Hematite and Mn oxide (U-Th)/He dates from the Buckskin-Rawhide detachment system, western Arizona: Gaining insights into hematite (U-Th)/He systematics

Hematite He diffusion kinetics.

As in most minerals, the lengthscale of the He diffusion domain in hematite appears to scale with the physical crystal or grain size (Bahr and others, 1994; Farley and Flowers, 2012). In polycrystalline hematite samples analyzed in this study, the size of hematite crystals varies by one to two orders of magnitude within each sample. The bulk He retentivity of these samples may be best described by a “poly-domain” model. We use the term “poly-domain” when referring to He diffusion from polycrystalline aggregates. In such samples where a range of crystal sizes are present, it may be expected that multiple domains inferred from diffusion data may correspond to diffusion from crystals of different sizes (Farley and Flowers, 2012). We use “poly-domain” instead of “multi-domain” with the rationale that the latter would describe diffusion from multiple domains contained within what appears to be a single crystal or crystal fragment. He diffusion is typically modeled using the Arrhenius equation (eq. 1): where D = diffusivity, a = the diffusion domain lengthscale (which we often refer to as domain size), T = temperature, R = the gas constant, E_a = activation energy, and D₀ = frequency factor. Bahr and others (1994) first proposed the idea that the size of diffusion domains scaled with physical crystal or grain size in hematite, so that crystal or grain size may be an important control on the effective bulk He retentivity of hematite samples. Farley and Flowers (2012) showed that results of step-heating ³He diffusion experiments on a polycrystalline hematite sample were consistent with the presence of multiple diffusion domains that display constant E_a, but with D₀/a² values that vary over 20 to 30 natural log units. Based on comparisons between their step heating diffusion results and crystal size variation observed in SEM images of their sample, Farley and Flowers (2012) speculated that crystallite sizes (r values, in μm) observed in a sample may correspond directly to diffusion domain sizes (a values), while D₀ and E_a are constant across these domains. If this is correct and applies to all hematite, then we can substitute D₀/r² for D₀/a² in equation 1. This substitution and natural log transformation yields equation 2: For a dataset containing D/a² values taken at one temperature from multiple hematite samples, the (−E_a/RT) and ln(D₀) terms in equation 2 are constants. A plot of the natural log of these apparent diffusivity values versus the natural log of the inverse square of the physical crystal or grain size (1/r²) that each diffusivity value corresponds to should yield a linear relationship.

To test this idea, we compiled apparent He diffusivities (D/a² values) at 350 °C and estimates of corresponding crystal size data (r values, equal to half widths or radii of crystals or crystal fragments) from hematite diffusion experiments in several previous studies (table 10). In datasets where no pronounced poly-domain or multi-domain diffusion behavior was evident, we calculated D/a² values by interpolating the Arrhenius trend between two data points or by extrapolating the trend (whichever was necessary to obtain a D/a² value at 350 °C). In datasets from samples described by other authors as polycrystalline aggregates where He diffusion from multiple domains was apparent (poly-domain diffusion), we obtained D/a² values by extrapolating single-domain lines fit to sub-linear segments of the Arrhenius trend in the early, low-T part of the dataset, and used the smallest crystal size identified in the sample when assigning r values to these D/a² values. We obtain data points from ³He diffusion data of Farley and Flowers (2012) as explained in figure A4. Figure 11 shows the data from table 10 in a plot of ln(D/a²) at 350 °C versus ln(1/r²). A line fit to the linear array that these data form has a slope of 0.817 ± 0.10 (R² = 0.90, 1σ error). Broadly speaking, this relationship is consistent with correspondence of diffusion domain size and crystal size, and a restricted range of E_a and D₀ for He in hematite. Similar relationships have been observed in the titanite and apatite (U-Th)/He systems (Reiners and Farley, 1999; Farley, 2000). This is consistent with the findings of Farley and Flowers (2012).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 11.

Plot of natural log-transformed apparent diffusivity at 350 °C versus natural log-transformed inverse square of observed hematite crystal or grain radius or half-width. The inner part of the x-axis shows the equivalent crystal or fragment radii for several ln(1/r²) values. Lower values of ln(1/r²) correspond to larger crystal or grain radii. Points on this plot were taken from data in this study and other studies, references and data shown on the plot and in table 10. Open symbols represent data taken from this study and Farley and Flowers (2012), filled symbols are from older studies. Several of the open data points are derived from ³He diffusion data, as shown on the plot. This figure shows that there is a direct relationship between D/a² and 1/r², which supports the argument that crystal or grain size corresponds to diffusion domain size, and diffusion domain size is the primary control on D/a² variation at a given temperature. The best-fit line for these data has a slope slightly lower than expected, and its intercept implies a D₀ value of 5.2 × 10⁻³ cm²/s (1σ error: +3.3 × 10⁻²/−4.5 × 10⁻³ cm²/s). A line with the expected slope of one (see eq. 2 in text) provides a similarly good fit to these data, and implies a lower D₀ value of 2.2 × 10⁻⁴ cm²/s (1σ error: +1.6 × 10⁻³/−1.9 × 10⁻⁴). We note the large uncertainty on these values, but consider the latter value as our best estimate of D₀ in hematite.

