⁴⁰Ar/³⁹Ar and cosmic ray exposure ages of plagioclase-rich lithic fragments from Apollo 17 regolith, 78461

Abstract

Argon isotopic data is used to assess the potential of low-mass samples collected by sample return missions on planetary objects (e.g., Moon, Mars, asteroids), to reveal planetary surface processes. We report the first ⁴⁰Ar/³⁹Ar ages and ³⁸Ar cosmic ray exposure (CRE) ages, determined for eleven submillimeter-sized (ranging from 0.06 to 1.2 mg) plagioclase-rich lithic fragments from Apollo 17 regolith sample 78461 collected at the base of the Sculptured Hills. Total fusion analysis was used to outgas argon from the lithic fragments. Three different approaches were used to determine ⁴⁰Ar/³⁹Ar ages and illustrate the sensitivity of age determination to the choice of trapped (⁴⁰Ar/³⁶Ar)_t. ⁴⁰Ar/³⁹Ar ages range from ~4.0 to 4.4 Ga with one exception (Plag#10). Surface CRE ages, based on ³⁸Ar, range from ~1 to 24 Ma. The relatively young CRE ages suggest recent re-working of the upper few centimeters of the regolith. The CRE ages may result from the effect of downslope movement of materials to the base of the Sculptured Hills from higher elevations. The apparent ⁴⁰Ar/³⁹Ar age for Plag#10 is >5 Ga and yielded the oldest CRE age (i.e., ~24 Ma). We interpret this data to indicate the presence of parentless ⁴⁰Ar in Plag#10, originating in the lunar atmosphere and implanted in lunar regolith by solar wind. Based on a chemical mixing model, plagioclase compositions, and ⁴⁰Ar/³⁹Ar ages, we conclude that lithic fragments originated from Mg-suite of highland rocks, and none were derived from the mare region.

Introduction

The regolith of planetary objects (e.g., Moon, Mars, and asteroids) can be defined as the boundary layer between the solid crust and outer space. The regolith evolution on planetary objects is a function of the continuous flux of impactors of various sizes, volcanism, as well as the constant bombardment by solar and galactic energetic particles. Regolith generally consists of fragmental and unconsolidated rock material. The hyper-velocity of the impacts and the continuous bombardment of energetic particles and micrometeorites reduce the initially rocky planetary surface to increasingly finer grain-sized regolith. For example, on the lunar surface, the mean grain size of lunar regolith ranges from 40 to 800 μm and averages between 60 and 80 μm (Lucey et al. 2006). A record of impact events and interactions with cosmic rays preserved in lunar regolith can be understood by geochemical and chronological studies of lunar regolith samples. The outcome of these studies provide constraints to understand the evolution of the Earth-Moon system and inner solar system (e.g., Lucey et al. 2006; Stöffler et al. 2006). The geochemical and chronological constraints developed based on lunar regolith studies can be applied to the surfaces of other solar system objects and used as a guide for future and ongoing missions like Hayabusa (Tsuda et al. 2013; Okada et al. 2015) to a C-type asteroid 1999 JU3, OSIRIS-REx sample return mission (Lauretta and Team 2012) to the asteroid 101955 Bennu, and the Mars Science Laboratory (Grotzinger et al. 2012), currently conducting experiments on the Martian surface. Kring (2015) describes how the success of NASA’s Orion crew vehicle, as well as the outcome of current lunar orbiters, has revitalized future interest in robotic and human exploration programs to the moon and beyond.

