Characteristic magnetic behavior of subaerial and submarine lava units from the Hawaiian Scientific Drilling Project (HSDP-2)



[1] This study presents rock magnetic properties and the magnetic mineralogy of subaerial and submarine lava flows of Mauna Loa and Mauna Kea volcanoes collected from the 3109 m deep HSDP-2 drill hole in Hawaii. Three different groups of magnetic behavior are recognized in the subaerial lava flows related to the degree of high temperature oxidation during extrusion. Group 1 shows homogenous titanomagnetite with low Xmt, low Curie temperatures (TC: 100°–200°C) and weak median demagnetizing fields (< 20 mT). Further subdivision into 1a and 1b subgroups is based on the low temperature behavior of magnetic susceptibility (MS) and hysteresis loops, which indicate a contribution from ferrimagnetic Cr-Al spinel below ca. −160°C in the 1b-type samples. Group 2 samples, with exsolution lamellae of ilmenite in the titanomagnetites, have higher TC (480°–580°C) and higher coercive forces (20–40 mT). Group 3, the highest oxidation stage, is characterized by titanohematite-bearing assemblages with enhanced median demagnetizing fields (35–85 mT) and a significantly different low-temperature MS behavior. MS core logging shows a systematic variation occurs in the subaerial lava flows, directly related to the degree of high temperature oxidation and their flow morphology. Aa lava flows have higher mean MS than other lava flow types. Besides these factors, MS appears to be also affected by the magma composition of the various shield-building stages. Mauna Loa subaerial lava flows generally show lower mean susceptibilities (4.6 ± 3 × 10−3 SI) than subaerial Mauna Kea lava flows (9.8 ± 5 × 10−3 SI). As submarine lava flows show no group 3 assemblages no high temperature oxidation influenced these rocks. Some hyaloclastites and pillow breccias show low MS (< 1 × 10−3 SI), small amounts of nearly pure magnetite (TC = 580°C) and high coercive forces up to 110 mT suggesting single domain and/or superparamagnetic behavior. The controlling mechanism of the magnetic properties in the submarine lava units is the cooling and quenching rate of lava flows, which creates large grain size variations in titanomagnetites of varying compositions. Hydrothermal alteration, as described from ocean floor or Icelandic basalts, is not an important process that influences the magnetic properties in the ocean island basalts from the HSDP-2 drill hole.

1. Introduction

[2] Hawaii, as the “archetype of ocean island volcanism” [Stolper et al., 1996] has been selected for an international scientific project, with the aim to enhance the general knowledge concerning the origin and development of hot spot volcanism. Within the framework of the Hawaii Scientific Drilling Project (HSDP), which is part of the International Continental Scientific Drilling Project (ICDP), two boreholes have been drilled into the flank of the Mauna Kea volcano, near Hilo on the Island of Hawaii. In 1993, a 1000 m deep well (KP-1) was drilled as a pilot hole, followed in 1999 by the 3109 m deep main borehole (HSDP-2, Figure 1). In 2003, the HSDP-2 hole will be drilled further to about 4500 m.

Figure 1.

Evolving stages of a typical Hawaiian shield volcano: (a) submarine/seamount with pillows, intrusives and sheet flows; (b) shallow water environment with the formation of hyaloclastites, pillows and massive basalts; (c) main subaerial shield building with eruption of pahoehoe and aa lava, major landslides and formation of massive submarine units as tube-fed flows. The latter extends from subaerial to submarine conditions and includes feeders for submarine flank eruptions [modified after Moore and Fiske, 1969, and DePaolo et al., 2001]; (d) lithologial profile of the HSDP-2 drill hole [from DePaolo et al., 1999] with magnetometer-borehole measurements from Steveling et al. (submitted manuscript, 2002).

[3] The HSDP provides an excellent opportunity to study the rock magnetic properties of the volcanic rock sequence in relation to the evolution of a hot spot related shield volcano. Due to the 95% recovery of core material, a nearly continuous profile through the subaerial and submarine lava flows is available over the last 600 ky. This allows detailed investigations on primary and secondary processes in relation to the long-time evolution of the volcano that controls the today's magnetic mineralogy.

[4] Borehole magnetometer measurements in the HSDP-2 drill hole revealed significant variations in the total magnetic field of subaerial and submarine lava flows within the measured interval from 600 to 1800 m (E. Steveling, J. B. Stoll, and M. Leven, Quasi-continuous depth profiles of rock magnetization from magnetic logs in the HSDP-2 borehole, Island of Hawaii, manuscript submitted to Geochemistry, Geophysics, Geosystems, 2002, hereinafter referred to as Steveling et al., submitted manuscript, 2002), which include the subaerial to submarine transition (Figure 1d). In the magnetic logs, this transition is marked by a distinct change from strongly disturbed (subaerial flows), to less disturbed sections (submarine flows) forming closely spaced magnetic anomalies.

[5] The magnetic behavior and Fe-Ti oxide petrology of subaerial and submarine basalts have been extensively studied over the last decades because of their importance in understanding Earth's magnetic anomalies. Combined magnetic and mineralogic studies from Grommé et al. [1969], on tholeiitic basalts sampled from the Alae and Makaopuhi subaerial lava lakes on Kilauea volcano, showed a complex behavior related to different Fe-Ti oxides. The studied material was sampled from in situ temperatures of 50° and 1020°C, and contained abundant hemoilmenite. Samples quenched from high temperatures (800° to 1000°C) displayed additionally a second component identified as unoxidized, homogeneous titanomagnetite. In contrast, samples taken from lower temperature conditions (50° to 400°–700°C) revealed oxidation of the originally Ti-rich titanomagnetite to Ti-poor titanomagnetite with ilmenite lamellae. Grommé et al. [1969] concluded that these titanomagnetites formed by subsolidus reactions at temperatures below their Curie temperature. The high-temperature or deuteric oxidation causes the production of Ti-poor titanomagnetite – ilmenite intergrowth, evolving to mixtures of hematite, pseudobrookite and rutile when highly oxidized.

[6] Titanomagnetites may experience different types of alteration ranging from high-temperature deuteric oxidation (> 500°–600°C), through hydrothermal alteration (200°–400°C), to low-temperature oxidation (< 100°C). Such conditions lead to modifications of the primary titanomagnetite composition and texture. The primary magnetic mineral in basalts of the seafloor is titanomagnetite with an ulvöspinel component of 60% [e.g., Bleil and Petersen, 1977]. Alteration processes occur in subaerial as well as submarine basalts, and changes due to fluids are usually related to the strength of hydrothermal activity. Regional hydrothermal alteration, taking place after burial by younger flows, has been reported from Icelandic basalt lavas [e.g., Ade-Hall et al., 1971] and from seafloor hydrothermal sites at oceanic ridges [e.g., Wooldridge et al., 1990]. Magnetic properties vary significantly with mineralogical changes caused by hydrothermal fluids [e.g., Pariso et al., 1995; Wooldridge et al., 1990]. Cooling rates and crystallization temperatures have been shown to effect magnetic properties, especially in submarine basaltic glasses [Pick and Tauxe, 1994] or in basaltic pillow lava [Zhou et al., 2000].

[7] In this study, we adopt a combination of rock magnetic (different kinds of low-field susceptibility and remanence measurements) and mineralogical methods (Fe-Ti oxide textures and composition) to characterize and classify the magnetic carriers within the subaerial and submarine lava flows of the HSDP-2. The magnetic behavior of titanomagnetite is discussed in terms of magma composition, viscosity, crystallization temperatures and alteration processes. Whereas the magnetic susceptibility data correlate very well with the magnetic anomalies measured in the drill hole (Steveling et al., submitted manuscript, 2002), our findings explain the different types of magnetic anomalies that are observed within the different levels of the Hawaiian shield volcano. This study documents that in the ocean-island setting of the Hawaiian volcanoes, primary textures and compositions are preserved, especially in submarine units, due to weak and localized alteration of the upper part of the shield volcano [dePaolo et al., 2001]. A remarkably low geothermal gradient (8°C/km between 600 and 1800 m, 18°C/km below 1800 m [Dannowski, 2002]) was measured from continuous monitoring of temperatures along the drill hole. This situation is clearly different to active plate margin settings such as typical ocean floor or spreading ridges (e.g., Iceland), which often show high geothermal gradients along with strong hydrothermal alteration.

2. Development of the Hawaiian Shield Volcano and the Lithoprofile of HSDP-2

[8] The submarine growth of a volcano proceeds with the formation and accumulation of pillow basalts and sheet flows, accompanied by the intrusion of dike systems (Figures 1a and 1b). During its development the submarine volcano breaches sea level and the shield volcano grows further by eruption of subaerial aa and pahoehoe lava flows (Figure 1c). Various conditions of lava extrusion and the complex geochemical fractionation processes that occur during the volcano evolution result in different magnetic signatures.

[9] The HSDP-2 drill hole recovered cores from a 3109 m deep submarine and subaerial section which allows study of a third of the 10 km thick Hawaiian shield volcano. From the surface to 246 mbsl, subaerial lava flows of the Mauna Loa volcano, and below that depth until 1079 mbsl, subaerial lava flows of the Mauna Kea occur (Figure 1d). Aa and pahoehoe flows are of equal abundance in the Mauna Loa core section and have a tholeiitic geochemical character. The uppermost 50 m of the underlying Mauna Kea section contains alkaline flows marking the end of the shield-building volcanic cycle [DePaolo et al., 2001]. About 75% of the Mauna Kea flows are aa lava and belong to the tholeiitic serie.

