Biotite 40Ar/39Ar ages older than corresponding muscovite 40Ar/39Ar ages, contrary to the diffusion properties of these minerals, are common in the Himalaya and other metamorphic regions. In these cases, biotite 40Ar/39Ar ages are commonly dismissed as “too old” on account of “excess Ar.” We present 32 step-heating 40Ar/39Ar ages from 17 samples from central Himachal Pradesh Himalaya, India. In almost all cases, the biotite ages are older than predicted from cooling histories. We document host-rock lithology and chemical composition, mica microstructures, biotite chemical composition, and chlorite and muscovite components of biotite separates to demonstrate that these factors do not offer an explanation for the anomalously old biotite 40Ar/39Ar ages. We discuss possible mechanisms that may account for extraneous Ar (inherited or excess Ar) in these samples. The most likely cause for “too-old” biotite is excess Ar, i.e., 40Ar that is separated from its parent K. We suggest that this contamination resulted from one or several of the following mechanisms: (1) 40Ar was released during Cenozoic prograde metamorphism; (2) 40Ar transport was restricted due to a temporarily dry intergranular medium; (3) 40Ar was released from melt into a hydrous fluid phase during melt crystallization. Samples from the Main Central Thrust shear zone may be affected by a different mechanism of excess-Ar accumulation, possibly linked to later-stage fluid circulation within the shear zone and chloritization. Different Ar diffusivities and/or solubilities in biotite and muscovite may explain why biotite is more commonly affected by excess Ar than muscovite.
where E is the activation energy, R is the gas constant, A is the grain geometry factor, τ is the time constant, D0 is the diffusion coefficient, and a is the radius of the effective diffusion domain. The concept of closure temperature is based on the assumptions that (1) the cooling history is characterized by a linear increase in 1/T; (2) Ar transport in the mineral is controlled by volume diffusion; (3) Ar escapes into an “infinite” reservoir, and the concentration of Ar at the grain boundary remains zero (“zero-concentration boundary condition”). Experimental diffusion data suggest that the Tc of Ar in muscovite and biotite is ∼400–500 °C and ∼300–400 °C, respectively, depending on grain size, mineral chemistry, and cooling rate (Harrison et al., 1985; Grove, 1993; Grove and Harrison, 1996; Harrison et al., 2009).
Mineral 40Ar/39Ar ages may deviate from the inferred time of cooling through Tc for several reasons. The 40Ar/39Ar ages can be reset by neocrystallization or dynamic recrystallization (e.g., Dunlap 1997; Mulch and Cosca, 2004). Hydrothermal fluids may displace radiogenic 40Ar through chemical reactions (e.g., Miller et al., 1991). Both mechanisms result in 40Ar/39Ar ages that may be younger than the expected cooling ages, depending on the timing of crystallization or fluid flow during cooling. In other cases, calculated 40Ar/39Ar ages are “too old”; a short metamorphic pulse may be insufficient to completely degas Ar from a mineral (e.g., Viete et al., 2011), or 40Ar may become trapped, for example, in fluid inclusions (Cumbest et al., 1994) or lattice defects (Camacho et al., 2012). The problem of anomalously old 40Ar/39Ar ages is particularly known from biotite (e.g., Roddick et al., 1980; Baxter et al., 2002) and is commonly attributed to the presence of excess Ar.
The 40Ar accumulating in a mineral may originate from different sources. Slightly different nomenclature is used throughout the literature; we follow the terminology of Dalrymple and Lanphere (1969) and McDougall and Harrison (1999): Radiogenic Ar is 40Ar produced within the mineral by radioactive decay of 40K. Inherited Ar is essentially radiogenic 40Ar that remained in the mineral during incomplete resetting (e.g., older core with younger rim) or was introduced in the form of older material into the mineral (e.g., older K-bearing particles become incorporated into younger volcanic rocks). Nonradiogenic Ar includes Ar of atmospheric composition (here we use 40Ar/36Ar = 295.5) and excess Ar. Excess Ar is parentless 40Ar, i.e., 40Ar that has been separated from its K-bearing source; it is incorporated in rocks and minerals by processes other than in situ radioactive decay. Trapped Ar is Ar that was incorporated during mineral formation or at a later event and can encompass atmospheric Ar as well as excess Ar. Inherited and excess Ar—collectively termed extraneous Ar—cause the age of the mineral to appear older than its “true” crystallization or cooling age.
In the Himalaya, many studies have reported biotite 40Ar/39Ar ages that were considered “too old” and were therefore excluded from the thermal history interpretations (e.g., Hubbard and Harrison, 1989; Catlos et al., 2001; Stüwe and Foster, 2001; Godin et al., 2006, and references therein; Horton et al., 2015; Adams et al., 2015). Other studies have investigated possible sources of extraneous Ar and tried to link it, for example, to host-rock composition (e.g., Foland, 1979; Boven et al., 2001; Baxter et al., 2002), presence or absence of fluids (e.g., Cumbest et al., 1994; Stüwe and Foster, 2001; Itaya et al., 2009; Halama et al., 2014), or (ultra)high-pressure metamorphism (e.g., Scaillet, 1998; Giorgis et al., 2000; Warren et al., 2011). If extraneous Ar is quantifiable in the system, it may provide additional information about the geologic history of the sample and the physicochemical conditions it experienced (e.g., Kelley and Wartho, 2000).
With the goal of constraining the geological reason(s) behind anomalously old biotite 40Ar/39Ar ages in the Himalaya and in metamorphic rocks in general, we present a data set of biotite and muscovite 40Ar/39Ar ages (BtAr and MsAr, respectively) from central Himachal Pradesh, NW Himalaya. Petrographic microscope thin section observations, electron microprobe and X-ray diffraction analyses, bulk-rock and mineral compositions, and the degree of biotite chloritization were used to investigate structural or chemical relationships to the BtAr ages. Isochron analyses of high-resolution step-heating experiments were applied to determine whether this method could usefully detect the presence of excess Ar. The process we outline for determining the likely sources of extraneous Ar is applicable to any metamorphic terrane.
