Introduction

K–Ar and 40Ar/39Ar step-heating data from micron- to submicron-scale illite/muscovite grains are commonly used to date low-grade metamorphism, deformation, and hydrothermal alteration (e.g., Hunziker et al. 1986; Reuter 1987; Dallmeyer and Takasu 1992; Kirschner et al. 1996; Jaboyedoff and Cosca 1999; Markley et al. 2002; Sherlock et al. 2003; Wyld et al. 2003; Ramírez-Sánchez et al. 2008; Wells et al. 2008; Egawa and Lee 2011; Zwingmann et al. 2011). Most analyses are conducted on aliquots that contain a large number of individual grains, and these mixtures of particularly small grains present challenges for interpretation that are not often encountered in other types of geo- and thermochronology. First, aliquots of clay-sized illite/muscovite from sedimentary rocks typically include both detrital and neo-formed grains (e.g., Hower et al. 1963; Hurley et al. 1963) and therefore include grains that clearly differ in crystallization age and also, potentially, in Ar retentivity. Second, growth of neo-formed illite/muscovite during prograde metamorphism can occur at the expense of pre-existing grains, a process referred to as Ostwald ripening or dissolution/precipitation (e.g., Inoue et al. 1988; Jaboyedoff and Cosca 1999). Third, although Ar diffusion in illite is poorly constrained experimentally, from estimates based on muscovite diffusion parameters (Harrison et al. 2009; Duvall et al. 2011; Rahl et al. 2011), formation of illite/muscovite crystallites during prograde metamorphism may occur first below and subsequently above temperatures at which these crystallites quantitatively retain 40Ar. Fourth, illite is susceptible to Ar loss both during neutron irradiation (Halliday 1978; Kapusta et al. 1997) and in nature (e.g., Dong et al. 1995; Hall et al. 2000). Finally, as with all thermochronometers, illite/muscovite separates that are sorted by grain-size will produce an age/grain-size correlation arising from differences in diffusion domain size (e.g., Dodson 1973, Harrison et al. 2009).

The cumulative effects of these processes for 40Ar/39Ar dating of clay-sized illite/muscovite are complicated step-heating spectra and a correlation between age and grain-size. In many cases, however, this correlation has been attributed solely to simple mixing between detrital and authigenic grains, each with their own distinct “end-member” age (e.g., Pevear 1992; Grathoff and Moore 1996; Clauer et al. 1997; Jaboyedoff and Cosca 1999; Środoń 1999; Grathoff et al. 2001; van der Pluijm et al. 2001). Indeed, controlled 40Ar/39Ar experiments on illite and fine-grained muscovite that recognize, or consider, all of the potential complications listed above are rare or non-existent. To help clarify these issues and test the utility of clay-sized illite/muscovite crystallites as geo- and/or thermochronometers, we collected 40Ar/39Ar step-heating data from illite-/muscovite-rich pelitic rocks sampled from a single stratigraphic unit that varies, regionally, in metamorphic grade. Our approach is modeled on a classic study from the Alps (Hunziker et al. 1986) and is analogous to borehole experiments that have been conducted for other thermochronometers (e.g., Gleadow et al. 1986; House et al. 1999; Wolfe and Stockli 2010), with the possible advantage that detrital mineralogy and provenance are expected to vary less within a single, relatively thin stratigraphic datum than in deep boreholes that penetrate multiple formations. Our results are used to infer (1) the temperature range of the illite/muscovite Ar partial retention zone; (2) the relationship between illite crystallite thickness, grain-size, and Ar retentivity; (3) metamorphic conditions leading to dissolution of clay-sized detrital muscovite and Ostwald ripening of neo-formed illite/muscovite; (4) the minimum age of clay-sized detrital muscovite in a sample of Middle Cambrian shale from the Colorado Plateau; (5) the timing of anchizonal metamorphism in the Panamint and Resting Spring Ranges of eastern California; and (6) the Cretaceous cooling rate of the footwall of a Cordilleran metamorphic core complex.

Considerations for illite/muscovite 40Ar/39Ar dating

Illite to muscovite transition

Illite is a phyllosilicate made up of packets of 1-nm-thick sheets (Fig. 1). Each sheet consists of an octahedral aluminosilicate layer positioned between upper and lower tetrahedral layers, and the sheets themselves are separated by potassium atoms (e.g., Moore and Reynolds 1997). In terms of crystal structure and chemical composition, illite is therefore similar to muscovite. For the purpose of this study, we consider the primary distinctions between illite and muscovite to be grain-size and packet thickness, illite being <2 μm and consisting of packets with comparatively few 1 nm sheets. We use the term “illite/muscovite” when drawing no distinction between the two or when describing particularly thick, micron-scale crystallites, in which case the only practical difference between illite and muscovite is grain-size. When referring to formerly coarse-grained muscovite that has been comminuted to micron scale, we retain “muscovite” for clarity, for example “clay-sized detrital muscovite.”

Fig. 1
figure 1

Illustration of one-dimensional model for Ar retention in illite/muscovite (Dong et al. 1995)

Full size image

Relative differences in packet thickness of illite/muscovite can be measured with X-ray diffraction (XRD) and are quantified with a parameter called illite crystallinity (IC; e.g., Kübler 1967; Warr and Rice 1994; Kübler and Jaboyedoff 2000). In pelitic rocks, low-grade metamorphic progression from the diagenetic zone to the epizone (roughly equivalent to the transition from zeolite to lower greenschist facies) is marked by a transition from thin illite crystallites to thick illite/muscovite grains that can be tracked with IC (e.g., Hunziker et al. 1986; Weaver 1989; Merriman et al. 1990; Warr and Nieto 1998; Merriman and Peacor 1999). As measured with transmission electron microscopy, the average number of 1-nm sheets per packet can range from <10 for thin illites in diagenetic zone shales to hundreds for thick illite/muscovite in epizonal slates (Merriman et al. 1990). As described below, the thickness of illite/muscovite crystallites has important implications for Ar retention and, by extension, K–Ar and 40Ar/39Ar geochronology.