If the value of E_a for He diffusion in hematite is known or estimated, then we can use this value and the y-axis intercept of the linear relationship in figure 11 to calculate an estimate of D₀ (eq. 2). Published E_a values for He diffusion in hematite range from ∼120 to 180 kJ/mol (Lippolt and others, 1993; Wernicke and Lippolt, 1994; Bahr and others, 1994; Farley and Flowers, 2012). Farley and Flowers (2012) used ³He diffusion data to obtain E_a = 157±6 kJ/mol. We suspect that many of the smaller values of E_a from ⁴He data in earlier studies (Lippolt and others, 1993; Wernicke and Lippolt, 1994; Bahr and others, 1994) may be a result of the contribution of He from multiple domains to early steps of the diffusion experiments from which E_a estimates are calculated (Lovera and others, 1997). Examples of this can be seen in the convex-up shapes formed by low-temperature steps in some Arrhenius trends from Bahr and others (1994). Sub-linear segments of the ³He Arrhenius trend from sample FW2-12-85-8 yield E_a = 160 to 180 kJ/mol (fig. 8A), slightly higher than those obtained by Farley and Flowers (2012). We note that this sample is texturally distinct from other samples analyzed in this study and the sample that Farley and Flowers (2012) examined in that it comprises an apparent single crystal fragment. Therefore, this sample may yield somewhat different kinetic parameters (fig. 2A). Sub-linear segments of low-T portions of the Arrhenius trend from sample FW11-29-11-1 in this study yield E_a = 147 and 148 kJ/mol (fig. 7). Because these values are similar to the value obtained by Farley and Flowers (2012), and are derived from a sample that is texturally similar to most other samples examined in this study, we shall proceed using the average of these two values as our assumed E_a value (E_a = 147.5 kJ/mol).

Using this E_a value, the fixed T value for the dataset (350 °C), and the intercept of the best-fit line (dashed, fig. 11), we calculate a ln(D₀) = −5.3 ± 2.0 (1σ error estimate from uncertainty on intercept; eq. 2). This yields a D₀ value of 5.2 × 10⁻³ cm²/s (+3.3 × 10⁻²/−4.5 × 10⁻³ cm²/s). We note that the line used to calculate this value has a slope of less than the expected value of one (eq. 2), therefore, we fit another line with a fixed slope of one to the data (solid line, fig. 11). This line is only a slightly poorer fit to the data than the best fit line (R² = 0.85, best fit R² = 0.90). Following the above calculation, the intercept of this slope = 1 line yields a D₀ value of 2.2 × 10⁻⁴ cm²/s (+1.6 × 10⁻³/−1.9 × 10⁻⁴). As this value is calculated from a line that is fit to the data while also matching the form expected from equation 2, we favor it over the higher value obtained from the best fit line. We recognize that this value of D₀ bears a large uncertainty, but for the purposes of this study, we suggest that it is the best available estimate.

Hematite He diffusion data from this study.

The size of the smallest domain calculated from the FW11-29-11-1 diffusion experiment using hematite E_a and D₀ values determined in this study and an assumed plane sheet domain geometry (r = 1.3 μm, from P1 segment in fig. 7B) is consistent with the smallest observed crystal size in this sample (table 1, fig. 2D). This match between calculated diffusion domain size and physical crystal size observations is consistent with the findings of the compilation of diffusion data described above, and with the hypothesis that crystal size is a primary factor controlling He diffusivity in hematite (Farley and Flowers, 2012).

⁴He/³He diffusion data from sample FW2-12-85-8 indicate that domains smaller than the observed sample chip size may be present within the sample. Because there are no clearly visible internal structures within the aliquot itself, or other crystals of this sample examined in SEM, we hypothesized that this sample held all He that it contained in a single large diffusion domain, with an a value approximately equal to r = 80 μm (the observed minimum half width of the sample chip).

The results of MDD modeling of diffusion data from sample FW2-12-85-8 (using kinetic values discussed above) suggest that >99 percent of the ³He in the sample was released from domains with a values of ∼1 to 34 μm, with about 80 percent from domains with a values of ∼14 to 34 μm (figs. A3A-A3B). These inferred a values are lower than the observed half width (r value), but given the uncertainties on E_a and particularly D₀, a difference between the a values obtained from MDD models and the measured crystal fragment radius of less than one order of magnitude is not unreasonable, and thus these data could be cited in support of the inferred relationship between observed crystal size and diffusion domain size (Farley and Flowers, 2012).

To summarize the ³He diffusion results, our interpretations from sample FW2-12-85-8 suggest that He diffusion occurs from domains similar to but slightly smaller than the observed crystal size. Although in detail this contradicts the findings of our compilation of diffusion data, and of Farley and Flowers (2012), we note that the ³He diffusion results imply that the majority of the ³He resides in a domain with a size that is of the same order of magnitude as the bulk crystal fragment size (fig. A3). Thus, if the observed crystal or grain size/s and diffusion domain size/s inferred from diffusion data are not identical, they may scale to first order.

