The Tiva Canyon Tuff contains dispersed nanoscale Fe-Ti-oxide grains with a narrow magnetic grain size distribution, making it an ideal material in which to identify and study grain-size-sensitive magnetic behavior in rocks. A detailed magnetic characterization was performed on samples from the basal 5 m of the tuff. The magnetic materials in this basal section consist primarily of (low-impurity) magnetite in the form of elongated submicron grains exsolved from volcanic glass. Magnetic properties studied include bulk magnetic susceptibility, frequency-dependent and temperature-dependent magnetic susceptibility, anhysteretic remanence acquisition, and hysteresis properties. The combined data constitute a distinct magnetic signature at each stratigraphic level in the section corresponding to different grain size distributions. The inferred magnetic domain state changes progressively upward from superparamagnetic grains near the base to particles with pseudo-single-domain or metastable single-domain characteristics near the top of the sampled section. Direct observations of magnetic grain size confirm that distinct transitions in room temperature magnetic susceptibility and remanence probably denote the limits of stable single-domain behavior in the section. These results provide a unique example of grain-size-dependent magnetic properties in noninteracting particle assemblages over three decades of grain size, including close approximations of ideal Stoner-Wohlfarth assemblages, and may be considered a useful reference for future rock magnetic studies involving grain-size-sensitive properties.
The magnetic remanence characteristics of magnetic particles are strongly size-dependent. Therefore, being able to constrain magnetic particle sizes and domain states through measurement of magnetic properties is important for determining a rock's capacity for carrying a stable remanence. In magnetite, the most stable remanence is typically held in fine, submicron-sized grains. However, particles in this size range are the most difficult to study because their concentrations are low in natural samples and it is often not straightforward to image them, even with high-resolution techniques, and because of difficulties in dispersing synthetic particles well enough to eliminate magnetic interaction effects.
The Tiva Canyon Tuff of the Paintbrush Group is a rhyolitic welded ash flow sheet that extends across Yucca Mountain, and through much of the nuclear test site in southern Nevada. The main part of the Tiva Canyon Tuff is compositionally zoned and covers a total area of more than 2,600 km2 with an average thickness of about 120 m [Byers et al., 1976]. 40Ar/39Ar dating puts the age of this sheet at 12.7 Ma [Sawyer et al., 1994]. The magnetic properties of the Tiva Canyon Tuff have long been of interest in rock magnetism due to the narrow grain size distribution and high dispersion of magnetic particles that greatly reduces magnetic interaction effects. Because these properties are rarely found in natural rocks, this tuff provides a unique opportunity to isolate the effects of magnetic grain size on size-sensitive magnetic properties, particularly in the very fine superparamagnetic (SP) and small single-domain (SD) domain size ranges for magnetite.
We present here a collection of rock magnetic results from a densely sampled section of the Tiva Canyon Tuff with continuously changing grain size above the base of the unit. These data complement earlier magnetic studies on a small number of these and similar samples [Schlinger et al., 1991; Rosenbaum, 1993; Worm and Jackson, 1999; Pike et al., 2001; Egli and Lowrie, 2002; Jackson et al., 2006]. The results of this work are intended to help elucidate the changes in magnetic behavior of very fine magnetic particles that occur with progressive increases in grain size for magnetic grain assemblages that closely approximate ideal Stoner-Wohlfarth populations.
The principal remanence carriers in various sections of the Tiva Canyon Tuff have been previously determined to consist of two distinct ferrimagnetic components [Schlinger et al., 1991; Rosenbaum, 1993]. One component comprises multidomain (MD) titanomagnetite phenocrysts that were present in the magma chamber of the source caldera prior to eruption. The other component consists of Fe-oxide microcrystals that precipitated from volcanic glass after emplacement of the pyroclastic material. However, Rosenbaum  determined that while the MD phenocrysts are important remanence carriers in the middle section, Fe-oxide microcrystals are the primary carrier of remanence and magnetic susceptibility in the two 15-m-thick zones at the bottom and top of the unit. Therefore, variations in magnetic properties in the basal section of the Tiva Canyon Tuff are directly related to the size of Fe-oxide microcrystals, whose growth is in turn dependent on different cooling rates within the unit. The composition of the oxide microcrystals has been characterized as slightly impure magnetite, with an average composition around Fe2.9A0.1O4 where A represents substitutional cations [Jackson et al., 2006]. The impurities are mainly Mn and Cr, with minor amounts of Ti according to semiquantitative compositional analysis by Schlinger et al. .
