light specks).Note the strong partitioning of the small-sized
magnetiteinto the darkest silicate phase,which is plagioclasefeldspar. (b) The small-sizedmagnetite (white specks)is again largely
restrictedto the plagioclase(dark gray), but also occursas inclusionswithin and along the boundariesbetweenthe pyroxenegrains
(medir.rmgray). Scalesameas Figure l.
Iavas, and results pertaining to them might be expected to apply equally to intermediate oxidation
states. Following preparation and preliminary examination of polished thin-sections, the balance of
the cores were crushed to -200 mesh, washed in
water and acetone, and the -200 + 325 mesh
(174p, >44p) portion recovered. The .,oxide',
fraction was concentrated by separating with Clerici
solution. A specific gravity (Sc) of 4.0 was used
for the unoxidized sample, but, to recover enough
material to work with, was lowered to 3.g for the
oxidized sample. Since the silicates all have densities <3.4, these concentratesconsist predominantly
of the oxide fraction even though they do contain
substantial silicate contaminants. The remaining
powder was split using bromoform (Sc 2.85), the
feldspar being concentrated in the light fraction and
ferromagnesian minerals in the heavv fraction.
These concentrates were further purifie.d using a
Frantz Isodynamic separator (forward tilt 20., side
tilt 15"), with current brackets of 0.5 to 0.6 amps,
and 0.4 to 0.5 amps for the ,,pyroxene,' fraction in
unoxidized and oxidized samples respectively; the
"feldspar" fraction used was ,'nonmagnetic,, at 1.6
amps in both cases.
Small aliquots of these mineral concentrates were
retained for high field magnetic property experiments, the remainder being cast in KBr pellets, using
the technique of Evans and McElhinny (1969) for
measurement of remanent magnetic (Rvr) properties.
The Ru measurements were made on a Princeton Applied Research SM-l spinner magnetometer
utilizing standard thermal and An' demagnetization
equipment and procedures (McElhinny, 1963).
High field temperature-dependent properties were
measured using a vertical-type magnetic balance
(25" to 700oC, up to 9000 gauss) and a horizontal
translationbalance ( - 190'C to 25"C,3500 gauss).
Results
(I ) Microscopic Stu:dies
In transmitted light, both basalts are fine grained,
with microphenocrysts of plagioclase and olivine
set in a groundmass of plagioclase laths, pyroxene,
and opaque mineral. The unoxidized basalt also
contains some interstitial, greenish, serpentinous
material, which may be an alteration product of
original glass. The olivine crystals in the oxidized
basalt are altered, and a reddish brown stain is
present within and around them, and also around
the opaque grains. The results of model analyses
are given in Table l.
The reflected light study confirmed the original
identification in terms of oxidation state of the
specimens and added several significant details:
The high deuteric oxidation state of Ul2-2a was
readily apparent with all original oxide grains now
altered to classes 4, 5, or 6 on the Ade-Hall et al
(1968) scale. These grains frequently contain variable but usually small amounts of magnetite sur-
)L
R. B. HARGRAYES, AND ], M, ADE-HALL
viving as small islandsbetweenthe broad lamellae
of rutile and titanonow consistingof fine aggregates
hematite (Fig. 1a). Another quite distinct oxide
occurrencein this specimenis the presenceof fine
veins of probable hematite identified by its high
reflectivity, within and along the borders of oxidized
olivine grains (Fig. 1b). As noted in transmitted
light, these altered olivines are frequently red
stainedor are surroundedby red stainedsilicates,
very
indicatingthe presenceof widely disseminated
fine hematiteparticles.No secondarymagnetitewas
definitely identified within the altered olivine grains.