The ⁴He diffusion data from sample FW2-12-85-8 suggest that radiogenic ⁴He has a significantly lower bulk retentivity than ³He in this sample. If this arises entirely from effective diffusion domain size differences, then our modeling suggests that significantly more of the radiogenic ⁴He than ³He is held in the small domains; more specifically, about 35 to 40 percent of the ⁴He is in domains with a < 1 μm (figs. 8 and A3C-A3D). We hypothesize that this difference between the ⁴He and ³He diffusion behavior arises from a spatial correlation between the source of ⁴He (U and Th) and locations of the small domains, such that ⁴He is preferentially implanted into smaller domains. We suggest that smaller domains of hematite and other phases that are eU-rich relative to surrounding hematite clustered together, perhaps surrounding cracks, inclusions, or other defects in the sample (fig. 12). These clusters will have a higher “domain surface area” per unit of sample volume and thus a potentially higher eU concentration than parts of the sample made up of larger diffusion domains of low-eU hematite (figs. 4A and 12B-C). In order for these clusters of smaller domains to capture a large fraction of the ⁴He produced by the decay of U and Th that they contain, the half-width of the clusters must be greater than or equal to the alpha stopping distance in hematite (13-18 μm, Ketcham and others, 2011).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 12.

Schematic illustration of the aliquot of hematite from sample FW2-12-85-8 analyzed in ⁴He/³He experiment. (A) Cross-section through a euhedral crystal showing the internal location of the analyzed sample chip, and the outline of the part of the chip that is illustrated in (B). (B) Cross-section through portion of analyzed hematite sample chip showing a hypothetical distribution of diffusion domains surrounding a central crack or other defect. (C) Curves showing hypothetical spatial distribution of eU, ⁴He, and diffusion domain sizes within the cross-section shown in (B). The clustering of smaller domains of hematite and perhaps other, eU-rich phases along the crack results in a higher eU concentration in this zone. Ejection of some of the ⁴He produced by decay of this eU into domains farther from the crack widens the zone of elevated ⁴He concentration, but ⁴He concentration is highest in the small domains surrounding the crack.

The ⁴He/³He diffusion data from sample FW2-12-85-8 may also be explained by models other than the one presented above. Radiation damage provides an alternative, though perhaps less likely explanation for the presence of very small, ⁴He-rich diffusion domains in hematite. It is possible that alpha-recoil-induced radiation damage contributes to the reduction of effective diffusion domain size in parts of the sample where eU is concentrated.

Without additional data that replicate the kind of inconsistency observed between ⁴He and ³He diffusion data from FW2-12-85-8, we conclude that, for most hematite samples, crystal size is a primary control on He diffusivity in hematite. Therefore, the findings of the compilation of diffusion data obtained above (including the estimated value for D₀ in hematite) are also considered valid. Additional studies of He diffusion in hematite, which include a careful examination of hematite crystal structure in analyzed samples, are necessary to better understand this process.

In order to compare hematite (U-Th)/He dates obtained in this study with cooling dates from other studies of lower-plate cooling in the Buckskin-Rawhide detachment system, we need to estimate the He closure temperature (T_c) of hematite samples. We use observed crystal half-width values as an estimate of diffusion domain radii (a values, see table 1), various domain geometries and cooling rates, and assumed diffusion kinetic values stated earlier in this subsection to calculate effective closure temperatures for the smallest and largest crystals observed in each hematite sample (Dodson, 1973; table 11). The cooling rate (80 °C/Myr) that we have applied above in calculating closure temperatures from He diffusion data (figs. 7 and 8) is taken from Scott and others (1998). Their study suggests that this rate was applicable to lower-plate cooling for much of the ∼300 to 150 °C temperature window, the range in which previous estimates of He T_c in coarsely crystalline hematite fall (Lippolt and others, 1993; Bahr and others, 1994). We also include T_c estimates that incorporate the highest cooling rate cited by Scott and others (1998, 280 °C/Myr). Due to the plate-like or tabular shape of hematite crystals in most samples, we suggest that T_c ranges that assume plane sheet domain geometry provide better estimates. Closure temperature ranges that assume an 80 °C/Myr cooling rate and plane sheet domain geometry fall largely between 140 to 240 °C. Given the uncertainty in the E_a and D₀ values used to calculate these estimated T_c ranges, we do not attempt to calculate more precise bulk closure temperatures.

Results of the FW2-12-85-8 ⁴He/³He step-heating experiment suggest that the true closure temperature of this sample may be <180 °C. MDD models of ⁴He diffusion data from this sample imply a bulk closure temperature of ∼150 to 180 °C (calculated for 80 °C/Myr cooling rate, spherical domain geometry, figs. 8B, A3C-A3D), much lower than closure temperature estimates based on crystal size (table 11). We have no clear evidence that the mismatch between observed crystal size and ⁴He diffusion domain size inferred from MDD models (fig. 8) of data from this sample extends to other hematite samples, which are made up of crystals that are much smaller, contain fewer inclusions, and are plate-shaped rather than rhombohedral (fig. 2).