[10] The upper part of the submarine Mauna Kea volcano (1079–1984 mbsl) mainly consists of hyaloclastite debris flows interleaved with about 10% massive submarine basalt flows with an average thickness of 3–4 m. Most of the massive basalts have been interpreted as subaerial flows that continued past the shoreline to produce submarine flows [dePaolo et al., 2001]. The hyaloclastites of this section are interpreted as debris flows from oversteepened, nearshore deposits containing glass and lithic clasts, formed when the subaerial lavas quenched and fragmented on entering the water. The submarine Mauna Kea between 1984 and 3098 mbsl is composed dominantly of pillow lavas (60%) with less abundant intercalated volcaniclastic sediments and few (about 4%) intrusive units.

3. Methods

[11] Magnetic susceptibility (MS) measurements have been performed on core slabs (diameter 6 cm) of the HSDP-2 cores which are stored at the American Museum of Natural History, N.Y. We used a hand-held kappameter (KT-5) from Geofyzika Brno that determines the susceptibility with a sensitivity of 1 × 10−5 SI based on frequency changes of the operating coil (10 kHz). The distance between single measurements was ca. 15 cm. Core slab geometry allows integration over a total volume of ca. 60 cm3. Approximately 5000 individual measurements were undertaken on a total of 22 different sections of the cores, a list of which is provided in the appendix. All MS values in this work are given in 10−3 SI units. The auxiliary material supports a data table behind Figure 2 and Figure 3. Based on these profiles representative samples were selected which reflect the observed MS variations. From these samples, plugs (2.5 cm in diameter, 2.1 cm high) were prepared for investigation by different analytical techniques (standard cylinders, polished sections for optical and scanning electron microscopy as well as electron microprobe analyses and powder for thermomagnetic experiments). The significance of routine MS measurements using a hand-held kappameter were tested by comparison with laboratory measurements using a kappabridge (KLY-2, sensitivity 8 × 10−8). This test indicates a linear relationship between both these methods with a high correlation coefficient (R2 = 0.96) for samples with k > 2 × 10−6 SI and a general shift toward lower values for the kappameter values (y = 0.63x + 0.66). On the basis of this correlation, variations in MS monitored during the core logging can be directly related to laboratory measurements.

Figure 2.

Frequency distribution of MS from (a) subaerial units of Mauna Loa (ML), (b) subaerial units of Mauna Kea (MK), (c) massive, pillow and intrusion units from the submarine part of MK and, (d) hyaloclastites from the submarine units of MK. Insert shows the MS-distribution of paramagnetic-dominated hyaloclastites in more detail. Colors relate to lithologies presented in Figure 1d.

Figure 3.

Examples of MS logs (core measurements, 15 cm resolution) for subaerial and submarine units of Mauna Loa and Mauna Kea. Variations in MS are observed between the individual lithologies (e.g., aa, pahoehoe lava and hyaloclastite, massive lava) as well as within single flows. Unit boundaries are based on the HSDP-2 core description [DePaolo et al., 1999]. Legend for lithologies, see Figure 1d.

[12] The temperature dependency of MS was measured over the temperature range of −192 to 700°C using the KLY-2 Kappabridge, fitted with the CS-2/CS-L furnace apparatus of AGICO [Hrouda, 1994]. During the cooling run, temperatures and susceptibilities were recorded as the sample warmed from −192° to 0°C. Heating/cooling cycles from room temperature to 700°C (heating rate 10°/min) were performed in an argon atmosphere flow (110 mL/min) in order to avoid mineral reactions with oxygen during the heating process. The raw data were corrected for the empty furnaces and normalized to the susceptibility magnitude at 0°C (low-temperature) and room temperature (high-temperature). For a few selected samples, the low temperature susceptibility was measured for the temperature interval from −263° to 27°C using a superconducting susceptometer (MPMS-XL) from Quantum Design at the Institute of Rock Magnetism (IRM) in Minneapolis.

[13] Hysteresis parameters (Jr, Js and Hc, Hcr) were determined from measurements at room temperature using a vibrating sample magnetometer at the IRM and at the Paleomagnetic Laboratory at Niederlippach. Low-temperature hysteresis loops were measured with an alternating gradient force magnetometer (MicroMag) from Princeton Measurements at the IRM. The measurements were performed in a 0.5 T field during heating up to 300°K, in increments of 10°K. Jr, Js and Hc are determined from the hysteresis loop.

[14] Measurements of remanent magnetization were made with a JR5A spinner magnetometer (AGICO). For stability tests, alternating field demagnetization (AF demagnetization) was performed in peak fields up to 160 mT with a MI AFD 1.1 from Magnon International.

[15] Oxide textures were characterized using reflected light and scanning electron microscopy and by adopting the high-temperature oxidation classification system for titanomagnetites (C1–C7 stage) of Haggerty [1976, 1991]. The ferrofluid method [e.g., Soffel, 1991] was used to discriminate magnetic and non-magnetic Fe-Ti oxides. Mineral chemical data for the Fe-Ti oxides were obtained by electron microprobe (CAMECA SX51) at the Institute of Mineralogy in Heidelberg. Standards used were periclase (Mg), Al2O3 (Al), wollastonite (Si), TiO2 (Ti), Cr2O3 (Cr), rhodonite (Mn), and hematite (Fe). The raw data were corrected with the PAP algorithm of Pouchon and Pichoir [1984]. An acceleration voltage of 15 kV and a sample current of 20 nA were used.

4. Results

4.1. Magnetic Susceptibility Variations Within the HSDP-2 Profile

[16] The frequency distribution of MS for the different levels of the volcanic sequence reveals distinct differences in their mean values and distribution (Figure 2 and Table 1). Subaerial as well as submarine units of the Mauna Kea volcano (Figures 2b and 2c) show near Gaussian distributions for the data with a slight tail toward higher values. However, there are significant differences in the mean susceptibility, when comparing the subaerial tholeiitic flows of the Mauna Loa (MS = 4.6 ± 3.0) and the Mauna Kea (MS = 9.8 ± 4.7). The F test showed that the scatter in MS for subaerial flows of these volcanoes differs significantly, indicating two different populations. Two alkaline lava flows from the Mauna Kea volcano (unit 42 and 48) show even higher MS values (MS = 20.8 ± 10.4). All the different lava types (pahoehoe, aa, transitional lava) of the Mauna Kea yield higher susceptibilities than the Mauna Loa lava types. In general, the subaerial aa and transitional lava flows erupted from the volcanoes have higher susceptibilities than the pahoehoe-type flows. This is especially true for aa flows which display higher values and create tails in the frequency distributions as shown in Figures 2a and 2b.

Table 1. Magnetic Susceptibility Characteristics for Different Lava Types of the HSDP-2a
 Mauna Loa (Subaerial)Mauna Kea (Subaerial) Tholeiitic/Alkaline
  • a

    MS measurements for two alkaline flows of the Mauna Kea volcano are also presented (unit 42, transitional lava, and unit 48, aa lava). pm, paramagnetic; fm, ferrimagnetic.

MS min.
MS max.6.028.512.431.395.5/60.57.5/23.7
mean4.6 ± 3.09.8 ± 4.7/20.8 ± 10.4
 Mauna Kea (Submarine)
MassivePillowIntrusive, < 2900 mIntrusive, > 2900 mHyaloclastite, pm-DominatedHyaloclastite, fm-Dominated
MS min.
MS max.23.622.89.429.91.019.7
Mean 4.7 ± 2.6    

[17] Additionally, the susceptibility distribution within subaerial lava shows large variations. Selected examples of susceptibility profiles are presented in Figure 3. These profiles show that some lava flows have only a narrow range of susceptibility variations (e.g., unit 0046), while others display a very strong scattering (e.g., unit 0032). Often, a significant decrease in susceptibility toward the top and bottom of individual lava flows can be observed (e.g., unit 0048 and 0049).

[18] The submarine units of Mauna Kea can be generally subdivided into hyaloclastites and massive lava flows (including pillows and intrusives), where the latter units show a similar distribution range of susceptibilities (Table 1 and Figure 2c), as for the subaerial lava flows from Mauna Loa. Pillow and massive units yield mean susceptibilities of 4.8 ± 3.7 and 4.1 ± 2.2, respectively. The intrusive units cover a broad range of susceptibilities. Down to 2900 mbsl intrusive units have mean susceptibilities of 5 ± 1.9, coinciding with the susceptibility values of the massive and pillow lava units. Intrusives below 2900 mbsl produced significantly higher values.

[19] According to bulk rock susceptibility and temperature dependency, hyaloclastites can be subdivided into a paramagnetic-dominated type 1 (MS < 1, with a maximum at 0.5; see insert in Figure 2d) and a ferromagnetic-dominated type 2 (MS > 1; Table 1). Susceptibilities of type 1 are dominated by the iron content of the glass and the paramagnetic olivine and antiferromagnetic Cr-Al spinel that form as early crystallization products. According to temperature-dependent susceptibility measurements, a minor ferrimagnetic contribution may occur (see below). Magnetic susceptibilities are always found to be below 0.5 when sedimentary structures are present (e.g., bedding and grading) with no or only few small clasts (< 1 cm). Strong variations of MS in type 2 hyaloclastites are mostly related to clasts of basalt that occur within a glassy matrix.