Regolith sample collected at station 8, Apollo 17

The Apollo 17 landing site is located in the center of the Taurus-Littrow valley. One of the prominent geological features of the valley is the highland unit found near the north massif, the Sculptured Hills. Apollo 17 astronauts sampled the regolith at the base of the Sculptured Hills during their last surface traverse at station 8 (latitude 20.278° N, longitude 30.848° S; Robinson and Jolliff 2002). They dug a 25-cm-deep trench (Wolfe et al. 1981) that revealed the best preserved lunar regolith stratigraphy collected during the Apollo missions (Mitchell et al. 1973; Fig. 1). Based on the chemical composition of the deepest layer of the trench (i.e., sample 78421), Korotev and Kremser (1992) concluded that the regolith represents six chemical components. Among them, three chemical components are of mare origin: (1) high-Ti (HT) mare basalt, (2) very-low-Ti (VLT) basalt, and (3) pyroclastic orange glass. The other three chemical components are of highland origin: (4) impact-melt noritic breccia (NB), (5) anorthositic norite (AN) breccia, and (6) Mg-suite (high Mg/Fe troctolites and norites). A similar study by Simon et al. (1981) found that mare volcanics and highland-derived fragments were present in equal proportions in Apollo 17 regolith sample 78221. A mixing model based on the major and trace-element composition of the sample from the top layer of trench regolith (i.e., 78481) suggest mixing of 40–50 % mare and 52–59 % highland components (Rhodes et al. 1974). Goswami and Lal (1974) performed nuclear track studies on the samples from the trench and observed that the density of tracks observed in the bottom sample (i.e., 78121) is only slightly less than the track density observed in the surface sample (i.e., 78481) and concluded that the track density distribution pattern, along with the knowledge of the geological setting at station 8, can be used to characterize the macroscopic lunar surface transport processes including regolith mixing and down slope movement. Prior to our study, the crystallization and metamorphic ages on this trench regolith sample were unknown (Meyer 2010). We present the first ⁴⁰Ar/³⁹Ar ages and ³⁸Ar cosmic ray exposure (CRE) ages on plagioclase-rich lithic fragments separated from the 78461 trench regolith sample collected at station 8. These chronology data are used together with plagioclase compositions and a mixing model to determine the origin and duration that the regolith resided on the lunar surface.

Fig. 1

Topographic image of the Sculptured Hills and an inset image of station 8 trench. The moderately inclined transition zone between the Sculptured Hills and the Taurus-Littrow Valley floor with Apollo 17 mission traverses indicated. Station 8 was located at the base of the Sculptured Hills. A 25-cm-deep trench (shown in inset) was sampled at station 8. The plagioclase-rich lithic fragments analyzed in this work were separated from 78460 (1 to 6 cm layer). Also, the crew climbed up towards the hills to sample the 0.5-m boulder (i.e., 78235). Background image: LRO Narrow Angle Camera (NAC) high-resolution image (0.5 m per pixel) laid over the wide-angle Camera morphologic basemap (100 m/pixel). Inset image source: Lunar and Planetary Institute, Houston (http://www.lpi.usra.edu/resources/apollo/frame/?AS17-142-21720)

Sample and experimental details

Sample details

Regolith sample 78460 was collected at a depth interval of 1–6 cm at station 8 at the base of the Sculptured Hills (Meyer 2010), and ~20 m above the Taurus-Littrow valley floor from the Apollo 17 landing site (Fig. 1). Plagioclase-rich lithic fragments comprise ~7–9 modal percentage of the 1000–90 μm (Heiken and McKay 1974) size fractions of station 8 regolith samples. The median grain size of regolith sample 78460 is 47 μm (Graf 1993). The eleven plagioclase-rich lithic fragments analyzed in this study were handpicked from the 1000–90-μm fraction of a 5-g aliquot of 78461, which is the <1 mm size fraction of 78460 (Butler 1973). The sieved fraction was ultrasonically cleaned in acetone, dried, handpicked, and weighed. Out of eleven handpicked lithic fragments, eight fragments were less than 0.52 mg. The other three fragments, Plag#21, Plag#23, and Plag#24 weighed 1.20, 1.08, and 0.72 mg, respectively (Table 1).

Table 1 Chemical composition of plagioclase-rich lithic fragments separated from Apollo 17, 78461 regolith

Chemical characterization of plagioclase-rich lithic fragments

Plagioclase-rich fragments were individually mounted in CrystalBond^© adhesive, ground, polished, and carbon coated. Samples were analyzed for major element composition using a CAMECA SX-100 electron microprobe at Rensselaer Polytechnic Institute with run condition of 15 kV, 20 nA, and 30-s counting times with a 10-μm-diameter beam. Samples were subsequently polished with 0.3-μm alumina powder to remove the carbon coating and warmed to ~100 °C for ~1 min to soften the CrystalBond^© adhesive for removal of each lithic fragment. These were then ultrasonically cleaned in acetone to remove adhering CrystalBond©. Major elemental compositions of the eleven plagioclase-rich lithic fragments are given in Table 1. The lithic fragments contain Ca-rich feldspar (i.e., anorthite), with compositions ranging from An₉₃ to An₉₆ (Table 1). The K₂O contents range from 0.03 to 0.28 wt.% suggesting 200 to 2300 ppm of K in the fragments. The Fe/(Fe + Mg) ratio varies from 0.19 (Plag#1) to 0.84 (Plag#24). Given the size of the plagioclase lithic fragments (<48 μm), we were unable to obtain a pure plagioclase separate using standard mineral separation techniques.