He diffusion in Mn oxide.

Using kinetic parameters from previous studies of He diffusion in Mn oxide and estimates of crystal size from SEM images, we estimate that Mn oxide samples dated in this study have closure temperatures of ∼36 to 57 °C. Lippolt and Hautmann (1995) performed a step-heating experiment on a microcrystalline sample of cryptomelane-hollandite composition. These phases have the same crystal structure as Mn oxide minerals identified in Shannon and Priceless Mine samples in this study (2×2 tunnel structure; Post, 1999), leading us to apply the He diffusion kinetic parameters from Lippolt and Hautmann (1995). Following the reasoning we applied to hematite samples, we assume that crystal size (in μm) is approximately equal to the diffusion domain lengthscale (a), and estimate a = 0.1 μm (a lower-end estimate of a, fig. 9; Farley and Flowers, 2012). Using the E_a and D₀ values calculated from the low temperature (<500 °C) steps of the diffusion experiment in Lippolt and Hautmann (1995; 134 kJ/mol and 3.981 × 10⁻³ cm²/s, respectively) and an assumption of spherical diffusion domain geometry, we calculate closure temperatures of 36, 49, and 57 °C for cooling rates of 10, 80, and 280 °C/Myr. These closure temperatures are consistent with the findings of Reiners and others (2014), who suggest that fine-grained Mn oxide of a similar composition and apparent crystal size will retain most of its He when exposed to temperatures of up to ∼55 °C for several million years.

Possible causes of date dispersion in hematite samples.

Systematic relationships among (U-Th)/He date, aliquot type, pre-treatment, and the molar ratios of eU and other elements (noted in the Results section above) lead us to a discussion of the causes of date dispersion in these samples. Thermal histories obtained by Scott and others (1998) suggest that lower-plate rocks in the Buckskin Mountains cooled through the estimated range of He closure temperatures in hematite (140-240 °C, table 11) over a relatively short period of time (∼1 Myr). This period of time is less than the variation in (U-Th)/He dates observed in aliquots from each hematite sample (fig. 3). The textural homogeneity of hematite crystals in most samples (euhedral, formed early in the paragenetic sequence—see table 1, fig. 2) leads us to infer that hematite formation, like cooling, occurred over less than 2 Myr. If these inferences are correct, then variations in the formation age or closure temperature of hematite crystals within a single sample are not likely to be the causes of (U-Th)/He date dispersion observed in some samples.

In several hematite samples, cleaned aliquots, particularly those made up of hematite crystals alone (PO aliquots), yielded lower molar ratios of Mg, Mn, and eU (figs. 4C-4E and fig. 5). We interpret the difference between cleaned and NC aliquot compositions in these samples to reflect disaggregation and removal of interstitial phases from hematite-rich parts of the sample material during ultrasonic cleaning. EPMA analyses of common interstitial phases yield higher molar ratios of Mg and Mn than hematite crystals, supporting this assertion (tables 2 and 3). Analyses of chrysocolla from sample FW3-23-85-8 indicate that this phase contains a large amount of U (table 5); we note that aliquots of the two hematite samples that contain chrysocolla (samples FW3-23-85-8 and FW3-21-85-12, figs. A1 and A2) also yield the highest eU molar ratios observed among all samples (fig. 4). While we did not obtain direct measurements of the U and Th contents of other common interstitial phases, it is plausible that some of these phases (chlorite, calcite) also display a high molar ratio of U and/or Th relative to hematite.

Removal of parent nuclides during ultrasonic cleaning has a clear effect on the (U-Th)/He dates obtained from cleaned aliquots. As a result of the long stopping distance of radiogenic ⁴He atoms (13-18 μm in hematite, ∼20-25 μm in less dense interstitial phases; Ketcham and others, 2011), much of the ⁴He produced by decay of U and Th in interstitial phases is likely to be implanted in neighboring hematite crystals and retained over geologic time. The loss of parent nuclides contained in these interstitial phases following a long (several Myr) period of He implantation will result in an increase in the (U-Th)/He dates obtained from the surrounding sample material. If this removed parent material was added to a sample at a near-zero date, it is possible that the resulting apparent (U-Th)/He dates obtained from the sample will be approximately equal to its true He closure age. We find little geologic or textural evidence for late addition of U or Th (via uptake on mineral surfaces or incorporation into growing interstitial phases) to hematite samples. We therefore interpret the observation that cleaned aliquots tend to yield older ages than NC aliquots (fig. 3) as an artifact of interstitial phase removal and accompanying parent nuclide loss. Under this interpretation, cleaned aliquots yield apparent (U-Th)/He dates that are older than the true He closure age, and NC aliquots (which contain a greater fraction of ‘initial’ U and Th) provide the best estimate of He closure age. The inverse relationships observed between (U-Th)/He date and the molar ratios of eU and other elements in some samples (figs. 4C-4E and fig. 5) can also be attributed in part to parent nuclide loss.