In central Himachal Pradesh, Cenozoic sediments of the Himalayan foreland are overthrusted along the Main Boundary Thrust by the Proterozoic to early Cambrian Lesser Himalayan Sequence. The Main Central Thrust (MCT), a middle to late Miocene ductile shear zone, separates the Lesser Himalayan Sequence from the Greater Himalayan Sequence. The Greater Himalayan Sequence consists of lower-greenschist-facies to upper-amphibolite-facies Haimanta metagraywacke (Neoproterozoic to Cambrian metapsammite and intercalations of metapelite and calc-silicate) and Ordovician granite (Figs. 1 and 2; Thöni, 1977; Steck, 2003; Webb, 2013). The dominant foliation, metamorphic isograds, and metamorphic zones based on index minerals (Bt + Chl, Grt, Ky ± St ± Sil; Wyss, 2000; Steck, 2003; mineral abbreviations after Kretz, 1983) outline a kilometer-scale, overturned to recumbent, SW-vergent antiform, resulting in an inverse metamorphic sequence N and NE of the Kullu-Rampur window (overturned limb) and in a normal metamorphic sequence in the Chandra valley (upper limb, northern study area) and S and W of the town of Kullu (Fig. 2; Thöni, 1977; Epard et al., 1995; Wyss et al., 1999). The antiform is spectacularly exposed along the western hillslopes of the Beas valley (Phojal fold; Fig. 2). Although the mechanism and tectonic setting are disputed, the folding is generally attributed to early Miocene exhumation of midcrustal rocks by SW-directed thrusting and folding and accompanying erosion and/or normal faulting (e.g., nappe emplacement—Epard et al., 1995; extrusion and channel flow—Searle et al., 2007; tectonic wedging—Webb et al., 2007).
Along most of the Himalayan range, the hanging wall of the MCT is cut by the normal-sense South Tibetan detachment system, which separates amphibolite-grade metamorphic rocks of the Greater Himalayan Sequence from the greenschist-facies to unmetamorphosed Tethyan Himalayan Sequence. In the western Himalaya, the Zanskar shear zone and the Sangla detachment were identified as strands of the South Tibetan detachment system, but in central Himachal Pradesh, the presence and location of shear zones equivalent to the South Tibetan detachment system are disputed (Thakur, 1998; Wyss et al., 1999; Webb et al., 2007; Stübner et al., 2014; Fig. 1). No strands of the South Tibetan detachment system are depicted in Figures 2A–2C, although top-to-the-NE shear does occur, localized along the contacts between intrusive and metamorphic rocks.
The highest-metamorphic-grade rocks occur in the core of the Phojal fold (600–700 °C, ∼6 kbar; Epard et al., 1995; Leger et al., 2013), below the Deo Tibba Ordovician granite (650–700 °C, ∼8 kbar; Wyss, 2000), and in the deeply incised Chandra valley (Ky, St, Hbl, Sil, indicating upper amphibolite facies; Epard et al., 1995; our observations). Here, pegmatite and aplite dikes and veins suggest melting in the latest Eocene and Oligocene (Wyss, 2000; Stübner et al., 2014). We will use the term “crystalline core” to refer to these amphibolite-grade metamorphic rocks (Ky ± Grt zones; Figs. 2A–2C). North and east of the Kullu-Rampur window, the base of the Haimanta sequence above the MCT has experienced a strong retrograde, greenschist-grade overprint with widespread chloritization of Bt and Grt (Wyss, 2000; Figs. 2A–2C, hatched area). Crustal thickening and prograde metamorphism occurred in the Eocene and Oligocene (ca. 41–26 Ma; Mnz and Grt geochronology; Walker et al., 1999; Thöni et al., 2012; Stübner et al., 2014). The onset of exhumation is recorded by decompression at 26–23 Ma, followed by rapid cooling to ∼250 °C at ∼30–60 °C/m.y., as constrained by Ms and Bt Rb-Sr and 40Ar/39Ar and zircon fission-track (ZFT) ages (Schlup et al., 2011, and references therein). Most MsAr ages are 22–20 Ma (total of 20 samples from Walker et al., 1999; Schlup et al., 2011; Stübner et al., 2014). Fewer BtAr dates are reported in the literature, and they are in general more variable and include contrasting ages (e.g., 42.6 ± 0.3 Ma and 16.7 ± 0.3 Ma from one locality with a MsAr age of 22.2 ± 0.2 Ma; Schlup et al., 2011). Walker et al. (1999) and Webb et al. (2011) dismissed their BtAr results as geologically meaningless on the grounds of excess Ar.
SAMPLES AND METHODS
Seventeen samples from different lithologies and structural levels of the MCT hanging wall in central Himachal Pradesh were collected for 40Ar/39Ar analysis (Table 1). Eleven Bt and 14 Ms separates were analyzed at the Noble Gas Laboratory, Dalhousie University, Halifax, Canada (DGC; MsAr ages are published in Stübner et al., 2014). Additional three MsAr and four BtAr ages were obtained using finely tuned step-heating experiments at the Argon Laboratory, Freiberg, Germany (ALF; Table 2). Analytical details of 40Ar/39Ar analysis at both laboratories are given in the Data Repository Item.1
Thin sections were investigated by optical microscopy to characterize the host rocks and micas, with special attention to grain size, texture, and deformation textures such as folding, kinking, undulose extinction, and fracturing, indicating intracrystalline strain (Passchier and Trouw, 2005, and references therein). Backscatter electron (BSE) imaging using the high-contrast setting on JEOL JXA 8900 electron microprobes at the Universities of Göttingen and Tübingen was used to characterize compositional variability of Bt in polished thin sections. The same microprobes were used to determine chemical composition (Si, Ti, Fe, Al, Na, K, Mg, Mn, Ca, Ba, Cr, F, Cl) of Bt by wavelength dispersive spectrometry (WDS), using an acceleration voltage of 15 kV, a probe current of 15–20 nA, a beam diameter of 10–20 μm, and counting times of 15 s and 60 s on the X-ray signal and total background, respectively. Both core and rim positions were analyzed. For separated grains, grain mounts were used for WDS analysis with a probe current of 7 nA and a beam diameter of 1–2 µm. Biotite chemical formulae were calculated stoichometrically based on 22 oxygens. In order to assess mineralogical purity of the Bt separates used for 40Ar/39Ar thermochronology, X-ray powder-diffraction data were collected by the Rietveld method to quantify the amount of Chl and Ms in Bt separates. Analytical procedures for Rietveld X-ray powder-diffractrometry are described in the Data Repository.
