r-srura-

 

APPLICATION OF A HIGH PRESSURE MBROBOMB
T0 HYDROSTATK} PRESSURIZATION 0F MAGNET” E

Thesis Io: tho Doom of 81.8.

MICHIGAN STATE UNIVERSITY

JOHN KELLY ”AH“!

1977

.t."'

 

ABSTRACT

APPLICATION OF A HIGH PRESSURE MICROBOMB

TO HYDROSTATIC PRESSURIZATION OF MAGNETITE

By

John Kelly Maher

This thesis presents a study of hydrostatic pressure
effects on the remanent magnetization of magnetite. Two
types of samples were used; a single crystal of magnetite,
and a synthetic "rock" composed of a resin matrix and a
powder of synthetic magnetite.

The samples were initially saturated in a 2 kilogauss
field, then pressurized to 4.0 kilobars in an oil-filled non-
magnetic chamber. Isothermal remanent magnetization was
measured during pressurization by means of a ballistic
magnetometer. Pressure was increased and decreased in a
continuous cycle during the tests.

During the first pressurization, the sample magnetization
dropped to about 70% of the initial saturation magnetization
with a pressure of 1.5 kilobars, and continued to drOp at a
slower rate to about 55% of the initial saturation with a
pressure of 3.0 kilobars. Pressure was reduced to 2.0 kilobars
with no further change in magnetization, then increased to
4.0 kilobars with another 5% drop in magnetization to about

50% of the initial value.

The thesis also presents some alterations made in a
high pressure microbomb used by earlier researchers (Car-
michael, et. al., 1968). The paper discusses both the
development and use of the bomb and the complimentary apparatus
required for pressurization and magnetic measurements. The
latter includes a ballistic magnetometer specifically designed

by R.S. Carmichael for use with this system.

ACKNOWLEDGEMENTS

I would like to thank my advisor, Dr. Robert 8. Car-
michael, for his advise and support during this work. He
not only designed and supplied a good deal of the equipment
used, but also furnished many of the ideas for this work.

Dr. Ted Vinson of the Civil Engineering Department
supplied the hydraulic press and other items of support.
Dr. Jerry Cowen of the Physics Department loaned a digital
voltmeter and integrator for the ballistic magnetometer.

I would like to extend special thanks to my colleague
Joon Kim, who assisted in much of the experimental work,
and frequently put up with my unorthodox working hours.

The figures were produced by Yvonne Plaisance and Jake

Keeser of Chevron Oil Company.

ii

APPLICATION OF A HIGH PRESSURE MICROBOMB
TO HYDROSTATIC PRESSURIZATION OF MAGNETITE

A THESIS
SUBMITTED TO THE DEPARTMENT OF GEOLOGY
OF MICHIGAN STATE UNIVERSITY
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF
MASTER OF SCIENCE

By
JOHN KELLY MAHER
JUNE, 1976

TABLE OF CONTENTS

List of Figures

I.
II.

III.

IV.

VI.

Introduction ,
Theory
A. Domain Energies
1. Exchange energy
2. Magnetocrystalline anisotropy energy
3. Domain wall energy
4. Magnetostrictive anisotropy energy
5. Magnetostatic energy
6. Field energy
B. Grain Size Effects
Equipment and Experimental Techniques
A. Previous work
B. Pressure microbomb
C. Operation
D. Ballistic Magnetometer
Sample Specifications and Experimental Results
A. General prOperties of magnetite
B. Samples
C. Magnetic results
Interpretation of Results
Conclusion
Appendix A - Bomb Calibration Data

Bibliography

iv

14
14
14
19
22

29
29
30
38
40
42
43

I-l
I-2
I-3

II-l

11-2
11-3
11-4

II-S
11-6
11-7

III-1
III-2
III-3
III-4

LIST OF FIGURES

Possible domain structures

Fringing field at a plane surface

Variation in rotation with distance from domain
wall center

Non-magnetic pressure chamber and ballistic
magnetometer

Piston assemblies

A typical calibration of pressure microbomb

us1ng NH4F

Hydraulic press force vs internal confining
pressure inside bomb

Sense coil field

Mutual cancellation of solenoid and lab fields

Schematic diagram of equipment set-up

Magnetization vs pressure - sample 7

Magnetization vs pressure - sample PH-3
Magnetization vs pressure - sample PH-3
Magnetization vs pressure - sample PH-3

12
12

15

18
21
23

24

26
27

34
35
36
37

I. INTRODUCTION

To date, laboratory research on magnetic properties of
earth materials under simulated crustal conditions has cen-
tered on high temperature experiments and tests involving
application of uniaxial pressure. Little work has been done
concerning properties of minerals and rocks under realistic
high hydrostatic pressure. The latter can be an important
factor in the behavior of physical parameters of rocks and
minerals. For example, Stesky and Brace (1973) have shown
that the electrical conductivity of oceanic basalts varies
with increased confining pressure, and Carmichael (1969) has
shown a variation of coercive force in magnetite with hydro-
static pressure. Breiner and Kovach (1966) used pressure-
induced changes in magnetization of rock to try to forecast
seismic events along the San Andreas fault. Other recent
studies of applications of geopiezomagnetism to earthquake
prediction include those by Golovkov (1969), Rikitake (1968)
andJohnston (1975). Davis and Stacey (1971) noted that a
local magnetic anomally was produced in the area of a dam,
as a result of crustal loading when the reservoir was filled.

Since most crustal rocks are, or have been, subjected
to confining (hydrostatic) pressures of at least several
kilobars, it is important in the study of geopiezomagnetism
to understand the effects of confining pressure on magnetic
materials. Both transient and permanent effects of pressure
need to be studied to understand the true nature of magnetic

remanence in rocks, and how it changes in the lithosphere.

Specifically, this paper is intended to study the
nature of the variation of the magnetization of magnetite
within the pressure range of one atmosphere to five kilobars.
Isothermal remanent magnetization was monitored as hydro-
static pressure on the magnetite samples was increased from
one atmosphere to about three kilobars, then decreased to
two kilobars, increased again to four kilobars and finally
released to about one kilobar (the pressure held within the
pressure bomb by the frictional force of the packing mater-
ial). A significant drop was noticed in the magnetization
with the first pressurization to two kilobars, J dropping to
about 70% of the initial saturation. Little change in J was
noticed beyond 2.0 kilobars.