40Ar loss from low-retention sites in nature

The natural decay of 40K to 40Ar releases 28 eV (Dong et al. 1995), corresponding to a 40Ar recoil distance of ~0.8 nm (calculated using the SRIM 2003 code, http://www.srim. org; Ziegler et al. 1985). Natural “recoil” energy is thus insufficient to eject 40Ar atoms from interior illite/muscovite interlayers to the packet exterior (a distance ≥1 nm in the c-axis-parallel direction), but 40Ar produced from 40K located on the edges of packets is highly susceptible to loss in nature (Fig. 1; Dong et al. 1995).

Production of 39Ar from 39K during neutron irradiation, on the other hand, is significantly more energetic (1–2 × 105 eV; Turner and Cadogan 1974) and has a correspondingly greater recoil distance of ~100 nm. Illite grains typically have thicknesses less than or roughly equal to this distance, so recoil during irradiation is expected to homogenize the 39Ar distribution in illite/muscovite separates (Dong et al. 1995). A key observation made during step-heating experiments of illite irradiated in vacuum-encapsulated tubes is that the proportion of 39Ar released before any heating occurs (i.e., the amount that is immediately released when the tube is broken open) is inversely correlated with crystallite thickness (Dong et al. 1995, 1997; Hall et al. 1997, 2000). Based on this finding, it has been inferred that 39Ar atoms that were recoiled to “unprotected” edges of illite/muscovite packets during irradiation were quickly lost into the encapsulating tube at room temperature and thus that illite edges are non-retentive sites, both for 39Ar produced during irradiation and for 40Ar produced in nature. Loss of 40Ar from the edges of illite packets via this mechanism is analogous to the loss of 4He from the edges of apatite and zircon crystals during the much more energetic (4–8 × 106 eV) alpha production from U and Th (e.g., Farley et al. 1996; Reiners 2005).

A one-dimensional model of illite/muscovite Ar retention based on the proportion of K residing on the exterior of packets (Fig. 1; Dong et al. 1995) implies that thin packets have lower Ar retentivity than thick packets, both during laboratory irradiation and in nature. Structural defects in illite/muscovite crystallites are also important pathways for Ar loss (e.g., Hames and Cheney 1997; Kramar et al. 2001; Mulch et al. 2002; Mulch and Cosca 2004; Cosca et al. 2011), and the density of crystallographic defects in illite/muscovite generally decreases as packet thickness increases (e.g., Lee et al. 1985; Jiang et al. 1997; Livi et al. 1997; Warr and Nieto 1998). The contrast in Ar retention between thin and thick crystallites is therefore exacerbated by Ar loss via defects. In this paper, we use “retentive” to describe illite/muscovite crystallites that, by virtue of their relatively large thickness and low defect density, are capable of retaining a large proportion of Ar produced either in nature or during irradiation. According to the one-dimensional retention model, a one-packet-thick illite crystallite is entirely non-retentive because it is predicted to lose 100 % of the 40Ar produced in nature and 39Ar produced during irradiation.

Illite “retention ages” were introduced to account for Ar loss from poorly retentive illite (Dong et al. 1995). These are encapsulated 40Ar/39Ar bulk ages that take into account all of the 40Ar released during step-heating but exclude the 39Ar that is released when the encapsulating tube is initially broken open. In principle, retention ages from encapsulated aliquots are equivalent to total gas ages produced from non-encapsulated experiments. Encapsulated total gas ages, on the other hand, are equivalent to K–Ar ages (Dong et al. 1995; Onstott et al. 1997). An assumption made during the introduction of retention ages (Dong et al. 1995) that we also make in interpreting our data is that the crystallographic sites in illite/muscovite that are most susceptible to loss of 39Ar during neutron irradiation are the same sites that are most susceptible to loss of 40Ar in nature.

Detrital and neo-formed grains

Siliciclastic rocks, especially pelitic rocks, contain both detrital muscovite and neo-formed illite/muscovite, a factor that introduces complications for geo- and thermochronology, particularly when these grains are mixed together in individual analytical aliquots. In this paper, neo-formed grains are referred to as “authigenic” when they form at particularly low metamorphic grade, as “metamorphic” when they form at approximately greenschist facies, and simply as “neo-formed” for general cases. Shales at very low metamorphic grade (e.g., diagenetic grade, or zeolite facies) contain mixtures of thick, retentive, detrital muscovite grains and thin, less-retentive authigenic illite grains. Separates of illite/muscovite from these low-grade mixtures typically lose significant 39Ar during irradiation, and 40Ar/39Ar step-heating experiments conducted on them often produce staircase-shaped spectra (e.g., Hunziker et al. 1986; Jaboyedoff and Cosca 1999; Dong et al. 2000; Verdel et al. 2011a). A key point in the interpretation of K–Ar and 40Ar/39Ar data from very low-grade shales is that there is not only a difference in crystallization age between detrital muscovite and authigenic illite but also differences in Ar retentivity between thick muscovite packets (high retentivity, characteristic of detrital muscovite) and thin illite crystallites (low retentivity, characteristic of authigenic illite). At higher metamorphic grade (e.g., anchizone–epizone), four important and related observations have previously been made: average packet thickness is greater (e.g., Merriman et al. 1990; Jiang et al. 1997), crystallographic defects are reduced (e.g., Livi et al. 1997), 40Ar/39Ar spectra are frequently flatter (Hunziker et al. 1986), and there is less 39Ar loss during irradiation (Dong et al. 1995, 2000). These observations suggest relationships between nanometer-scale crystallographic structure, Ar retention, temperature, and the low-grade metamorphic transition of illite to muscovite.