Petrographic sample descriptions including photomicrographs are provided in the Data Repository (text; Fig. S1). A brief summary is given here and in Table 1. Metasedimentary samples 807D1, 807E1, 811B1, 812D2, and 823G3 are medium- to coarse-grained mica schist to metapsammite with a typical assemblage of Qtz + Bt + Ms + Pl ± Grt. Biotite is more abundant than Ms, aluminosilicates are not present, and accessory minerals include Tur + Ap + Zrn ± Aln ± Mnz ± Ilm ± Rt ± Czo. Quartz is abundant in all metasedimentary samples. Most samples show minor post-tectonic (across-foliation) growth of Chl or chloritization of Bt and/or Grt. The foliation is pervasive, straight or wavy, and parallel to compositional banding. Biotite lacks evidence of intracrystalline strain (e.g., kinks, undulose extinction). Sample 807E1 has a millimeter-scale crenulation cleavage with Bt both along the crenulation foliation and within microlithons; the Bt shows no evidence of intracrystalline strain. Sample 814G1 is a fine-grained, banded metapsammite with abundant calcite; large skeletal Ms grains at high angle to the foliation are probably detrital. Sample 827B1 is a fine-grained metapelite with a wavy crenulation cleavage.
Granite and orthogneiss samples are quartz-rich and contain both Bt and Ms, except for 823G2, which lacks Ms (Table 1). Garnet occurs in orthogneiss samples 823G2 and 819A2 and in leucogranite samples 803B3 and 815E1. Similar to the metasedimentary samples, Bt shows no evidence of intracrystalline strain. In augengneiss samples 823G2 and 819A2, Bt forms the main foliation and is intergrown with Grt, suggesting no or little deformation since (re-)crystallization of Bt + Grt. The same textures are observed in the psammitic Haimanta metasedimentary rocks (e.g., 811B1).
Table 2 summarizes the 40Ar/39Ar results, age spectra and inverse isochron plots are shown in Figure 3, and raw data are reported in Supplemental Tables S1 and S2 in the Data Repository. Many samples yielded high to medium percentages of atmospheric 40Ar throughout the whole experiment (Fig. 3; Table S1). Therefore, the age spectra and the weighted mean ages (WMA) calculated from all heating steps (e.g., Figs. 3E and 3F) or the specified temperature range (e.g., Figs. 3A and 3B) are strongly influenced by the atmospheric correction, such that even small fluctuations in the initial 40Ar/36Ar ratio can cause relevant changes in the calculated step ages. For most samples, the interpreted age (rounded values with ±2σ error; Table 2) thus relies on the inverse isochron age (IIA), which is unaffected by this type of correction.
A typical experiment at DGC consisted of 12–16 heating steps. The age spectra of both Ms and Bt often show complex patterns with only minor plateau sections (e.g., Figs. 3B, 3K, 3P, and 3Q). Likewise, data in the inverse isochron plots are only partly arranged along single mixing lines between trapped Ar (36Ar/40Ar intercept) and radiogenic Ar (39Ar/40Ar intercept; e.g., Figs. 3X, 3a, 3b, and 3c). For two experiments, Ms of sample 814G1 and Bt of sample 827B1, the scatter of data was too high to calculate reliable ages, and therefore probable age ranges are suggested in the age interpretation (Fig. 3V and 3g; Table 2). For other samples, IIAs were calculated from all data points, or sometimes excluding the highest- or lowest-temperature heating steps, and these typically yielded mean square weighted deviates (MSWD) of ≤5 (Fig. 3, gray error boxes and ellipses and specified temperature range). Initial 40Ar/36Ar ratios ranged from 210 to 420 and were mostly ≥295.5 (Table 2).
Additional Ms and Bt analyses, using a heating schedule of typically 18–20 steps, were obtained at ALF. All analyses resulted in plateau ages that included 100% of the 39Ar with MSWD <0.6 (Figs. 3E, 3F, 3N, 3O, 3S, 3T, and 3U; Table 2). These WMAs are identical to the IIAs, which have MSWD <0.3. Initial 40Ar/36Ar ratios ranged from 270 to 340, and all were within uncertainty of the atmospheric value.
Muscovite ages of most samples ranged from 19 ± 1 Ma to 23 ± 2 Ma (Table 2; Figs. 2 and 3). Muscovite from sample 807E1—analyzed in both laboratories—yielded indistinguishable ages (23 ± 2 Ma and 22 ± 1 Ma; Figs. 3L and 3N). Older Ms ages—up to 64 ± 5 Ma—originated from lower-grade metamorphic rocks in the southern and northern parts of the study area (Fig. 2A). Sample 807D1 yielded a complex age spectrum with step ages increasing from ca. 33 Ma to 102 Ma with increasing temperature. An inverse isochron, fit to the lowest-temperature degassing steps (800–900 °C), yielded 27.3 ± 0.7 Ma (Fig. 3J). Muscovite ages in the Chandra valley increase with elevation from 20 ± 1 Ma (807A1) through 24 ± 2 Ma (807C1) to 28 ± 5 Ma (807D1; Figs. 3E, 3G, 3J, and 2A, elevation profile marked with white box frame).
The BtAr ages are generally older than the MsAr ages from the same locality (cf., Figs. 3T and 3U). Most of them range from 20 ± 2 Ma to 43 ± 2 Ma and are thus up to 16 m.y. older than the respective MsAr ages (Figs. 2A and 4; Table 2). Samples from the MCT zone (823G2, 823G3, and 827B1) yielded pre-Cenozoic BtAr ages (84 ± 6 Ma; 106 ± 5 Ma; Fig. 2B). This inverted age relationship between Bt and Ms was observed irrespective of lithology (cf. Haimanta metasedimentary rocks vs. orthogneiss), structural position (upper limb vs. core of the Phojal fold), and metamorphic zone (Ky vs. Grt zones; Figs. 2A, 2C, and 4). The elevation profile in the Chandra valley records a positive age-elevation trend (Figs. 2A and 4). A “normal” age relationship between Bt and Ms was revealed for sample 804C1 from the Bt + Chl zone granite (Ms 64 ± 5 Ma, Bt 31 ± 2 Ma) and for sample 807A1 from the Ky zone augengneiss (Ms 20 ± 1 Ma, Bt 20 ± 1 Ma).