Irradiation of plagioclase-rich lithic fragments

Following electron probe analysis, each plagioclase-rich lithic fragment was individually packed in Al foil and irradiated for 100 h in the cadmium-lined in-core irradiation tube (CLICIT) facility of the TRIGA reactor at Oregon State University, USA. Cadmium shielding was used to minimize the production of ³⁸Ar from ³⁷Cl and undesirable isotopic interference reactions. The neutron flux was monitored using Hb3gr (hornblende, PP-20) hornblende (Turner 1971; Jourdan and Renne 2007; Schwarz and Trieloff 2007). These flux monitor standards were placed between the plagioclase-rich lithic fragments to monitor the fast neutron dose. In addition, CaF₂ and KSiO₃ glasses were also placed between the samples to derive correction factors for reactor-produced ³⁷Ar and ³⁹Ar.

Gas extraction and mass spectrometric measurements

Following irradiation, the plagioclase-rich lithic fragments, international standard Hb3gr (hornblende, PP-20) flux monitors, CaF₂, and KSiO₃ were unwrapped from Al foil. The samples were outgassed using a CO₂ laser (Synrad model J48) with beam diameter of 3.5 mm and ~8 W of power. Extracted Ar from each fused sample was measured using a VG5400 noble gas mass spectrometer in Syracuse University’s Noble Gas Isotope Research Laboratory. Based on the measurement of 8 PP-20 grains (1073.6 ± 4.6 (1σ); Schwarz and Trieloff (2007)), the mean J value for the samples is calculated to be 0.02643 ± 0.00021 (1σ). These values are similar to those reported for 100 h irradiation at the CLICIT facility (Fernandes et al. 2013; Gombosi et al. 2015). The correction factors for reactor-produced isotopes are as follows: (³⁹Ar/³⁷Ar)_Ca = (7.57 ± 0.34) × 10⁻⁴, (³⁶Ar/³⁷Ar)_Ca = (3.22 ± 1.26) × 10⁻⁴, and (⁴⁰Ar/³⁹Ar)_K = 0.01576 ± 0.0004. After extraction, reactive gases were removed using SAES ST-707 getters. The purified gas was then expanded into the VG5400 mass spectrometer, and 11 cycles at m/e = 40, 39, 38, 37, 36, and 35 were measured in peak jumping mode. Blank measurements were performed before and after every sample analysis to determine the background for Ar isotopes in the extraction system. Typical blank values in volts, during the course of these experiments, including 1σ errors are as follows: ³⁵Ar = (1.064 ± 0.330) × 10⁻⁸, ³⁶Ar = (1.521 ± 0.004) × 10⁻⁷, ³⁷Ar = (1.933 ± 0.414) × 10⁻⁸, ³⁸Ar = (2.849 ± 1.156) × 10⁻⁸, ³⁹Ar = (8.923 ± 7.049) × 10⁻¹⁰, and ⁴⁰Ar = (1.320 ± 0.001) × 10⁻⁵. For ⁴⁰Ar and ³⁹Ar, the blank seldom exceeded greater than 5 % of sample signal during the experiments. Air standards were analyzed before and after the sample analyses to determine the Ar sensitivity, as well as mass discrimination. The Ar sensitivity determined during the course of these experiments was 1.340 × 10¹¹ V/mol. The mass discrimination correction factors for ³⁶Ar, ³⁷Ar, ³⁸Ar, and ³⁹Ar relative to ⁴⁰Ar were 0.973 ± 0.15 %, 0.981 ± 0.33 %, 0.987 ± 0.16 %, and 0.993 ± 0.57 %, respectively. Based on the present Ar sensitivity and low K content in the plagioclase-rich lithic fragments, individual fragments were outgassed in a single-step total fusion. Multiple step extraction would have resulted in the Ar signals below the present detection limit of the VG5400 mass spectrometer.

Sources of Ar in lunar regolith: radiogenic, cosmogenic, and trapped argon components

