Impact thermochronology and the age of Haughton impact structure, Canada



[1] Most successful efforts to determine the ages of impact events are based on the isotope geochronology of crystalline or glassy impact melts. Studies of impact sites on Earth show that many form without significant melt production, meaning that traditional geochronologic approaches can yield unsatisfying results. We describe here an alternative approach based on theoretical calculations that even brief thermal events related to impact can reset the isotopic systematics of unmelted target rocks. Thermochronometers based on the production of radiogenic 4He in accessory minerals are particularly amenable to complete resetting by impact and should yield robust impact ages if helium systematics are not further disturbed during post-impact thermal events. We illustrate the utility of this method through a presentation of a new zircon (U-Th)/He date for the Haughton impact structure, Canada, of 23.5 ± 2.0 Ma.

1 Introduction

[2] Bolide impact is among the most important planetary surface processes in the inner solar system, but the precise and accurate radiometric dating of impact sites remains challenging. Only about 30% of the 183 confirmed impact structures on Earth have been dated with precisions of ≤ 10% [Planetary and Space Science Centre, 2013]. Most reliable ages are based on U-Pb or 40Ar/39Ar isotopic methods as applied to impact glass or neoblastic minerals from crystallized impact melts. However, many impact events are too small to have produced significant amounts of impact melt and are thus not tractable for geochronology of melt products. Here we explore an alternative approach to impact dating based on the resetting of isotopic clocks in target rocks due to rapid, impact-related heating. Many isotopic thermochronometers are susceptible to partial or complete resetting during meteorite impacts, especially those for which the radiogenic isotope is a noble gas such as 4He or 40Ar. We report here an application of the (U-Th)/He zircon thermochronometer to samples from the Haughton impact structure in northern Canada, with results that suggest a 23.5 ± 2.0 Ma age for the impact at the approximately 95% confidence level.

[3] The thermal evolution of an impact structure has three stages. The first and second comprise catastrophic thermal events directly related to the impact, referred to here as the contact/compression and excavation stages [Melosh, 1989]; the third stage involves post-event mineral alteration or metamorphism. The contact/compression stage refers to the actual impact of a bolide, an event lasting on the order of 0.1–1 s. At the point of contact, some target rocks may be superheated to temperatures of 10,000°C–15,000°C [Melosh, 2013]. Heating occurs over a broader area during the excavation stage as the shock energy propagated into the target is released by decompression [Grieve et al., 1977]. Target temperatures during this stage are highly variable across an impact site but likely range from approximately 500°C to at least 2000°C over time scales of no more than a few minutes [Keil et al., 1997; Ferrière and Osinski, 2013]. Finally, large impacts on hydrous planets frequently trigger the development of local hydrothermal systems [Abramov and Kring, 2004], and related metamorphism (typically at temperatures ranging from 100°C to 500°C over time scales of 1 kyr to 3 Myr) represents a possible third stage in the thermal evolution of an impact structure [Kirsimäe and Osinski, 2013].

[4] While the temperatures obtained during the contact/compression and excavation stages are sufficient to cause melting in the target area, the amount of melt produced and preserved at an impact structure scales with crater dimension, and the proportion of target rock clasts within an impact melt sheet decreases with the volume of melt produced [Grieve and Cintala, 1977]. This means that large impact sites are more straightforward to date isotopically than small ones as the melt products suitable for sampling are more abundant, and there is less contamination of those samples by unmelted xenocrysts or xenoliths that could complicate the robust interpretation of geochronologic data. However, there is good empirical evidence that isotopic clocks in the minerals of unmelted target rocks that experienced high degrees of shock metamorphism can be reset [Spray et al., 1999; Min et al., 2004; Schwenzer et al., 2008; Moser et al., 2011]. This provides an alternative “thermochronometric” approach to impact dating.

2 Thermal Modeling

[5] The science of thermochronology traditionally deals with establishing the temperature at which mineral-isotopic systems become effectively closed to daughter isotope loss as a consequence of cooling. The mathematics of this calculation—based on laboratory-determined diffusion properties, diffusion dimension, and cooling rate—were established by Dodson [1973]. However, “closure temperatures” are not the same temperatures at which mineral-isotopic systems are reset due to heating, which is the case of interest here. We explore the resetting phenomenon in a quantitative way using Watson and Cherniak [2013, equation (27)], which describes the diffusive response of a mineral grain—here in terms of radiogenic daughter loss—to a parabolic T-τ path defined by a maximum temperature (Tmax) and an event duration (τ). We have reproduced Watson and Cherniak [2013, equation (27)] here.