The samples used were a magnetite crystal and a synthe-
tic magnetite powder. The powder was mixed with a resin
matrix to simulate a rock with a high magnetite content.

A fundamental part of this type of study is the develop-
ment of technology capable of both simulating a high pres-
sure environment and allowing for convenient measurement of
the magnetic properties of the sample.

This work is intended to contribute something both to
the technology of current high pressure experimentation and

to the understanding of rock magnetism.

II. THEORY

Domain theory describes the magnetization of individual
mineral grains; the magnetization of such grains in rock
governs the magnetic behavior of the rock as a whole and is
therefore of prime importance in the study of rock magnetism.
The limit of how well the magnetic properties of minerals are
understood is the limit to how well paleomagnetism may be
understood and relied upon.

Grain size is a prime factor in the magnetization of
mineral grains and as such is an important factor in how long
a rock will hold its natural remanent magnetization, and to
what extent changing conditions affect that remanence.

The effect of grain size may be better understood by
viewing the energies involved in rock magnetization. Both
the energies and sizes of grains are interrelated with the
structure of domains within ferromagnetic mineral grains. A
number of good reviews of the subject may be found in the
literature (Neel, 1955; Stacey, 1963; Kittel, 1949, Chikazumi,
1964; Stacey and Banerjee, 1974).

DOMAIN ENERGIES

 

The purpose of domains in ferromagnetic minerals is to
minimize the total magnetic energy of the particular speci-
men. Since ferrites possess spontaneous magnetization below
the Curie temperature, magnetic fields will be set up within
the material, and these will extend beyond the sample sur—

face. This increases the magnetic energy of the specimen,

causing it to be in an unstable state-~hence the formation
of domains. The total magnetic energy of a ferrite grain
may be divided into six components, which will be briefly

described.

1. Exchange Energy, E This is produced by the interaction

x.

 

of electron spins of adjacent atoms, and was derived initial-
ly by Heisenberg. This energy can be minimized when the
magnetic moments of adjacent spins are parallel. If a domain
wall between two antiparallel domains is considered to be of
infinitessimal thickness, then the angle between the spin
axes will be 180°, with a maximum exchange energy. But by
making the domain wall relatively thick, with many contained
spin axes, then the angle between any two spin axes can be

reduced, thus reducing the exchange energy density.

2. Magnetocrystalline Anisotropy Energy, Ek- Sometimes

 

called the anisotropy energy or crystal magnetic anisotropy
energy (Chikazumi, 1964), this energy is a measure of the
preference of a ferromagnetic crystal for spontaneous magne-
tization along particular crystallographic axes (Kittel, 1949;
Chikazumi, 1964). The direction of preferred magnetization

is called the direction of "easy” magnetization; the most
difficult directions are called the hard directions. The
excess energy required to magnetize a grain in the hard di-
rection, as opposed to the easy direction, is the anisotropy

energy, Ek (Kittel, 1949).

'5
Expressions for Ek, in terms of the direction cosines
of the internal magnetization, have been worked out in most
discussions (Chikazumi, 1964; Kittel, 1949; Stacey, 1963).
For a cubic crystal system, two anisotropy constants are
used, K1 and K2; the value of Ek for a cubic crystal, such
as given below (01, 62, 03 are the direction cosines of the

magnetization vector with respect to the three cube edges):
2 2 2 2 2
2 + “2 “3 I 0‘3 0‘1) I K2 0‘1 “2 “3

Both anisotropy constants are temperature-dependent,
and can vary enough to change sign of the net Ek' K1 is
generally more important than K2 (Stacey, 1963).

In magnetite, the easy axes are the <lll> axes, and the
anisotrOpy constants are negative, so we are primarily con-
cerned with a magnetization inclined at some angle 0 to the
<lll> axis. To a second-order function in sin 0 (Stacey and

Banerjee, 1974) the above equation reduces to

K .
E.= e +3.9) - as! +212) .,

the second term dropping out as the magnetization aligns

itself with the easy axis. For pure magnetite at 290°K,

K1 = -1.36 x 105 ergs/cc and K2 = -0.44 x 105 ergs/cc, making
at -1.80 x 105

E ergs/cc, lowest along <111>.

k!
A more gradual rotation of spins in domain walls, while

decreasing the exchange energy, increases Ek by forcing some

spins out of the easy direction of magnetization. However,

‘while EX is affected by this movement of spins within the

crystal, the exchange energy itself does not produce any anis-
otropic effects. The origin of magnetocrystalline anisotropy
is due rather to atomic interactions: "the spin interacts
with the orbital motion (of the electrons) by means of the
spin—orbit coupling and the orbital motion in turn interacts
with the crystal structure by means of the electrostatic
fields and overlapping wave functions associated with neigh-

boring atoms in the lattice" (Kittel, 1949).

3. Domain Wall Energy, Ew. This is a function of the mini-

 

mized sum of the exchange and anisotropy energies (Ex + Ek).
In a uniaxial model the exchange energy per unit area of wall

is given by

(2) EX = fA(6¢/6X)2dx (Craik and Tebble, 1965)

where A is the exchange constant and 4 is the angle of spin
rotation as a function of x (see Fig. I-3). In the same

model, Ek per unit area is given by
= ' 2
(3) ER le Sln (¢)dx

where K is the anisotropy coefficient. The energy of the

wall using the above values for Ex and Ek is then given by

(4) Ew = 2 (AK1)% (Stacey, 1963)

this being approximately 1 erg/cmz, while letting A equal
the exchange energy for spin S and lattice spacing "a" --

A = J is the exchange integral.

7

4. Magnetostrictive Anisotropy Energy, EA . When a crystal

 

o
is stressed, the domains and walls within it are physically

moved. This movement causes subsequent changes in the direc~
tions of the domain magnetization vectors, and a change in
the net magnetization of the crystal. The energy associated
with this change in magnetization is the magnetostrictive
anisotropy energy (also referred to as the magnetoelastic
energy or the magnetostrictive strain energy). BAG is defined
to be zero for an unstrained lattice (Kittel, 1949). It
should be kept clear that when the magnetization induces a
physical change in the shape of a specimen the effect is
termed magnetostriction; when applied stress on the specimen
induces a change in magnetization, the effect is termed in-
verse magnetostriction, or piezomagnetism.