The top layer of lunar regolith is constantly bombarded by energetic particles originating from solar wind, solar cosmic rays, and galactic cosmic rays. The three stable isotopes of Ar (³⁶Ar, ³⁸Ar, and ⁴⁰Ar) are affected in different ways due to the interaction of lunar regolith with outer space. For example, ³⁶Ar resides close to the surface of individual grains and is mainly implanted by solar wind; this component is the trapped component (e.g., Wieler and Heber 2003). Cosmogenic ³⁸Ar is produced by interaction of energetic protons from solar and galactic cosmic rays with target elements such as Fe, Ca, K, and Ti. ⁴⁰Ar is produced by radioactive decay of ⁴⁰K; however, an “excess or parentless ⁴⁰Ar” component has also been observed in lunar regolith samples (Manka and Michel 1971). On the lunar surface, ⁴⁰Ar/³⁶Ar is 2–14 (e.g., Norman et al. 2003), and solar wind ⁴⁰Ar/³⁶Ar is 10⁻⁴ (Lodders 2010). The ⁴⁰Ar atoms present in the lunar atmosphere is photoionized and then accelerated by solar wind. About 50 % of the ⁴⁰Ar ions escape to space, and the remaining 50 % of ⁴⁰Ar atoms are driven back to the lunar surface (Signer et al. 1977). Only a few percent of ions are trapped in the lunar regolith while most ⁴⁰Ar ions are recycled to the lunar atmosphere. The small fraction of trapped lunar atmospheric ⁴⁰Ar has been proposed to be the source of excess ⁴⁰Ar in lunar regolith (e.g., Manka and Michel 1971; Wieler and Heber 2003). Below, we discuss procedures for calculating the abundance of radiogenic ⁴⁰Ar and cosmogenic ³⁸Ar to determine ⁴⁰Ar/³⁹Ar ages and ³⁸Ar cosmic ray exposure ages, respectively, for the plagioclase-rich lithic fragments.

Calculating the abundance of radiogenic ⁴⁰Ar

In the case of lunar regolith samples, the non-radiogenic ³⁶Ar originates from solar wind. Measured ⁴⁰Ar (⁴⁰Ar_m) is corrected for non-radiogenic ⁴⁰Ar using the non-radiogenic trapped (⁴⁰Ar/³⁶Ar)_t and assuming the measured ³⁶Ar is of trapped Ar. Generally, a standard isochron plot (⁴⁰Ar/³⁶Ar vs. ³⁹Ar/³⁶Ar) is used where the y-intercept of a linear regression is assumed to be the trapped argon composition (e.g., Merrihue and Turner 1966; McDougall and Harrison 1999). We use two approaches to determine the trapped ⁴⁰Ar/³⁶Ar composition from total fusion analyses of plagioclase-rich lithic fragments. The first approach assumes that all the fragments contain the same trapped (⁴⁰Ar/³⁶Ar) composition. The second approach assumes that a trapped ⁴⁰Ar/³⁶Ar = 0.

Calculating the abundance of cosmogenic ³⁸Ar

The measured ³⁸Ar abundances in the plagioclase-rich lithic fragments result from (1) nucleogenic (i.e., reactor produced), (2) trapped, and (3) cosmogenic components. During neutron irradiation, ³⁸Ar can be produced from Ca (i.e., ⁴²Ca(n, nα)³⁸Ar) and K (i.e., ³⁹K(n,d)³⁸Ar). Based on CaF₂ crystals that were also irradiated with plagioclase-rich lithic fragments, the (³⁸Ar/³⁷Ar)_Ca is (4.50 ± 0.02) × 10⁻⁵, a negligible contribution to total ³⁸Ar. To estimate the ³⁸Ar produced from K, we determined (³⁸Ar/³⁹Ar)_K = 0.0120 ± 0.0002, on KSiO₃ glasses irradiated with the plagioclase-rich lithic fragments. ³⁸Ar can also be produced during irradiation from reactions on Cl (i.e., ³⁷Cl(n, γ)³⁸Cl(β⁻)³⁸Ar). However, in lunar regolith, the amount of Cl is very low, averaging 20 ppm (e.g., Taylor and Hodges 1981). Because cadmium shielding was used during irradiation, ³⁸Ar produced from ³⁷Cl can be neglected.

The trapped ³⁸Ar in lunar surface samples results from solar wind implantation ((³⁸Ar/³⁶Ar)_t = 0.188; Eberhardt et al. (1972)). Following previous studies (e.g., Fernandes et al. 2013), the cosmogenic ³⁸Ar (³⁸Ar_c) in individual plagioclase-rich lithic fragments was determined using the following equation:

$$ {}^{38}\mathrm{A}{\mathrm{r}}_{\mathrm{c}}{=}^{38}\mathrm{A}{\mathrm{r}}_{\mathrm{m}}{-}^{36}\mathrm{A}{\mathrm{r}}_{\mathrm{m}}\times {\left({}^{38}\mathrm{A}\mathrm{r}/{}^{36}\mathrm{A}\mathrm{r}\right)}_{\mathrm{t}}-\left[{}^{39}\mathrm{A}{\mathrm{r}}_{\mathrm{K}}\times {\left({}^{38}\mathrm{A}\mathrm{r}/{}^{39}\mathrm{A}\mathrm{r}\right)}_{\mathrm{K}}\right], $$

