The magnetic compass was one of mankind's first high-technology devices. Possession of the compass gave the Islamic world an early edge in navigation and led to the rapid eastward spread, by sea, of their trade, religion and civilization. But man was a comparative latecomer in magnetically aided navigation. Birds, fish, insects, and even bacteria had evolved efficient compasses millions of years earlier.

Magnetic memory, whether of a compass needle, a lava flow, or a computer diskette, is a remarkable physical phenomenon. The magnetic moment is permanent. It requires no expenditure of energy to sustain. Yet it can be partly or completely overprinted with a new signal. Nowhere is this more strikingly demonstrated than in rocks. A single hand sample can record generations of past magnetic events. This family tree can be decoded in the laboratory by stripping away successive layers of the magnetic signal.

Paleomagnetism is the science of reading and interpreting the magnetic signal of rocks. Rock magnetism is more concerned with the writing or recording process. The principles are no different from those of fine-particle magnetism as applied in permanent magnet and magnetic recording technology. But the physical parameters are rather different. Weak magnetic fields are involved, on the order of the present geomagnetic field (0.3–0.6 G or 30–60 μT), much less than the switching fields of the magnetic particles. Temperatures may be high: thermoremanent magnetization of igneous rocks is acquired during cooling from the melt. Times are long, typically millions of years.

Introduction

There are two views of how magnetism originates: microscopic current loops and magnetic dipoles. The latter view is actually the older and is the basis of magnetostatics, the analog of electrostatics. It was discredited when it was demonstrated that magnetic dipoles cannot be separated into isolated + and − magnetic charges (‘monopoles’). Yet the criticism is hardly damning. The component charges of electric dipoles (nuclei and electron clouds, respectively, expressing the polarization of atoms by an electric field) cannot readily be isolated either. Admittedly there is no analog of the free or conduction electron in magnetism, but in many ways magnetically polarized or polarizable materials are close analogs of electrically polarizable materials (dielectrics) and we shall make considerable use of this analogy. Ultimately neither current loops nor charge pairs can explain ferromagnetism. Ferromagnetic moments arise from a non-classical phenomenon, electron spin (§2.3).

A current loop or a dipole produces a magnetic field H or B. What is this magnetic field? H and B are defined not by their cause but by their effect, the forces they exert on physical objects. A magnetic field exerts a torque on a compass needle or pivoted bar magnet (macroscopic dipoles) tending to align their axes with H or B; this is the magnetostatic definition. A magnetic field also exerts a Lorentz force on a moving charged particle, either in free space or when channeled through a conductor as current I, at right angles to both the field and the particle velocity or the current flow; this is the electrodynamic definition.

Introduction

When magnetic minerals cool in a weak field H0 from above their Curie temperatures, they acquire thermoremanent magnetization (TRM) in the direction of H0 (or rarely, in the opposite direction) with an intensity proportional to H0. The fidelity of TRM in recording directions and intensities of ancient geomagnetic fields is the justification for paleomagnetism. TRM is the primary NRM of igneous rocks and some high-grade metamorphic rocks. The NRM of individual detrital grains in sediments and sedimentary rocks is frequently also a TRM inherited from the eroded source rocks.

TRM is much more intense than isothermal remanence (IRM) acquired in the same weak field H0 at room temperature. TRM is also very stable over long periods of time against changes in field (e.g., polarity reversals) or reheating. The reason for this high intensity and stability compared to room-temperature remanence is that TRM is acquired at high temperatures, where energy barriers and coercivities are low, and stabilized by cooling to ordinary temperatures, where barriers and coercivities are high.

TRM is a frozen-in high-temperature equilibrium distribution achieved by thermally excited transitions among different magnetic states (cf. §7.10). Transitions cease below the blocking temperature, TB, because in the course of cooling, the energy barriers Eb between different magnetization states grow larger than the available thermal energy (≈25kT for experimental times of a few minutes; ≈60kT for long geological times). In the last chapter, we discussed transdomain TRM, a partition between different domain structures governed by nucleation or denucleation of domains.