When a ferromagnetic body is placed in a magnetic field,
interaction between the atomic magnetic moments causes a
change in bond lengths, since E10 is a function of the inter-

atomic distance 3. For a cubic lattice, EA may be expressed

o
in terms of the lattice strain tensor components (eij) and
the direction cosines of the domain magnetization (ai) as

(Chikazumi, 1964) for the one-dimensional case:
(5) Bio = B1<§xx(a - l/3)+ eyy(a - 1/3) + ezz(a - 1/3))

+ Bz(exya1a2 + eyzaZa3 + ezxa3al)

_ 91L _
where B1 — N(dr ro , B2 - ZNL

and N = demagnetization factor, L = specimen length.

 

8

If a stress 0 is applied to a ferromagnetic material,
the energy due to that stress may be expressed by giving the
strain tensor components of the magnetoelastic energy in
terms of the direction cosines gi, the stress, and the elas-
tic constants SijI

as,

_ p 2 2 2
(6) exx — o(sllg1 + 512(g2 + g3))....

By substituting (6) and (5), we arrive at a new expres-

sion for E :
Ao

_ _ 2 2 2 2 2 2 _
(7) Exo ‘ B10(511 S12)(31g1 I 3282 I 3333 1/3)

I B2°544(31a2g1g2 I azasgzgs I a3alg3g1)

If, for example, the magnetization of the domain is
parallel to the (111) face, Elo may be expressed in terms of
0, the angle between the direction of tension in the crystal

and the (111) face:
(8) cos(0) = (1/3)2 (g1 + g2 + g3) and

3
(9) BAG = + 3 Alllocosz(0)

The same holds for other domain directions, with the
substitution of the correct value for A, the saturation
magnetization strain (Chikazumi, 1964).

The dependence of the magnetostrictive energy on the
domain configuration becomes complex in the threeédimensional

real case. However, a two-dimensional example computed for

9
the four domain case of Fig. I-1.c (Stacey and Banerjee, 1974)

serves to give an order of magnitude value for EA
3 S

o in magne-

tite of 6 x 10 ergs/cc, compared to -2 x 10 ergs/cc for Bk.

5. Magnetostatic Energy, Em. When free poles exist at the

 

surface of a grain, they give rise to the magnetostatic

energy, given by

(10) Em = %NJ: (for an ellipsoid) (Stacey, 1963)

where JS is the saturation magnetization parallel to the long
axis of the ellipsoid and N is the demagnetization factor.
The strength of Em is dependent upon the shape of the sample

and the number and strength of free surface poles.

6. Field Energy, Eh. A grain may also acquire a field

 

energy when in a magnetic field H, with magnetization l of

the sample. The energy per unit volume is given as

(11) Eh = HJ cosO (0 = angle between J a H) (Stacey, 1963).

This may cause the enlargement of domains oriented in the
general direction of H, while opposing domains will be re-

duced.

GRAIN SIZE EFFECTS

 

The importance of grain size in the study of domains is
evidenced by the several properties dependent upon it: such
properties include coercive sorce, susceptibility, and transi-
tions from the superparamagnetic state to single domain and

multidomain states.

10

Consider first a spherical isotropic grain of sufficient
size to contain only one domain. The energy of the grain is

then given by the magnetostatic energy only (Nagata, 1961):

(12) Em = l/zNJ:(4/3)1TR3 , where N = “24/3.

As grain size increases, a point is reached where the
inclusion of a domain wall will decrease the energy of the
grain, by decreasing the magnetostatic energy by a greater
amount than is gained by the addition of the wall energy.
The new grain energy is given by the sum of BW and Em, the

new total energy now being reduced by 50%:

2

(13) E = %(%J:R3)(4n/3)2 + nR a

total w

where aw is the wall energy per unit area.
The theoretical critical radius, R2, may be found by
equating the total energies of the single and multidomain

states. In this instance, the value for R2 is (Nagata, 1961)

_ 2
(14) R2 - 9aw/41TJS

Single domain particles range in size from 0.1 microns
in diameter to 0.5 microns. From 0.5 to 20.0 microns grains
undergo a transition from the single domain state to the
multidomain state. Grains in this range are termed pseudo-
single domain particles. Grains larger than 20.0 microns
zare multidomain only. (These values are for magnitite).

Single domain particles have coercive forces that are

ggreater than either SPM or MD particles. Neel (1955) states

11

that this occurs because wall displacements, necessary in

the movement of a multidomain particle through the hysteresis
cycle, require a release of energy, while this is not a fac-
tor with single domain grains. SPM particles have smaller
values of HC because below a certain volume, 1, the thermal
agitation energy kT becomes large with respect to the energy
terms dependent upon X and the anisotropy constant E, such

that kT exceeds Kv Sin2

(0). This perturbs the precession
of the atomic moments, altering O and causing the remanence
to decay with time, according to Jr - Joexp(-t/t0), to being

the relaxation time (Neel, 1955).

 

 

 

 

 

 

 

POSSIBLE DOMAIN STRUCTURES
(After Stacey, 1963)

 

 

 

 

 

 

 

 

FRINGING FIELD AT A PLANE B
(After Stacey, 1963)

13

 

 

 

180 1 I l

150v /

¢ 90

 

 

 

 

 

 

 

FIGURE l-3
ABOVE IS THE VARIATION IN ROTATION WITH DISTANCE
FROM THE DOMAIN WALL CENTER.

BELOW IS THE MAGNETIZATION DIRECTION WITHIN THE
DOMAIN WALL. (After Craik and TobeOI.

III. EQUIPMENT AND EXPERIMENTAL TECHNIQUES

A large part of this study involved the development of
new high-pressure apparatus and subsequent modifications of
this equipment. The following is a discussion of the equip-

ment and techniques used.

PREVIOUS WORK

 

The study of geophysical properties of earth materials
under hydrostatic pressure has previously been largely re-
stricted to the pressure range of less than two kilobars,
with the exception of some velocity studies and some work in
electrical conductivity (Stesky and Brace, 1973). Experi-
ments in solid-state physics have gone into hundreds of
kilobars, but these studies usually begin at approximately
10 kilobars, and are not done on earth materials. The pur-
pose of the present study is to partially fill the 2-—10 Kb
gap.