(1)

where ³⁹Ar_K = ³⁹Ar* = ³⁹Ar_m − ³⁷Ar_m × (³⁹Ar/³⁷Ar)_Ca; (³⁹Ar/³⁷Ar)_Ca = (7.57 ± 0.34) × 10⁻⁴; (³⁸Ar/³⁹Ar)_K = 0.012 ± 0.0002; and c, m, and t subscripts denote cosmogenic, measured, trapped (i.e., solar wind), respectively; and K and Ca subscripts indicate reactor-produced components.

Results

Argon data for the plagioclase-rich lithic fragments, corrected for blank contribution, and reactor-produced interferences are given in Table 2.

Table 2 Ar isotope data, cosmic ray exposure ages, ⁴⁰Ar/³⁹Ar ages for individual plagioclase-rich lithic fragments from Apollo 17, 78461 sample

Solar wind-implanted ³⁶Ar

On the lunar surface, solar wind is the most dominant source of ³⁶Ar (i.e., ions of ³⁶Ar directly implanted in lunar regolith). Since the implantation of solar wind is depth dependent and can penetrate only a few microns on the lunar surface, the variation in abundance of ³⁶Ar is a function of solar wind implantation duration and/or the depth at which fragments were exposed to solar wind. The measured ³⁶Ar abundance (³⁶Ar_m) in eleven plagioclase-rich lithic fragments, corrected for blank contribution and mass discrimination, varies from 0.41 × 10⁻⁶ cm³ STP/g (i.e., Plag#23) to 4.1 × 10⁻⁶ cm³ STP/g (i.e., Plag#10) (Table 2).

Argon isotopic composition

The comparison of (⁴⁰Ar/³⁶Ar)_m and (³⁸Ar/³⁶Ar)_m ratios for eleven plagioclase-rich lithic fragments, corrected for blank and mass discrimination, is shown in Fig. 2. Generally, ⁴⁰Ar/³⁶Ar values are explained as a result of the mixture between non-radiogenic trapped and radiogenic argon compositions. Non-radiogenic trapped argon on the lunar surface is mostly solar wind-implanted argon as discussed above. The (⁴⁰Ar/³⁶Ar)_m varies from 0.62 (i.e., Plag#14) to 15.5 (i.e., Plag#23) with higher values suggesting the presence of radiogenic and/or excess parentless ⁴⁰Ar in the samples. The (³⁸Ar/³⁶Ar)_m value for individual plagioclase-rich lithic fragments varies from ~0.191 to 0.6 (Fig. 2) and can be explained as a mixture of non-radiogenic trapped (i.e., ³⁸Ar/³⁶Ar = 0.188; Eberhardt et al. (1972)) and cosmogenic Ar (i.e., ³⁸Ar/³⁶Ar = 1.54; Hohenberg et al. (1978)). For the Plag#23 sample, the concentration of ³⁶Ar is lowest when compared to all other plagioclase-rich lithic fragments. We argue that the high ³⁸Ar/³⁶Ar ratio is a result of ³⁶Ar loss from Plag#23. The retentivity of noble gas components generally have an order: solar wind implanted (e.g., ³⁶Ar) <radiogenic (e.g., ⁴⁰Ar) <cosmogenic (e.g., ³⁸Ar). In general, solar wind-implanted gases are released at low temperature when stepwise heating is performed on lunar regolith samples (e.g., Hohenberg et al. 1970). This observation supports the argument that the solar wind-implanted Ar is more susceptible to loss compared to radiogenic and cosmogenic components of Ar.

Fig. 2

Argon three isotope plot for plagioclase-rich lithic fragments. (⁴⁰Ar/³⁶Ar)_m vs. (³⁸Ar/³⁶Ar)_m for eleven plagioclase-rich lithic fragments separated from Apollo 17, 78461 regolith sample. The observed variation in (³⁸Ar/³⁶Ar) values can be explained by mixing of non-radiogenic trapped Ar implanted via the solar wind (vertical dashed line ³⁸Ar/³⁶Ar = 0.188 (Eberhardt et al. 1972)) and cosmogenic (³⁸Ar/³⁶Ar = 1.54 (Hohenberg et al. 1978)) ³⁸Ar in lunar plagioclase-rich lithic fragments. m = measured isotopic composition corrected for blank contribution and mass discrimination. The numbers beside the symbols represent the sample names (e.g., 10 represents sample Plag#10)