Electron microprobe analyses of Bt revealed chemical compositions typical for metamorphic rocks (Guidotti, 1984; Fleet et al., 2003), with 0.21–0.55 Mg/(Mg + Fe), 3.10–3.47 Al per formula unit (p.f.u.), 0.15–0.33 Ti p.f.u., and low Mn, Ca, Na, Ba, and Cr (Table 3). The value of F was <1.5 wt%, and Cl was close to or below the detection limit of 0.1 wt%. Total oxides were 93–97 wt%. Within each sample, chemical compositions were narrowly defined without detectable variations between Bt from different textural contexts, i.e., Bt within foliation, in Grt strain shadows, from inclusions or embayments in Grt, with different grain size, or between cores and rims. We recognized Chl and marginally chloritized Bt in BSE images in some of the samples (see Table 1), but with no effects on bulk chemistry and with no correlation to the K content.
Mineralogical Composition of Biotite Separates
Powder XRD analyses suggest that the Bt separates used for 40Ar/39Ar dating comprise 84%–98% Bt (Table 4; Fig. 5). Trace amounts (0.6%–2.4%) of Chl were present in all analyzed samples; up to 15% Ms was found intergrown with Bt in some samples. Repeated rinsing of the separates in H2Odest. reduced the Ms content (e.g., 812D2: 14% Ms in untreated separate vs. 5% Ms in rinsed separate; Fig. 5), but a significant amount of Ms remained after rinsing (e.g., 807E1: 15% Ms).
The Ti-in-biotite geothermometer of Henry et al. (2005) yielded Bt equilibration temperatures of ∼360–520 °C; 827B1 did not yield a valid result (not enough Ti; Table 3). Regional trends in Ti-in-biotite temperatures correspond to the mapped metamorphic zones: The highest temperatures (500–520 °C) were obtained from the center of the crystalline core exposed in the Chandra valley (807E1, 807A1) and in the core of the Phojal fold (815C1); lower temperatures (400–500 °C) were obtained from lower-metamorphic-grade rocks (Grt ± Ky zones); the lowest temperatures (<400 °C) were obtained from the structurally highest samples in the upper limb of the Phojal fold (811B1, 807D1; Fig. 2B). The elevation profile in the Chandra valley records up-section decreasing temperatures (807A1, 807C1, 807D1; Fig. 2C).
Do Muscovite and Biotite 40Ar/39Ar Ages Reflect Cooling through Their Closure Temperatures?
In the metasedimentary rocks (807D1, 807E1, 811B1, 812D2, 815C1, 823G3), foliation-forming Bt and Ms are interpreted to have grown during Oligocene prograde metamorphism (see petrographic descriptions and Figure S1). Only the large skeletal Ms grains in sample 814G1 are likely relict grains. In lower-greenschist-grade metapelite 827B1, some mica may be detrital, but most of the coarser-grained, foliation-defining Bt and Ms are interpreted as metamorphic. In the orthogneiss and augengneiss samples (806D3, 807A1, 807C1, 810B1, 819A1, 823G2), all mica appears to have recrystallized during the Cenozoic tectonometamorphic cycle. In undeformed and weakly deformed granitic samples, mica grew during magmatic crystallization, i.e., in the Ordovician (Bt and Ms in 804C1) or possibly the Cenozoic (Ms in leucogranites 803B3 and 815E1). With the exception of the detrital grains, there is no indication that one mica phase crystallized later than the other, e.g., as a result of dynamic recrystallization in shear bands. The inverted age relationship between MsAr and BtAr ages is therefore not a result of postcooling neocrystallization or recrystallization of Ms.
Metamorphic and magmatic samples from the crystalline core, with 600–700 °C peak metamorphic temperatures (Epard et al., 1995; Wyss, 2000), are expected to yield 40Ar/39Ar cooling ages (Warren et al., 2012). With increasing distance from the crystalline core, samples are less likely to have resided at temperatures above Ar closure for a sufficiently long period for complete thermal resetting. Figure 6 provides a compilation of all published as well as our new thermochronologic ages from central Himachal Pradesh. Closure temperatures are 480 ± 30 °C for MsAr and 350 ± 30 °C for BtAr, calculated after Equation 1 based on grain radii of 100–400 µm (Table 1), cooling rates of 20–100 °C/m.y., and the diffusion parameters of Harrison et al. (2009) and Grove and Harrison (1996). Monazite geochronologic data indicate prograde metamorphism during ca. 41–26 Ma and onset of decompression between 26 and 22 Ma (Stübner et al., 2014, and references therein).
Rb-Sr and ZFT thermochronologic data from the crystalline core record early Miocene cooling from 550 to 200 °C at an average rate of ∼20 °C/m.y. (Fig. 6, open red symbols and arrow). The MsAr ages from the crystalline core (23–19 Ma) are consistent with these independent thermochronologic and geochronologic constraints (Fig. 6, red solid triangles); they are younger than the Rb-Sr Ms ages and lie on the cooling path defined by the Rb-Sr and ZFT chronometers. In contrast, most of the BtAr ages coincide with the timing of prograde metamorphism and are 10–20 m.y. older than what would be expected for cooling ages (Fig. 6, red solid circles).
Samples from the northern (Chandra valley; Fig. 6, blue symbols) and southern study areas (black symbols) yield more variable Rb-Sr and ZFT ages and are, in general, a few million years older than those from the crystalline core. The MsAr ages range from ca. 60 Ma to ca. 20 Ma (Fig. 6, black and blue solid triangles). The MsAr age spectrum of 807D1 (Fig. 3G) may represent a loss profile resulting from partial resetting at ≤28 ± 5 Ma (Forster and Lister, 2004; Viete et al., 2011). The MsAr age of 804C1 (64 ± 5 Ma; Fig. 3A) predates the Himalayan prograde metamorphism and may reflect partial resetting of Ms from this Ordovician granite. The BtAr ages from these areas are—with two exceptions—5–10 m.y. too old to be interpreted as cooling ages (Fig. 2).