In spite of the generality of the physical principles outlined in previous chapters, the natural remanent magnetization (NRM) of rocks is enormously variable in its intensity, its time stability and its resistance to thermal and AF demagnetization. This chapter and the two that follow will study magnetism in the context of the formation and subsequent alteration of igneous, sedimentary, and metamorphic rocks. In each case, the environment determines which magnetic minerals form, their composition and grain size, and microstructure such as exsolution intergrowths and internal stress.

The oceanic lithosphere and linear magnetic anomalies

Our treatment of igneous rocks begins with the oceanic lithosphere, the largest continuous igneous body available for paleomagnetic study. Linear anomalies in the geomagnetic field over the oceans (‘magnetic stripes’) are replicas of the regular pattern of NRM in the seafloor and provide tangible evidence of seafloor spreading (§1.2.1). In turn, it was the obvious fidelity of the seafloor paleomagnetic record that, in the minds of most earth scientists, proved the validity of the paleomagnetic method as applied to much older continental, and even extraterrestrial, rocks.

The Vine and Matthews' model

Figure 1.4 summarizes the Vine and Matthews' (1963) model. A similar interpretation of magnetic stripes was put forward about the same time by L. J. Morley (see Cox, 1973, p. 224). At a mid-ocean ridge, basaltic magma rising from shallow depths in the upper mantle is extruded as pillow lavas (seismic layer 2A, 0.5–1 km thick), intruded at shallow depths as sheeted dikes and sills (layer 2B, 1–2 km) or cools at depth as massive gabbroic plutons (layer 3, 2–5 km).

Introduction

Chapters 2 and 3 were concerned mainly with material properties like spontaneous magnetization, Curie temperature, magnetocrystalline anisotropy and magnetostriction. If there were no other factors to consider, exchange coupling would cause ferromagnets and ferrimagnets to be magnetized to saturation along a magnetocrystalline or magnetoelastic easy axis throughout their volumes, apart from thermal disordering at high temperatures. Such single-domain (SD) grains do exist, but in most magnetic minerals they are quite small, often < 1µm in size. Larger multidomain (MD) grains spontaneously subdivide themselves into two or more domains (Fig. 1.2). Ms is uniform within each domain (at least on a macroscopic viewing scale) but Ms vectors are in different directions in different domains.

Why do domains exist? Landau and Lifschitz (1935) recognized that the longrange effect of dipole–dipole interactions between atomic moments is to generate a magnetostatic or demagnetizing energy which eventually outweighs the tendency of exchange forces and magnetocrystalline anisotropy to produce a uniform magnetization. Without the guidance of domain observations, Landau and Lifschitz predicted the basic pattern of body and closure domains which were later (in the 1950's) observed experimentally.

Because dipole–dipole interactions are a long-range effect, involving each pair of dipoles in a crystal, magnetostatic calculations can be computationally gruelling. Visualizing the fields involved is not easy, even for quite simple magnetization distributions. Ultimately the goal is to generate magnetization structures which resemble the domains actually observed in large grains by starting from first principles, without imposing initial constraints.

Introduction

Extraterrestrial rocks formed in quite different environments from those on earth. In practically all cases, oxygen fugacity was very low during crystallization and later shock or thermal metamorphism. Iron-nickel alloys, rather than iron–titanium oxides, are the principal magnetic minerals.

Erosion and sedimentation occur to a very limited extent on the other terrestrial planets and their satellites. Lack of a shielding atmosphere has favoured another secondary process, bombardment of surface rocks by meteorites, principally 4000 to 3300 Ma ago.

The global fields of Mercury, Venus and the moon are orders of magnitude less than the present earth's field. The moon does have magnetic anomalies of regional extent, bearing witness to underlying magnetized crust, but none have a lineated or other regular pattern that would suggest plate tectonic processes. Except on Venus (Solomon et al., 1992), there is no surface evidence for vertical or horizontal tectonic movements except those related to volcanism or meteorite impacts.