[20] Examples showing the susceptibility distribution of profile sections typical for the submarine part of the HSDP-2 are shown in Figure 3. Massive units, which are intercalated with hyaloclastite units, mostly show constant susceptibility values except for a distinct decrease at the flow margins (e.g., unit 0224, Figure 3). The small variation indicates a homogeneous susceptibility distribution within these flows. Pillow units, however, display similarly strong variations in susceptibility as the ferrimagnetic hyaloclastites. Higher values are related to the interior and low values to the pillow margins. Massive pillow units (e.g., unit 0291, Figure 3) produce higher susceptibilities than the pillow units that are composed of dm-sized pillows (e.g., unit 0340e).

[21] Overall, the MS measurements of the HSDP-2 cores reveals that individual lava flows and lithologic boundaries can be well recognized by variations in their magnetic properties, especially in the submarine portion of the drill core. Moreover, changes in MS across single flows can be observed. In all units, except the hyaloclastites that show dominantly paramagnetic behavior, the MS indicates ferrimagnetic minerals control the magnetic behavior of the rocks, and can be attributed to the presence of titanomagnetite in these rock types. However, as it is seen by the wide variation in MS, the magneto-mineralogy is complex and the characteristics of the different units are presented in the following.

4.2. Temperature-Dependent Susceptibility and Magnetic Mineralogy in Subaerial Units

[22] Based on changes in MS during cooling and heating, three different groups (1 to 3) of magneto-mineralogical behavior can be distinguished within the subaerial units of the Mauna Loa and Mauna Kea. These groups all show characteristic mineral textures, compositions (Table 2) and remanence behavior (Table 3).

Table 2. Chemical Composition of Titanomagnetite in wt% Oxide (n = Number of Analyses) From Subaerial and Submarine Lava Units (Sample Localities See Figure 3)a
  • a

    Formula is calculated assuming stoichiometry (3 cations, 4 oxygen). Magnetite (Xmt) and hematite (Xhm) component are calculated according to Stormer [1983]. Xmt (calc) is the magnetite component calculated from measured Curie temperature (TC (meas)).

  • b


SampleSR0132-5.3SR0102-1.3  SR0101-7.0SR0131-4.4SR0101-3.0SR0101-3.0 SR0127-0.4SR0623-1.7SR0799-16.0SR0956-15.5SR0957-3.9
Lithologyaaaa  aaaatransitionaltransitional transitionalhyaloclastitemassive pillowintrusiveintrusive
Texture in ol-rimin matrixil-mt-xtal mt-grainrim around cspcruciformgirlandes  il-mt-grainil-mt-grainil-mt-grain
SiO20.07 (0.03)0.10 (0.02)0.10 (0.03)0.14 (0.04)0.08 (0.04)0.12 (0.09)0.06 (0.01)0.12 (0.04)0.28 (0.11)0.07 (0.02)0.17 (0.11)0.39 (0.82)0.13 (0.05)2.11 (1.69)
TiO222.85 (0.66)23.59 (1.10)22.16 (0.35)20.05 (0.07)20.42 (1.52)16.35 (1.55)18.35 (2.53)21.48 (3.59)15.50 (0.18)8.00 (2.25)13.11 (2.32)20.18 (1.61)18.34 (0.77)15.89 (1.90)
Al2O31.67 (0.10)2.31 (0.87)1.71 (0.17)1.95 (0.13)1.78 (0.68)1.75 (0.13)3.88 (0.85)1.61 (0.16)0.95 (0.05)2.56 (0.16)3.30 (0.96)2.13 (0.37)1.90 (0.32)2.39 (0.53)
Cr2O30.13 (0.03)0.16 (0.15)0.11 (0.07)0.13 (0.08)1.14 (1.51)0.14 (0.04)14.19 (7.39)0.01 (0.01)0.02 (0.03)0.01 (0.02)3.05 (7.15)0.02 (0.03)0.02 (0.02)0.01 (0.02)
Fe2O323.00 (1.09)19.09 (2.94)22.40 (0.47)26.13 (0.35)24.05 (4.13)35.12 (2.73)13.56 (2.97)22.86 (7.38)35.03 (0.47)84.46 (4.71)36.24 (3.61)26.24 (2.25)30.52 (1.29)29.87 (5.11)
FeO46.10 (0.68)47.67 (1.97)48.04 (0.49)46.75 (0.36)45.60 (0.71)38.83 (0.91)42.84 (2.54)48.56 (3.12)44.02 (0.05)5.79 (1.80)37.15 (2.97)48.15 (1.22)46.04 (0.85)45.79 (1.86)
MnO0.62 (0.05)0.38 (0.02)0.38 (0.04)0.28 (0.05)0.47 (0.15)0.54 (0.09)0.28 (0.08)0.35 (0.12)0.30 (0.01)0.21 (0.09)0.29 (0.04)0.47 (0.10)0.46 (0.08)0.51 (0.13)
MgO3.22 (0.33)2.55 (1.44)1.49 (0.07)1.11 (0.07)1.78 (0.90)3.87 (0.40)3.01 (0.54)0.51 (0.11)0.20 (0.03)0.61 (0.07)3.45 (0.77)0.82 (0.47)0.81 (0.22)0.48 (0.34)
Total97.66 (0.40)95.85 (1.08)96.38 (0.61)96.54 (0.41)95.34 (1.22)96.73 (0.38)96.18 (1.1695.50 (0.57)96.30 (0.34)101.70 (1.08)97.04 (0.46)98.40 (0.85)98.21 (1.18)97.06 (1.19)
Si0.00 (0.00)0.00 (0.00)0.00 (0.00)0.01 (0.00)0.00 (0.00)0.00 (0.00)0.00 (0.00)0.00 (0.00)0.01 (0.00)0.00 (0.00)0.01 (0.00)0.01 (0.03)0.00 (0.00)0.08 (0.06)
Ti0.64 (0.02)0.67 (0.03)0.64 (0.01)0.58 (0.00)0.59 (0.04)0.46 (0.04)0.52 (0.08)0.63 (0.11)0.45 (0.01)0.15 (0.04)0.37 (0.07)0.57 (0.04)0.52 (0.02)0.45 (0.06)
Al0.07 (0.01)0.10 (0.04)0.08 (0.01)0.09 (0.01)0.08 (0.03)0.08 (0.01)0.17 (0.03)0.07 (0.01)0.04 (0.00)0.08 (0.00)0.14 (0.04)0.09 (0.02)0.08 (0.01)0.11 (0.02)
Cr0.00 (0.00)0.00 (0.01)0.00 (0.00)0.00 (0.00)0.03 (0.05)0.00 (0.00)0.42 (0.21)0.00 (0.00)0.00 (0.00)0.00 (0.00)0.09 (0.20)0.00 (0.00)0.00 (0.00)0.00 (0.00)
Fe30.64 (0.03)0.54 (0.08)0.64 (0.01)0.75 (0.01)0.70 (0.13)0.99 (0.08)0.38 (0.09)0.67 (0.22)1.03 (0.01)1.61 (0.08)1.02 (0.11)0.74 (0.07)0.87 (0.04)0.85 (0.15)
Fe21.43 (0.02)1.51 (0.07)1.53 (0.01)1.49 (0.01)1.47 (0.04)1.22 (0.03)1.34 (0.10)1.57 (0.10)1.43 (0.01)0.12 (0.04)1.16 (0.10)1.51 (0.04)1.45 (0.01)1.44 (0.06)
Mn0.02 (0.00)0.01 (0.00)0.01 (0.00)0.01 (0.00)0.02 (0.01)0.02 (0.00)0.01 (0.00)0.01 (0.00)0.01 (0.00)0.00 (0.00)0.01 (0.00)0.02 (0.00)0.01 (0.00)0.02 (0.00)
Mg0.18 (0.02)0.14 (0.08)0.08 (0.00)0.06 (0.00)0.10 (0.05)0.22 (0.02)0.17 (0.03)0.03 (0.01)0.01 (0.00)0.02 (0.00)0.19 (0.04)0.05 (0.03)0.05 (0.01)0.03 (0.02)
Xmt0.34 (0.02)0.27 (0.05)0.32 (0.01)0.38 (0.00)0.35 (0.07)0.54 (0.04)0.15 (0.05)0.33 (0.11)0.52 0.81 (0.04)b0.57 (0.02)0.38 (0.04)0.44 (0.02)0.46 (0.07)
Xmt (calc)0.35 0.48 0.98/10.85/0.96 0.92/1 0.88b0.610.390.490.58
TC (meas)102 200 568/584467/554 520/590 569295/526135/580204273
Table 3. Rock Magnetic Data From Subaerial and Submarine Samples of the HSDP-2a
SampleLithologyUnitDepth, mbslNRM, A/mMS (30 A/m), 10−3 SIQTC, °CgroupMDF, mTMres, %
  • a

    NRM, natural remanent magnetization; MS (30 A/m), magnetic susceptibility measured in a field of 30 A/m (x = no standard cylinder available); Q, Königsberger ratio; TC, Curie temperature; group, sample grouping due to temperature dependent susceptibility behavior; MDF, median demagnetizing field; Mres, residual magnetization after demagnetization in maximum field of 160 mT (* M at 70 mT, those samples were provided by J. Kirschvink and have a smaller sample volume of 1.7 cm3).