(⁴⁰Ar/³⁶Ar)_m vs. (³⁹Ar/³⁶Ar)_m correlation

The (⁴⁰Ar/³⁶Ar)_m and (³⁹Ar/³⁶Ar)_m ratios were determined by correcting the measured abundances of ³⁶Ar, ⁴⁰Ar, and ³⁹Ar for blank contributions. In addition, the abundance of ³⁶Ar was corrected for mass discrimination, and the abundance of ³⁹Ar was corrected for radioactive decay since the time of sample irradiation. The (⁴⁰Ar/³⁶Ar)_m and (³⁹Ar/³⁶Ar)_m for eleven plagioclase-rich lithic fragments yield a strong linear correlation (r ² = 0.99) (Fig. 3). Based on this linear regression, the ⁴⁰Ar/³⁶Ar intercept is 0.146 ± 0.002 and the slope is 335.44 ± 16.77. For the x-y error-weighted linear regression (York 1968), the intercept is 0.050 ± 0.104 and the slope is 336.49 ± 17.78. A linear correlation is generally observed in the case of isochron analysis of step heat experiments on undisturbed samples (e.g., Merrihue and Turner 1966; McDougall and Harrison 1999). A similar correlation is also observed for grain size fractions separated from regolith (e.g., Bogard and Nyquist 1975; Eberhardt et al. 1976), where the y-intercept represents the non-radiogenic composition of trapped (⁴⁰Ar/³⁶Ar)_t and slope is used to determine the isochron age. Here, we discuss whether the apparent linear correlation observed in the case of the plagioclase-rich lithic fragments analyzed can be used to argue for the lithic fragments having (a) the same ⁴⁰Ar/³⁹Ar age and (b) the same (⁴⁰Ar/³⁶Ar)_t (i.e., implanted solar wind).

Fig. 3

The correlation between ⁴⁰Ar/³⁶Ar and ³⁹Ar/³⁶Ar. Measured ⁴⁰Ar/³⁶Ar and ³⁹Ar/³⁶Ar for eleven plagioclase-rich lithic fragments separated from 78461 trench regolith sample. Plagioclase-rich lithic fragments show linear fit with a slope of ~335. The ⁴⁰Ar/³⁹Ar age estimated based on ⁴⁰Ar/³⁹Ar = 335.44 is 4.33 ± 0.09 Ga. The dashed lines show 2σ deviations for the linear regression. (⁴⁰Ar/³⁶Ar)_m and (³⁹Ar/³⁶Ar)_m ratio were determined after correcting measured abundances of ³⁶Ar, ⁴⁰Ar, and ³⁹Ar for blank contribution. In addition, the abundance of ³⁶Ar was corrected for mass discrimination, and the abundance of ³⁹Ar was corrected for radioactive decay since the time of sample irradiation. The numbers beside the symbols represent the sample names (e.g., 10 represents sample Plag#10)

⁴⁰Ar/³⁹Ar ages

⁴⁰Ar/³⁹Ar ages are calculated for the plagioclase-rich lithic fragments using three different approaches as described below. For all age determinations, a half-life of 1.250 ± 0.002 × 10⁹ for ⁴⁰K is used (Steiger and Jäger 1977; Renne et al. 2010).

Assuming the same trapped (⁴⁰Ar/³⁶Ar)_t composition

In this first approach, we assume that all the plagioclase-rich lithic fragments contain the same trapped (⁴⁰Ar/³⁶Ar)_t = 0.146 (i.e., the y-intercept value on the (⁴⁰Ar/³⁶Ar) _m versus (³⁹Ar/³⁶Ar)_m regression (Fig. 3)). We prefer to use a slope obtained through normal regression as the correlation is observed for individual samples and the correlation between errors of ³⁹Ar/³⁶Ar (x-axis) and ⁴⁰Ar/³⁶Ar (y-axis) is not expected unlike the case of an age isochron where x-y error-weighted linear regression is required. The ⁴⁰Ar/³⁹Ar ages determined by this approach range from 3.96 to 5.03 Ga. (Table 2, Fig. 4).