In cases where large accumulations of eU-rich interstitial phases are present in analyzed sample material, ⁴He loss from these phases may also contribute to the observed inverse relationships between date and eU or other elements. The fraction of ⁴He produced in a given body of interstitial material that is implanted in neighboring hematite crystals is inversely related to the width of the space between hematite crystals, and directly related to the alpha stopping distance in interstitial phases. Working under the simplifying assumptions of parallel orientation of hematite crystals and a uniform alpha stopping distance of 25 μm in interstitial phases, we estimate that >10 percent of ⁴He produced in interstitial phases that fill large spaces (>10 μm wide) between platy hematite crystals will not be implanted into hematite (fig. A5). This fraction increases rapidly as the spacing of neighboring hematite crystals rises. Given the high He diffusivities of common interstitial phases such as calcite (Cros and others, 2014), quartz (Shuster and Farley, 2005b), and chrysocolla (which displays an amorphous structure analogous to glass and is therefore unlikely to retain He; Shelby, 1972; Jambon and Shelby, 1980; Frost and Yi, 2013), it is likely that most ⁴He that is not implanted in hematite will be lost over million-year timescales. Among all samples, four aliquots yield apparent (U-Th)/He dates younger than the cessation of extension along the detachment fault (estimated at ∼8 Ma, see Geologic Setting). Of these aliquots, two from sample FW3-23-85-8 yield high molar ratios of eU (fig. 4B), Mg (fig. 5A), and Ca (table A2) and low Th/U ratios (table 4), all characteristics that support the presence of a relatively high volume fraction of chrysocolla in these aliquots (tables 3, 5). The 5.8 Ma aliquot from sample FW11-29-11-1 displays similar characteristics relative to other aliquots in that sample. We suggest that young dates obtained from these aliquots result from the loss of large fractions (30-50%) of ⁴He produced in 10 to 50 μm-scale pockets of highly eU-enriched chrysocolla and/or chlorite, that was not implanted in neighboring hematite crystals (fig. A1). We exclude the dates of these <8 Ma aliquots from further consideration (noted as ‘discarded’ in table 4).

In the two preceding paragraphs, we discussed ways that interstitial phases (and the removal thereof) can contribute to inverse relationships between date and the molar ratio of eU and other elements. It is possible that volatilization of parent nuclides during He extraction may also contribute to these relationships. During high temperature He extraction, parent nuclides and other elements may be volatilized and lost from the sample. The difference in the molar ratios of Cu and Pb between undegassed (not heated to high temperatures) and dated hematite aliquots indicate that these elements are volatilized during He extraction (fig. 6). In contrast, the molar ratios of Mg (figs. 5A-5B) and Ca (table A2) measured in most dated aliquots are consistent with those measured in undegassed aliquots and greater than or equal to those measured in hematite alone (table A1), which suggests that these elements are typically retained through the degassing process. These observations are consistent with relative volatilities of these elements (Larimer and Anders, 1967; Lamoreaux and others, 1987; Lodders, 2003). In these studies, both Th and U are found to be nearly as refractory as Mg and Ca. Based on this information and the data presented above, we suggest that U volatilization does not have a large effect on most dated aliquots, and that the effects of cleaning on (U-Th)/He dates of aliquots (as manifested in inverse relationships between date and element molar ratio) are typically of a greater magnitude than volatilization.

A few aliquots display characteristics that we interpret as signs of the loss of large fractions of parent nuclides. Three aliquots (from samples FW3-23-85-8 and FW2-12-85-8) yield dates that are older than the earliest radiometric date of detachment activity (26 Ma; Lucchitta and Suneson, 1993), along with low molar ratios of eU (figs. 4B-4C) and Mg (below the minimum molar ratio in hematite crystals; tables 2 and 4) and high Th/U ratios (a sign of U volatilization identified by Danišík and others, 2013 and Vasconcelos and others, 2013, table 4). It is difficult to determine the extent to which these aliquots were affected by volatilization; however, the low Mg molar ratio in these aliquots might imply that refractory elements (including U and/or Th) have been lost. Evidence of inclusions was observed in sample FW2-12-85-8 (fig. 2A, table 1). Inclusions in analyzed aliquots of this sample may have contained ⁴He present at the time of sample formation, which would result in older (U-Th)/He ages. Whether physical removal of interstitial phases, volatilization, or the presence of ⁴He in inclusions has had a larger effect on the measured dates of these aliquots, it seems that some parent material has been lost. We use the characteristics that these too-old aliquots display to identify several other aliquots that, despite yielding geologically reasonable ages (<26 Ma), are also believed to have also been affected by parent nuclide loss (induced by either cleaning or volatilization) to a large extent. We exclude the dates obtained from these aliquots (six in total) from consideration when interpreting the significance of hematite dates, and mark them as discarded in table 4.