SR0102-1.3aa32190.0-10.4-−165/200 (470)1b1812*
SR0104-4.7aa32197.3-3.9-(−105) 2101b4934*
SR0104-6.5aa32197.9-4.6-(−105) 2451b3222*
SR0106-1.8aa35202.11.32.415.8(250) 510/570211037
SR0132-5.3aa48277.25.850.43.3102 (580)1a142
SR0443-6.5aa, baked1661066.4xxx5853xx
SR0454-4.8massive basalt1791095.−155/164/4001b/2278
SR0456-1.2massive basalt1791099.−160/155 (470/590)1b185
SR0623-1.7hyaloclastite2181582.214.519.221.5(10?) 295/5261a/27712
SR0626-11.0massive basalt2211594.4-4.9-−165/2701b/3244
SR0630-4.2massive basalt2241604.06.26.626.7−165/2861b173
SR0799-16.0massive pillow2912315.38.928.28.9135 (580)1a61
SR0800-0.9massive pillow2912317.22.926.73.1110 (213)1a197
SR0800-1.0massive pillow2912317.37.519.111.1−155/265/4701b92
SR0800-6.2massive pillow2912318.−175/215/4801b124
SR0800-19.8massive pillow2912323.−160/240/5051b134
SR0801-13.4massive pillow2912327.66.412.614.7−175/180 (505)1b133
SR0801-15.1pillow breccia2922328.218.03.2159.6para/(290) 57546114
SR0957-3.9intrusive341b3020.313.325.413.5−145/273 (480)1b81

[23] Examples of representative heating-cooling curves for the subdivided groups 1 to 3 are presented in Figure 4. Group 1 samples display Curie temperatures between 100° and 200°C and reversible heating and cooling legs. A further subdivision into group 1a and 1b is based on different low-temperature behavior. Group 1a samples show an increase in susceptibility with increasing temperature until the Curie point between 100–200°C is reached. Group 1b shows an initial decrease in susceptibility followed by an increase with increasing temperature above about −160°C.

Figure 4.

Typical curves showing the temperature-dependence of MS and Curie temperatures from subaerial lava flows (TC). Determination of TC from the heating curve (solid line) is described in Grommé et al. [1969] and shown in Figure 4b. (a) group 1a, (b) group 1b, (c) and (d) group 2, (e) and (f) group 3. The green curves in Figures 4a, 4c, and 4e presents MS measurements in the temperature interval −263°–27°C using a superconductiong susceptometer. Further explanations are given in the text.

[24] Optical observations (Figure 5) and microprobe analyses (Table 2) of group 1 samples indicate the presence of homogeneous titanomagnetite, and primary ilmenite either as two-phase grains or single grains in the matrix of the basalt (C1 stage). Grain shapes are either idio- to hypidiomorphic with grain sizes up to about 30 μm for group 1a samples (Figure 5a). Group 1b samples often show similar textures as group 1a samples but also contain euhedral dendritic, cruciform and girlande-like quenching structures, producing a wide grain size variation. Some grains already show ilmenite lamellae, suggesting the beginning of the C2 oxidation stage (Figure 5b). These samples additionally contain Cr-Al spinels, either as inclusions in olivine or as xenocrysts in the matrix, and sometimes with Cr-bearing titanomagnetite rims. Titanomagnetites of both groups 1a and 1b have a low magnetite component (Xmt) between 0.27 and 0.38 (Table 2) which is within the limits (0.25 < Xmt < 0.55) of typically extruded basalt [e.g., Creer et al., 1976]. However, Fe-Ti oxides with different textures display different chemical composition. Titanomagnetite formed in pyroxene rims around olivine have the lowest Xmt and grains with ilmenite-titanomagnetite intergrowth in the matrix show the highest Xmt (e.g., sample SR0102-1.3). The total oxide percent determined by our microprobe analysis, especially in the group 1b sample, is less than 98 wt%, which may indicate some degree of cation-deficiency within the titanomagnetite crystals. The discrepancy between calculated Curie temperature assuming a linear relationship between chemical composition and Curie temperature [e.g., Dunlop and Özdemir, 1997], and the measured Curie temperature, using low-field MS in group 1b samples (Table 2) support such an interpretation.

Figure 5.

BSE-images of Fe-Ti oxide textures from subaerially extruded lava samples. (a) homogeneous titanomagnetite (tm) with ilmenite (il) lamellae (group 1a), (b) tm with il-lamellae, representing the beginning of C2 oxidation (group 1b), (c) homogeneous hypideomorhic tm and il grains (group 2), (d) tm with fine network of il-lamellae, and homogeneous il-crystals (group 2), (e) skeletal to lath-like il, and hematite (hm) with il-lamellae (group 3), (f) exsolved tm with different generations of il-lamellae and indistinct mottling, hm-near phase at grain boundaries (possibly titanomaghemite; group 3).

[25] Group 2 samples show significantly higher Curie temperatures in the range between 480 and 590°C indicating magnetite-rich titanomagnetite or titanomaghemite. Often, two Curie temperatures occur within this interval (Figures 4c and 4d). In the low-temperature course a peak at about −160°C is observed, which in some samples is strongly pronounced (e.g., SR0101-7.0, Figure 4d) while it is less pronounced in others (e.g., SR0131-4.4, Figure 4c). According to the experimental work of Senanayake and McElhinny [1981], a peak increase in MS for pure or nearly pure magnetite at low temperatures indicating decreasing grain size down to 1 μm. However, below 1 μm, the peak flattens off until it reaches a constant line typical for single domain grains. As similar features are observed in our samples, a distinctly smaller grain size is inferred for the magnetite-rich titanomagnetite grains of sample SR0131-4.4 (Figures 3 and 4c) when compared to sample SR0101-7.0 (Figures 3 and 4d).

[26] Optical observations of group 2 samples reveal significant textural differences. Sample SR0101-7.0 shows idio-hypidiomorphic titanomagnetite grains with abundant fine ilmenite lamellae (C3 stage; Figure 5d), whereas sample SR0131-4.4 contains homogeneous titanomagnetite and ilmenite similar to group 1 samples (Figure 5c). Occasionally Cr-Al spinel occurs. Microprobe analyses show a significantly higher Xmt (0.54 versus 0.35) for the homogeneous sample SR0131-4.4, in contrast to the exsolved sample SR0101-7.0. Both compositions are not in agreement with the Curie points measured, which indicate a higher Xmt in titanomagnetite (Table 2).

[27] Group 3 samples show similar high Curie temperatures as group 2 samples (500°–600°C) but at low temperatures, a continuous decrease of susceptibility with increasing temperature occurs (Figures 4e and 4f). A common feature of all the material showing this behavior is their red color, indicating alteration under oxidizing conditions. Compared to fresh basalts, these samples always have lower susceptibilities, indicating that titanomagnetite is decomposed (oxidized). Under the microscope, titanohematite is abundant and the amount of titanohematite and ilmenite is commonly higher than titanomagnetite. Often an indistinct mottling is observed within the altered titanomagnetite. Optical observations show different textures for samples of this group, which are indicative for high-temperature deuteric oxidation as well as for low-temperature oxidation. Sample SR0127-0.4 contains Fe-Ti oxides with grain sizes up to 100 μm and skeletal to lath-like ilmenite (Figure 5e). The hematite component in ilmenite reaches up to 0.1 (Table 2). Original titanomagnetite identified by the ferrofluid method is strongly oxidized and occurs as relics in grains that consist of titanohematite and rutile (C5 stage). Such alteration of titanomagnetite is typical for high-temperature deuteric oxidation [e.g., Haggerty, 1991]. Additionally, titanohematite grains with ilmenite lamellae occur showing a hematite component of 0.81 (Table 2, Figure 5e). In contrast, sample SR0101-3.0 shows Cr-Al spinels with Cr-bearing titanomagnetite-rims, girlands of ilmenite intergrown with a ferrimagnetic phase and exsolved titanomagnetite. Within the latter grains, broad ilmenite lamellae are developed and grain boundaries are transformed into a hematite-near (possibly titanomaghemite) phase. The development of titanomaghemite at grain boundaries is often attributed to low-temperature oxidation [e.g., Dunlop and Özdemir, 1997; Sherwood, 1990]. Microprobe investigations again provide inconsistent results with low totals of weight percent for all titanomagnetite types (Table 2). It is not clear whether high-temperature deuteric oxidation or low-temperature maghemitization is the driving force responsible for significant cation-deficiency within these samples.

4.3. Remanence Behavior of Subaerial Units

[28] Samples from subaerial groups 1 to 3 were tested for characteristics in their natural remanent magnetization (NRM; Table 3). Examples of typical demagnetization curves measured at room temperature are shown in Figure 6. Most subaerial samples have less than 15% of their initial NRM remaining after demagnetizing to peak alternating fields (AF) of 160 mT. The demagnetization field required to remove half of the NRM is referred to as the median demagnetizing field (MDF). Samples of group 1 have MDFs < 20 mT. Group 2 samples are more resistant to demagnetization with MDFs between 20 and 45 mT. Group 3 samples show similar behavior to group 2 samples but with higher MDF values in the range between 35 and 80 mT.