Excluding the BtAr ages, the data set indicates regional rapid exhumation and cooling of the crystalline core since ca. 25 Ma (Fig. 6). Age gradients between the younger crystalline core and the slightly older lower-grade rocks may be attributed to diachronous exhumation. The samples from the northern and southern areas include rocks from the Grt and Bt + Chl zones. There, residence at peak temperatures may have been insufficient (too short and/or too cold) for complete resetting of the MsAr system, and these analyses may reflect partially reset ages (e.g., 804C1, 803B3, 807D1, 827B1; Fig. 6, blue and black solid triangles). Although this interpretation implies that the BtAr ages from these samples are likewise partially reset, it does not offer an explanation for the BtAr ages to be 5–20 m.y. older than the corresponding MsAr ages (Fig. 2B). Therefore, we interpret the MsAr data as early Miocene cooling ages, with the possible exception of a few samples from the low-grade metamorphic rocks (804C1, 803B3, 807D1, 827B1). BtAr ages are—with the possible exception of samples 814G1 and 807A1—not cooling ages.
Do Biotite 40Ar/39Ar Ages Correlate with Biotite Equilibration Temperatures?
The temperature and the portion of the pressure-temperature (P-T) path along which Bt crystallized are important for determining links between the BtAr age and metamorphic evolution. Although the Ti-in-biotite temperatures are consistent with regional metamorphic zoning in the study area, the temperatures are generally 100–200 °C lower than the results from Bt-Grt thermometry on samples from the same region (600–700 °C; Epard et al., 1995; Wyss, 2000). Several samples do not meet the criteria for which the geothermometer was calibrated, namely graphitic, peraluminous metapelites that contain ilmenite or rutile and have equilibrated at ∼4–6 kbar (Henry et al., 2005), suggesting that the temperature estimates may be inaccurate. Biotite in metaluminous samples generally incorporates higher amounts of Ti compared with peraluminous samples (Henry et al., 2005); therefore, the calculated temperatures may be upper estimates. Our Ti-in-biotite temperature estimates reveal two characteristics (Fig. 7A): (1) The oldest ArBt ages correspond to temperatures ≤∼400 °C. (2) There is no correlation between ArBt age and Ti-in-biotite temperature above ∼400 °C.
Effects of Biotite and Host-Rock Chemical Composition on Biotite 40Ar/39Ar Ages
The Bt chemical composition of our samples is typical for high-Al metapelitic Bt (Guidotti, 1984; Fleet et al., 2003) and similar to the samples used by Grove and Harrison (1996) for the characterization of Ar diffusion in Bt (Table 3). In particular, Mg/(Fe + Mg) ratios and anion compositions (F/[F + OH]) are comparable to our samples. Increased Ar retentivity—proposed for F-rich and Fe-rich mica (Harrison et al., 1985; Grove and Harrison, 1996)—is thus an unlikely cause for the anomalously old Bt. Similarly, the other elements analyzed (e.g., Ti, Cr, Mn, Ca, Ba, Na) do not show any unusual values. Although the Bt in metasedimentary samples (807D1, 807E1, 812D2, 811B1, 814G1, 815C1, 823G3, 827B1) has slightly higher Mg and lower F contents compared to the magmatic samples (807A1, 807C1, 810B1, 823G2; Table 3), the differences are insignificant. Neither the samples that yielded the oldest BtAr ages (823G2, 823G3, 827B1 ≥100 Ma) nor those with BtAr ages consistent with the regionally established cooling paths (807A1, 814G1) reveal any distinct chemical characteristics.
Figure 4 shows BtAr and MsAr ages color-coded by rock type and sorted by their location. There is no obvious correlation between the BtAr age and lithology. Chemical or mineralogical variations within each of the distinguished rock types are minor: The magmatic rocks have 20%–50% quartz, 15%–40% plagioclase, 20%–40% orthoclase; leucogranite lacks biotite, in contrast to granite/orthogneiss, which have biotite and muscovite (for chemical characterization of the magmatic rocks in central Himachal Pradesh, see Kreidler, 2014). The metasedimentary samples were obtained from the psammitic layers within the graywacke sequence and are, therefore, quartz-rich, too. Thus, quartz as a potential local Ar sink (see following discussion) occurs in all sampled lithologies in similar proportions. Quartz-poor metapelitic layers have not been sampled.
The discrepancy between the BtAr and MsAr ages tends to be larger in the southern part of the study area compared to the north, suggesting that proximity to the MCT shear zone may be a factor responsible for the anomalous BtAr ages (Fig. 4). This conclusion is further supported by samples 823G2 and 823G3 from the MCT zone, which yielded Cretaceous BtAr ages (despite Ti-in-biotite thermometry indicating Bt equilibration at ∼430–490 °C during Cenozoic metamorphism) and a MsAr cooling age from sample 823G3 at 22 ± 4 Ma (Table 2; Fig. 2A). However, distance to the MCT shear zone does not account for the anomalously old Bt ages from the elevation profile in the Chandra valley, where BtAr ages increase faster with elevation than do the MsAr ages (Fig. 4).
Effect of Chlorite on Biotite 40Ar/39Ar Ages
The effect of chloritization on BtAr data is not well understood, but it is known that chloritization might severely disturb age spectra (e.g., Lo and Onstott, 1989). Handpicking of Bt separates is an effective method to remove Chl and partly chloritized Bt in samples that contain significant amount of Chl. However, our XRF data from four optically Chl-free Bt samples show that ∼2% of Chl remained in all separates even after multiple rounds of ultrasonic cleaning of the grains (Fig. 5; Table 4). Because there is no difference in Chl content between the “well-behaved” (807A1) and “too-old” samples (807E1, 811B1, 812D2; Fig. 5), chloritization is unlikely to be the main cause for the anomalously old Bt ages. On the other hand, samples 823G2, 823G3, and 827B1 from the MCT zone, which yielded pre-Cenozoic BtAr ages, were strongly affected by retrograde metamorphism, accompanied by partial to complete chloritization of Grt. We cannot exclude the possibility that chloritization of Bt contributed to the anomalously old ages of these samples.
The Neoproterozoic to Cambrian Haimanta sedimentary rocks were intruded by Ordovician (486 ± 6 Ma; Stübner et al., 2014) granite and affected by local contact metamorphism and, possibly, a regional metamorphic overprint (Gehrels et al., 2003). Minerals that have not been reset during Himalayan orogeny may therefore be as old as ∼500 Ma, and even small relicts of such old K-bearing minerals and fluid inclusions in Himalayan metamorphic Bt may lead to too-old 40Ar/39Ar ages (Fig. 8, point 2). The probability that mineral phases and fluid inclusions with an early Paleozoic 40Ar component survived Cenozoic metamorphism is expected to decrease with the intensity of the metamorphic overprint (e.g., Copeland et al., 1991; Viete et al., 2011; Mottram et al., 2015). We argued earlier that in samples 804C1, 807D1, 814G1, and 827B1, the MsAr system and therefore likely also the BtAr system may have been partially reset, i.e., they may include a component of inherited Ar. Older-than-expected BtAr ages, however, are observed throughout the study area from the lowest- to the highest-grade metamorphic zones (Figs. 2B and 6), suggesting that inherited Ar is unlikely to be the main cause of the too-old BtAr ages.