2. Samples and Measurements
Samples were collected from a 5-m-thick section of the basal zone of the Tiva Canyon Tuff. The section is located on the west flank of Yucca Mountain, in the Topopah Springs Quadrangle (36.82°N, 116.47°W) and corresponds to a site sampled by Rosenbaum et al.  and Site IV of Schlinger et al. . Samples were primarily collected as drilled cylindrical cores of 2.5 cm diameter, and were taken directly adjacent to the drill core holes left by previous studies. Hand samples were also collected in intervals with poor exposure for drilling (Table 1). The three-digit suffix on the end of each sample name reflects the stratigraphic height above the base of the unit in cm. Transmission electron microscope (TEM) observations by Schlinger et al.  demonstrate that Fe-oxide microcrystals in glass in the ash flow are spatially well-dispersed with a narrow grain size distribution and elongated shapes. Various magnetic analyses by Worm and Jackson , Pike et al. , Egli and Lowrie , and Shcherbakov and Fabian  determined that magnetostatic interactions were detectable but weak in samples from the lower 2 m of the unit.
Table 1. Sampling Information for the Studied Specimens
The Fe-oxide grain sizes and shapes measured by Schlinger et al.  that correspond to our samples taken at equivalent heights within the bottom 2.5 m of the unit are listed in Table 1. From measurements reported by Schlinger et al.  there is evidently a trend toward increasing grain aspect ratios with height in the studied section (Figure 1b). A small number of specimens from sampling heights above 2.5 m were polished using progressively finer grades of diamond lapping films and were analyzed with a scanning electron microscope (SEM) to verify the continuous increase in microcrystal dimensions. These observations are also included in Table 1. Although the resolution of the images is not great enough to make precise particle size measurements, the Fe-oxide laths appear to reach a size of around 1 μm long and 0.1 μm wide at the top of the section. Oxide particles in samples below 3.95 m could not conclusively be identified in SEM images. Magnetic particle volumes calculated from the published TEM measurements have an approximately logarithmic increase in grain volume with sample height, z (Figure 1a). Thus, the variations in magnetic properties that we report as a function of stratigraphic height may be equivalently viewed roughly as a function of log-grain size.
Schlinger et al.  observed a transition from what they termed the basal subzone to the middle subzone of the Tiva Canyon Tuff several meters above the base of the unit, noting a color change associated with this transition. In both TEM and optical microscope observations of samples from the middle subzone, Schlinger et al.  saw Fe-oxides in the form of tangles and linear aggregates of elongated magnetite. They suggested that these nonuniform distributions of Fe-oxide particles in glass signify a transition from homogeneous to heterogeneous nucleation between the basal and middle subzones. For the samples in the present study, we observe elongated microscale oxides in glass in sample TC04-395 with approximate particle lengths around 0.5 μm (Figure 2a). In this sample, two-particle intergrowths appear occasionally, but magnetite primarily occurs as dispersed, isolated grains. Above this height, samples TC04-455 and TC04-502 contain some dispersed Fe-oxide grains, but magnetite occurs primarily as clusters of elongated particles identical to those shown in Figure 11 of Schlinger et al. , that appear to have nucleated heterogeneously. Sample TC04-502 is also noticeably darker in color than the lower samples. Therefore, it seems likely that the transition from the basal to the middle subzone in our sampled section lies between 4 and 4.5 m above the base of unit. Because the clustered magnetite particles in the middle subzone are closely spaced, we may expect the magnetic properties in the highest part of the section to be influenced by magnetic interactions.