Until recently, research into the magnetic properties
of minerals has generally been restricted to tests under
uniaxial pressures. However, effects of high hydrostatic
pressure on the Curie temperature of magnetite (Schult, 1970),
and on coercive force and saturation remanence (Carmichael,

1969) have been observed.

PRESSURE MICROBOMB

 

The pressure vessel used in this study is essentially
that described in previous studies (Carmichael, et al, 1968).

It is a piston apparatus (Fig. II-l) of beryllium copper

14

15

NO N- MAGNETIC PRESSURE CHAMBER

AND

BALLISTIC MAGNETO METER

PISTON

SENSE COIL

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

/ LOCKING CAP

I“

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

VT
>-
m
J
J
O
D:
*—
SAMPLE
I I‘ "I
I I |
SOLENOID __I |__ _I
. N
\
_ ‘Q
X
‘ 3
T l'
I I
‘K
PISTON
SUPPORT
CYLJNDER

FIG. ”-1

16

alloy (1.81% Be by weight) and is capable of sustaining
internal pressures to 11 kilobars. The bomb is diamagnetic,
with a susceptibility of -0.6x10.7 emu/gm at 20°C, so magne-
tic changes due to the bomb with variations in applied fields
are negligible compared to changes in magnetite and other
samples.

Aside from its non-ferromagnetic material, the bomb has
a number of other unique attributes. It is capable of pro-
ducing both hydrostatic pressure and a uniaxial pressure on
the sample simultaneously, and needs only a simple uniaxial
press to achieve this. The bomb is small enough that magne-
tic measurements may be made with a small solenoid during
pressurization, and it is easily portable. The pressure
may be held in the bomb for long periods of time by means
of the locking caps, which may be screwed down against the
mushroom pistons on either end, thus making possible the
study of long-term pressure effects. The present version of
the bomb has a larger sample space than the original, allow-
ing the study of samples up to 1.2 cm in diameter and 2.5 cm
in length.

After machining, the bomb used here was heat treated at
3160C for four hours. After heat treating, the parts were
work hardened by the method outlined by Paul, et al (1959).
All materials were work hardened up to 5.5 Kb.

Following several pressure runs to four kilobars, the
large piston's mushroom cap sheared around the base. The

outer ring of the cap had been supported by the washers and

l7

shoulder of the main piston, while the inner portion of the
cap was forced down into the recess in the outer piston by

the confining pressure (see Fig. 11-2). This was corrected
by modifying the piston and cap: the flanged ends of each

cap were increased in thickness, and the length of the neck
of the cap was increased.

The modification corrected the problem with the mushroom
caps, but the stress previously absorbed by the caps was now
transferred to the shoulders of the main pistons. No ill
effects were noted at pressures below four kilobars, but at
pressures between 4.0 and 5.5 kilobars the main piston of
the large piston assembly began to deform, expanding the
portion of the piston between 0.2 and 0.5 centimeters below
the bottom of the recess. This expansion made removal of the
piston assembly difficult, and greatly increased the friction
between the cylinder wall and the piston during pressure runs,
thereby adding an uncertainty in the value of the internal
confining pressure with increasing applied press force. The
piston was remachined to its original diameter, but this
weakened it enough to cause shearing in the previously ex-
panded area.

The large piston was again modified, this time by re-
ducing the diameter of the recess and corresponding neck of
the mushroom cap. This allowed the stress to be absorbed by
a greater area of the piston, and to be concentrated more
toward the center of the piston instead of at its perimeter.

The final versions of the pistons are shown in Figure II-Z.

 

LARGE PISTON ASSEMBLY PISTON

 

 

[11

Be— Cu WASHERS (Ill /.<-— LEAD GASKET
‘— RUBBER GASKET

, 96-h

Hm

 

 

ALL PIECES HAVE 4— MUSHROOM CAP
CIRCULAR CROSS—SECTION

 

 

 

 

 

 

 

SMALL PISTON ASSEMBLY <— PISTON
/"D
Be—Cu WASHERS \h-EI 84— Pb GASKET
‘3 9"- RUBBER GASKET

<—- MUSHROOM CAP

 

SCALE
‘22::
0 1,200 1"

PISTON ASSEMBLIES

FIG. “—2

19

One final modification of the previous bomb was made.
When extracting the pistons and sample from the bomb after
pressurization, a force of several hundred pounds is required
to drive out the large piston. Single crystal samples gen-
erally cannot sustain such large directed stress without
crushing. To prevent destroying the samples after each pres-
surization, a brass bushing was made to fit around the sample.
Thus, when the large piston is driven out by forcing the small
piston through the entire length of the bomb, this force is
absorbed by the bushing and not the sample. Small notches
were filed in each of the bushings to allow free fluid flow

around the sample during pressurization.

OPERATION

 

Pressurizing techniques for the current research closely
followed tyose outlined for the original bomb (Carmichael,
et al, 1968). A fluid pressure-transmitting medium of a 1:1
mixture of kerosene and transformer oil was used. A fluid
pressure medium makes for more difficult sealing, but some
advantages are gained over solid pressure-transmitting medi-
ums: uniform confining pressure is assured, so there is no
concern about possible pressure gradients (Paul, et al, 1959)
and though the oil mixture congeals at low temperature, there
is a low fractional change in volume and therefore little
hydrostatic pressure loss (Carmichael, et a1, 1968).

The sample was first coated with resin to prevent oil
from entering any small cracks and causing the sample to

disintegrate under pressure. The sample was then glued to

20
the base plate and inserted into the microbomb, One piston
was then driven into the cylinder far enough to allow the
corresponding locking cap to be secured. The chamber was
then filled with oil, the other piston enserted and the
second locking cap secured.

Seals around the pistons were provided by two gaskets,
one of rubber and one of lead. The rubber gasket expands
against the wall of the cylinder upon slight pressurization,
and may be tightened by hand to a sufficient degree to pre-
vent leakage. As pressure is applied by a press, the lead
gasket is compressed between the two Be-Cu washers, flowing
out around the piston to form a seal sufficient to withstand
the pressures involved (Fig. II-Z).