Figure 6.

Alternating field demagnetization in peak fields up to 160 mT of natural remanent magnetization for samples from subaerial lava flows. Displayed are examples for group 1 to 3 behavior of samples from subaerial lava flows. Samples show an increase of MDF from homogeneous (group 1a and b) to exsolved titanomagnetite (group 2). Stronger oxidation with the formation of titanohematite (group 3) shows similar MDFs as the samples showing exsolution textures. Colors relate to lithologies presented in Figure 1d.

[29] The increase in MDFs can be directly related to change in titanomagnetite composition or even to phase changes into titanohematite. The oxidation of homogeneous, magnetite-poor titanomagnetite to magnetite-rich compositions is related to a grain size reduction related to the formation of ilmenite exsolution lamellae, which causes a significant hardening in magnetization. However, the difference between magnetite-rich titanomagnetite of group 2, and samples with dominantly titanohematite/titanomaghemite and minor titanomagnetite of group 3, is small. According to our measurements the increase in hardening that is related to this oxidation process is transitional in nature.

[30] The presentation of room temperature hysteresis parameters in correlation plots of Jrs/Jr versus Hcr/Hc [Day et al., 1977], as displayed in Figure 7, show group 1a samples differ significantly with lower Jrs/Jr and higher Hcr/Hc ratios, from all other groups (2, 3) and subgroups (1b; Table 4). This is in agreement with the magneto-mineralogical investigations, according which group 1a is related to homogeneous magnetite-poor titanomagnetite with medium grain sizes at about 10 μm, whereas group 2 is characterized by smaller magnetic domains (few μm and less) of magnetite-rich titanomagnetite that are subdivided by ilmenite lamellae. There is no clear distinction between group 1b, group 2 and group 3 behavior, which shows up the limitations of the Day-plot presentation. The samples of all groups (1, 2, 3) fall into the same field for a pseudo-single domain-state. However, a comparison between our measured hysteresis ratios and the values for synthetic titanomagnetite used in the study of Day et al. [1977] shows a reasonable correspondence. The results for group 1a samples (e.g., SR0132-5.3) fit favorably with compositions of Xmt of 0.4 and a grain size of 9 μm. Samples of group 2 (e.g., SR0101-7.0) fit to a Xmt of 0.8, as indicated by the Curie temperatures, and a grain size of about 3 μm.

Figure 7.

Hysteresis parameter at room temperature for group 1 to group 3 samples for subaerial lava flows presented in a Day-plot. All groups fall into a relatively narrow area of the PSD field. Distinction between different groups, especially of group 2 and 3, is not possible on this plot.

Table 4. Hysteresis Data From Subaerial and Submarine Samples of the HSDP-2
SampleLithologyJs, Am2/kgJrs, Am2/kgHc, mTHcr, mTGroup
SR0694-10.9massive basalt0.0240.01213203/1a
SR0799-16.0massive pillow0.5810.095491a
SR0800-0.9massive pillow0.4050.0975111a
SR0800-1.0massive pillow1.0590.1896131b
SR0800-19.8massive pillow0.2800.0497171b
SR0801-13.4massive pillow0.3210.0737151b
SR0801-15.1pillow breccia0.0660.05260714

[31] Generally, the magnetic properties of subaerial units are strongly controlled by high temperature oxidation alteration of primary homogeneous titanomagnetite (group 1) with transitions from the originally homogeneous titanomagnetite to magnetite-rich titanomagnetite + ilmenite (group 2) and hematite-bearing samples (group 3). In minor amounts, low-temperature oxidation occurred with formation of titanomaghemite. Figure 15 summarizes the characteristic magnetic properties for all three groups. Additionally to these “end-member” groups a number of transitions were observed which complicate the interpretation of the rock magnetic data (see Table 3).

4.4. Temperature-Dependent Susceptibility and Magnetic Mineralogy of Submarine Units

[32] The defined mineral groups can also be recognized within the submarine units with exception of group 3. Group 1 (a and b) and mixtures of group 1 and 2 behavior (called group 1a/2 or 1b/2 when group 1 dominates) occur in all massive lava units and in related ferrimagnetic hyaloclastites. A further group (group 4) is restricted to the paramagnetic dominated hyaloclastites and pillow breccias, which consist mainly of glass with olivine and Cr-Al spinel xenocrysts. We observed transitions between hyaloclastites with typical paramagnetic behavior to those with dominantly ferrimagnetic behavior. This transition is related to the crystallization of minerals in the glassy matrix.

[33] Typical MS-T curves from hyaloclastites with paramagnetic-dominated susceptibilities (group 4) are shown in Figure 8a. They are characterized by an exponential decrease in susceptibility with increasing temperatures. Despite the paramagnetic dominance during heating, many of the analyzed samples display a drop in susceptibility around 580°C, indicating minor contribution from a ferrimagnetic component. The Curie temperature of 580°C indicates almost pure magnetite. Cooling back to room temperature produces exactly the same Curie point but the susceptibility strongly increases as the temperature decreases. We suggest that the heating produces additional magnetite that is responsible for the increased MS.

Figure 8.

Typical curves showing the temperature-dependence of MS and evaluated Curie temperatures (TC) from submarine units. (a) paramagnetic-dominated hyaloclastite, group 4; (b) ferrimagnetic-dominated hyaloclastite, group 1a/2; (c) massive pillow-lava, group 1a; (d) pillow-breccia, group 4; (e) interior of intrusive unit, group 1a; (f) marginal part of intrusive unit, group 1b; (sample location see Figure 3). Further explanations see text.

[34] Although the low-temperature leg is similar to group 3 behavior from the subaerial lava units, the mineralogical origin of the submarine curve is significantly different. Group 3 samples from the subaerial lava units contain titanohematite, whereas reflected light microscopy using the ferrofluid method and scanning electron microscopy failed to resolve the magnetic mineralogy of the hyaloclastite samples (Figure 9a).

Figure 9.

BSE-images of oxide and sulfide textures from submarine samples. (a) chromian spinel (csp) as inclusions in olivine (ol), (b) hypideomorphic tm and very small cruciform tm, (c) homogeneous skeletal tm and il intergrown with xenomorphic pyrrhotite (po), (d) csp-inclusion in ol, very small oxide phases in matrix, (e) homogeneous tm (hypideomorphic, skeletal), il (dendritic), and xenomorphic po, (f) homogeneous cruciform tm and lath-like il.

[35] Ferrimagnetic hyaloclastites often show a mixture of group 1 and group 2 rock magnetic characteristics with two or three Curie temperatures in the range between 200° and 580°C. A characteristic feature is a broad hump before reaching the Curie temperatures (Figure 8b). In some samples exsolved titanomagnetite is found within basaltic fragments, with magnetic behavior similar to the group 2 samples of the subaerial units. But other samples, for example SR0623-1.7 (Figure 9b), also show different grain size populations, whereby the titanomagnetite with grain sizes above the diameter of the electron beam (about 1 μm) yield a homogeneous Xmt of 0.57 (Table 2). This composition is similar to that determined from the first Curie temperature at 295°C (Figure 8b). We assume that the higher Curie temperature at about 530°C is related to the very small grains, and not to exsolution textures as we described for group 2 subaerial samples.

[36] Pillow units often show a very inhomogeneous MS distribution with high values in massive units and medium-to-low values in units containing small pillow structures. Figure 8c presents an example of a temperature-dependent susceptibility curve for the medium to high susceptibility pillow samples. A major Curie point occurs at 135°C and a subordinate one at 580°C. The latter one could indicate minor amounts of magnetite. This phase is assumed to be responsible for the increase in susceptibility during the cooling leg, similar to that observed in the hyaloclastite samples. This type of curve is typical for the whole unit 0291, although susceptibilities show high variation. Microprobe and back-scattered electron (BSE) imaging reveal homogeneous titanomagnetite (Xmt between 0.3 and 0.4; Table 2), ilmenite and minor pyrrhotite (Figure 9c). Pyrrhotite is not visible in the MS-T curves. The variation in susceptibility (Figure 3) is mainly related to variable amounts of titanomagnetite but it also corresponds with a significant change in the vesicularity of the lava flow. A high amount of vesicles occur in the upper part of the flow, which decreases where MS decreases.

[37] Pillow breccias typical for the margins of pillow units (Figure 3) have very low susceptibilities, and Curie temperatures at about 575°C, similar to the paramagnetic hyaloclastite samples (Figure 8d). Non-reversibility of the high-temperature cooling leg probably indicates that heating caused the formation of titanomagnetite or titanomaghemite of varying composition. At low temperatures, the nearly constant path is in agreement with single domain grains of magnetite [e.g., Muxworthy, 1999]. The spinifex texture of the matrix (Figure 9d) suggests rapid quenching, similar to what is observed in the hyaloclastite samples. Scanning electron microscopy again failed to resolve the magnetic mineralogy of the very small bright grains in Figure 9d but it is reasonable to assume that it is a magnetite-near phase.