The complete absenceof visible signs of deuteric
oxidation in U16-1a was confirmed.The most interestingfeature for this specimenwas the occurrenceof magnetitein two distinctforms. The larger,
primary accessoryoxide consists of anhedral to
ophitic titanomagnetitesin the 1Op to 100p size
assorange.A finer grainedform, characteristically
plagioclase,
ciated with
consists of material from
0.3p at leastdown to the limits of visibility at about
0.1p (Fig. 2). The nature of such fine, apparently
opaque grains cannot be determined wholly satisfactorily at the maximum magnificationavailableof
1350. However, the apparentcolor and relatively
low reflectivity are consistentwith magnetiterather
than hematiteor sulfides.Again the typically equidimensionalform of the grainssupportstheir identification as magnetite rather than ilmenite, which
frequently crystallizesas elongate grains. It was
consideredthat these tiny grains might consist of
fine grains of polishing agentsleft on the surfaceof
the section.However, they could not be removed
by carefulcleaningand are not distributedrandomly
but are strongly associatedwith the plagioclase
feldspar grains; these indicate that they are not
extraneous.This fine magnetiteprobablyformed by
exsolutionfrom the plagioclase.There is evidence
from the Curie point measurementsthat the two
forms of magnetite are of very different composimagnetitebeing
tions,with the plagioclase-associated
nearly pure magnetite and the larger accessory
oxidesbeingulviispinelrich (-65Vo Usp).
(2) Magnetic Studies
E 80
5
E
o
4.O
4
H, Kilogouss
cl
PYROXENE
E
Eoe
4
b
H, Kilogouss
PLAGIOCLASE
E
E04
E
o246
C
H,Kitogouss
Frc. 3. Specific magnetization (J) us field (H) curves for
(a) oxide, (b) pyroxene, and (c) plagioclaseconcentratesfrom
oxidized basalt (open symbols) and unoxidized basalt (closed
symbols).
The magnetic property measurementsare summarized in Table 1, and the J/H and J"/Z runs making allowance for the amount of oxide in the
shown in Figures 3 and 4; the significantpoints are whole rock, indicatesthat the Tnvr and presumably
the Nnu is principally associatedwith the largediscussed
below:
oxide grains.
(a) Comparison of the oxide and whole rock sizedprimary accessory
(b) The saturationmagnetizationof the oxide
TnM's for both oxidized and unoxidizedsamples,
MAGNETIC
PROPERTIES OF MINERALS
fraction in the oxidized sampleis almost twice that
for the unoxidizedsample.The Curie point of 580"
for the oxide in the oxidized sample indicates the
presenceof pure magnetite,while that in the un-
t
l,
\
b
q
o
IN BASALTS
33
oxidizedsamplehas a value of 150'C which indicatesthe presenceof ulvtispinel-richtitanomagnetite.
Thesefactsemphasizethe importantcontributionto
the magnetic properties of the oxidized sample
(Ul2-2a) by fine-grainednear-pure magnetite remaining in the primary oxide fraction even of highly
oxidized samples.
(c) As indicated by the Curie point measurements, the magnetic phase in the plagioclaseand
pyroxene fractions of the unoxidized basalt is pure
magnetite, but the large, primary oxide phase is
-65 percent Usp. fn contrast, whereas the Curie
point of the oxide fraction in the oxidized basalt
indicates pure magnetite, a hematite- or titanohematite-richphase is indicated for the plagioclase
and pyroxene fraction (in keeping with the red
staining and the decompositionproducts of olivine
actuallyobserved).Hematiteis presentin the oxide
fraction, but the surviving magnetiteis sufficientto
dominate its high-field magnetic properties; oxida-
2.5
t
PYROXENE
{
PLAGIOCLASE
\
\A
\
\
\
t.5
q
b
J/ Jo
\
\
1
\
\
t
J/Jo
\
\
I
1
t.o
t.o
h
\
o.8
\ -i r L
\
\
o.5
\
\
o.4
I
I
o.8
t
o.6
\..
\'
bJ
o.2
b
o.2
200
400
600
TEMPERATUR
E
,
"C
c
400
TEMPERATURE,'C
Frc. 4. Magnetization (J) us temperaturecurvesfor (a) oxide, (b) pyroxene,and (c) plagioclaseconcentrates.Magnetizing field :
3000gauss; initial vacuum : 2 X l0- 5 torr. Curves normalized to value at room temperature.Circles and triangles pertain to unoxidized and oxidized samples,respectively;solid symbols for heating and open symbols for cooling cycles.
34
R. B. HARGRAVES, AND I. M. ADE.HALL
tion of any magnetite originally present in the silicateshas been more complete.