40Ar loss by volume diffusion

There are currently no experimentally based Ar diffusion parameters for illite, but parameters for muscovite may be good proxies because illite and muscovite are structurally similar (e.g., Parry et al. 2001). Duvall et al. (2011) and Rahl et al. (2011) used muscovite diffusion parameters from Harrison et al. (2009) to calculate theoretical closure temperatures (Dodson 1973) of approximately 250–350 °C for illite/muscovite grains with radii ranging from 0.05 to 2 μm, using cooling rates of 1 to 10 °C/My. Detailed experiments indicate that diffusion of Ar and other elements occurs perpendicular to the c-axis in muscovite (van der Pluijm et al. 1988; Hames and Bowring 1994). Similar experiments are impractical for illite crystallites because of their small size, but we infer that the same diffusion process is true for illite given the structural similarities with muscovite. Cumulatively, these observations suggest that Ar escapes from illite/muscovite in nature primarily via three processes: c-axis perpendicular volume diffusion, loss from low-retention sites on packet exteriors, and loss along microstructural defects. The geometries of grains therefore have important implications for Ar retentivity. Although measurements of the dimensions of individual crystallites is impractical for mixtures containing numerous grains, bulk estimates can be made in the laboratory: grain-size separates based on settling velocity will sort grains primarily by diameter, and thickness variations can be quantified with IC. Diameter and thickness are typically positively correlated (e.g., Hnat 2009), and, as noted earlier, the density of crystallographic defects is inversely correlated with packet thickness.

Ostwald ripening

Ostwald ripening (Ostwald 1900) refers to the growth of new mineral grains by transfer of material from pre-existing grains. Two common examples of Ostwald ripening are the formation of monazite (e.g., Kingsbury et al. 1993; Ayers et al. 1999; Catlos et al. 2002) and illite/muscovite (e.g., Eberl and Środoń 1988; Inoue et al. 1988; Eberl et al. 1990) during prograde metamorphism. The clearest evidence that Ostwald ripening of illite/muscovite occurs during low-grade metamorphism is a shift to larger particle sizes with increasing metamorphic grade (e.g., Eberl et al. 1988; Jaboyedoff and Cosca 1999; Kim and Peacor 2002). The transfer of 40Ar during Ostwald ripening and other “illitization” mechanisms is an issue of considerable debate (e.g., Aronson and Hower 1976; Hunziker et al. 1986; Eberl and Środoń 1988; Clauer and Chaudhuri 1996; Velde and Renac 1996; Wilkinson and Haszeldine 2002), but one logical hypothesis is that the process of Ostwald ripening could, under certain conditions, produce large, neo-formed grains with younger ages than the small remnants of older, detrital grains. This relationship would be the opposite of the more frequently encountered positive correlation between age and grain-size that arises from volume diffusion and daughter-isotope recoil/ejection, as described above.

In short, there are competing factors that influence K–Ar and 40Ar/39Ar results from illite separates of siliciclastic rocks, all of which lead to correlations between age and grain-size. Two of the most important considerations are that increased temperature induces both growth of progressively thicker crystallites (e.g., Merriman et al. 1990) and diffusion of 40Ar out of the crystallites. The first effect increases Ar retentivity, which, in isolation, will lead to older ages. It is countered by the second effect, which produces younger apparent ages. The net result can be evaluated with 40Ar/39Ar step-heating data from samples of illite-/muscovite-rich sedimentary rocks that have experienced a range of metamorphic conditions.

Field experimental setup

We conducted an experiment utilizing samples collected from a regionally extensive stratigraphic unit in the western United States that has spatial variations in metamorphic grade. This stratigraphic datum consists of Cambrian pelitic rocks that are referred to as the Bright Angel Formation (Fig. 2a) east of the Cordilleran craton–miogeocline hingeline (for example, in locations such as the Grand Canyon) and the Carrara Formation west of the hingeline (for example, in the vicinity of Death Valley). The hingeline separates relatively thin (~1.5 km) Paleozoic stratigraphic sections to the east from much thicker (>6 km) correlative sections to the west (e.g., Wright et al. 1981) and generally corresponds with the leading edge of the Sevier fold-and-thrust belt (Burchfiel and Davis 1972; Burchfiel et al. 1992). The Bright Angel/Carrara Fm. lies near the base of the westward-thickening Paleozoic passive margin sequence and was susceptible to large gradients in metamorphic grade arising from both stratigraphic and tectonic burial (Verdel et al. 2011b).

Fig. 2
figure 2

Sample positions and experimental results. a Highly simplified reconstruction of the Cordilleran miogeocline, after Wright et al. (1981) and Wernicke et al. (1988). Sample positions are shown relative to present-day geographic locations. b Illite crystallinity results, after Verdel et al. (2011b). c Step-heating results, plotted at equivalent scales. Depositional age of the Bright Angel Fm. from Middleton et al. (2003)

Full size image

Samples of Bright Angel/Carrara pelitic rocks were collected along a transect with endpoints at Frenchman Mtn., NV, and Death Valley, CA (Fig. 2a). Clay-mineral characterizations of these samples have previously been reported (Verdel et al. 2011b), and here we focus on the results of encapsulated 40Ar/39Ar step-heating experiments conducted on micron to submicron, illite-/muscovite-rich grain-size fractions of key samples and discuss the implications for illite/muscovite 40Ar/39Ar geo/thermochronology.

Methods

Illite crystallinity

Samples were crushed in a mortar and pestle, and <2-μm-grain-size fractions were separated using a centrifuge. This material was mixed with water and pipetted onto glass slides to prepare oriented mounts. These mounts, along with IC standards (Warr and Rice 1994), were scanned from 2 to 50° 2θ with a Scintag X-1 X-ray powder diffractometer at the University of Michigan. IC was determined from XRD patterns using the MacDiff software program (Petschick et al. 1996).

40Ar/39Ar step-heating procedure

Grain-size separates of 0.75–2, 0.2–0.75 μm, and <0.2 μm were prepared with a centrifuge. Approximately 0.1–1 mg of each separate was placed in a quartz glass tube and vacuum-encapsulated prior to irradiation at the McMaster Nuclear Reactor. The MMhb-1 hornblende standard (520.4 Ma; Samson and Alexander 1987) was used as a neutron-fluence monitor. After irradiation, the tubes were broken open under vacuum and step-heated with a continuous argon-ion laser from 0 to 4 W at the University of Michigan. Ar isotopic data were measured on a VG1200S mass spectrometer. Additional details of the analytical procedure are available in Verdel et al. (2011a).