Fig. 4

⁴⁰Ar/³⁹Ar ages of plagioclase-rich lithic fragments. ⁴⁰Ar/³⁹Ar ages of individual plagioclase-rich lithic fragments. The filled symbols represents the ⁴⁰Ar/³⁹Ar ages calculated after correcting measured ⁴⁰Ar for non-radiogenic trapped component using (⁴⁰Ar/³⁶Ar)_t = 0.146 (i.e., the y-intercept, Fig. 3). The open symbols represents ⁴⁰Ar/³⁹Ar ages calculated without any correction in measured ⁴⁰Ar for trapped contribution (e.g., Norman et al. 2006; Fernandes et al. 2013). The horizontal line represents the age of 4.13 ± 0.09 Ga, determined using the slope obtained in Fig. 3. The uncertainty shown for ⁴⁰Ar/³⁹Ar ages are 1σ

Assuming the absence of trapped ⁴⁰Ar

For this second approach, we assume that the plagioclase-rich lithic fragments are devoid of any trapped ⁴⁰Ar (i.e., (⁴⁰Ar/³⁶Ar)_t = 0). In this case, it is assumed that all the measured ⁴⁰Ar is radiogenic ⁴⁰Ar (e.g., Norman et al. 2006; Fernandes et al. 2013). The calculated ⁴⁰Ar/³⁹Ar ages determined in this way range from 4.13 to 5.26 Ga (Table 2, Fig. 4).

⁴⁰Ar/³⁹Ar age based on (⁴⁰Ar/³⁶Ar) _m vs. (³⁹Ar/³⁶Ar)_m correlation

A third approach does not assume a trapped (⁴⁰Ar/³⁶Ar)_t composition. As discussed in the “(⁴⁰Ar/³⁶Ar) _m vs. (³⁹Ar/³⁶Ar)_m correlation” section, we obtained a ⁴⁰Ar/³⁹Ar age of 4.13 ± 0.09 Ga for all the lithic fragments based on the slope of the linear regression (i.e., ⁴⁰Ar*/³⁹Ar_K = 335.44 ± 16.77) observed on the correlation plot (Fig. 3). If we use the slope of linear regression (based on the York method), the ⁴⁰Ar/³⁹Ar age =4.135 ± 0.086 Ga, similar to the age derived using a slope obtained through linear regression. In Fig. 4, this age is compared with individual ⁴⁰Ar/³⁹Ar ages determined using the approaches outlined in the “Assuming the same trapped (⁴⁰Ar/³⁶Ar)_t composition” and “Assuming the absence of trapped ⁴⁰Ar” sections above. The comparison of ⁴⁰Ar/³⁹Ar ages derived from three different approaches clearly indicate the sensitivity of age determination to the choice of trapped (⁴⁰Ar/³⁶Ar)_t.

CRE ages

On the lunar surface, ³⁸Ar is produced by solar as well as galactic cosmic rays. With the knowledge of ³⁸Ar production rate, it is possible to estimate the duration of cosmic ray exposure for individual plagioclase-rich lithic fragments. However, the production rate depends on the shielding depths. Hohenberg et al. (1978) calculated the production rates of ³⁸Ar on the lunar surface, as well as various shielding depths. These production rates are for 2π exposure geometry and give a production rate of ³⁸Ar, produced predominantly from the target element Ca, with smaller amounts contributed from Fe, Ti, and K. Assuming that the plagioclase-rich lithic fragments received the cosmic ray exposure on the lunar surface, we calculated production rates for ³⁸Ar using the measured major elemental compositions following surface production rates given by Hohenberg et al. (1978). These production rates are the maximum production rate estimates and range from 0.73 to 0.79 × 10⁻⁸ cm³ STP g⁻¹ Ma⁻¹. However, the depths from which the plagioclase-rich lithic fragments were sampled is not known precisely. As noted previously, the samples analyzed were picked from ~410 g of regolith sampled from 1 to 6 cm depth. We calculated production rates using average production rates of 2, 5, and 10 g/cm² corresponding to 1.3, 3.3, and 6.6 cm depths, respectively (Hohenberg et al. 1978). The production rate for surface (P₃₈ ^sf) and average production rate for 1 to 6 cm depth (P₃₈ ^avg) along with respective ³⁸Ar cosmic ray exposure ages (T₃₈ ^sf and T₃₈ ^avg) are listed in Table 2 and plotted in Fig. 5. Because the abundance of the major target element (i.e., Ca) is similar for all the plagioclase-rich lithic fragments (Table 1), the range of the production rates is narrow as listed in Table 2. The surface exposure ages (T₃₈ ^sf) for plagioclase-rich lithic fragments vary from ~1 to 24 Ma representing the minimum exposure duration for the fragments. On the other hand, the exposure ages based on the average production rate for 1 to 6 cm depth (T₃₈ ^avg) range from ~5 to 110 Ma.