This discussion has implications for future studies that utilize hematite (U-Th)/He dating. Results from cleaned hematite aliquots analyzed in this study illustrate how the use of ultrasonic cleaning or any other pre-treatment meant to remove U and Th sited outside of the hematite crystal lattice (for example, dissolution of phases other than the target phase, Fanale and Kulp, 1962) will produce the desired result (a date closer to the age of He closure/mineral formation) only in cases where removed U and Th was added very recently. It is more probable that cleaned samples will yield too-old ages that contribute to age dispersion (see Orme, ms, 2011 and Murray and others, 2014 for an exploration of an analogous problem in the apatite (U-Th)/He system). Such an outcome is especially likely in cases where the fraction of eU in phases vulnerable to removal is greater than that contained in hematite. In future studies, pre-treatments that aim to remove extra-hematite-lattice parent nuclides should be used carefully, and several aliquots from each dated sample should be excluded from cleaning and/or other pre-treatments to control for their effects.

Two hematite samples analyzed in this study do not display date-element relationships, and yield overlapping cleaned and NC dates (figs. 3 and 4A-4B). We attribute these characteristics to the relative scarcity of interstitial phases in these samples (figs. 2A and 2D), which may in turn indicate that eU is held within hematite crystals in these samples, where it is not easily removed by ultrasonic cleaning or other physical processes. As a result, there is less potential for eU loss during cleaning or other sample preparation procedures, and a higher likelihood that He produced by decay of eU within each aliquot is implanted into He-retentive hematite crystals.

Assessment of Mn oxide (U-Th)/He and compositional data.

(U-Th)/He dates from Mn oxide samples are compatible with the inferred ages of sedimentary units that are cross-cut by host veins. The oldest (U-Th)/He date from the Priceless Mine sample (15.7 Ma) is younger than the maximum depositional age of the host Chapin Wash Formation (∼18.5-16.2 Ma). The oldest of the Shannon Mine aliquots (8.5 Ma) is younger than the 9.6 to 9.5 Ma Manganese Mesa basalt, which underlies the sampled part of the Sandtrap Conglomerate at the Shannon Mine locality (Shafiqullah and others, 1980; Shackelford, 1980).

Comparisons of undegassed and dated aliquot compositions suggest that He-extraction-induced eU loss has not affected dates obtained from Mn oxide samples. eU molar ratios measured in dated aliquots from the Priceless Mine sample are approximately equal to or greater than those obtained from undegassed aliquots. This comparison provides no evidence for loss of U or Th from dated aliquots of this sample. No undegassed aliquots from the Shannon Mine sample were analyzed; in lieu of these data, we extend our conclusion regarding an absence of eU loss from the Priceless Mine sample to this sample.

EPMA-measured compositions of Priceless Mine Mn oxide overlap with the ICPMS-measured compositions of dated and undegassed aliquots of this sample, but neither group of analyses plots precisely on the hollandite-coronadite mixing line (fig. 10). Volatilization of Pb during He extraction (as observed in hematite, fig. 6) might explain why most dated aliquots contain little or no Pb, and thus plot below the mixing line near the horizontal axis of figure 10. In the case of undegassed aliquots (which were never heated to high temperature prior to dissolution) and EPMA analyses, low Ba and Pb contents might be explained in part by the substitution of other elements for Ba and/or Pb (see table 9; Post, 1999), or the presence of phases other than Mn oxide in sample material. One such phase may be an Fe oxide (fig. 9A), the presence of which is supported by Fe/Mn molar ratios of 0.05 to 0.11 in Mn oxide from the Priceless Mine sample (table A2). The closure temperatures calculated for Mn oxide samples in the previous sub-section do not account for the presence of phases other than 2×2 tunnel-structure Mn oxides (Post, 1999). Given the similarity of closure temperatures calculated for fine-grained Mn and Fe oxides by Reiners and others (2014), we suggest that the T_c values calculated for Mn oxides in this study (∼36-57 °C) remain a reasonable approximation of the actual T_c.

Comparison of hematite and Mn oxide (U-Th)/He dates to other radioisotopic dates of Buckskin-Rawhide detachment system activity.

Apart from the westernmost sample (FW3-23-85-8), mean dates of NC aliquots of hematite samples decrease from SW to NE (table 12, fig. 13). This is consistent with data from a number of other studies of footwall cooling in the study area (Foster and others, 1993; Scott and others, 1998; Brady, 2002). As discussed above, NC aliquots from FW3-23-85-8 appear to be prone to He loss from eU-rich interstitial phases with high He diffusivities, which may result in a shift to a younger mean sample date (fig. 13).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 13.