After ruling out several factors that have previously been suggested to account for anomalously old 40Ar/39Ar ages or reverse age relationships between MsAr and BtAr ages, we suggest that excess Ar is the main factor producing too-old BtAr ages in the NW Himalaya. Many previous 40Ar/39Ar studies of Himalayan rocks have come to a similar conclusion, sometimes leading authors to dismiss or not even present their BtAr ages (e.g., Hubbard and Harrison, 1989; Macfarlane, 1993; Vannay and Hodges, 1996; Walker et al., 1999; Godin et al., 2001; Stüwe and Foster, 2001; Stephenson et al., 2001; Horton et al., 2015; Adams et al., 2015), whereas other Himalayan studies have reported BtAr ages that are unaffected by excess Ar (e.g., Hubbard and Harrison, 1989; Sorkhabi et al., 1996; Guillot et al., 1994; Wang et al., 2006; Leloup et al., 2015).
Inverse Isochron Diagrams Do Not Reveal Excess Ar
The 40Ar/39Ar step-heating experiments offer, besides age determinations, the opportunity to study the Ar degassing behavior of a specific mineral with increasing heating temperature. In the simplest case, Ar isotopes in the mineral are a mixture between radiogenic and atmospheric Ar, so that the data from the different heating steps plot in the inverse isochron diagram along a straight mixing line with an intercept at the 36Ar/40Ar axis defined by the atmospheric Ar isotope composition (atmospheric 40Ar/36Ar = 295.5). Excess Ar as a third component (commonly stated as its reciprocal, the initial 40Ar/36Ar ratio) becomes noticeable if it changes the 36Ar/40Ar intercept due to its nonatmospheric composition. Usually, excess Ar is enriched in 40Ar, so that the initial 40Ar/36Ar ratio becomes higher. Increased initial 40Ar/36Ar values are not uncommon in our experiments, for both Bt and Ms, with values up to 553 ± 2, but usually scattering between 300 and 350 (Table 2). No correlation between inverse isochron ages and initial 40Ar/36Ar ratios is detectable, either for all data together or for Ms or Bt alone (Fig. 7B). Moreover, the high-resolution BtAr analyses from ALF, for which all heating steps defined a plateau age and which show significantly less scatter in the inverse isochron diagrams (Figs. 3E, 3F, 3N, 3O, 3S, 3T, and 3U), yielded trapped 40Ar/36Ar values between 270 and 340, all indistinguishable within error from the atmospheric value (Table 2).
The discussion about the occurrence of excess Ar and its detection from the Ar degassing behavior during step-heating experiments is hampered by the fact that during step heating, excess 40Ar may be released together with the radiogenic Ar, the trapped Ar, or with both (Kuiper, 2002). If only the trapped Ar component is contaminated with excess 40Ar, the 36Ar/40Ar intercept of the inverse isochron will yield initial 40Ar/36Ar >295.5, but the 39Ar/40Ar intercept will reflect only the radiogenic 40Ar, and the inverse isochron age will be unaffected by excess Ar. Conversely, if only the radiogenic component is contaminated by excess 40Ar, only the age is changed, but not the initial 40Ar/36Ar ratio, even though the latter is usually regarded as an indication for excess Ar (Kuiper, 2002). This means that the nonexistence of increased initial 40Ar/36Ar ratios is not a proof for the absence of excess Ar. Therefore, excess Ar may explain the too-old BtAr ages in the Himachal Pradesh despite the fact that initial 40Ar/36Ar ratios are not consistently higher than the atmospheric ratio.
Which Processes May Have Led to Excess Ar in the Biotite?
The 40Ar/39Ar age dating technique relies on the assumption that Ar concentrations outside of the mineral grain remain close to zero, so that Ar that diffuses out of the mineral grain escapes into an “infinite reservoir” (Fig. 8, point 1; e.g., Dodson, 1973). This reservoir may be a local sink, such as a fluid or a host-rock mineral with a high Ar solubility, or an external reservoir—ultimately the atmosphere. The “total local sink capacity” is a function of the modes of mineral and fluid phases in the vicinity of the source mineral and their respective Ar partition coefficients; specifically, it has been suggested that the presence or absence of quartz in a rock affects the total local sink capacity and hence controls excess Ar accumulation (Baxter, 2003). In addition, efficient Ar transport from the Ar-releasing mineral grain into the external reservoir is a critical aspect in 40Ar/39Ar thermochronology (e.g., Baxter, 2003). Transport of Ar in a rock volume depends on the presence or absence of fluids in the intergranular space, and it is necessary to distinguish “open systems,” in which fluid circulation facilitates rapid transport of Ar over large distances (meters to kilometers), and dry “closed systems,” in which transport of Ar is limited to a few centimeters over millions of years (Scaillet, 1996; Kelley, 2002; Baxter, 2003). In dry systems, 40Ar may accumulate along grain boundaries and partition into minerals, resulting in “internally” derived excess Ar; in fluid-rich systems, if the fluid is enriched in 40Ar, partitioning of 40Ar into the mineral may result in “externally” derived excess Ar (see Kelley, 2002, and references therein; Baxter et al., 2007). Here, we consider the following (not mutually exclusive) options for excess Ar accumulation:
High 40Ar Partial Pressure in the Intergranular Medium
Biotite and other K-bearing minerals from the Neoproterozoic–Cambrian Haimanta metagraywacke and/or Ordovician granite broke down and/or outgassed during Eocene–Oligocene prograde Barrovian metamorphism, releasing radiogenic 40Ar (40Ar*) that had accumulated over ∼500 m.y. A typical Haimanta rock with 3–4 wt% K2O (Vance and Mahar, 1998; Chambers et al., 2009; Stübner, personal data) could result in ∼100 ppb 40Ar* in the host rock, assuming 40Ar* accumulation over 500 m.y. and no Ar loss (see Appendix 1). For granite with 6 wt% K2O (Kreidler, 2014), the 40Ar* concentrations would add up to ∼200 ppb in a closed system. These 40Ar* concentrations are 10–20 times higher than the ∼11 ppb 40Ar* that would have accumulated in Bt with 10 wt% K2O over ∼20 m.y. and could significantly disturb the 40Ar* budget of Bt and other mineral phases. High Ar partial pressure in the intergranular space in combination with the lower Ar solubility in Ms as compared to Bt may have resulted in partitioning of 40Ar into Bt in preference to Ms (Fig. 8, point 3; for a recent summary of Ar solubility, see Kelley, 2002). This mechanism predicts that metasedimentary and magmatic rocks are both affected by excess Ar, depending on their respective mineralogical composition (e.g., quartz content) and K concentration (Baxter, 2003). This may explain why both magmatic and metasedimentary samples are affected by excess Ar (Fig. 4), but it does not provide a straightforward explanation for why BtAr ages from the MCT zone are significantly more affected by excess Ar than samples from the MCT hanging wall.