Results

Illite crystallinity

Full results of the illite crystallinity study are available in Verdel et al. (2011b), so here we summarize the results that are most pertinent to interpretation of the 40Ar/39Ar results. All of the samples have a prominent illite/muscovite diffraction peak at 1 nm, and XRD patterns indicate that illite/muscovite is the predominant K-bearing phase in the <2 μm fraction of each sample (Verdel et al. 2011b). Average IC from two measurements of sample FM09 from Frenchman Mtn., normalized to the standards of Warr and Rice (1994), is 1.1° 2θ (Fig. 2b; Table 1), corresponding with the shallow diagenetic metapelitic zone. Average IC from the Mesquite Pass sample (MP05) is 0.7° 2θ, within the deep diagenetic zone. The Panamint Range sample (PR10) has IC of 0.34° 2θ (low anchizone), a slightly greater value than the Resting Spring Range sample (RS03), which has IC of 0.26° 2θ (high anchizone). The average IC of sample 0902 from Bare Mtn. is 0.23° 2θ (epizone), the lowest IC measured during the study. The total range in IC from these five samples is 1.1 to 0.23° 2θ, spanning from the shallow diagenetic zone to the epizone and corresponding to estimated peak temperatures of ≤100 °C to ≥300 °C (Fig. 2b; e.g., Merriman and Frey 1999).

Table 1 Summary of illite crystallinity and 40Ar/39Ar data
Full size table

40Ar/39Ar step-heating experiments

40Ar/39Ar step-heating results are in Table S1, illustrated in Fig. 2c and summarized in Table 1. For each sample, we discuss below the maximum age reached during step-heating and the fraction of total 39Ar released when the encapsulating tubes were broken open.

Frenchman Mountain

The 0.75–2 μm grain-size from the Bright Angel Formation at Frenchman Mtn. (FM09) produced a staircase-shaped 40Ar/39Ar spectrum that reaches a maximum age of 967 Ma (Table 1). Of the total 39Ar in this grain-size, 24 % was released upon breaking open the capsule. The two finer grain-sizes also produced staircase-shaped spectra, but with younger maximum ages (831 and 578 Ma) and with roughly equal or greater 39Ar loss during irradiation (23 and 34 %).

Mesquite Pass

Step-heating data were collected from two grain-sizes of the Mesquite Pass sample (MP05). The larger size (0.75–2 μm) produced a staircase-shaped spectrum with a maximum age of 518 Ma and 17 % 39Ar loss during irradiation. The spectrum of the finer size (<0.2 μm) is also staircase-shaped, reaches a maximum age of 337 Ma, and lost 28 % 39Ar during irradiation.

Panamint Range

The coarsest grain-size from the Panamint Range sample (0.75–2 μm) produced a staircase-shaped spectrum reaching a maximum age of 486 Ma and lost 12 % 39Ar. The 0.2- to 0.75-μm fraction has a maximum age of 442 Ma and lost 14 % 39Ar. The spectrum from the finest size (<0.2 μm) reaches a maximum age of 285 Ma and lost 38 % 39Ar.

Resting Spring Range

For the Resting Spring Range sample, the 0.75–2 μm grain-size has a maximum age of 297 Ma and recoiled 13 % 39Ar. The 0.2–0.75 μm size has a slightly younger maximum age (294 Ma) and recoiled slightly more 39Ar (17 %). Both grain-sizes produced “plateau-like” segments between approximately 280–297 Ma at the highest temperature steps. The finest grain-size (<0.2 μm) produced a much older maximum age of 441 Ma and recoiled 15 % 39Ar.

Bare Mountain

Three grain-sizes from Bare Mtn. sample 0902 were analyzed. None produced true plateaus according to most definitions, but they are all characterized by relatively flat spectra in the middle part of the degassing experiment. In the largest size (0.75–2 μm), this middle portion has an age of approximately 80 Ma, and ages are slightly younger in the plateau-like segments of the two smaller sizes. Maximum ages are 99–108 Ma, and 39Ar recoil increases from 11 to 35 % with decreasing grain-size.

Discussion

General features

40Ar/39Ar step-heating behavior of our samples is highly correlated with both grain-size and IC. There is considerable variation between step-heating results for different grain-size separates from the same sample, suggesting that the separates did not simply sort the broken fragments of an originally uniform population of grains. There was undoubtedly some breaking of grains during crushing, however, which will clearly lead to an overall reduction in the grain-size of the separates compared with in situ grain-size. Nevertheless, the fact that the narrow ranges of grain-size separates produced distinct step-heating results suggests that the separates isolated narrow ranges in the physical dimensions of illite/muscovite crystallites that are at least related to their in situ dimensions, i.e., relatively large in situ grains were concentrated in relatively large grain-size separates.

Maximum ages, total gas ages, and retention ages all become younger as IC decreases (packet thickness increases; Fig. 2c; Table 1). 39Ar recoil is correlated with IC increases and grain-size decreases, both of which reflect larger proportions of thin crystallites. With one exception (the Resting Spring Range sample), ages decrease with decreasing grain-size. These general findings are consistent with numerous previous studies (e.g., Perry 1974; Aronson and Hower 1976; Hunziker et al. 1986; Dong et al. 1995, 1997, 2000; Hall et al. 1997, 2000; Jaboyedoff and Cosca 1999; Haines and van der Pluijm 2010).

Our results can be interpreted in terms of the factors discussed above: the influence of detrital grains, loss of Ar from low-retention sites on the edges of illite/muscovite packets, the metamorphic transition from thin to thick crystallites, thermally activated Ar diffusion, and dissolution/Ostwald ripening. Given that our samples are all from the same stratigraphic interval and probably included roughly equivalent detrital muscovite populations, sample-to-sample variation in 40Ar/39Ar results primarily reflects the latter four processes.