[38] Figures 8e and 8f show MS-T curves for homogeneous titanomagnetite from the central and marginal part of the intrusive unit 0341b (Figure 3). Although Curie temperatures are similar (204° and 273°C, respectively), the low-temperature behavior below about −150°C is significantly different from the behavior of 1a for the central part (Figure 8e) and 1b toward the margin (Figure 8f). Optical observations reveal significantly lower grain sizes at the margins than in the interior (Figures 9e and 9f). Again, as for the pillow unit, titanomagnetite, ilmenite and pyrrhotite make up the opaque phase assemblage. The Xmt of titanomagnetite is similar in both parts of the flow, but good agreement with the measured Curie temperature is only attained in the interior, whereas the margin sample has a difference of 80°C (Table 2).

[39] Exsolution textures that are typical for group 2 are only observed down to a depth of about 1900 mbsl, which is close to the deepest occurrence of subaerially degassed lava [DePaolo et al., 2001]. Below about 2300 mbsl, where glasses contain a higher water content typical of submarine eruptions [DePaolo et al., 2001], no exsolution has been observed in the profiles investigated until now, and antiferromagnetic or mixed types of pyrrhotite occur in the opaque assemblage.

4.5. Remanence Behavior of Submarine Units

[40] During AF demagnetization experiments of NRM, most submarine samples show less than 20% of the initial remanent magnetization at 160 mT (Figure 10). Group 1 samples from the pillow and intrusive units generally show weak magnetic behavior with a MDF below 20 mT. Samples with a mixture of group 1 and 2 behavior, as typical for the ferrimagnetic hyaloclastites and pillows with intermediate to low susceptibilities, have distinctly higher MDF values between 20 and 80 mT. MDFs from paramagnetic-dominated hyaloclastites with group 4 behavior can show values up to 110 mT (Table 3).

Figure 10.

Alternating field demagnetization of natural remanent magnetization for group 1 to group 4 samples from submarine lava units. All group 1 samples show an exponential decrease during AF demagnetization, typical for large MD titanomagnetites. Group 4 samples show an initial plateau and then development toward a sigmoidal shaped demagnetization curve typical for SD behavior. Colors relate to lithologies presented in Figure 1d.

[41] One striking feature observed in submarine units is the minor influence of oxidation alteration, which seems to disappear with depth. On the other hand, the grain-size of magnetic minerals plays an increasing role. This is confirmed by room temperature hysteresis parameters (Table 4), which are presented in a Day-plot (Figure 11). All submarine samples fall in the PSD field of the Day-plot, with a few exceptions that show SD behavior. Whereas the intrusive sample from the interior of a flow has higher Hcr/Hc ratios and lower Jrs/Jr ratios, the samples from the rim displays the opposite ratios, indicating smaller grain sizes in agreement with optical observations and the MS-T curve at low-temperature. The pillow samples plot close together with Hcr/Hc ratios between 2.0 and 2.5 and Jrs/Jr ratios ranging between 0.15 and 0.25, but the pillow breccia has a significantly higher Jrs/Jr ratio and clearly falls in the SD field. This indicates significantly smaller grain sizes for the pillow breccia than the pillow interiors. This observation is in agreement with the low temperature behavior of susceptibility and reflected light microscopy study. The hyaloclastite samples scatter across the PSD field. This behavior is attributed to the different kinds of magneto-mineralogic types that may occur in these lithologies. Nearly all samples from the submarine units follow an exponential trend in the Day plot, which is clearly related to the grain size variations of titanomagnetite.

Figure 11.

Hysteresis parameters at room temperature of submarine samples presented in a Day-plot. Most submarine samples with a mixture of MD and SD grains fall into the PSD field. Pillow breccia and some pillows with magnetic phases below the resolution of SEM fall into the SD field.

[42] In the submarine units the Xmt of homogeneous titanomagnetite can be observed to range between about 0.25 and 0.5. Here, magnetic variation is stronger related either to the amount of titanomagnetite (e.g., unit 0291) or to the grain size distribution within a flow, and not to the oxidation processes which control the magnetic mineral assemblages of the subaerial units. We therefore assume that the formation of the magnetite-rich submicroscopic grains is related to the fast cooling of the lava and not to secondary, low-temperature alteration processes. This is a surprising feature considering rapidly cooled basalt, which normally contains titanomagnetite with a low Xmt between 20 and 50 per cent [e.g., Haggerty, 1976].

5. Discussion

[43] Variations in rock magnetic properties are generally related to (1) the amount of ferrimagnetic minerals within samples, (2) the chemical composition of ferrimagnetic components, especially chemical variations of titanomagnetites, and (3) the grain size of ferrimagnetic particles. In volcanic rocks these parameters may depend on magma composition, viscosity, temperature, and on the oxygen fugacity during magma extrusion and subsequent high- and low-temperature alteration processes. Such processes can lead to complex MS signatures on a range of different scales. In the following section, the characteristic magnetic behavior that we observed within the subaerial and submarine lava sequences of the HSDP-2 cores are discussed in relation to these influences.

5.1. Correlation Between Magnetic Susceptibility and Magma Composition of the Hawaiian Lava Flows

[44] MS variations from the different subaerial and submarine HSDP-2 core profiles can be related to the evolution of the Hawaiian shield volcano (Figure 1). Especially the subaerial flows of Mauna Loa and Mauna Kea display significant differences in their mean susceptibility values (Table 1), with low values for the former and high values for the latter. Such differences in MS may reflect the influence of different magma compositions from the two volcanoes and their stages of evolution. The Mauna Loa basalts are classified as tholeiites based on major and trace element analyses [DePaolo et al., 2001], whereas the uppermost ca. 50 m of the underlying subaerial Mauna Kea lavas (> 246 mbsl) contain interbedded nepheline-normative and hypersthene-normative lavas, which are generally more alkaline in composition. These flows mark the end of the shield-building phase of the Mauna Kea volcanic cycle. One example of these more alkaline flows is unit 0048 (Figure 3), characterized by very high magnetic susceptibilities attributed to its titanomagnetite content of ca. 20 vol%.

[45] According to DePaolo and Stolper [1996], Mauna Loa lava flows are related to more central parts of the plume whereas the Mauna Kea lava flows are assumed to originate from the margin. One characteristic trace element ratio used to distinguish the magma source is the Nb/Y ratio, which is significantly lower for Mauna Loa basalts than the Mauna Kea lavas (J.M. Rhodes, personal communication, HSDP workshop, 2000). In order to correlate magnetic and compositional data we plotted the mean Nb/Y ratios (from X-DIS database) versus the mean MS from defined sections (Figure 12). When comparing tholeiitic subaerial units from the Mauna Loa and Mauna Kea it is notable that lower Nb/Y ratios and lower susceptibilities are typical for the Mauna Loa, whereas the Mauna Kea subaerial units have higher and more variable Nb/Y ratios along with higher susceptibilites. This is evident although a large susceptibility variation within both subaerial lava sequences is caused by the high degree of oxidation of the homogeneous titanomagnetite. Still higher Nb/Y ratios corresponding to high MS were observed within the alkaline flows of the Mauna Kea volcano.

Figure 12.

Mean MS versus mean Nb/Y ratios (from X-DIS database) for different subaerial and submarine lava of the Mauna Loa (ML) and Mauna Kea (MK) volcanoes. A significant trend from tholeiitic Mauna Loa (initial stage) to tholeiitic Mauna Kea units (main stage) and toward the more evolved alkaline Mauna Kea subaerial units (final stage) is displayed showing a correlation between magma composition and MS. Bars indicate standard deviation.

[46] For the submarine lava units the Nb/Y ratio is relatively constant, whereas the MS shows strong variation. This variation is related to contrasting grain size and concentration of magnetic minerals formed during the cooling of the lava (see below).

5.2. Variation of Magnetic Susceptibility Within Subaerial Lava Flows

[47] Various subaerial lava types show a clear difference in their MS. Aa lava generally displays higher susceptibilities than transitional and pahoehoe lavas (Table 1). According to Ho and Cashman [1997], aa and pahoehoe flows are characterized by groundmass textures that result from different flow transport and cooling history. The smaller crystal sizes of the aa groundmass result from the higher rates of cooling, and thus higher rates of crystal nucleation than in the better insulated, often slower moving pahoehoe flows. Larger grain sizes in the pahoehoe than in the aa lava flows (e.g., unit 0047 and unit 0046, respectively) are also observed in the HSDP cores. It is suggested that an additional controlling parameter on the MS is the amount of Fe-Ti oxide nuclei, which strongly depends on viscosity and temperature of the lava. Bücker et al. [1999] also reported that aa lava from the ODP-hole 990A, drilled on the Southeast Greenland margin, show higher magnetic susceptibilities (16.94 ± 0.62) than transitional and pahoehoe types (12.69 ± 0.35 and 12.60 ± 0.35). They proposed the MS as a diagnostic petrophysical parameter for distinguishing different subaerial lava types. However, within the subaerial lava flows of the Mauna Loa, aa and transitional lava flows have similar but relatively low mean values compared to the Mauna Kea subaerial types and the susceptibility difference of the pahoehoe flows are less pronounced (Table 1).