Dry Intergranular Medium and Cenozoic Partial Melting Disturb the 40Ar Balance
The solubility of Ar in hydrous fluids is 4–5 orders of magnitude higher than in most minerals (see compilation in Kelley, 2002), providing an effective local sink for Ar as well as serving as a pathway to a more external sink (Baxter, 2003; Baxter et al., 2007; Fig. 8, point 1). For dry systems, it has been shown both experimentally (Baxter et al., 2007) and in field studies (Foland, 1979; Scaillet, 1996) that restricted mobility of Ar can cause accumulation of 40Ar in the intergranular medium and can result in excess Ar in minerals.
In the Himachal Himalaya, there is ample evidence for fluid circulation throughout the Cenozoic. Examples include the following: (1) Ubiquitous quartz segregation veins are associated with early to late Himalayan deformation and metamorphism (e.g., Epard et al., 1995; Wyss et al., 1999). (2) Tourmaline is one of the main accessory phases in the Haimanta metasedimentary rocks. (3) Ky + Qtz segregation veins are common and indicate the presence of aqueous fluids during amphibolite-facies metamorphism. (4) Pegmatite dikes intruded along the axial surfaces of folds are related to the late stages of thrusting along the MCT and formation of the Phojal fold in the late Oligocene–early Miocene (Epard et al., 1995; our observations). We therefore consider it unlikely that dry conditions existed throughout the entire Cenozoic metamorphic cycle.
It is, however, conceivable that partial melting and melt extraction consumed hydrous fluids, resulting in locally dry pore spaces (Fig. 8, point 4). The NW Himalaya was affected by a protracted history of crustal melting between ca. 36 and 18 Ma (e.g., Dézes et al., 1999; Robyr et al., 2006; Langille et al., 2012; Lederer et al., 2013; Stübner et al., 2014), similar to other locations in the Himalaya (e.g., Rubatto et al., 2013; Zeiger et al., 2015). Removal of hydrous fluids by late Eocene–Oligocene migmatization could have significantly reduced the total local sink capacity (Baxter, 2003), resulting in accumulation of excess Ar. However, most, if not all, of our Bt samples are from rocks that did not undergo partial melting; for dry conditions to have affected the BtAr ages, migmatization and melt extraction must have affected the fluid budget on a several-kilometer scale. In this case, it is theoretically possible that the excess-Ar Bt ages provide a maximum age estimate of migmatization.
Silicate melts have Ar solubilities intermediate between those of hydrous fluids and most minerals (e.g., Kelley, 2002, and references therein). During migmatization, melts are likely to serve as an additional sink for intergranular 40Ar. During melt crystallization, 40Ar may be released back into the intergranular medium, increasing its 40Ar concentration. Consequently, excess Ar may diffuse from the intergranular medium into the mineral phases for as long as the rocks reside at temperatures above Tc (Fig. 8, point 5). Some Himalayan Miocene leucogranite yields geologically reasonable BtAr ages, reflecting the timing of emplacement and cooling (e.g., Copeland et al., 1990; Hodges et al., 1998; Wang et al., 2006), whereas in other locations, the BtAr ages are apparently affected by excess Ar (Horton et al., 2015). Therefore, migmatization does not offer a straightforward explanation for excess BtAr ages in the study area or in the Himalayan orogen in general.
Migmatization and melt emplacement may be related to accumulation of excess Ar in minerals through two effects: (1) It affects the availability of hydrous fluids in the pore space, and (2) the high solubility of Ar in the melt affects the distribution of 40Ar in the melt, hydrous fluid, and mineral phases. A better understanding of these mechanisms requires determination of partition coefficients of Ar between silicate melt, hydrous fluids, and mineral phases, in particular, for Ms and Bt.
Fluid Circulation along the Main Central Thrust
In the NW Himalaya, the MCT was active between ca. 26–23 Ma and ca. 17–15 Ma (e.g., Stephenson et al., 2001; Vannay et al., 2004). This thrust activity was associated with intense fluid circulation, causing retrograde greenschist-facies metamorphism with chloritization of Bt and Grt and formation of centimeter- to decimeter-sized quartz veins within the ∼3-km-wide shear zone (e.g., Wyss, 2000). Circulation of fluids has been documented along the entire MCT shear zone, for example, in Nepal (Copeland et al., 1991; Evans et al., 2008; Derry et al., 2009) and Bhutan (Stüwe and Foster, 2001; Mottram et al., 2015), where it was proposed to account for excess Ar in Bt and a reverse relationship between MsAr and BtAr ages. In central Himachal Pradesh, two MsAr ages from the MCT shear zone and its immediate hanging wall are geologically reasonable: The age for 823G3 (22 ± 4 Ma) is indistinguishable from other MsAr cooling ages from the hanging-wall crystalline rocks; the age for 827B1 (32 ± 9 Ma) is consistent with other thermochronometric ages from the same locality (Fig. 2A). In contrast, Bt from these samples yielded pre-Cenozoic 40Ar/39Ar ages, significantly older than any other BtAr ages from the study area and predating Cenozoic prograde metamorphism.