Internal hydrostatic pressure was determined using a
sample of NH4F. This crystal undergoes a change of phase at
3.65 Kb, and reduces in volume by about 30%. The sample of
NH4F was sealed by an impermeable rubber membrane, and pis-
ton displacement was monitored versus applied press pressure.
A typical calibration curve Of several tests for this micro-
bomb is given in Figure 11-3. The sudden change in the slope
of the curve indicates the point where internal pressure has
reached 3.65 Kb; the sample has suddenly reduced its volume,
and the pistons have moved in rapidly to maintain a constant
pressure. The half-width of the hysteresis loop of each
pressure cycle is approximately equal to the packing friction
between the large piston and the cylinder wall (Paul, et al,

1959). Internal hydrostatic pressure is assumed to be a

21

“3.22 0253
mEOmOmQE wmammmcn. “.0 20.53.5943 4(03: (

 

 

T: 9.50:.
.3 82x memo“. is
m. E m. a. : o. a m A o o
_ _ _ _ _ L _ _ _
o
I I _
386 new I oz_m<mmoz_ I
386 EN. 0
386 .20 o

OIEEm o ).9
o x m
I. I m mm
0 . . N
$28.5. men I. w
wing $335 A
0 3.2.58 $255. .1

I I e

o
oo o
ozafimoz. / oo o
I o 0 oil! 3.935% I m
o o
GIG. o o O
_ _ _ _ _ _ _ _ _ o

 

 

 

22

linear function of the applied pressure on the piston(s),
and a plot of internal hydrostatic pressure versus applied
force is given in Figure II-4. Applied force in Figures
11-3 and 11-4 is accurate to withinlOOpounds. This implies

that the hydrostatic pressure is accurate to i 0.1 Kb.

BALLISTIC MAGNETOMETER

 

Measurements of magnetic properties involved the use of
a ballistic magnetometer, consisting of a solenoid, a pair
of sense coils (coil A wound in the opposite direction from
coil B), a D-C power supply, and an integrator and a digital
voltmeter (DVM). The ambient lab field inhomogeneities may
be either nulled out by the solenoid, or a large field (30
to 60 gauss) may be produced. By moving the sense coils up
and down within the solenoid (see Figure 11-1), changes in
magnetic flux within the solenoid induce electrical current
changes within the coils. These changes in current are trans-
mitted through the integrator and read as a total voltage
change on the DVM. Thus, a change in total magnetic flux
from the bottom to the top of the solenoid is read as a rela-
tive change in voltage. When a magnetic sample, such as
magnetite, is introduced into the center of the solenoid, the
flux lines are concentrated around the sample. Since the
coils are wound oppositely, a maximum signal is obtained when
the sample is effectively moved from one coil to the other.
A sample curve of the field produced by a small current in

the sense coils is shown in Figure II-S.

23

920m No.2,: umbmmmmm 02.2.n—ZOU 4<2¢mh2.
.w> womOu 8mm... 03:42.32...

 

        

T: 230:
8.. 8N. + memo". 33%? u . 3.. 82x. memo“. is
a «N cm 2 2 3 «F o. m o v N o o
— L _ _ _ A _ A _ a

I P
w
I N 3
H
N
V
fil )1
m mm
0 4.
8 |
.50 02.93.. w m
I m> / is. m
a
U
a
V m>mao 02.96.23 I m m
\ \ H
\ \ 3

\
\ \
I \ \ \ I o
\ \
\ \ \
[ _ _ L.\ _ _ _ _ _ _ _ _ h

 

 

 

24

I: .21 224.. me do Emmmao :53.
3m... .:oo 32%

.20 2: mice “5 :mhzmu
Pr momzwm mh<mlx=._u_ mo :thmo Soc”. muz<hw5

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

: S a m m S m N _ a T NI T T T I T T T 2.. :I
_ _ _ _ _ _ _ _ _ _ _ _ _ . _ L _ _ _
I I T
I I @I
I I T
I \ I/ I T
I \ E I ”I
\ \ 3:5;
I \ \ I NI
I \ \ - -
\ Vt w “2.5%” a
I \ \ I — IN:
I x x .. N m
I \ \ I m m
I \ I/ I .
I 2.8 .6 22:8. \« I m
I I m.
I I A
I I a
I I m
_ _ _ _ _ _ _ _ _ _ _ _ _ F _ _ _ _ _ E

 

 

 

25

The lab field within the press contained a small in-
homogeneity. This could be either cancelled or strengthened
by the solenoid field, since the solenoid field was strongest
at the center, and the strength of the field dropped sharply
toward the ends of the coil. An example of the solenoid
field cancelling the lab field is shown in Figure II-6.

The equipment set-up used is shown schematically in
Fugure II-7. The bomb, with sample, was placed in the center
of the solenoid, which was positioned in the center of the
hydraulic press, and the sense coil was manually moved ver-
tically within the solenoid. Voltage changes were noted on
the DVM.

The samples were first saturated in a 2 kilogauss field,
and their initial saturation magnetization measured in a
nulled field outside of the press before loading in the bomb.
The field within the press was then checked to determine if
the ambient field was nulled out by the solenoid field. This
being done, the bomb, with sample, was placed within the coil,
and the magnetization of the sample was measured in the press
at one atmosphere. The sample was then pressurized, and
readings of magnetization were taken at intervals of 0.5 Kb.

After cycling the pressure, a measurement of magnetiza-
tion was made with no press pressure, leaving the sample
under only the pressure due to the friction of the packing in
the bomb (approximately 1.0 Kb). Final measurements were
made with all pressure released, but with the sample and bomb
still in the press, and outside the press and bomb. Magnetic

data values are accurate to within i4.0% ofJ} sat

 

7////////4/

 

 

 

 

F L m
m m M
m I as
U. m. an
I.—
w 4 um
I AF
I CB
I—
A
m .u...
m mm
F MA

27

bl: .07.”.

 
    

mama...

z. mzomomoi

a mmhwzohmzmzz
o_.rm_._..<m mo .:<._.mo

h2m2m>o2

+ mmmmm

   

 

 

 

 

 
 

   

) m0h<mowhz_

E.

\

miomomnzs
3.8 32% $25.. 2
90233 < o...
mmwmm
USD<mD>I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

malhwm .PstESOm
".0 E<m0<a U_._.<s_mIom

ambus—
._.._o>

mow<momhz_ 4.4.555

Pam \‘fi.

mmhwszz< olo

\Qo
@ mm.