[48] Measurements on the HSDP cores reveal MS variations between lava flows as well as within lava flows, especially in subaerial units. According to Dunlop and Özdemir [1997], oxide mineral assemblages can vary from point to point within a single lava flow because of differences in oxygen fugacity, temperature and cooling rate. Grommé et al. [1969] suggest that only flows quenched from very high temperatures escape oxidation entirely and had single low Curie points. The interiors and margins of flows may, however, become oxidized by their contained fluids. The reported variations in MS from some flows presented in our study support such a scenario. The high MS within flow 0048 (MS up to 60) is related to titanomagnetite characterized by an intergrowth of magnetite-rich titanomagnetite and ilmenite with high Curie points (Figure 4c) and hard magnetic behavior (MDF: 45 mT). The intermediate susceptibility values of about 30 within this flow can be related to homogeneous titanomagnetite with a low Curie point (100°C) and weak magnetic behavior (MDF: 14 mT). At the top of this lava flow, MS decreases dramatically to values < 5 (Figure 3), which relates to even higher deuteric oxidation during the cooling of the lava, creating titanohematite and rutile at the expense of titanomagnetite. Such baked tops are typical for the subaerial lava flows and show group 3 characteristics in their low-temperature behavior. According to our investigations, high temperature oxidation can be recognized as the most important process influencing the magnetic mineralogy and magnetic properties of the subaerial lava flows. Such an oxidation alteration is probably favored by the lower crystallization temperatures in the subarial units compared to the submarine units. Applying the titanomagnetite-ilmenite thermometer of Ghiorso and Sack [1991] for homogeneous titanomagnetite - ilmenite pairs, significantly higher temperatures are indicated for the submarine units. Here, temperatures of 960°–1050°C are calculated for submarine unit 0249, in contrast to temperatures of 815°–870°C for subaerial unit 0048.

5.3. Variation of Magnetic Susceptibility Within Submarine Lava Flows

[49] The submarine Mauna Kea units in the depth interval between 1079 and 1984 mbsl are dominated by hyaloclastite debris flows (about 90%), intercalated with massive submarine basalts. The subaerial-submarine transition is marked by a decrease in MS (Figure 3) related to the volcaniclastic sediments and glassy lavas. The lithic basalt clasts that occur within the hyaloclastites may be the cause of strong variations in MS. Hyaloclastite intervals, which are well-bedded and/or poor in lithic clasts have very low magnetic susceptibilities (about 0.5). The submarine basalts of this depth interval are considered to originate as subaerial flows that continued past the shoreline as submarine flows [DePaolo and Stolper, 1996]. This explains why lithic clasts within the hyaloclastites and massive flows down to that depth contain exsolved titanomagnetite reflecting high-temperature oxidation typical for the subaerial basalts. The degree of oxidation does not reach the high-grade stage characterized by the formation of titanohematite. Only exsolution of homogeneous titanomagnetite into magnetite-rich titanomagnetite and ilmenite was observed (C2–C3 stage). However, there also occur hyaloclastite units like the profile section 0218–0225, characterized by homogeneous titanomagnetite with group 1/2 behavior. Here, grain size variations and the absence of exsolution textures indicates rapid cooling due to direct extrusion into seawater. The variation in grain size is assumed to be the most important mechanism controlling magnetic behavior within the submarine extruded basalts.

[50] The varying magnetic properties of submarine basalts have been largely interpreted until the 1990s as the result of seafloor alteration. Submarine basalts usually display Xmt of 0.4 (TM60) and low Curie temperatures. Kent and Gee [1994] argued that fine-grained titanomagnetite from oceanic basalts, small enough constitute single domains, tend to become oxidized to a cation-deficient titanomaghemite. These phases have significantly higher Curie temperatures, which accounts for the variation in remanence properties. These authors, among others, [see, e.g., Dunlop and Özdemir, 1997, and references therein] attribute the oxidation to low-temperature alteration (< 100°C) during the initial stages of burial. However, recent studies of pillow-lava by Zhou et al. [1997, 2000] document abundant titanomagnetite, submicrometer in size, contained within globules present in a glassy matrix, with a broad compositional range (Xmt of 0.2–1). Additionally, multidomain-sized titanomagnetite occurs within these samples with a narrow compositional range (Xmt of about 0.4). This chemical variation can also explain the hard magnetic behavior and high Curie points that were observed in seafloor pillow basalt and eliminates the need of rapid alteration.

[51] Our observations on the submarine basalts from the HSDP-2 cores are in excellent agreement with the studies of Zhou et al. [1997, 2000]. Pillow units below 1984 mbsl show a variation in MS (Table 1) that can be related to the position within an individual pillow. Interior regions mostly show the highest susceptibilities, which then decreases toward the rapidly quenched, glassy margins or brecciated flow tops (Figure 3). Titanomagnetite grains large enough to be analyzed by the microprobe show a very narrow compositional range (Table 2) and samples that are dominated by this grain size population behave magnetically soft (Figure 10). An increase in the volumetric portion of small grain sizes, as observed in Figure 9b for samples showing dominantly a group 1 behavior, increases the hardness of magnetization (Figure 10, sample SR0623-1.7). The hardest magnetization was found in pillow breccia and hyaloclastites with dominantly paramagnetic and minor ferrimagnetic behavior. Hysteresis loops confirm single domain behavior for the pillow breccia (Table 4, Figure 11) and a potbellied loop, as described for submarine basaltic glass by Tauxe et al. [1996, see Figure 1b] for the hyaloclastite sample SR0689-8.0 of superparamagnetic behavior.

[52] In accordance to Zhou et al. [2000], we assume that the Ti-content of titanomagnetite varies as a function of cooling rate and crystal-melt fractionation, either from the interior to the rim of a pillow or from the interior to the rim of a submarine flow. Ti-poor magnetite is also reported as an initial component of submarine basalt glasses by Pick and Tauxe [1994]. Such a variation in the Xmt of titanomagnetite requires small scale variation in the Fe/Ti ratio of the melt. Zhou et al. [2000] explains such behavior by the lack of a homogeneous equilibrium state within the residual melt, which occurs in small, isolated interstices among the crystals.

5.4. Significance of Sample Grouping and Their Mineralogical Interpretation

[53] The subdivision of the HSDP-2 samples into different magneto-mineralogical groups is similar to the categories proposed by Senanayake and McElhinny [1981] from low temperature susceptibility behavior of basalts, except their category 3 and 2 correspond to our group 2 and 3. Gonzales et al. [1997] reported a fourth type equivalent to our group 1b from lava of the Transmexican Volcanic Belt, with an initial decrease in susceptibility as the temperature rises. We switched group 2 and 3 described by Senanayake and McElhinny [1981], because magnetic properties of our groups 1 to 3 are related to increasing degrees of high-temperature oxidation (stages C1 to C5 given by Haggerty [1991]) within the subaerial units. Group 1 behavior is restricted to homogeneous, unoxidized titanomagnetites, which already show C2 oxidation stages, but still consist of magnetite-poor titanomagnetite with low Curie points. Group 2 behavior is characterized by magnetite-rich titanomagnetite grains that are subdivided by exsolved ilmenite lamellae into magnetically independent small regions, giving rise to high Curie temperatures and a peak at low temperatures (C2–C3 high-temperature oxidation stage). MS is significantly increased compared to samples with magnetite-poor titanomagnetite. This increase is related to the grain size reduction that occurs as a consequence of the formation of ilmenite lamellae and the increase of Xmt in titanomagnetite. Group 3 is related to the highest degree of oxidation, including the formation of titanohematite and the decomposition of titanomagnetite. This group is characterized by its decrease in susceptibility with increasing temperature on the low-temperature leg and high Curie temperatures. MS is similar or distinctly lower than that of the original titanomagnetite-bearing samples, which is an effect of the decomposition of titanomagnetite. The resistance to AF demagnetization in group 2 and 3 samples is significantly higher than in group 1 samples, but the former groups can not be discriminated by this method.

[54] The subdivision of group 1 samples into 1a and 1b subgroups is related to their differences in low-temperature behavior as shown in Figures 4a and 4b. These subgroups are documented from different localities of basaltic lava [e.g., Gonzales et al., 1997; Urrutia Fucugauchi et al., 1991; Sherwood, 1990], and are abundant in the HSDP-2 profile in subaerial (Figures 4a and 4b) as well as in submarine units (Figures 8c and 8f). The cause of these differences in low-temperature behavior is still under debate. The increase in susceptibility at temperatures <−160°C has either been related to (1) the Néel temperature of an antiferromagnetic mineral such as impure ilmenite [e.g., Moskowitz et al., 1998; Gonzales et al., 1997] or (2) to the contribution of superparamagnetic grain fractions [e.g., Urrutia Fucugauchi et al., 1991; Radhakrishnamurty et al., 1977]. From our observations, ilmenite of similar composition occurs in both group 1a and group 1b samples. Susceptibility measurement down to −263°C for group 1a sample SR0132-5.3 (Figure 4a) indicates no magnetic phase is present other than titanomagnetite, although ilmenite could be observed optically (Figure 5a). We therefore infer that ilmenite does not contribute to the differences observed in the low-temperature behavior.