display math

[6] In this equation, D0 and Ea refer to the pre-exponential constant and the activation energy in the Arrhenius law of diffusion, respectively; R is the gas constant; and ζ is a value that is uniquely related to the total fractional loss from the crystal (for more details, see Watson and Cherniak [2013]). For the present study, the Watson-Cherniak approach was confirmed to give reasonably accurate results even for very brief but intense thermal pulses (for more details on this approach and its validity in this study, see the supporting information). In the present application, we define a resetting threshold as the loss of 99% of the radiogenic daughter isotope that had accumulated in a crystal prior to the heating event. Figure 1a is a plot of resetting threshold curves for a variety of commonly used thermochronometers. Values of Tmax and τ to the right of a curve correspond to conditions under which the chronometer would be expected to be reset by impact heating and thus would record the age of impact, barring subsequent disturbance of the isotopic systems by unrelated heating events. Superimposed on the plot are established log (Tmax)-log (τ) ranges for the contact/compression, excavation, and hydrothermal stages in the thermal evolution of impact structures.

Figure 1.

Diffusive loss behavior of radiogenic daughter isotopes in response to impact-related thermal events based on the methods described by Watson and Cherniak [2013]. Temperature-time histories for heating events are modeled as parabolic and described in terms of maximum temperature (Tmax) and event duration (τ). (a) Numbered curves correspond to 99% preimpact daughter isotope loss (F = 0.99, defined here as the threshold for resetting) for a variety of thermochronometers. Rectangular boxes correspond to the nominal ranges of log (Tmax) and log (τ) for stages in the thermal history of impact sites. See Table S1 for experimental daughter element diffusion data, as well as assumed diffusion dimensions. (b) Impact heating-related loss of 4He diffusion in zircon. Curves represent 10%, 50%, and 99% fractional loss (F = 0.1, 0.5, and 0.99). Rectangular boxes are the same as those in Figure 1a.

[7] For the geologically brief intervals of heating associated with impact, the maximum temperatures necessary to reset the chronometers in Figure 1 range over hundreds to thousands of degrees. Most minerals in target rocks subjected to extreme temperatures during the contact/compression stage would be expected to have their isotopic clocks completely reset, assuming the stage was brief enough that the minerals could persist metastably at temperatures far above their nominal melting points. In rocks subjected to heating at lower temperatures and for slightly longer periods during excavation, several of the thermochronometers shown in Figure 1a would be reset. However, the zircon and titanite U-Pb and the biotite and muscovite 40Ar/39Ar chronometers are not likely to be reset unless the pre-impact minerals dynamically recrystallized during the shock event, a phenomenon that has been documented for zircon at the Vredefort impact site in South Africa [Moser et al., 2011]. The comparatively long duration of hydrothermal activity (denoted by the height of the upper grey box in Figures 1a and 1b) requires that most of the chronometers, excluding the U-Pb systems, would be reset if temperatures of several hundred degrees Celsius were attained. However, the lifetimes of post-impact hydrothermal systems are frequently less than the 95% confidence interval (i.e., the uncertainty) for thermochronometric dates, such that the apparent ages of a chronometer reset by hydrothermal activity often would be statistically indistinguishable from the age of impact. We conclude that the thermochronometry of pre-impact minerals in the target region is a valuable tool for dating impact events. Chronometers plotting farthest left in Figure 1a—feldspar 40Ar/39Ar and zircon, titanite, and apatite (U-Th)/He—should be the most generally useful chronometers for small impact structures where target heating is of limited extent. This study engages the use of (U-Th)/He thermochronology due to the system's low resetting temperature and ability to date events of varying ages. It should be noted that (U-Th)/He thermochronology is only of use in geologic regimes where the target rocks contain appropriate accessory minerals (e.g., zircon, apatite, titanite, etc.) and where there have been no subsequent heating events with the potential to overprint the signal from the impact event itself.

[8] One complication of impact thermochronology is that heating to temperatures below those necessary for 99% daughter isotope loss can still result in partial resetting. Figure 1b illustrates curves for 10%, 50%, and 99% loss of preimpact radiogenic 4He loss from relict zircons, again calculated using Watson and Cherniak [2013, equation (27)]. While the zircon (U-Th)/He thermochronometer is well suited for impact studies, as even small impacts can reset it completely, variable heating at an impact site may result in a dispersion of measured zircon (U-Th)/He dates between the pre-impact cooling ages of the zircons and the age of impact that reflect different degrees of resetting. For this reason, successful impact thermochronologic studies require the acquisition of large data sets (n> ~25), and the best interpretation of the results is that the youngest mode of dates represents the timing of the thermal stage responsible for resetting.