 

 

 

 

 

 

 

 

28

It should be noted that absolute magnetization was not
measured during pressurization. Relative changes in magne-
tization, J, were measured, and all plots of data give rela-
tive changes in the normalized magnetization, since the
voltage changes registered on the DVM are directly proportion-

al to magnetization changes (J/JO = dV/dVO).

29

IV. SAMPLE SPECIFICATIONS AND EXPERIMENTAL RESULTS

Two types of magnetite samples were used in the present
work: a single crystal, and a synthetic "rock" composed of
magnetite powder cemented in a resin matrix. A description of
the samples is given below, followed by the magnetic results

obtained.

GENERAL PROPERTIES OF MAGNETITE

 

Magnetite, Fe304, is one of the most widespread of the
oxide minerals. It is found in igneous bodies, contact-
metamorphic deposits, as replacement deposits, some high-
temperature sulfide veins, and in detrital beach deposits.
It is a major ore mineral for iron; specimens showing strong
natural polarity are called lodestone.

The magnetite unit cell is of cubic, inverse spinel
structure. The mineral has no cleavage, but will show parting
along the (111) faces.

It is ferrimagnetic at room temperature, and has a
saturation magnetization of 92—93 emu/gm (Clark, 1966)
hold radius from the single domain to multidomain state is
0.57 to 0.72 microns (Soffel, 1971; Dunlop, 1972). Other

parameters are given in Table l.

SAMPLES
The single crystal sample used was a magnetite rod, cored
perpendicular to the (111) face. It is denoted as "Sample

PH-3." Specifications are given in Table l. The sample was

30

coated with an epoxy glue prior to any pressurization, to
prevent the fluid from seeping into any cracks and damaging
the crystal.

The magnetite used in the synthetic sample was a syn-
thetic magnetite powder derived from an aniline chemical
process. It had been previously roasted at 900 C to remove
water, then it was ground and sieved to remove particles with
diameters greater than 149 microns. Most particles were about
6 microns in diameter.

Thegrains were mixed into a resin compound at a ratio
of one part magnetite to three parts resin. This mixture was
poured into a plastic tube, allowed to harden, then cut into
sections one centimeter in length.

Both samples (PH-3 and No. 7) were saturated in a 2
kilogauss field before any measurements were made. This pro-

duces a saturation isothermal remanence, Jr sat
, O

MAGNETIC RESULTS

 

Results of pressurized tests within the solenoid are
presented in Figures III-1 through III-4. Samples were
pressurized to 3.0 kilobars initially, resulting in a magnet-
ization drop to around 70% of the initial saturation magnet-
ization. Pressure was then reduced to 1.5 kilobars, while
noting a slight drOp in magnetization of 2 to 3% Jr,sat.
Then pressure was increased to 4.0 kilobars, causing another

2-3% decrease in J. The final reduction in pressure from the

4.0 Kb maximum to standard pressure showed the magnetization

31

to remain at the 4.0 Kb value to about 500 bars, at which
point the magnetization increased by about 5%. The point of
interest here is not so much the exact pressures at which J
decreases or increases, but rather the trend of the samples
to decrease with initial pressurization and thento stabilize
somewhat during increasing stress.

The effect of the packing friction within the bomb should
benoted here. Once pressurized to over 2.0 Kb, the frictional
force of the lead packing worked to hold in approximately 1.0
Kb. This is shown by the two pressure curves of Figure II-4.
As a result, less press force was needed on a decompression
cycle than on a compression cycle to achieve a given internal
pressure.

The factors most notably prone to error in measurement
were press force and J. Press force oscillations on the order
of 100. 1b. were noted on all tests, apparently as a result
of the developed inertia, causing the press mass to continue
moving. Since the magnetic affects of interest occurred at
press forces of approximately 4,000 to 14,000 pounds, an
allowable upper limit of percent error in applied pressure is
:2.5%. These oscillations were probably damped out to some
degree by the packing friction of the bomb.

Relative values of J were read as emf output in micro-
volts from the magnetometer. A single movement of the sense
coil past a typical unpressured sample produced a voltage

change on the order to 100. microvolts. Each measurement

32

was made three times, and the average value recorded. Values

were reproducible to within 4. microvolts, or 4.0% Of Jr,sat.

33

Properties pf Magnetite (Fe304)

 

 

Saturation magnetization (J5) 98 emu/gm at 00K

92 emu/gm at 20°C

Single domain--MD critical

diameter 0.5 - 20.microns
Kl , at 170C -l.36 x 105 ergs/cc
K2 , at 170C -0.44 x 105 ergs/cc
Curie temperature, TC 575°C
N00' 20 x 10-6 cm/cm
M10 60 x 10"6 cm/cm
M11 78 x 10-6 cm/cm

(from Carmichael, 1971; Stacey + Banerjee,l974;)
(Soffel, 1971; Dunlop, 1972)

 

Sample PH-3
Magnetite, [111] rod from Port Henry, N.Y.
length 1.33 cm
weight 1.70808 gm

Trace analysis:

Ti02 = 1.06% (Ti = 0.636%)
Ni = 400 ppm
Sn = 160 ppm
Cu = 150 ppm
CaO = 280 ppm
K20 = 60 ppm
(from Carmichael, 1975)
Sample 7 -- Magnetite powder

 

Synthetic Fe304; from aniline chemical process

Average particle size grouped around 6 microns; 90% of
particles between 1.5 microns and 20 microns.

Very small trace amounts of Si, Al (Carmichael, 1975)

Table 1.

34

mmamwmmm .m) ZO_._.<N_._.mZO<E
pl... manor.

Amm<m04 C: memmwmn.
mi 06 md o.m 9N ON m; o.— md o

 

_ _ . . _ _ . . . c

 

 

h .02 wIEE<m /L;.o

I I I mm<m 10mm.
mm<wmoz_

 

 

3S

mmammmm... m) ZO.h<N..—.m20<$

 

NI... umber.
5143045.. mmammmmn.
m6 oé m.m o.m mN ON m.— o.— md o .
_ _ A _ 4 q . . . cc
I md

 

 

mlIm uni—21$

I I I wm<mmomo
wm<mmuz_

 

 

 

36

wmsmmmmn. .m> 20....<N.hm2mu<2

 

 

   
   

 

«I... mane...
.mm<mo.._v.. $83....
3 m...” 9m 3 3. m... 3 no .3 .
_I q . q . . A _ m0
I m...
D
I IIIII I I I I I I I
o o IOI I I I g
-3
38....
I-a
mi... 3.2% m...
..I I $553
$552. a...