Hematite (U-Th)/He dates and dates from other studies of lower-plate cooling, plotted against distance in slip direction. Sample locations and references for dates from other studies are shown in figure 1. The horizontal axis shows the distance of the sample location northeast of the dashed reference line on figure 1, measured in the upper-plate transport direction (N50E). A schematic cross section that shows the present day configuration of lower and upper-plate rocks is shown along the x-axis. Hematite and Mn oxide dates plotted here are mean NC aliquot dates from each sample (table 12). We estimate that the closure temperatures (T_c) of hematite samples analyzed in this study fall between 140–240 °C (table 11). The closure/annealing temperatures of the other thermochronometers in the plot (calculated for elevated cooling rates appropriate for lower-plate rocks) are shown below the plot. Error bars on dates from other studies excluded to reduce visual clutter. 1σ uncertainties on most of these dates are ∼10–15% of date, except for AFT uncertainties, which are ∼25% of date. All hematite dates (except for the westernmost, FW3-23-85-8) overlap with biotite ⁴⁰Ar/³⁹Ar /⁴⁰K/³⁹Ar and ZFT dates from nearby samples, though these systems typically have higher T_c values than those estimated for hematite. The faded hematite data point above the westernmost hematite sample data point shows a hypothetical age for this sample if inferred He loss had not occurred.

We interpret evidence from fluid inclusions to suggest that hematite samples record cooling. Textural evidence from this and other studies suggests that the timing of quartz formation immediately preceded or overlapped with hematite formation in hematite-bearing mineral deposits in the Buckskin Mountains and other nearby detachment-hosted ore deposits (Halfkenny and others, 1989; Schuiling, ms, 1978). Fluid inclusion data indicate that quartz in Buckskin-Rawhide hematite-bearing deposits formed at temperatures >225 to 325 °C (Wilkins and others, 1986). These temperatures fall at the upper end of the estimated hematite T_c range (140-240 °C, table 11). We interpret this evidence to indicate that hematite formed at a temperature greater than its closure temperature, and therefore, hematite (U-Th)/He dates should record cooling.

Hematite (U-Th)/He dates overlap with dates obtained from various thermochronmeters in other studies of lower plate cooling in the Buckskin-Rawhide detachment system. In figure 13, hematite dates are plotted with dates from samples mapped in figure 1. In the western end of the study area, the mean date from sample FW2-12-85-8 (sample FW3-23-85-8 is excluded from consideration for reasons outlined in the first paragraph of this subsection) overlaps with biotite ⁴⁰Ar/³⁹Ar and ⁴⁰K/³⁹Ar dates [T_c ≈ 325-375 °C, Harrison and others (1985); fig. 13], as well as apatite fission-track dates from nearby samples. The average dates from the group of hematite samples located near the eastern end of the study area fall more clearly below biotite K-Ar dates and overlap within analytical uncertainty with ZFT dates from nearby samples (ZFT 1σ errors = 10-15%). While it is encouraging that hematite (U-Th)/He dates can replicate dates obtained from other studies of footwall cooling, the large amount of uncertainty on the hematite closure temperature and the small range of dates obtained from other thermochronometers make it difficult to use these data to test whether hematite dates record cooling or mineral formation.

Regardless of which process hematite dates record, we suggest that the decrease in mean hematite sample dates in the direction of upper-plate transport reflects the migration of isostatic uplift and exhumation of lower-plate rocks (fig. 14, see models presented in Spencer, 1984; Spencer and Reynolds, 1991). Our estimate of the depth of hematite formation is ∼3 to 7 km, based on a >225 to 325 °C formation temperature estimate (Wilkins and others, 1986), and an elevated geothermal gradient in the upper crust during active extension (perhaps as high as ∼50 to 100 °C/km, Scott and others, 1998). Following hematite (±quartz) formation, interstitial phases crystallized at shallower depths and lower temperatures (similar to the findings of, Schuiling, ms, 1978). Owing to rapid rates of exhumation evidenced by cooling rates documented in previous studies of the Buckskin-Rawhide detachment (Scott and others, 1998), lower-plate rocks were advected from depths of hematite formation to shallower depths in a short period of time (relative to the ages of hematite obtained in this study), and very little time elapsed between the formation at depth and the cooling of hematite samples through the estimated closure temperature range. Therefore, while it is difficult to be certain whether hematite dates from this system record cooling or formation (as noted in the previous paragraph), we suggest that the difference between hematite formation ages and the hematite cooling ages is very small (less than 2 million years, based on a cooling rate of 80 °C/Myr; Scott and others, 1998). Northeastward migration of the locus of isostatic uplift resulted in rocks that are now exposed in the Buckskin Mountains passing through the zones of hematite formation and cooling prior to those now exposed in the Rawhide Mountains, yielding the observed age-distance trend in hematite dates and in other thermochronometers (figs. 13 and 14B-14D). Hematite samples from the upper-plate yielded mean dates similar to those obtained from nearby lower-plate samples (table 12). We interpret this to indicate that the host rocks of these samples were transferred from the upper-plate to the lower-plate and were mineralized at the same time as nearby lower-plate samples.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 14.