[55] To obtain more diagnostic information we performed low-temperature hysteresis measurements for group 1a and 1b samples. Figure 13 presents the difference in the low-temperature course of the coercive force (Hc) for the two sample groups. Group 1a samples show a behavior with strongly increasing Hc with decreasing temperature. In contrast, group 1b samples display a continuous increase in Hc until a peak at ca. −200°C, followed by decreasing Hc. The increase in Hc recorded until liquid nitrogen temperature, as observed in both sample groups, is consistent with experiments and modeling of Tucker [1981]. He described titanomagnetite, with a similar composition as the homogeneous titanomagnetite of this study, as a phase where magnetocrystalline-controlled domain-wall pinning becomes dominant with decreasing temperatures. Hysteresis loops determined at different low temperatures (Figure 13) shed further light on the variations in the domain state of both the sample groups. Hysteresis loops shown in Figure 13 deviate from typical examples of pure SD (single domain) and MD (multidomain) behavior. At −23°C both groups show low Hc (Table 5), due to a significant contribution of MD behavior. During cooling, Hc increases significantly for both samples indicating an increase in SD contribution. This may point to interactions of domain walls with crystal defects when the large increase in magnetocrystalline anisotropy with decreasing temperature are considered [e.g., Clark and Schmidt, 1982]. Below liquid nitrogen temperature, the shapes of the loops become significantly different. Increasing Hc results in a “pot-bellied” shape of the hysteresis loop of group 1a samples, which is indicative for very small grain sizes with SP (superparamagnetic) behavior (about 8 nm, according to Tauxe et al. [1996]). Group 1b samples show a “wasp-waisted” shape below ca. −200°C, which can either result from a population spanning the SP/SD threshold size or from two populations with distinct coercivities. Both types can be discriminated by ΔM curves [Tauxe et al., 1996]. The difference between ascending and descending loops (ΔM) for H > 0 mT from hysteresis loops measured at different low temperatures are plotted versus the magnetic field H (Figure 14). While monotonic decrease should indicate a single population of mixed SP/SD particles, “roller coasters” are related to multiple single domain populations with different coercivity spectra (e.g., hematite and magnetite). All 1a group hysteresis loops measured at temperatures between 23 and −243°C and 1b group hysteresis loops measured at temperatures from 23 to −163°C show a monotonic decrease indicative for one single titanomagnetite population (Figure 14). This is confirmed by the derivative of the ΔM curve (Figure 14), which shows a single hump (for comparison, see Figure 9 in Tauxe et al. [1996]). Measurements below −163°C reveal a slightly unsteady decrease for group 1b, and the derivative of the ΔM curve additionally displays a strong increase at low fields (Figure 14) indicative of a second mineral phase. The tight peak shape points to a multidomain behavior for this second phase, which we assume to be a member of the chromian-bearing spinel group. This is based on the observation that in addition to titanomagnetite and ilmenite, all the group 1b samples contain Cr-Al spinels. The Cr/(Cr + Al) ratios range between 0.65 and 0.73, and the Fe/(Fe + Mg) ratios from 0.5 to 0.6. The Néel temperature for Fe3−xCrxO4 is reported at −193°C for x = 2 and −30°C for x = 1.3 [Robbins et al., 1971], and for MgCr2O4 at −258°C [Blasse and Fast, 1963]. Therefore cations like Al and Fe3+ shift the magnetic transition temperatures and may explain the observed Curie temperature of ca. −165°C in group 1b samples (Figure 4b, Table 3). Further studies on the low temperature magnetic behavior of Cr-Al spinel bearing basalts with different compositions of the Cr-Al spinel and different degrees of titanomagnetite oxidation are needed to confirm this hypothesis.

Figure 13.

Normalized coercivity forces versus temperature for group 1a and 1b. Low-temperature hysteresis loops for both groups show MD-dominated behavior at −23°C. With decreasing temperatures SD behavior prevails. Below −203°C hysteresis loops for group 1a display “potbellied” shapes indicating a superparamagnetic contribution, whereas group 1b sample show “wasp-waisted” shapes.

Figure 14.

ΔM curves (a), (c), (e) and (g) and their derivatives (b), (d), (f) and (h) for group 1a and 1b samples at different low temperatures. Below −163°C the curves for group 1a sample still shows a behavior typical for only one titanomagnetite population, whereas the curve for group 1b sample indicates a second ferrimagnetic phase is present. This phase is responsible for the “wasp-waisted” hysteresis loop observed in Figure 13.

Table 5. Coercive Force (Hc) and Saturation Magnetization (Ms) at Different Temperatures for 1a and 1b Group Samples Determined From Low-Temperature Hysteresis Loops
Temperature, °C−240−203−163−23
Hc, mT
Ms, Am2/kg

[56] Except for group 1a samples, all groups (1b, 2, 3) show an increase in susceptibility at very low temperatures (<−100°C) in the MS-T curves (Figures 4 and 8). Although the ferri-/antiferromagnetic properties of oxide phases different to titanomagnetite may account for this MS increase, low-temperature SD to SP behavior of the grain population forming fine exsolution structures could also be the cause.

[57] If comparing the significance of the methods applied in this study for discriminating the different groups of magnetic behavior, we found the low-temperature susceptibility measurements most useful, although group 3 and 4 could not be discriminated (Figure 15). The major advantage of the low-temperature susceptibility method is its sensitivity to compositional changes and the domain-state of titanomagnetite. However, this method alone cannot solve the complex rock magnetic behavior observed in the HSDP-2 samples. Additional remanence experiments and the mineralogy determined by optical methods are required. We found temperature-dependent susceptibility measurements to be more useful than microprobe analyses for determing the chemical composition of homogeneous and exsolved titanomagnetite, but different compositions due to short range melt fractionation and oxidation are difficult to obtain. A widening of the concave upward shape in MS-T curves largely indicates a range of chemical compositions. High-resolution transmission microscopy studies would be most helpful to confirm this interpretation.

Figure 15.

Magnetic characteristics of sample groups 1 to 4 discriminated by this study.

6. Conclusions

[58] A study of basalt cores from the HSDP-2 drilling at Hawaii reveals that lava extruded subaerially or under submarine conditions is characterized by a variety of magnetic properties related to different origins. The following characteristics were recorded in the HSDP-2 section:

  1. The mean MS in subaerial lava flows is generally lower in the Mauna Loa compared to the Mauna Kea units, which may reflect the influence of different magma compositions.
  2. Different subaerial lava types (aa, transitional, pahoehoe) show clear differences in their MS with generally higher susceptibilities in aa lava than in transitional and pahoehoe lavas.
  3. In subaerial lava flows, three different groups of magnetic behavior (Figure 15) are distinguishable, related to the degree of high-temperature oxidation during the lava extrusion. MS shows systematic variation along with differences in these magneto-mineralogical assemblages. In relation to group 1, group 2 behavior is associated with increased MS whereas group 3 behavior accompanies a significant decrease.
  4. Submarine lava flows from Mauna Loa can similarly be subdivided into groups based on their rock magnetic properties. Lava flows and volcaniclastic sediments between 1079 and 1984 mbsl in the drill core have been interpreted as subaerial flows that continued past the shoreline as submarine flows. Clasts and massive flows from this depth range, either have homogenous and/or exsolved titanomagnetites and display group 1 or 2 behavior. No samples displayed group 3 behavior indicating that these submarine lava flows experienced no high temperature oxidation. However, a fourth rock magnetic group was recorded in these rocks. This group is restricted to hyaloclasites and pillow breccias and is characterized by a dominance of paramagnetic behavior on the temperature versus MS curves, with small amounts of nearly pure magnetite with single domain and/or superparamagnetic behavior. Samples below 1984 mbsl in the drill core showed no exsolution textures with group 1 and 4 behavior dominating.
  5. High-temperature oxidation of primary homogeneous titanomagnetite controls the texture and mineralogy of the magnetic minerals in the subaerial lava flows.
  6. The controlling mechanism of the magnetic properties in the submarine lava section is the cooling and quenching rate of the lava flows, which results in large grain size variations and titanomagnetites of different compositions.
  7. Low-temperature alteration, which is of importance in ocean floor rock magnetism, is of minor relevance in this setting. Similarly, hydrothermal alteration as described from the ocean floor or from Icelandic basalts, is unimportant for the magnetic properties of the 3109 m deep section of the HSDP-2 drill hole.

[59] In general, our findings on subaerial lava flows are similar to what is described by Grommé et al. [1969] for samples from the Alae and Makaopuhi subaerial lava lakes of the Kilauea volcano. Our observations additionally confirm the new aspects concerning the magnetic behavior of submarine basalt as given by Zhou et al. [1997, 2000], and are important to understand the deeper magnetic structure of the Hawaiian shield volcano, as monitored, for example, by magnetometer borehole measurements (Steveling et al., submitted manuscript, 2002).


[60] This work was funded by a grant from the Deutsche Forschungsgemeinschaft (DFG), Germany (KO1514/1). H.d.W. gratefully acknowledges a visiting fellowship grant at the Institute of Rock Magnetism, Minneapolis. Critical comments and suggestions by the reviewers J.F. Diehl and L. Tauxe, and by L. Warr improved the paper. We thank C. Seaman and J. Kirschvink for support during sampling at California Institute of Technology in Pasadena, CA, and C. Dietl and T. Pelzer (both Heidelberg) for their laboratory assistance.