3 Dating the Haughton Impact Structure

3.1 Geologic Background

[9] We illustrate this approach with (U-Th)/He zircon data from the relatively small (approximately 23 km) Haughton impact structure on Devon Island, Nunavut, Canada (Figure 2; 75°22′N, 89°40′W). Target rocks for the Haughton impact were Lower Paleozoic sedimentary units (primarily carbonate strata with subordinate clastic rocks) that unconformably overlie Precambrian granitic and gneissic basement rocks [Frisch and Thorsteinsson, 1978; Osinski et al., 2005b]. Exposures of impact breccia are widespread within the crater and dominated by Paleozoic carbonate clasts, but scattered gneissic and granitic clasts indicate that the impact penetrated down to the Precambrian basement. These breccias are overlain by post-impact lacustrine sediments of the Haughton Formation that are poorly biostratigraphically dated at between ca. 14 and > 29 Ma [Osinski and Lee, 2005]. Apatite fission track dates for shocked basement clasts have been interpreted to suggest a 22.4 ± 2.8 Ma (2σ) age for the impact [Omar et al., 1987]. 40Ar/39Ar data for Haughton samples (both whole rock and in situ) are difficult to interpret unambiguously due to substantial isotopic heterogeneity and variable excess 40Ar contamination. Impact age interpretations based on these data range from 23.4 ± 2.0 [Jessberger, 1988] to ca. 39 Ma [Sherlock et al., 2005]. The most widely accepted estimate is the Eocene age favored by Sherlock et al. [2005], but the inconsistency of that estimate and the fission track estimate is troubling.

Figure 2.

Simplified geologic map of the Haughton impact structure showing (circles) collection locations for samples H1–H4. Two samples—H2 and H3—are from the same locality. Small black squares indicate mapped locations for various features (mineralized vugs and fossil vents) indicative of impact-related hydrothermal activity (figure redrawn after Osinski et al. [2005a]).

3.2 Methods

[10] In an effort to better constrain the age of the Haughton impact, we collected four samples of basement clasts from the crater-fill breccia (Figure 2). All four hand samples were similar in appearance and were all collected from the allochthonous breccia unit within the structure. Mineral assemblages included quartz, feldspar, and biotite, all of which were clearly shocked when viewed under a microscope. There were no discernible differences between breccia samples at all four locations. Thirty-seven euhedral zircon crystals were selected from separates from these four samples for conventional (U-Th)/He thermochronology (see supporting information for tabulated data). All 37 zircons showed rounded pyramidal terminations and a slight cloudy, opaque texture and none showed visible fractures or breaks. These impact breccia samples were crushed and sieved, and minerals were separated using standard density and magnetic techniques. After microscopic examination of the separation products, 37 euhedral, transparent, and visibly inclusion-free zircon crystals were selected for study. In order to enable post-analytical alpha ejection corrections of raw (U-Th)/He dates [Hourigan et al., 2005], dimensions of each zircon were measured from photomicrographs prior to loading into Nb tubes for analysis. Helium measurements were made via isotope dilution techniques using an Australian Scientific Instruments (ASI) Alphachron system at the Noble Gas Geochronology and Geochemistry Laboratories (NG3L) at Arizona State University. Gasses were extracted in vacuuo using a 980 nm diode laser, scrubbed of reactive gasses, spiked with 3He, and measured by an electron multiplier on a Balzers QMS 200 quadrupole mass spectrometer. To measure U and Th isotopes, the crystals and tubes were spiked with a solution of 230Th and 235U in 50% HNO3 and digested at elevated temperature and pressure in a mixture of HF, HNO3, and HCl. Uranium and Th measurements were performed using isotope-dilution techniques on a Thermo Electron X-Series inductively coupled, plasma-source mass spectrometer in the W. M. Keck Foundation Laboratory for Environmental Geochemistry at Arizona State University. Raw and alpha ejection-corrected dates are reported in Table S1 in the supporting information, with 2σ errors based on propagated analytical uncertainties only. Further details on NG3L procedures are available elsewhere [van Soest et al., 2011].

4 Results

[11] The dated zircon crystals yielded a range of (U-Th)/He dates from 419 ± 13 to 18.56 ± 0.57 Ma (2σ) as illustrated in Figure 3. We interpret the dispersion of dates as direct evidence that the impact breccia represents a physical mixture of materials that experienced a wide range of shock metamorphic temperatures ranging from high enough to completely reset (U-Th)/He systematics in zircon to low enough to only cause partial resetting. Furthermore, reexamination of high magnification photographs of dated zircon crystals, taken prior to analysis, shows no discernible correlation in morphological characteristics (i.e., fractures, color alteration, etc.) and the degree to which crystals have been reset. The complexity of the zircon data set helps explain the difficulty encountered in deducing an unambiguous age of impact from published Ar data sets [Sherlock et al., 2005]. However, the helium data are far more easily interpretable. The spectrum of dates in Figure 3 displays a prominent Neogene mode, suggesting that the most complete resetting occurred more recently than the Eocene. This mode, representing 22 of the 37 dates, has a weighted mean of 23.53 Ma with a 2σ uncertainty of 0.16 Ma based on propagated analytical uncertainty. However, as is frequently the case with (U-Th)/He data, the dispersion of dates within the reset population is greater than that which can be explained by propagated analytical uncertainties alone, even though the reset population represents a single heating event. To account for the internal analytical uncertainty and constrain the external uncertainty associated with dating a population of thermochronometers, we utilize a method put forth by Wendt and Carl [1991] of multiplying the calculated uncertainty by the square root of the mean square weighted deviation (MSWD) of the weighted mean (MSWD: 151.7926 ± 0.6172 2σ). This yields a more robust estimate of ± 2.0 Ma for the approximately 95% confidence interval. Our preferred age for impact resetting—23.5 ± 2.0 Ma—agrees well with the interpreted age based on Jessberger's 40Ar/39Ar data (23.4 ± 2.0 Ma; 1988), and it is statistically indistinguishable from the 22.4 ± 2.8 Ma estimate of Omar et al. [1987] based on apatite fission track dating.