Studies on the composition of modern hydrothermal fluids along the MCT suggest a large component of meteoric water (e.g., Derry et al., 2009), but the Ar isotopic composition of these fluids is unknown. High partial pressures of Ar of any isotopic composition can slow or inhibit diffusive loss of radiogenic 40Ar from the mineral, thus leading to accumulation of internally derived excess Ar (e.g., Baxter, 2003). In the end-member scenario of high Ar partial pressure in the intergranular medium blocking diffusional escape of radiogenic 40Ar, the excess Ar Bt age could theoretically record the time when the intergranular medium became saturated in Ar. Therefore, the pre-Cenozoic BtAr ages from three samples from the MCT zone (827B1, 823G2, 823G3) cannot be attributed to an influx of atmospheric Ar in the late Oligocene–Miocene alone. If anomalous BtAr ages from the MCT zone result from fluid circulation within the MCT shear zone, as suggested by the spatial correlation in the study area and by 40Ar/39Ar thermochronology studies in other localities in the Himalaya (Hubbard and Harrison, 1989; Stüwe and Foster, 2001), these hydrous fluids must have been enriched in 40Ar (Fig. 8, point 3). The source of this 40Ar may, for example, be prograde metamorphic Ar release (see earlier herein). Possible mechanisms leading to excess Ar in Bt include partitioning of 40Ar from the fluid phase into the mineral or trapping in fluid inclusions. Because the Bt samples from the MCT zone are an order of magnitude more strongly affected by excess Ar than the hanging-wall samples (Cretaceous vs. Cenozoic BtAr ages), and excess Ar in the hanging-wall samples does not vary systematically with distance to the MCT (Fig. 4), we suggest that various, although possibly related mechanisms account for excess Ar in the MCT zone and in the hanging wall, respectively.
Most of the BtAr ages from granitic and metapelitic samples from Himachal Pradesh are older than the MsAr ages from the same samples and are 5–20 m.y. older than the expected timing of cooling through the BtAr closure temperature derived from independent geochronologic evidence. Apart from BtAr ages from the MCT zone, which are pre-Cenozoic, BtAr ages tend to cluster at 20–40 Ma. The “too-old” BtAr ages do not correlate with lithology or structural position. The Bt separates contain only ∼2% Chl, suggesting that chloritization is not the main cause for these anomalously old BtAr ages.
We suggest that the likely cause for the “too-old” BtAr ages in the Himachal Pradesh is excess Ar in Bt but not, or to a lesser extent, in Ms. Possible sources of excess Ar are:
(1) 40Ar released from older K-bearing minerals during Cenozoic prograde metamorphism and mineral breakdown, which partitioned into Bt and/or inhibited diffusive loss of radiogenic 40Ar at high temperature;
(2) 40Ar that accumulated in Bt at temperatures above Tc because a temporarily dry intergranular medium, possibly resulting from late Eocene–Oligocene migmatization and melt extraction, restricted 40Ar transport toward an external sink;
(3) 40Ar released from melt into a hydrous fluid phase during Oligocene melt crystallization, resulting in high 40Ar concentrations in the rock and partitioning of 40Ar into Bt.
Samples from the MCT shear zone yield significantly older BtAr ages than samples from the hanging-wall rocks and may be affected by a different mechanism of excess-Ar accumulation, possibly linked to fluid circulation within the MCT zone and strong chloritization.
The proposed mechanisms of excess-Ar accumulation in Bt depend on the solubility of Ar in minerals, hydrous fluids, and melts, and further experiments that determine partition coefficients between these phases will provide further insight into the problem of excess Ar; in particular, although it is accepted that solubility of Ar in minerals is an order of magnitude lower than in hydrous fluids, the partition coefficients between different mineral phases are poorly known (see compilation in Kelley, 2002). Excess Ar is more commonly detected in Bt than in Ms, possibly due to higher solubility of Ar in Bt compared to Ms. This makes Bt more susceptible to excess-Ar accumulation if the zero-Ar boundary condition is violated in an open or closed system. The higher susceptibility of Bt to excess Ar compared to Ms could also be due to its diffusion characteristics, with its higher Ar diffusivity responsible for incorporation of more excess Ar in cases where a surplus is available in the intergranular medium. This surplus can be provided by Ar-enriched fluids or might be an effect of Ar accumulation under dry conditions (closed system), but it can also be caused by an increasing partition coefficient between mineral and fluid during postmetamorphic cooling (Kelley, 2002). The decreasing solubility of Ar in fluids at decreasing temperatures triggers its diffusion into the solid phase, preferentially into minerals with high Ar diffusivity, and thus more readily into Bt compared to Ms. These considerations suggest that in most sample sets in which Bt is affected by excess Ar, it is most likely that Ms, too, contains excess Ar, although its concentrations are probably smaller and may not affect 40Ar/39Ar ages.
This study was funded by Deutsche Forschungsgemeinschaft (DFG) grant STU525/1. Field work and analytical work by Grujic was supported by the Natural Sciences and Engineering Research Council of Canada (Discovery Grant RGPIN 04297). Operation of the Argon Laboratory, Freiberg, Germany, was supported by DFG grant Ra 442/37. Stübner acknowledges financial support from the Excellence Initiative of University of Tübingen and the Athene Program. Tashi Tsering assisted with field work. Andreas Kronz (University of Göttingen) and Thomas Wenzel (University of Tübingen) helped with electron microprobe analyses; Heiner Taubald (University of Tübingen) conducted whole-rock X-ray fluorescence analyses. We are thankful for the constructive comments of Y. Kuiper, P.H. Leloup, E. Baxter, and an anonymous reviewer, which improved an earlier version of this manuscript, and we thank Damian Nance for editorial handling.
APPENDIX 1. CALCULATION OF 40Ar ACCUMULATION
Assuming 3.5 wt% K2O in Haimanta sediments and using molar masses K2 = 2 × 39.1 g/mol and K2O = 94.2 g/mol, the amount of K in Haimanta sediments is 3.5 × (2 × 39.1/94.2) = 2.91 wt% K2. Using the molar fraction of 40K/K = 0.000117, the amount of 40K in the sediment is 0.000117 × (2.91/100) × 1E+6 = 3.4 ppm 40K or 3.4 mg 40K per 1 kg rock. The decay of 40K to 40Ar* is calculated after McDougall and Harrison (1999, their Eq. 2.11). For t = 500 m.y., the amount of 40Ar* is 0.114 ppm (0.114 mg 40Ar per 1 kg rock). For Ordovician granite with typically 6 wt% K2O (Kreidler, 2014), the same calculation yields 0.684 ppm 40Ar accumulated over 500 m.y.