Age of detrital muscovite

Our interpretation of the staircase-shaped spectra that characterize 40Ar/39Ar results from illite-rich, low-grade shales (e.g., the Frenchman Mtn. sample) is that they reflect mixtures between illite and muscovite crystallites that vary in Ar retentivity. Particularly low-grade shales contain both micron-scale detrital muscovite grains, which tend to be quite Ar-retentive despite their small size, and authigenic illite crystallites, which are characterized by thin packets and low Ar retentivity (Fig. 1; Dong et al. 2000). Initial low-T steps from 40Ar/39Ar step-heating of these mixtures extract Ar from the least-retentive grains, which necessarily have the youngest apparent ages. Subsequent higher-T steps degas crystallites with progressively greater retentivity, producing progressively older ages. The highest-T steps extract Ar from the most retentive grains, which, in particularly low-grade samples, are clay-sized detrital muscovite grains. The staircase-shaped spectra are therefore related to a continuum mixture and wide range in Ar retentivity. This explanation is consistent with 40Ar/39Ar step-heating results from artificial mixtures of coarse-grained muscovite and diagenetic illite (Onstott et al. 1997). Our interpretation of the old apparent ages often encountered in the high-temperature portions of step-heating spectra of clays is an alternative to the model of Kapusta et al. (1997), which attributes them to 39Ar recoil. In contrast to the Kapusta et al. (1997) model, we suggest that the old ages have geologic significance.

When multiple grain-size fractions are prepared from a low-grade shale, the largest and most retentive detrital grains, which will produce the oldest apparent ages, are concentrated in the coarsest fraction. Finer fractions concentrate smaller detrital grains and progressively more of the thin, authigenic, poorly retentive illite crystallites. As a result, maximum age decreases and % 39Ar recoiled increases with decreasing grain-size in illite-/muscovite-rich mineral separates from low-grade sediments.

According to this interpretation, the maximum age reached on a step-heating spectrum from a low-grade shale is a minimum estimate for the age of detrital muscovite within the sample. The age is a minimum because any illite/muscovite separation with an upper limit on grain-size (such as those analyzed in this study) is expected to produce a younger maximum age than a hypothetical separation with a slightly coarser upper limit. In theory, at sufficiently coarse size, a plateau-like segment would be produced from high-temperature steps, and this plateau would represent the age of a population of detrital muscovite. In practice, however, this is usually not observed for the grain-sizes typically utilized for illite/muscovite 40Ar/39Ar step-heating (Hunziker et al. 1986; Dong et al. 2000; Verdel et al. 2011a).

If the interpretation of step-heating results outlined above is correct, the minimum age of detrital muscovite in the Bright Angel/Carrara stratigraphic datum can be estimated from the maximum age reached during step-heating of the coarsest grain-size separate from the sample that experienced the lowest peak temperature. As determined with IC, this is the Frenchman Mtn. sample. The coarsest grain-size from this sample (0.75–2 μm) reached a maximum step-heating age of 967 Ma (Fig. 2c and Table 1). We suggest that 967 Ma is a reasonable minimum age estimate for detrital, clay-sized muscovite in Cambrian shales of the western United States because K–Ar and 40Ar/39Ar ages of ≥1,000 Ma have been determined from coarse-grained muscovite in basement exposures of the Basin and Range (Wasserburg and Lanphere 1965; Reiners et al. 2000), and basement rocks of the western United States were likely the source of detrital minerals, including muscovite, in Cambrian strata of western North America (Gehrels et al. 1995, 2011; Stewart et al. 2001).

Illite Ar partial retention zone

Only the Frenchman Mtn. sample and the coarsest grain-size from the Mesquite Pass sample have maximum step-heating ages that are older than the depositional age of the Bright Angel Formation, which is ~499 Ma (Fig. 2c; Middleton et al. 2003). This observation, in conjunction with the interpretation of step-heating spectra outlined above, implies that 40Ar/39Ar illite/muscovite ages have been partially or entirely thermally reset in the other, higher-grade samples. Maximum ages from the Mesquite Pass sample are 293–449 My younger than equivalent grain-sizes from the Frenchman Mtn. sample, suggesting that the Mesquite Pass sample is, itself, partially reset. Importantly, the Mesquite Pass sample was collected four meters from the edge of a 13-Ma felsic sill (Friedmann et al. 1996). While it is quite possible that elevated temperature in the proximity of the sill partially degassed this sample, 40Ar/39Ar spectra of the two Mesquite Pass grain-size fractions reach maximum ages of 518 and 337 Ma, suggesting that detrital muscovite in the shale was not completely reset at 13 Ma.

Temperature sensitivities of various thermochronometers have been estimated empirically using sampling profiles in boreholes and exposed crustal sections (e.g., Gleadow et al. 1986; House et al. 1999; Reiners et al. 2000; Stockli et al. 2000; Wolfe and Stockli 2010). Ages decrease with depth in these studies, often revealing an intermediate depth range in which the retention of daughter nuclides is only partial. We evaluate our data in a similar manner, but in the absence of depth measurements, we substitute IC, a parameter that varies with temperature (Ji and Browne, 2000). By analogy to a plot of age versus depth, we plot maximum ages from 40Ar/39Ar spectra vs. IC (Fig. 3). Maximum age is used because it is the apparent age of the most retentive illite/muscovite crystallites in a given grain-size fraction, which, for low-grade samples, correspond with clay-sized detrital muscovite grains. We assume that the detrital muscovite population was equivalent in each of our samples, although the more salient observation is simply that detrital muscovite ages must be older than the depositional ages of the Bright Angel/Carrara Fm., except in cases where those detrital ages have been partially or entirely reset due to heating. Thus, by tracking the decrease in maximum ages with increasing metamorphic grade, we empirically estimate the Ar partial retention zone (PRZ) of clay-sized muscovite.

Fig. 3
figure 3

“Borehole-like” plot of maximum ages reached on 40Ar/39Ar spectra (Fig. 2C and Table 1) versus illite crystallinity, a measure of greenschist to subgreenschist metamorphic grade. Theoretical clay-sized illite/muscovite closure temperature based on muscovite diffusion parameters from Harrison et al. (2009). Presumed age of detrital muscovite is based on ages of ~1,000 Ma from basement exposures in the western United States (e.g., Reiners et al. 2000). See Fig. 2 of Reiners et al. (2000) for an analogous borehole-like muscovite 40Ar/39Ar plot from the western United States. Abbreviations: FM Frenchman Mtn, MP Mesquite Pass, PR Panamint Range, RSR Resting Spring Range, BM Bare Mtn