 

 

o.—

mcammwm... .m) 20....<N.hm20<2

 

 

I mm<mmoz.

CI... mtDGE
5.73043: mmmewmn.
O.m ON ON m.— O.— md O .
. _ . . q — mo
I 0.0
lUIIMIIIIOIIIIOIIIIOI O

/ / A. no

$3....

   

Olin. m..n..2<m

II I mw<mm0mc
0.0

 

. O.—

V. INTERPRETATION

All samples experienced a sharp drOp to approximately
66% of the initial saturation isothermal remanence from zero
to 1.5 kilobars of applied pressure. This is similar to
results obtained by Carmichael (1968) using applied uniaxial
stress.

With the application of pressure, a certain amount of
mechanical energy was input into the sample. Essentially, this
added energy allows some domain walls to move over small
potential energy barriers and relocate in new positions such
that the domain energies more effectively cancel, thus de-
creasing the total grain energy. Since the pressurization
took place in a zero field, the alteration of the sample in
decreasing its total energy was also tending toward a net
demagnetization.

The movement of domain walls under pressure occursin both
reversible and irreversible fashion. The net drop in J is
a measure of the irreversible change; the difference between
the minimum value of J (at maximum pressure) and the final value
of J (at 1.0 kb) is a measure of the reversible change.

Some of the applied hydrostatic stress may have built
up unevenly along existing flaws in the sample, such as along
microcracks and irregularities in the lattice structure. This
locally increased pressure would add a measure of directed
stress, cauSing anisotropic effects. Such effects would

induce altering the anisotropic energy terms K1 and K2

38

39

(in ER). While an alteration of the "isotropic" energy, KO,
would not affect the domain configuration, changes in K1 and K2
would. Both amplitude and physical location of domain energy
barriers in the region of locallized stress could be expected
to change.

While such anisotropic changes in domain magnetization
could occur, a random orientation of flaws within the sample
would be expected to cancel any net change in the direction
of magnetization. The resulting new domain configuration
(after pressurization) was of lower energy than the initial

configuration, with a greater degree of closure permitting

a lower magnetization.

VI. CONCLUSION

Samples of magnetite were magnetically saturated and
ubjected to hydrostatic pressurization to 4.0 kilobars. This
is equivalent to the hydrostatic pressure in the crust at a
depth of about 12 km. The isothermal remanent magnetization
decreased with increased pressure to approximately 70% of the
saturation value near 2.0 kb, at which point the magnetization
appeared to stabilize. Subsequent cycling of the pressure to
higher values failed to effect any noteworthy change in magnet-
ization.

The results noted were probably due to the effects of
the additional energy supplied to the samples by the pressur-
ization. This new energy input to the magnetite grain
allowed some domain walls to overcome small local potential
energy barriers. These potential barriers had previously
prevented the domains from aligning themselves in such a way
as to minimize the total magnetic grain energy. Realignment
of the domain walls and the subsequent alterations in domain
configuration, allowed the grain energy to decrease, as noted
in the magnetization drop. The irreversible effects are
apparent in the net loss in J after pressurization (about
a 30% Jr,sat loss). Reversible effects may be noted in the
rebound of the magnetization upon decompression Of the last
pressure cycle.

The results of the pressurizations have shown that the

remanent magnetization of magnetite may be altered if the mineral

40

41

is or has been subjected to deep burial. Most of the pressure
effect was observed at pressures less than 2 kilobars, cor-
responding to 6 km or about 20,000 feet (a depth to which some
oil wells are drilled). Sampling magnetic properties of mag-
netic minerals cored from such wells could be worth future
study.

Sufficient technology has now been developed and tested
for measuring magnetic properties of rocks and minerals during
pressurization. This is a distinct advancement over methods
involving pressurization followed by separate measurements of
magnetic prOperties. A suggested improvement would be a
continuous monitoring of changes in magnetic properties during

pressure cycling, as opposed to the discrete sampling

currently in use.

APPENDIX

APPENDIX A

BOMB CALIBRATION DATA

Column A Column B
Dial Force on Large Large Piston Position from
Piston (x 1000 lb.) Strain Gague

(decreasing into chamber)
(1.0 unit = 0.02 inches)

 

 

A B A B
11.5 15.10 7.5 15.74

12.0 14.82 7.0 16.12
12.6 14.30} TRANSITION 6.5 16.50

12.7 12.97 6.0 17.00

12.9 12.83 9.0 16.00

13.0 12.74 11.5 15.15

13.1 12.65 12.1 14.85
13.5 12.45 12.5 14.58

}TRANSITION

14.0 12.20 13.1 13.16

14.3 12.10 10.0 13.41

13.5 12.11 9.7 13.45
13.0 12.14 9.2 13.52

12.5 12.18 8.7 13.68

12.0 12.23 8.3 13.90

}TRANSITION

11.5 12.27 8.0 14.85
11.0 12.35 7.7 15.18

10.5 12.50 7.5 15.30
10.0 12.70 7.0 15.68

9.5 12.94 6.5 16.08

9'0 13'ZZITRANSITION

8.5 15.15

8.0 15.37

42

B I BL IOGRAPHY

 

BIBLIOGRAPHY

Banerjee, S.K., New grain size limits for paleomagnetic
stability in hematite: Nature Physical Sci., v. 232,
pp. 15-16, 1971.

 

Brace, W.F., A.S. Orange, and T.R. Madden, The effect of
pressure on the electrical resistivity of water-
saturated crystalline rocks: J. GeOphys. Res., v. 70,
pp. 5669-5678, 1965.

 

Breiner, 8., and R.L. Kovach, Local geomagnetic events
associated with displacements on the San Andreas
Fault: Science, v. 158, no. 3797, pp. 116-118, 1966.

Carmichael, R.S., Remanent and transitory effects of elastic
deformation Of magnetite crystals: Phil. Mag., v. 17,
no. 149, p. 911-927, 1968.

 

-—---, Stress control of magnetization in magnetite and
nickel, and implications for rock magnetism: J. Geomag.
Geoelec., v. 20, no. 3, 1968a.