(d) The vacuum employed in the J"/Z runs
(-2
X 10-5 torr) preserved the pure magnetite
of the silicates in the unoxidized specimen, and the
oxide fraction of the oxidized specimen, but otherwise was unable to prevent significant changes in
the ferromagnetic phases. In the pyroxene and
plagioclase fractions of the oxidized sample Curie
points were reduced to about 580oC, and the room
temperature saturation moments increased two to
three times. (Note: a J./T run on this pyroxene in air
showed no such increase.) These changes are consistent with the reduction of hematite to magnetite.
In contrast the Curie point and saturation magnetization of the oxide phase of the unoxidized basalt both
increase as the result of thermal cycling. Presumably
the vacuum employed was inadequate to prevent
some oxidation of the ulvospinel component, or the
original titaniferous magnetite was somewhat cation
deficient (titanomaghemite), and heating caused its
inversion to a less titanium-rich spinel, plus ilmenite.
(e) The Curie point of the unoxidized plagioclase indicates magnetite, in keeping with the identi
fication of the fine grains observed microscopically.
The Curie point of the oxidized plagioclase suggests
hematite; yet, other than the reddish staining, no
discrete opaque grains could be resolved. It is a
surprise therefore that the original saturation moment of the oxidized plagioclase should be more
than five times stronger than that of the unoxidized
feldspar. This suggeststhat the oxidized plagioclase
must contain a considerable amount of submicroscopic superparamagnetic hematite of which the
reddish brown pigmentation is one manifestation.
The white unoxidized feldspar, on the other hand,
contains far fewer, but relatively larger grains of
exsolved magnetite.
(f) Oxidation respectively increases and improves the overall specific remanence and microcoercivity of all major mineral phases in basalt.
Conclusions
We conclude that, for these two specimens, the
natural remanence is associatedwith original Fe-Ti
oxide grains, and not, in the case of the highly
oxidized specimen, with the breakdown products of
ferromagnesian silicates. Minor amounts of oxide
phases are associated with the silicates and have
interesting compositions, but contribute little to
natural remanences. Since Tnrvr and saturation
moments per unit oxide content are similar in the
two specimens, the great reduction in spinel oxide
content in the highly oxidized specimen must be
largely compensated by the reduction in effective
grain size and increase in saturation magnetization
of the surviving magnetite that result when titanium
is expelled from the structure of the original titanomagnetite.
References
(1969)
Opaquepetrologyandthe stability
M.
J.
ADE-HALL,
of natural remanent magnetism in basaltic rocks'
Geophys.J. R. Astrotr.Soc.lE, 93-107.
eNp R. L. Wrrsox
M. A., KneN, P. D. DAGLEY,
(196S) A detailedopaquepetrologicaland magnetic
investigationof a single Tertiary lava from Skye, Scotland, I. Geophys.I. R. Astron.Soc. 16, 375.
Ev.cNs.M. E.. euo M. W. McErnrNNv (1969) An investigationof the origin of stableremanencein magnetite
bearingigneousrocks. '/. Geomag.Geo-EIec'21, 757Hencn.rvrs,R. B., eNo W. M. YouNc (1969) Sourceof
stable remanent magnetismin Lambertville diabase'
.
Am. I . Sci.267, 116l-1177
McEr-nrNNv,M. W. (1963) Theory and operatingprocedures:rock magnetisminstrumentsat PrincetonUniversity. Princeton Unia. Geol. Eng. Rep. No. 63'1,24 ppSrneNcwlv. D. W., E. E. LensoN, lNo M. GolosrprN
(1968) A possible cause of high magnetic stability in
volcanic rocks. .I. Ge:oph. Res. 73, 3787-3796'
WATKTNs,N. D., eNo S. E. HeccEnrY (1965) Some magnetic properties and the petrogenic significance of
oxidized zones in an Icelandic olivine basalt. Nalure
(London), 206, 797-800.
Ztrorrvero, J. D. A. (1967) A. C. demagnetizationof
rocks: analyses of results. ln, Methods in Paleomag'
netism, Ed. D. W. Collinson, K. M. Creer, and S. K.
Runcorn, Elsevier, New York.
Manuscript receiued,Septemberfi, 1973; accepted
lor publication, Iune 18, 1974.