For all samples, room temperature low-field bulk magnetic susceptibility, χ0, was measured on a KLY-2 Kappabridge AC susceptibility meter. For selected samples, low-temperature frequency dependent magnetic susceptibilities were measured from 10 to 300 K or from 10 to 400 K using a Quantum Designs Magnetic Properties Measurement System at frequencies between 1 and 1000 Hz.
All samples were subjected to a 3-axis alternating field demagnetization with a maximum field of 200 mT. After demagnetization, samples were given an anhysteretic remanent magnetization (ARM) in an arbitrary direction using a direct current bias field of 0.01 mT. The anhysteretic susceptibility, χARM, was determined by normalizing the remanent magnetization by the strength of the bias field. Remanence measurements were performed on a 2-G Enterprises superconducting 760-R SQUID magnetometer. Hysteresis loops were measured for representative samples on a Princeton Measurements Corporation MicroMag 2900 vibrating sample magnetometer with a saturating field of 1.0 T.
3.1. Magnetic Susceptibility Measurements
Bulk magnetic susceptibility rises approximately linearly with stratigraphic height to a sharp maximum in samples TC04-075a, TC04-075b, and TC04-080 (Figure 3a), at approximately 0.75–0.80 m above the base of the unit. Above this level, χ0 decreases continuously with increasing stratigraphic height and becomes nearly constant above 3.0 m. This pattern is similar to the variation in χ0 theorized for an ideal collection of particles with increasing grain volume near the transition from a SP magnetic domain state to a stable SD state [e.g., O'Reilly, 1984, chap. 4]. The saturation magnetization, Ms (see next section) does not vary strongly as a function of stratigraphic position, and therefore the concentration of ferrimagnetic material is relatively uniform and does not account for the χ0(z) pattern. The peak in χ0 at around 0.80 m corresponds to a χ0/Ms ratio of about 8 × 10−5 m/A, which indicates a dominant magnetic grain size near the SP-SD boundary at room temperature [e.g., Hunt et al., 1995a].
Low-temperature (T < 400 K) measurements of alternating current (AC) susceptibility have a significant frequency dependence at low temperatures, and a unique AC susceptibility signature for each stratigraphic level (Figure 4). Sample numbers TC04-255 and lower (z = 2.55 m) have pronounced frequency dependence over a limited temperature spectrum, the range of which varies according to sample position. The temperature range of frequency dependence shifts to higher temperatures with increasing stratigraphic height, and the frequency dependence becomes negligible over the measured temperature range at ∼3.25 m (sample TC04-325). For samples with local maxima in low-temperature susceptibility, the location of the peaks represents blocking temperatures at that particular frequency. The highest blocking temperatures are indicated for the highest AC frequencies and they shift to lower temperatures with lower frequencies. Within the temperature range of observable frequency dependence, an out-of-phase component of susceptibility is also evident for each sample. The room temperature frequency dependence (300 K; see Figure 4) is strongest in sample TC04-080 (0.80 m), which coincides with the location of maximum χ0 in the section.
3.2. Remanence and Hysteresis Measurements
χARM decreases slightly with stratigraphic height up to 0.75 m (the location of the sharp peak in χ0), then increases to a peak at 3.25 m, and then decreases above 3.25 m (Figure 3a). The maximum in χARM is thought to represent the upper limit of stable SD behavior in the section. The samples have continuously varying χ0 and χARM values, therefore we display these data on a bivariate plot using the method of relative magnetic grain size determination described by Banerjee et al.  and King et al.  (Figure 5a). In this plot we see three distinct trends corresponding to three magnetic domain states. The lowest samples from the studied section have small values of χARM that are typical of SP material and plot along the abscissa. Samples in the SD region are characterized by a smooth progression from a maximum in χ0 and a minimum in χARM at the low end of the SD size range, to lower χ0 and higher χARM with increasing grain volume. Samples above 2.55 m have relatively low χ0 and decreasing χARM (corresponding to increasing stratigraphic height and grain size) and are inferred to exhibit pseudo-single-domain (PSD) behavior. Samples from each stratigraphic height cluster tightly in discrete, and largely unique, areas on the plot, which reflects the narrowness of the grain size distribution at each stratigraphic level.