Of course, crustal spreading from centres of igneous activity on the moon might not leave the distinctive magnetic signature it does on earth. There is no compelling evidence that the moon ever possessed a substantial global field, although comparatively strong fields must have existed locally to produce the observed anomalies, and no evidence at all for reversal of either global or local fields. If a global field existed, it may not have been dipolar. For these reasons, it is doubtful whether paleodirectional and linear magnetic anomaly studies of the terrestrial variety would be successful on the moon, even if conditions permitted such studies.

Introduction

The erosion products of igneous, sedimentary or metamorphic rocks are sources of detrital particles that go to make new sedimentary rocks. The detrital and post-depositional remanent magnetizations (DRM and PDRM) acquired when sediments are deposited and consolidated are no more than a reconstitution of the NRM's of detrital magnetic grains from the source rock (or rocks). Depending on the size and remanence mechanisms of grains in the source rocks, DRM may inherit the temperature and time stability and AF hardness of TRM, CRM, TCRM or VRM of SD, PSD, or MD grains.

DRM and PDRM are inherently weak, often < 10−3 A/m (10−6 emu/cm3). There are two reasons for this low intensity. First, dense oxide grains are less readily transported than silicate grains of similar size. Secondly, DRM represents only a partial realignment of original NRM vectors. With the advent of cryogenic magnetometers (Collinson, 1983), measuring the weak magnetizations of sediments and sedimentary rocks is no longer a major problem.

Red sedimentary rocks – so-called red beds – possess, in addition to depositional remanence, a CRM (10−3–10−1 A/m) carried principally by hematite pigment and cement. The CRM is useful paleomagnetically if the time at which the hematite formed is known.

The great attraction of sedimentary rocks is the comparative continuity of their magnetic record. Major igneous activity and metamorphism occur intermittently, sometimes at long intervals.

Introduction

Most magnetite grains in rocks are much larger than the critical SD size d0 of ≈0.1 μm. Yet these rocks possess a TRM that is both harder and more intense than MD theory predicts. In magnetite there is no abrupt change from SD to MD TRM intensity at any grain size (Fig. 8.4). Instead TRM intensity decreases continuously above d0, reaching MD levels around 10–20 μm. In high-titanium titanomagnetites, the corresponding range is ≈0.5–35 μm (Day, 1977, Fig. 9). This pseudo-single-domain (PSD) (Stacey, 1963) size range incorporates most of the magnetite or titanomagnetite carrying stable TRM in igneous rocks. Therefore it is important that we understand the mechanism of PSD remanence.

The size dependence of TRM is not well documented except in titanomagnetites. However, strong-field remanence parameters like Mrs and Hc vary gradually over broad size ranges in a great many minerals, rather than changing sharply around d0 (Fig. 12.1). In §11.9.4, we saw that most measured values of Mrs/Ms and Hcr/Hc are intermediate between SD and ideal MD values. Pseudo-single-domain behaviour seems to be an intrinsic feature of small MD grains rather than a special property of certain minerals.

The mechanism of PSD behaviour is still far from certain (for reviews of experimental data and theories, see Day, 1977; Dunlop, 1977, 1981, 1986b, 1990; Halgedahl and Fuller, 1983; Fuller, 1984; Halgedahl, 1987). Verhoogen (1959) proposed regions of deflected spins surrounding dislocations, while Stacey (1963) preferred ‘Barkhausen discreteness’ of domain-wall positions.

Introduction

In the last fifteen years, there has been a resurgence of interest in observing domain patterns on naturally occurring minerals, as well as a revolution in ideas about how the observed structures control magnetic properties. The strained surface layer that results when sections are mechanically polished for domain observations by the Bitter method can now be efficiently removed by final polishing with a suspension of amorphous silica microspheres (Hoffmann et al., 1987). The previously used method, ionic polishing, required many hours' exposure to an ion beam and still failed to remove the stressed layer completely. Surface domains can also be observed using the scanning electron microscope (SEM), whose greater depth of field eliminates the need for precision polishing. The interior of domains and crystals can be imaged directly with the transmission electron microscope (TEM) and the magneto-optical Kerr effect (MOKE). The magnetic force microscope (MFM) provides the ultimate in spatial resolution, allowing us to view fine structures, e.g., the structure of domain walls.