Full size image

At Frenchman Mtn., the sample site at lowest metamorphic grade, the maximum age from the coarsest size fraction (0.75–2 μm) is 967 Ma. By analogy with a borehole experiment, the Frenchman Mtn. sample represents the upper, low-temperature part of the borehole. Maximum ages decrease in our samples as IC decreases, in a manner similar to the decrease in age with depth in a borehole. The Resting Spring Range sample has an unusual age–grain-size relationship that is discussed in the following section. Maximum ages from the Resting Spring Range and Panamint Range samples (IC = 0.26 and 0.34° 2θ, respectively) vary from 285 to 486 Ma. These contrast markedly with maximum ages from Bare Mtn. (IC = 0.23° 2θ) of only 99–108 Ma (Fig. 3). There are similar large discrepancies in total gas ages and retention ages for these samples (Table 1). We suggest that this jump to much younger ages, which occurs at metamorphic conditions between those exemplified by the Bare Mtn. and Resting Spring Range samples, represents the base of the illite/muscovite Ar PRZ. This stage of low-grade metamorphism is approximately the anchizone–epizone boundary, which has been empirically calibrated at ~300 °C (e.g., Merriman and Frey 1999). Our empirical estimate for the base of the PRZ is thus consistent with (1) theoretical predictions based on muscovite diffusion parameters, assuming that our nominal grain-size ranges approximate the diffusion length scale (Harrison et al. 2009; Duvall et al. 2011; Rahl et al. 2011); (2) previous comparisons of 40Ar/39Ar data from epizonal (Dong et al. 1997) and anchizonal (Dong et al. 1995) pelitic rocks; and (3) general correspondence between illite/muscovite 40Ar/39Ar retention ages and zircon fission-track ages (Ramírez-Sánchez et al. 2008). The zircon fission-track partial annealing zone is between about 260° and 310 °C (Tagami and Dumitru 1996), which overlaps with our estimate for the base of the illite/muscovite Ar PRZ.

Our interpretation is that samples from the Resting Spring Range and Panamint Range are within the PRZ. The Mesquite Pass sample is also partially degassed, though presumably from proximity to the Tertiary sill. The pattern of decreasing ages illustrated in Fig. 3 is a direct analog of the finding that muscovite 40Ar/39Ar ages decrease with structural depth in an exposed crustal section in the western United States (Reiners et al. 2000). In fact, there is similarity between the overall age ranges obtained in each of these Cordilleran-based empirical thermochronology experiments (compare Fig. 3 with Fig. 2 of Reiners et al. 2000).

Illite/muscovite dissolution and Ostwald ripening

With one exception, our samples show a relationship between age and grain-size that is typical for thermochronometers, namely younger ages for smaller grain-sizes. The exception is the Resting Spring Range sample, in which the finest grain-size (<0.2 μm) has much older ages, by any measure, than the 0.2–0.75 and 0.75–2 μm fractions (Table 1). The mechanisms described above for Ar retention that we use to evaluate results from the other samples cannot explain this result. We suggest that the combined processes of progressive dissolution of detrital grains and Ostwald ripening of neo-formed crystallites could explain this finding, however. Our interpretation is that for this sample, the finest fraction (<0.2 μm) concentrates the relict cores of detrital muscovite grains, while the coarser fractions concentrate neo-formed illite/muscovite, which, by this stage of metamorphism, have grown to be larger than the remains of the detrital grains (Fig. 4). The finest grain-size separate therefore has the oldest age. As approximated by our IC measurements, this crossing-point in relative crystallite size seems to occur at roughly the transition from low to high anchizone conditions. The transition is superseded at just slightly higher-grade conditions (anchizone–epizone boundary) by complete resetting of 40Ar/39Ar ages in micron-scale illite/muscovite.

Fig. 4
figure 4

Formation of illite/muscovite during low-grade metamorphism. a At particularly low metamorphic grade, shales contain mixtures of thick, Ar-retentive, clay-sized detrital muscovite grains and thin, poorly retentive authigenic illite. A wide range of Ar retentivity results in staircase-shaped 40Ar/39Ar step-heating spectra. This stage in the formation of metamorphic muscovite is represented by the Frenchman Mtn. sample. b With increased temperature, neo-formed illite/muscovite crystallites become thicker and more Ar-retentive. Growth of these grains occurs at the expense of detrital muscovite, so the detrital grains become smaller. At high anchizone conditions, neo-formed illite/muscovite crystallites become larger than the remnants of detrital muscovite, which is a reversal of the size relationship at lower metamorphic grade. Elevated temperature partially degasses both detrital and metamorphic grains at these conditions, though if Ostwald ripening involves reduction in the diameter of detrital grains (as drawn), it would also lower their theoretical closure temperature. The net effect could be that at high anchizone conditions, neo-formed grains have higher closure temperatures than detrital grains. This stage is represented by the Resting Spring Range sample. c At epizone conditions, all clay-sized illite/muscovite is fully degassed. K–Ar and 40Ar/39Ar results are cooling ages at these and higher-grade conditions. The Bare Mtn. sample is representative of this stage. Circle diameters reflect relative crystal size and indicate the formation of large metamorphic muscovite concurrent with a reduction in the size of detrital grains

Full size image

This overall interpretation is consistent with the finding of Jaboyedoff and Cosca (1999) that grain-sizes of insoluble minerals (including illite/muscovite) increased in a suite of samples during the progression from diagenetic zone to anchizone metamorphism. A subtlety in our interpretation is the inference that detrital muscovite grains are reduced in size during low-grade metamorphism (Fig. 4), a transition that was also noted by Jaboyedoff and Cosca (1999), but which is nevertheless at odds with a continual increase in grain-size arising from Ostwald ripening. Our interpretation is that dissolution of detrital muscovite grains is distinct from Ostwald ripening of neo-formed grains, at least during the earliest stages of metamorphism. This is consistent with the results of detailed laboratory experiments conducted on monazite to stimulate Ostwald ripening under controlled conditions (Ayers et al. 1999). These experiments produced an initial grain-size reduction that was followed by grain-size increase, suggesting an initial phase during which experimental starting minerals (analogous to detrital grains) were partially dissolved, followed by, and possibly overlapping with, an Ostwald ripening phase during which neo-formed grains became larger. An important point is that unless dissolution preferentially removes 40Ar over K from detrital grains, partial dissolution, alone, will not directly reduce maximum ages reached on 40Ar/39Ar step-heating spectra. Clearly, however, heating of muscovite, either in the laboratory (e.g., McDougall and Harrison 1999) or in nature (e.g., Reiners et al. 2000), will decrease maximum ages via thermal diffusion, independent of the effects of Ostwald ripening.