Hysteresis results (Figure 3c) for most samples from the lower 1.0 m of the unit are characterized by low ratios of saturation remanence, Mrs, to Ms (Mrs/Ms < 0.1), low coercivity, Bc (Bc < 10 mT), and relatively high coercivity of remanence, Bcr (Figure 3b), which is typical of SP particles with some admixture of larger particles. There is some variation in Ms values, but the variation is not a function of height (Figure 3d). The scatter in Ms likely represents heterogeneity in magnetite concentrations due to varying proportions of pumice or lithic fragments on the scale of the specimen size. Within the lowest 0.80 m of the section, there is a small but distinct decrease in both Mrs/Ms and Bc, resulting in minimum values that spatially coincide with the peak in χ0. Mrs/Ms and Bc increase with height up to what is probably the location of the upper limit for stable SD behavior in the unit, near the peak in χARM, and both values decrease above 4.0 m, as expected for PSD-sized grains. Hysteresis parameters viewed in the form of a Mrs/Ms vs. Bcr/Bc plot [Day et al., 1977] have considerably more overlap between samples (Figure 5b) than in the χ0 vs. χARM plot, as the changes in hysteresis properties are less pronounced than those in χ0. Additionally, on the Mrs/Ms vs. Bcr/Bc plot, several samples from nonadjacent layers occur in similar positions and the boundaries between regions dominated by different magnetic domain states are not as clearly delineated.
It is important to note that for samples inferred to contain grains in the SD region, the general trend with increasing stratigraphic height (and thus increasing grain size) in Figure 5a is for the line connecting each point with the origin to have steeper slopes for increasingly coarse grains. The opposite trend was suggested by King et al.  for grains in the SD to MD range, where steeper slopes are associated with increasingly fine grains. King et al.  also cautioned that narrow size distributions of SP and MD grains would look similar on this plot, and could be a source of ambiguity in determining relative grain size if both grain size fractions were present in a set of samples. Another complicating factor noted by King et al.  is that both χ0 and χARM are affected by grain shape anisotropy, where increasing shape anisotropy shifts χ0 to lower values and χARM to higher values [Stacey and Banerjee, 1974]. The aspect ratios measured by Schlinger et al.  (Figure 1b) and observed in this study indicate increasing particle elongation with particle volume, and therefore the trends with stratigraphic height in Figure 5a may be slightly different for particles with constant shape anisotropy, or with less elongation.
Magnetic properties of samples in the current study are compared with previously reported values for various sized magnetites in Figure 6. The mean magnetite concentration in our samples was approximated by taking the average Ms value for all measured samples and assuming a saturation magnetization value of 80 Am2/kg for magnetite with a TM10 composition [Hunt et al., 1995b], which results in an estimated mean concentration of 0.28 wt% magnetite for all samples. This value was used to normalize bulk magnetic susceptibility and χARM values for comparison with values for pure magnetite samples. Because the magnetic particles in the Tiva Canyon Tuff are variably elongated while most studies of synthetic magnetite involve cubic or equant particles, magnetic properties are plotted as a function of grain volume.
The lower SD grain volume threshold delineated by a peak in χ0 occurs at a slightly lower grain volume in the Tiva Canyon samples than in the synthetic magnetites measured by Maher  (Figure 6a). This is in agreement with calculations by Butler and Banerjee  and Muxworthy and Williams , which predict magnetic blocking volumes for magnetite with aspect ratios of 0.3–0.4 to be slightly lower than for equidimensional magnetite. Bulk magnetic susceptibility in our samples is also notably higher than reported values for synthetic magnetites over the lowest size range, presumably due to interaction effects in incompletely dispersed synthetic samples. This was also suggested by Worm and Jackson , who noted that the χ0 values measured by Maher  were well below the theoretically predicted values. Data for Bc and Bcr are more scattered (Figures 6c and 6d), although our data appear to be roughly consistent with the pattern of size dependence seen in several synthetic sample sets.