Traditional Bitter-pattern observations use a colloidal suspension of ultra-fine magnetite particles to make visible the walls between domains, where flux leakage from the surface is maximum. These observations will be discussed in §6.2–6.6. SEM, TEM, MOKE and MFM observations are described in §6.7–6.9.

The observation of domains in minerals of paleomagnetic interest, in the ≈ 1μm grain sizes that probably carry the strongest and most reliable NRM in rocks, coupled with direct magnetization measurements on the same domains (e.g., Metcalf and Fuller, 1987a), should soon establish a direct link between the magnetic response of a rock sample and the nature of the domains responsible.

Introduction

The theory of domain structures in magnetite and titanomagnetite has made great strides in the last decade. Most earlier calculations used Kittel or Amar models like those of §5.4. These were deterministic; no interesting and novel structures could emerge. Block-like walls and domains without internal structure were imposed in advance. Only their numbers and widths were adjustable. We will review some of these calculations in §7.2.

About the mid-1980's, less constrained micromagnetic calculations (Brown, 1963a) were introduced in rock magnetism. In micromagnetism, model crystals are divided into cells and the magnetization direction of each cell is varied independently until an overall structure emerges that minimizes the total energy. Most such structures are local energy minimum (LEM) states rather than the global energy minimum (GEM) state. That is, the equilibrium number of domains has usually not been found. However, at least domain structure can be expected to emerge in a natural way instead of being imposed at the outset. Fine structures within domains and domain walls also appear without forcing.

In model magnetite crystals larger than ≈0.5 μm, the calculations do converge on structures resembling classic domains. Closure domains and finer structures such as vortices feature in the solutions. In smaller grains, the fine structures fill much of the particle. Domains as such are barely recognizable until the SD range (< 0.1 μm) is reached. These structures are described in §7.3–7.5. Changes in particular LEM states or families of states with changes in applied field H0, grain volume V, and temperature T are the subject of §7.6–7.8.

Introduction

The most important terrestrial magnetic minerals are oxides of iron and titanium. Their compositions are conveniently represented on a Ti4+−Fe2+−Fe3+ ternary diagram (Fig. 3.1). The titanomagnetites (TM for short) are cubic minerals with inverse spinel structure (see §3.2). The mole per cent of Ti4+ is measured by the composition parameter x. TMO (i.e., titanomagnetite with x = 0) is magnetite and TM100 is ulvöspinel. The titanomaghemites are also spinel minerals. They are cation-deficient oxidation products of titanomagnetites. The degree of oxidation is measured by the oxidation parameter z. The titanohematites (often called hemoilmenites after the end-member minerals hematite and ilmenite) are also oxidized equivalents of the titanomagnetites, but their crystal structure is rhombohedral. Notice that minerals of the same composition but different structures occupy the same point on the ternary diagram. For example, maghemite (cubic or γFe2O3) and hematite (rhombohedral or αFe2O3) both plot in the lower right corner.

The titanomagnetites form a complete solid-solution series for all values of x at very high temperatures (see Lindsley (1976, 1991) for an equilibrium phase diagram) but intermediate compositions can only be preserved as single-phase minerals at ordinary temperatures if they cooled very rapidly. Submarine pillow lavas, for example, have been quenched by extrusion into seawater. Their primary magnetic oxides are single-phase TM60 grains. If the same basaltic magma cools more slowly in an oxygen-poor setting, the primary oxide will not be a single-phase mineral but an exsolution intergrowth of low-x (near-magnetite) and high-x (near ulvöspinel) cubic minerals.

Magnetism has fascinated mankind since the invention of compasses that could track invisible magnetic field lines over the earth's surface. Much later came the discovery that rocks can fossilize a record of ancient magnetic fields. Unravelling this record – the ‘archeology’ of magnetism – is the science of paleomagnetism, and understanding how the microscopic fossil ‘compasses’ in rocks behave has come to be known as rock magnetism.