Figure 3.

Probability density function (PDF) and kernel density estimator (KDE) plots for the Haughton zircon (U-Th)/He data in Table S2. Measured dates are shown in black and white circles. The PDF curve (orange filled) is the sum over all probabilities for measured dates assuming Gaussian uncertainties. The KDE curve (yellow filled), constructed using a bandwidth calculated following the approach of Botev et al. [2010], is regarded as a more accurate reflection of the dispersion in mixed data sets [Vermeesch, 2012]. The prominent mode in the KDE curve emphasizes a tight clustering of of 22 zircon dates (black circles). The error-weighted mean of those dates (23.5 ± 2.0 Ma at the approximately 95% confidence level) is interpreted as the age of the Haughton impact event.

5 Discussion

[12] The response of various thermochronometers to heating events of different durations and magnitudes suggests that it may be possible to interpret the stage in the thermal evolution of an impact site to which a particular resetting date corresponds. If the Haughton zircons that yielded the youngest (U-Th)/He dates had been reset during the contact/compression stage, they would not have been able to accumulate new radiogenic 4He during the negligible timespan between contact/compression and excavation. Due to the extremely short duration of the excavation phase at Haughton, maximum post-shock temperatures of pre-impact zircons in the impact breccias are likely to have been in the range of 850°C–1250°C, or roughly half the plausible temperature range for carbonate-rich impact melts in the crater [Osinski et al., 2005a]. These conditions would have been sufficient to explain the fully and partially reset (U-Th)/He data set even if the zircons had not been reset during contact/compression. A moderate- to low-temperature hydrothermal system did develop at Haughton after impact, but the uneven distribution of hydrothermal mineralization across the structure suggests localized and variable heating within the structure [Osinski et al., 2005b]. The zircons we dated were collected in areas with no evidence of hydrothermal activity, but the nearest hydrothermal features (Figure 2) contained minerals (e.g., selenite) indicative of precipitation at temperatures below 100°C. Such temperatures, even if sustained for tens of thousands of years, as suggested by Osinski et al. [2005a], would not have induced sensible diffusive 4He loss from our zircons (Figure 1b). We thus interpret the 23.5 ± 2.0 Ma date as corresponding to the age of the excavation stage in the thermal evolution of the Haughton structure.

6 Conclusion

[13] Although this contribution has emphasized the application of the (U-Th)/He thermochronometer to impact dating, the application of multiple thermochronometers with a range of resetting characteristics to target rocks could yield detailed information regarding the thermal evolution of an impact site. Such multi-chronometer studies would be even more enlightening for extraterrestrial samples that may have experienced multiple impact heating events. For example, individual (U-Th)/He and 40Ar/39Ar data sets for meteorites and returned lunar samples show evidence of polyphase impact in the form of multiple apparent age modes [Hudgins et al., 2008; Min et al., 2013] or disturbed 40Ar/39Ar spectra [Shuster et al., 2010]. Even more remains to be learned from such samples through thermochronologic research employing a variety of mineral-isotopic systems with well-characterized diffusion characteristics.


[14] We would like to thank John Schutt, Jesse Weaver, Travis Oaks, and all those involved in the 2010 Haughton Mars Project for their logistical support in the field. We acknowledge the NASA Ames Research Center's Intelligent Robotics Group and thank them for the field and travel support. Alka Tripathy and Cameron Mercer are acknowledged for their help in laboratory analysis. We would also like to thank NASA for funding travel to Haughton (grant #NNX09AW29G), providing analytical funds (grant #NNX11AB31G), and granting additional funding for K.E.Y. (grant #NNX10AK72H). We also thank Kyle Min and Tom Shoberg for their thoughtful reviews.

[15] The Editor thanks Kyoungwon Kyle Min and Tom Shoberg for their assistance in evaluating this paper.