Evolution of the Buckskin-Rawhide core complex and associated mineral deposits, shown in schematic cross-sections (modeled after Spencer and Reynolds, 1989). The approximate location of these sections is shown in figure 1. The approximate depth range of hematite formation in the detachment zone (“Hem”, ∼3–7 km) and the brittle-ductile transition (“B–D”, ∼12–15 km) are shown as dashed horizontal lines. Approximate locations of hematite sample host rocks are shown as filled and open circles. (A) Extension and supra-detachment basin development in the Buckskin-Rawhide detachment system began as early as ∼26 Ma (Lucchitta and Suneson, 1993). (B) Before ∼19 Ma, the rate of extension along the detachment was moderate. During this time, isostatic uplift of the footwall began. Deposition of the Artillery Formation and other age-equivalent strata occurred from ∼26 Ma until ∼18.5 Ma. (C) At ∼19-18 Ma, the rate of extension increased (Nielson and Beratan, 1995). Lower plate rocks (now exposed in the western Buckskin Mountains) were rapidly exhumed to the surface. At ∼16 Ma, the lower-plate rocks in this part of the system were uplifted through the depth of hematite mineralization, and hematite (U-Th)/He dates were set in these samples. The surface exposure of these lower-plate rocks provided a source for mylonitic debris in the Chapin Wash Formation. By ∼16-15 Ma, Mn oxide vein mineralization in Chapin Wash strata had begun. (D) The locus of isostatic uplift migrated NE as extension progressed. The uplift of lower-plate rocks now exposed in the Rawhide Mountains was recorded by hematite mineralization in upper and lower-plate rocks at 14.5-12.8 Ma. The surface exposure and development of relief in these rocks provided a sediment source that fed coarse mylonitic debris into the Sandtrap Conglomerate between ∼12 and ∼9 Ma. After extension along the detachment ceased at ∼10-8 Ma, Mn mineralization in the upper-plate continued until at least 7 Ma.

The model of hematite mineralization presented in figure 14 is compatible with geologic constraints on the evolution of the Buckskin-Rawhide core complex established by previous studies. The timing of uplift and initial exposure of footwall rocks in the Buckskin Mountains implied by the presence of mylonitic debris in the upper Chapin Wash Formation (max depositional age = 18.5-16.2 Ma, minimum age uncertain, probably between ∼14 and ∼12 Ma, fig. 14C) is consistent with hematite dates on mineralization and rapid cooling that occurred during related lower-plate uplift (16-15 Ma, Spencer and Reynolds, 1989; Singleton and Stockli, 2013). Exposures of the Sandtrap Conglomerate northeast of the Rawhide Mountains contain abundant, locally derived clasts of lower-plate rocks (fig. 14D). This observation signals lower-plate exposure and the development of significant relief in the Rawhide Mountains prior to deposition of the Sandtrap Conglomerate (Spencer and others, 1989b; Yarnold, 1994). The depositional age of the Sandtrap Conglomerate (max depositional age uncertain), as implied by interbedding with 10 to 9 Ma Manganese Mesa basalt (Shafiqullah and others, 1980; Shackelford, 1980), is compatible with dates from hematite samples from lower-plate rocks in the Rawhide Mountains (14.5-12.8 Ma) that imply earlier lower plate uplift and mineralization in this area.

We speculate that Mn oxide dates represent the timing of vein formation in shallow hydrothermal settings. The relatively low closure temperature range (∼36–57 °C, see He diffusion in Mn Oxide above for T_c estimation) estimated for these samples indicates that they are susceptible to resetting by heat from overlying volcanic rocks or hydrothermal sources. The thickness of younger Neogene and Quaternary sediments is negligible, thus, reheating related to reburial is not probable (Spencer and others, 1989b). Some younger dates obtained from the Priceless Mine sample may have been affected by partial resetting, possibly due to a heating pulse related to the Manganese Mesa basalt. Mn oxide from the Shannon Mine sample postdates the flows of this basalt unit (age ∼9.5 Ma, Shafiqullah and others, 1980; Shackelford, 1980; Suneson and Lucchitta, 1983), therefore, we suggest that thermal resetting of the Shannon Mine sample is not likely, and that dates from this sample record Mn oxide formation.

These Mn oxide (U-Th)/He formation ages may provide a record of hydrothermal circulation through supra-detachment basin sediments driven by heat from buried portions of the detachment. Salinities and homogenization temperatures obtained from fluid inclusions in accessory phases in Mn oxide vein deposits (0–3 wt% NaCl equivalent, 165–170 °C; Spencer and others, 1989b; Spencer, 1991) are low relative to those observed in Fe oxide deposits (Wilkins and others, 1986). Based on this evidence, we speculate that the parent fluids of Mn oxide vein deposits are a mixture of deep-sourced, saline, hot fluids that transported heat from the detachment zone and shallow, cooler ground waters, which may have been enriched in Mn by interaction with detrital Mn deposits in the Chapin Wash Formation. If this proposed model is accurate, then the dates obtained from Mn oxide samples, particularly the younger sample from the Shannon Mine, provide evidence that buried lower-plate rocks continued to feed heat into the upper-plate hydrothermal system as extension along the detachment slowed, and perhaps after extension ceased at ∼10 to 8 Ma.