Given the preceding discussion, similarities in ages between the Resting Spring Range and Panamint Range grain-sizes may be significant. Step-heating results from the <0.2 μm fraction from the Resting Spring Range sample are comparable with results from the 0.75–2 μm and 0.2–0.75 μm fractions from the Panamint Range (Fig. 2c). Likewise, the <0.2 μm Panamint Range fraction has results that are similar to the 0.75–2 and 0.2–0.75 μm Resting Spring Range fractions. These similarities are consistent with the dissolution and Ostwald ripening hypothesis. The 0.75–2 and 0.2–0.75 μm grain-sizes from the Panamint Range sample contain large fractions of partially degassed detrital muscovite and produce maximum step-heating ages of 486 and 442 Ma, respectively (Table 1). Comparable detrital grains, diminished somewhat in size via dissolution, are concentrated in the <0.2-μm fraction of the slightly higher-grade Resting Spring Range sample, which produces a similar (though slightly younger) maximum age of 441 Ma. The <0.2 μm grain-size of the Panamint Range sample, on the other hand, contains a large proportion of relatively small, neo-formed illite/muscovite and produces a younger maximum age of only 285 Ma. The Resting Spring Range counterparts of these neo-formed grains are enhanced in size via Ostwald ripening and therefore concentrated in the larger 0.75–2 μm and 0.2–0.75 μm fractions, but nevertheless produce similar (though slightly older) maximum ages of 297 and 294 Ma, respectively.

In summary, the metamorphic transition from the lower-grade Panamint Range sample to the slightly higher-grade Resting Spring Range sample involves four changes to the step-heating results and their interpretation. First, detrital grains are concentrated principally in coarse grain-size separations at lower metamorphic grade, but in finer size fractions at higher grade. Second, apparent age of the detrital grains decreases with increasing metamorphic grade. Third, there is a parallel increase in the apparent age of neo-formed grains. The cumulative effect is a slight narrowing of the gap in apparent age between detrital muscovite and neo-formed illite/muscovite, consistent with concurrent degassing of the former and increasing Ar retentivity of the latter. A subtle but important consequence of the dissolution/Ostwald ripening process is that decreased crystallite diameter of detrital grains lowers their theoretical closure temperature (Fig. 4), while increased diameter of neo-formed grains has the opposite effect. Therefore, at anchizone conditions partially degassed detrital grains may coexist with undegassed neo-formed grains (Fig. 4). Fourth, step-heating spectra of the grain-size separations that principally concentrate neo-formed grains become flatter with increasing metamorphic grade (compare the spectrum of the <0.2-μm fraction of the Panamint Range sample with spectra from the 0.75–2 and 0.2–0.75 μm grain-sizes in the Resting Spring Range sample), an effect that reflects a narrowing of the range of Ar retentivities in a mixture of crystallites.

The Resting Spring Range sample lies near the base of the illite/muscovite Ar PRZ, so some, or all, of the grain-sizes from it may be partially degassed, in which case their ages may be geologically meaningless, as seems to clearly be the case for the Mesquite Pass grain-sizes. Given the preceding discussion, however, and the form of step-heating results for the 0.75–2 and 0.2–0.75 μm grain-sizes of the Resting Spring Range sample (Fig. 2c), these particular separates seem to contain retentive neo-formed illite/muscovite that could, in fact, preserve meaningful 40Ar/39Ar ages. The highest-temperature steps have plateau-like segments that range in age from about 280 to 297 Ma for the 0.75–2 μm grain-size and 280–294 for the 0.2- to 0.75-μm fraction. The final temperature steps from the <0.2 μm fraction of the Panamint Range have comparable ages of 269–285 Ma. These ages may represent prograde muscovite formation in the Resting Spring Range and Panamint Range during early Permian tectonic burial associated with the Sierra Nevada–Death Valley thrust system (SNDVTS; Stone and Stevens 1988; Snow 1992; Stevens and Stone 2005a, b; and see Fig. 4 in Verdel et al. 2011b). The absence of similar ages from the Mesquite Pass and Frenchman Mountain samples suggests that these locations, which are in the foreland of the SNDVTS, were less affected by early Permian tectonism. As discussed below, any indication of early Permian metamorphism has been erased from illite/muscovite 40Ar/39Ar results by subsequent heating at Bare Mountain, the highest-grade location we studied.

At epizone conditions, there is a return to the normal relationship of older ages for larger grain-sizes, but with significant differences in 40Ar/39Ar step-heating results compared with low-grade samples (Fig. 2c). At epizone conditions and above, metamorphic illite/muscovite crystallites are larger than, and have retentivities at least as great as, detrital muscovite grains (Fig. 4). Relatively flat spectra reflect a comparatively narrow range in Ar retentivity in mixtures within analytical aliquots. Illite/muscovite grains are also degassed at epizone conditions, so we interpret the Bare Mtn. ages as cooling ages. Next, we describe how these data can be used to infer thermal histories.