----- , Hydrostatic pressurization of magnetite: Geophysics,
v. 34, pp. 775-779, 1969.

 

----- , Apparatus and results for pressure research on mag-
netic prOperties of magnetite, nickel and cobalt:
Coll. Intern. Centre Nat. Recherche Sci (Paris)., no. 188,

pp. 4997503, 1970.

 

 

----- ”Paleomagnetism: A Laboratory Manual", Ex loration
Research Center, Shell Development Company, ouston, 1971.

----- , Personal communication, 1975.

----- , A. Sawaoka, and N. Kawai, A multipurpose high pressure
microbomb: Jap. J. App. Phys., v. 7, pp. 1120-1124, 1968.

Chikazumi, 8., Physics 9f Magnetism, Wiley, New York, 1964.

 

Clark, S.P., ed., Handbook of Physical Constants, G.S.A. Mem.
97, 1966. "" “ ‘—

 

43

 

44

Craik. D.J., and R.S. Tebble, Ferromagnetism and Ferro-
Magnetic Domains, Wiley, New York, 1965.

 

Davis, P.M., and F.D. Stacey, Geomagnetic anomalies caused
by a man-made lake: Nature, v. 240, 1971.

Dobrin, M., GeOphysical Prospecting, McGraw-Hill, New York,
1960. ‘

Dunlop, D.J., Magnetite: behavior near the single domain
threshold: Science, v. 176, pp. 41-43, 1972.

----- , Monodomain theory: experimental verification: Science,
v. 162, no. 3850, pp. 256-258, 1968.

----- , Superparamagnetic and single domain threshold sizes in
magnetite: J. Geophys. Res., v. 78, pp. 1780al793, 1973.

 

----- , M. Ozima, and H. Kinoshita, Piezomagnetization of

single domain graints: a graphical approach: J. Geomag.
Geoelec., v. 21, no. 2, pp. 513-518, 1969.

Golovkov, V.P., Anomalous geomagnetic field variations in a
seismically active region: Geomag. and Aeronom., v. 6,
1969.

Irving, E. Paleomagnetism and its application pp geological
and geophysicaI’pyOElems, Wiley, New YOrk, 1964.

 

 

 

Johnston, M.J.S., Tectonomagnetic effects: in Proc. pf
T. Nagata Conference, Goddard Space Flight Center,
GreenEeIt Md., I975.

Kawai, N., and A. Sawaoka, Magnetic measurements under
hydrostatic pressure: intensity of J and anisotrOpy:
Rev. Sci. Inst., v. 38, no. 12, pp. 1770-1772, 1967.

 

 

Kim, K.T., Analysis of pressure generated in a piston-
cylinder type apparatus: J. Geophys. Res., v. 79,
pp. 3325-3333, 1974.

Kinoshita, H., and T. Nagata, Dependence of magnetostriction
and magnetocrystalline anisotrOpy of magnetite on
hydrostatic pressure: J. Geomag, Geoelec., v. 91, no. 1,
pp. 77-79, 1967. _

Kittel, C., Introduction £9 solid state physics, Wiley,
New York, 1965.

 

----- , Physical theory of ferromagnetic domains: Rev. Mod.
Phys., v. 21, no. 4, pp. 541-583, 1949.

45

Mason, B., and L.G. Berry, Elements pf mineralogy, Freeman,
San Francisco, 1968.

 

Mirwald, P.W., I.C. Getting, and G.C. Kennedy, Low-friction
cell for piston-cylinder high-pressure apparatus,
J. GeOphys. Res., v. 80, pp. 1519-1525, 1975.

 

Nagata, T., Rock magnetism, rev. ed. Maruzen, Tokyo, 1961.

 

----- , Principles of the ballistic magnetometer for the
measurements Of remanence: Methods 13 paleomagnetism,
pp. 105-114, 1967.

 

----- , Tectonomagnetism: IAGA Bull. N9. 26, 1969.

 

Neel, L., Some theoretical aspects of rock magnetism:
Adv. Phys., v. 4, pp. .91-255, 1955.

Paul, W., G. Benedek, and D. Warschaurer, Non-magnetic high-
pressure vessels: Rev. Sci. Inst., v. 30, no. 10,
pp. 874-880, 1959.

Rikitake, T., Geomagnetism and earthquake prediction:
Tecton0physics, v. 6, p. 59, 1968.

 

Sawaoka, A., and N. Kawai, Change of magnetic anisotropy
constant K of magnetite under hydrostatic pressure:
Phys. Let., v. 24a, no. 10, p. 503, 1967."

 

----- , Effect of hydrostatic pressure on the magnetic aniso-
tropy of ferrous and ferric ions in ferrites with spinel
structure: J. Phys. Soc. Japan., v. 25, no. 1968.

 

Schult, A., Effect of pressure on the curie temperature of
titanomagnetites: Earth and Plan. Sci. Let., v. 10, 1970.

 

Shimuda, M., S. Kume, and M. Koizumi, Demagnetization of
unstable remanent magnetization by application of
pressure: Geophys. J. pr. Astr. Soc., v. 16, pp. 369-
373, 1968.

Soffel, H., The single domain--mu1ti-domain transition in
natural intermediate titanomagnetites: Zeit. Geophys.,
v. 37, pp. 451-470, 1971.

 

----- , Stress dependence of the domain structure of natural
magnetite: Zeit. GeOphys., v. 32, pp. 63-77, 1966.

 

Stacey, F.D., The physical theory of rock magnetism: Adv.
Phys., v. 12, pp. 45-133, 1963.

46

Stacey, F.D., and Banerjee, S.K., Physical principles of
rock magnetism, Elsivier, New York, 1974.

 

 

Stesky, R., and W.F. Brace, Electrical conductivity of
serpentinized rocks to 6 kilobars: J. Geophys. Res.,
v. 78, pp. 7614-7621, 1973.

Wyllie, P.J., High-pressure techniques: Methods and techniques

 

in geophysics, S.K. Runcorn, ed., Wiley, New York, 1966l

Zijlstra, H., Experimental methods in Magnetism, (2. Measure-
ment of Ma netic quantitIEs), NOrth-HOIIand, -
AmsteFHam, 1967.

 

 

 

 

MICHIGAN STATE UNIVERSITY I

I III III III

31293 3 45 044

ARIES

 

II