χARM reaches a maximum at slightly higher particle volumes than most synthetic magnetites shown in Figure 6b, although the χARM reported for whole cells of magnetotactic bacteria containing elongate chains of SD magnetite by Moskowitz et al.  is slightly higher than for any of our samples. As with χ0, the higher χARM values in our samples may be attributed to the absence of interactions effects, although the rough normalization performed on the data may also account for some of the discrepancy.
Model calculations indicate that the SD size range is extended to larger volumes with higher particle elongation, which predicts an upper SD size limit at particle lengths over 1 μm for an aspect ratio of 0.1 [Butler and Banerjee, 1975; Muxworthy and Williams, 2009]. However, while the peak in χARM almost certainly marks the upper limit of SD behavior in the section studied here, it is unclear whether this change in domain state is due to a particle volume threshold in independent elongated grains, or whether it simply marks a transition to inhomogeneous nucleation of magnetite grains in the studied section (Figure 2). Intergrowths of magnetite that produce larger effective particle volumes may cease to exhibit SD behavior. The decreases in Mrs/Ms and Bc above 3.25 m also suggest PSD or metastable SD behavior. Since ARM is known to be especially sensitive to even weak interactions [Egli, 2006], the clustered and intergrown magnetite observed here (Figure 2) and by Schlinger et al.  in the upper part of the section suggest that the sharp drop in χARM may reflect the presence of magnetostatic interactions.
If we conversely assume that the non-SD characteristics in samples above 3.25 m are attributed to isolated magnetite particles, the transition in magnetic properties at that height might indicate that transdomain or local energy minima (LEM) domain states become dominant in individual particles in the upper part of the section. Micromagnetic modeling by Fabian et al.  indicates that LEM states exist in elongated magnetite at smaller particle lengths that those predicted as the upper SD limit by the Butler and Banerjee  model. Dunlop et al.  also predicted multiple possible domain states, including LEM states, between grain lengths of 140 and 500 nm for magnetite with an aspect ratio of 0.67 at room temperature, although neither of the models mentioned above include grains with elongations as extreme as those in our uppermost samples. To our knowledge, there are not sufficient data on highly elongated magnetite covering a broad range of sizes or volumes to know precisely where the limit of stable SD behavior lies in elongated grains.
The data presented here demonstrate the degree of variation in rock magnetic properties for populations of dispersed, noninteracting magnetic particles with uniaxial anisotropy and narrow grain size distributions covering the span of SP and small SD domain ranges. Magnetic particle volumes increase progressively upward above the base of the unit. In the lowermost 2.5 m of the unit, mean particle volumes increase exponentially with height from ∼500 nm3 to ∼100,000 nm3. Particle size continues to increase upward over the lowermost 5 m of the unit, with magnetic properties indicating a SP-SD threshold at about 0.8 m and a transition away from stable SD behavior at 3.25 m. For these samples, plotting χ0 vs. χARM provides a clear indication of relative magnetic grain size, while domain regions and differences in grain size are less easily distinguished on a plot of hysteresis parameter ratios. The samples studied here afford unique insight into the size-dependent behavior of fine magnetic particle assemblages.
The authors thank G. Skipp and J. Honke for assistance with field work. A. Roberts, L. Chang, M. Hudson, and R. Reynolds are thanked for providing helpful and constructive reviews of this manuscript. This work was supported by grants 0218384 and0317922 from the Instruments and Facilities Program of the Earth Science Division of the National Science Foundation. The National Science Foundation also provided an REU opportunity in conjunction with this project. This is IRM contribution 1007.