Rock magnetism is both a basic and an applied science. Its fundamentals concern ferromagnetism and magnetic domains and were developed most authoritatively by Néel. Its applications continue to expand, giving impetus to new research into the mechanisms and fidelity of rock magnetic recording. Some of the history and applications are described in this chapter.

A brief history

Earth magnetism

Compasses were used in China and the Arab world for centuries before Petrus Peregrinus in 1269 gave the first European description of a working compass. The earliest compasses were lodestones, naturally occurring ores of magnetite (Fe3O4). Particular areas, or poles, of one lodestone would attract or repel the poles of another lodestone. This magnetic polarization is the key to their use as compasses in navigation. A suspended lodestone will rotate until its axis of magnetization or polarization, joining north and south poles of the lodestone, lines up with imaginary field lines joining the north and south geomagnetic poles. In modern terminology, the magnetization M aligns with the field H.

Introduction

All magnetizations are produced by an applied magnetic field, but certain magnetization processes, for example isothermal remanent magnetization (IRM) and alternating-field (AF) demagnetization, are field-driven in a more restricted sense. They are produced isothermally, usually at or near T0, over a time scale of at most a few minutes. They result, in other words, from the sole influence of an applied field. Even this definition is not entirely accurate: τ0 is so much less than ordinary measurement times that thermal fluctuations play some role in practically all field-induced processes. For weak applied fields and small V, Hq or Hf are > H0, and thermal excitation plays a major role (§8.7, 9.4.2).

Isothermal remanences do not carry useful paleomagnetic information. An IRM produced by the geomagnetic field is easily reset by later weak fields of similar magnitude: it lacks paleomagnetic stability. Only isothermal remanences due to strong fields, for instance saturation isothermal remanence (SIRM) and anhysteretic remanence (ARM) produced by the combination of a steady field and a strong but decaying AF, have the requisite stability, and these occur in nature only when outcrops have been struck by lightning and remagnetized. However, the stepwise acquisition or removal of magnetizations in the laboratory is used as a means of erasing NRM's of low stability (AF demagnetization) and of determining the composition and domain structure of mineral magnetic carriers (hysteresis, IRM acquisition, ‘DC demagnetization’, Preisach analysis).

Introduction

Crystallization remanent magnetization (CRM) results from the formation of a new magnetic mineral in the presence of a magnetic field, either by nucleation and growth to a stable blocking volume VB (single-phase or growth CRM) or through alteration of an existing magnetic phase (two-phase or parent-daughter CRM). The commonly used term chemical remanent magnetization is not always strictly accurate, e.g., in the γFe2O3 → αFe2O3 (spinel → rhombohedral) transformation, where no chemical change occurs, only a restacking of the lattice.

CRM is usually thought of as being blocked when grains grow from superparamagnetic (SP) to thermally stable SD size at V = VB. This simple picture breaks down in the case of two-phase CRM because the growing daughter phase is influenced not only by an external field H0 but also by its magnetic parent phase, to which it may be magnetostatically or exchange coupled with varying degrees of efficiency. Exchange coupling can occasionally result in self-reversed CRM (§13.4.6, 14.5.3).

CRM is usually regarded as a contaminant by paleomagnetists because the time of secondary mineral formation is difficult to date. Unfortunately CRM is not always easy to recognize because its unblocking temperatures and coercivities overlap those of primary TRM and DRM.

Many processes generate growth CRM in low-temperature sedimentary environments: precipitation of hematite cement from iron-rich solutions in red beds (Larson et al., 1982); microbially mediated production of authigenic magnetite in the iron-reducing zone of recent marine sediments (Karlin et al., 1987); biogenic magnetite production in calcareous sediments that eventually form limestones (Chang et al., 1987); and inorganic authigenesis of magnetite in soils (Maher, 1986; Maher and Taylor, 1988).