Cretaceous cooling at Bare Mountain, NV

Proterozoic to Paleozoic sedimentary rocks were exhumed in the lower plate of the Bullfrog/Fluorspar Canyon detachment at Bare Mountain, NV (Fig. 5; Hoisch et al. 1997; Hoisch 2000). A high-angle normal fault (the Gold Ace fault) within the lower plate of the detachment separates strata metamorphosed up to amphibolite facies in its footwall from greenschist to subgreenschist facies sediments in the hanging wall (Hoisch 2000). Previous muscovite K–Ar and 40Ar/39Ar ages from the lower plate of the Bullfrog/Fluorspar Canyon detachment are in the footwall of the Gold Ace fault and range from Miocene to Paleocene (Hoisch et al. 1997). To our knowledge, our illite/muscovite 40Ar/39Ar data from the Carrara Fm. are the first thermochronology data from the hanging wall of the Gold Ace fault. Muscovite grains large enough to recognize with a petrographic microscope were not observed in our sample, and in general, low metamorphic grade within the hanging wall of the Gold Ace fault may have precluded previous muscovite 40Ar/39Ar thermochronology. As in other locations, however, the Carrara Fm. in the hanging wall of the Gold Ace fault is rich in illite/muscovite. Retention ages from three grain-sizes of a sample of the Carrara Fm. are 81.8 ± 0.4, 77.6 ± 0.5, and 67.6 ± 1.6 Ma for grain-sizes of 0.75–2, 0.2–0.75, and <0.2 μm, respectively. Based on the low IC value of this sample, our interpretation is that it was at sufficiently high temperature to completely diffuse 40Ar from illite/muscovite, or at least from all but the most retentive grains. The age/grain-size correlation in this case therefore arises from progressive cooling below temperatures at which 40Ar is retained in each grain-size.

Fig. 5
figure 5

Simplified geologic map of Bare Mountain and vicinity (after Hoisch et al. 1997) showing previous muscovite and new illite/muscovite results

Full size image

To estimate a cooling rate for the Bare Mtn. sample, we utilize retention ages and closure temperatures, both of which are simplifications as elaborated below, but nevertheless illustrate a point about the apparent sensitivity of the grain-size separates to different temperatures. Based on diffusion parameters for muscovite (Harrison et al. 2009), closure temperatures for the grain-sizes analyzed are approximately 290, 272, and 249 °C (for domain diameters of 2, 0.75, and 0.2 μm, respectively, cooling rate of 5 °C/My, and spherical diffusion geometry). Using the retention ages as our best estimate of the bulk cooling age for each grain-size, these results indicate a Late Cretaceous period of cooling of the hanging wall of the Gold Ace fault below ~250–290 °C. Furthermore, cooling rate appears to have decreased from about 5 °C/My from approximately 82–78 Ma to ~0.8 °C/My from about 78–68 Ma. Using age differences from just the medium- and high-temperature steps of each grain-size produces similar cooling rate estimates. Our illite/muscovite 40Ar/39Ar ages thus overlap with geo- and thermochronology results from the kinematically linked Funeral Mtns. to the SW (Applegate et al. 1992; Beyene 2011) but suggest cooling rates that are slow when compared with Late Cretaceous rates from elsewhere in the Cordilleran orogen (Wells and Hoisch 2008; Chapman et al. 2012).

The use of retention ages and theoretical closure temperatures in the preceding cooling rate estimate might be an inferior approach to modeling continuous thermal histories using Ar release data (e.g., Harrison et al. 2005), particularly given that muscovite appears to be characterized by multiple diffusion domains (Harrison et al. 2009). However, our mineral separates nominally have extremely narrow ranges in grain-size (≤1 μm), so it is not clear that a particular size fraction includes significant variation in domain size. On the contrary, each size fraction may isolate a particular diffusion domain size, in which case the systematic variation in ages from these grain-sizes place an upper limit (200 nm) on the minimum size of muscovite diffusion domains. A mineral separate with a wider range of grain-sizes (and presumably a wider range of domain sizes) may be conducive to thermal modeling and could, in fact, contain the same age–temperature information as multiple grain-size separates. It is important to note, however, that the highest temperature steps from the Bare Mtn. grain-sizes may represent detrital and/or particularly retentive metamorphic illite/muscovite grains that have not been entirely reset, in which case those particular steps are not suitable for continuous thermal modeling. These uncertainties point out the need for additional tests of illite/muscovite 40Ar/39Ar thermochronology, but data from Bare Mtn., as well as the other sample sites, illustrate that illite/muscovite grain-size separates that differ in diameter by ≤1 μm preserve distinct age information.

Conclusions

40Ar/39Ar step-heating results from illite/muscovite in greenschist to subgreenschist pelitic rocks are closely tied to thermal histories. At very low metamorphic grade (e.g., zeolite facies), spectra are typically staircase-shaped, reflecting a mixture between relatively large (but still clay-sized), high-retentivity detrital muscovite grains and smaller, less-retentive authigenic illite grains. Maximum temperature steps from these samples place minimum constraints on the apparent age of detrital muscovite. In the case of the Middle Cambrian Bright Angel Formation at Frenchman Mtn., NV, this minimum age is 967 Ma, consistent with ≥1,000 Ma ages that have been determined from muscovite in basement rocks of the US Cordillera.

As metamorphic grade increases, neo-formed illite/muscovite crystallites become thicker and more Ar-retentive. At approximately the transition from low to high anchizone conditions, retentivity of neo-formed illite/muscovite appears sufficient to produce geologically meaningful ages and plateau-like 40Ar/39Ar step-heating spectra. In the Resting Spring Range and Panamint Range of eastern California, the Cambrian Carrara Fm. is at approximately this stage of low-grade metamorphism and contains illite/muscovite that seems to record early Permian metamorphism associated with the Sierra Nevada–Death Valley thrust system.

At greenschist (epizone) conditions, metamorphic illite/muscovite is completely degassed and difficult to distinguish from coarse-grained muscovite in terms of XRD characteristics, although individual grains may still be too small to observe with a petrographic microscope. Under such conditions, illite/muscovite is useful for 40Ar/39Ar thermochronology. An example of illite/muscovite 40Ar/39Ar thermochronology from the epizonal Carrara Fm. at Bare Mtn., NV, reveals a Late Cretaceous period of cooling at rates of ~1–5 °C/My in the hanging wall of the Gold Ace fault.

Illite/muscovite separates that differ in size by ≤1 μm have distinct 40Ar/39Ar step-heating spectra and can vary in apparent age by hundreds of millions of years. These observations suggest that the separates isolate narrow ranges in diffusion domain size, implying that illite/muscovite is characterized by multiple diffusion domains with dimensions of order hundreds of nanometers, if not smaller. Although significant additional work is needed to reliably interpret complex 40Ar/39Ar step-heating results from fine-grained illite/muscovite in low-grade sediments, this type of data appears to offer, in combination with illite crystallinity, a geo-thermochronometer/peak-temperature thermometer for low-grade metamorphism.