Fig. 5

The cosmic ray exposure ages along with ³⁸Ar production rates for plagioclase-rich lithic fragments. Cosmic ray exposure (CRE) ages (left y-axis) and ³⁸Ar production rates (right y-axis) for individual plagioclase-rich lithic fragments from 78461 regolith sample. The CRE ages and production rates are determined assuming that fragments were exposed either on lunar surface (i.e., T₃₈ ^sf, P₃₈ ^sf, a) or at depth between surface to 10 g/cm² (i.e., T₃₈ ^avg, P₃₈ ^avg, b). P₃₈ ^avg is the average of production rates at 2, 5, and 10 g/cm². The production rates are calculated using the major elemental compositions following production rates given by Hohenberg et al. (1978). Since plagioclase-rich lithic fragments have similar chemical compositions (as shown in Table 1), the production rates are also similar for the fragments. The variation in cosmic ray exposure ages is interpreted to result from the variation in abundances of cosmogenic ³⁸Ar. Dashed line indicates the average of ³⁸Ar production rates

Interpretation of ⁴⁰Ar/³⁹Ar ages

Although the small sample weights (0.06–1.20 mg) precluded step heat experiments from being performed on the plagioclase-rich lithic fragments, the laser total fusion ages can be viewed as equivalent to total integrated ages based on step heat results (e.g., Lister and Baldwin 1996) assuming no loss or implantation of ⁴⁰Ar. The relatively tight clustering of inferred total fusion ⁴⁰Ar/³⁹Ar ages (i.e., 3.96 to 4.28 Ga) for eleven plagioclase-rich lithic fragments as well as older age range (~4 Ga) for the fragments indicate a modest disturbance among the entire set of samples with negligible loss of radiogenic ⁴⁰Ar. However, low temperature diffusive loss of Ar cannot be ruled out completely. Low temperature loss in the lunar samples have apparently occurred at a time closer to the formation ages. For example, Shuster et al. (2010) reported concordance in the initial step ages (ranging from 3.3–3.4 Ga) for 63503 samples which have different ⁴⁰Ar/³⁹Ar plateau ages and argued that the last significant loss of ⁴⁰Ar* occurred 3.3–3.4 Ga ago. It should be noted that the initial step age provides an estimate for the timing associated with the last significant ⁴⁰Ar* loss from a sample. Boehnke et al. (2015) also observed a broad peak at ~3.7 Ga for the Apollo 16 impact-melt samples based on initial minimum ages. This observation is inconsistent with the proposed late heavy bombardment (LHB) type event at ~3.85 Ga (e.g., Tera et al. 1974). Based on these observations, we argue that the lunar regolith most likely experienced impact events that peaked at ~3.7 Ga. Also, the loss of radiogenic ⁴⁰Ar at a younger time would result in greater difference between the total fusion age and the plateau age derived by step heat experiments. For example, the ⁴⁰Ar/³⁹Ar data reveal partial resetting resulting in ~29 % of ⁴⁰Ar loss at ~2.3 Ga for 77,017 samples (Hudgins et al. 2008). The total fusion age for 77017 is 3.741 ± 0.083 Ga while the plateau age is 4.018 ± 0.04 Ga, respectively. Similarly, the loss of around ~49 % ⁴⁰Ar at 3.2 Ga was estimated for sample 60035 (Hudgins et al. 2008) while the total fusion age is 4.088 ± 0.100 Ga. We infer that the difference between total fusion age and the plateau age is proportional to the (1) fraction of ⁴⁰Ar* loss and (2) time difference between crystallization of sample and time of partial resetting. In the present study, we find total fusion ⁴⁰Ar/³⁹Ar ages from 3.96 to 4.28 Ga for eleven plagioclase-rich lithic fragments. Despite a tight clustering and older ⁴⁰Ar/³⁹Ar ages, we cannot rule out the possibility of partial resetting of ⁴⁰Ar/³⁹Ar ages in the plagioclase lithic fragments from regolith.

The two approaches used to determine the ⁴⁰Ar/³⁹Ar ages (sections “Assuming the same trapped (⁴⁰Ar/³⁶Ar)_t composition” and “Assuming the absence of trapped ⁴⁰Ar”) highlight the fact that the trapped non-radiogenic ⁴⁰Ar/³⁶Ar chosen for age calculation impacts the calculated ⁴⁰Ar/³⁹Ar ages (Fig. 4) of plagioclase-rich lithic fragments (c.f., Norman et al. (2003)). For example, Plag#14 has the lowest ⁴⁰Ar/³⁹Ar age (~3.96 Ga, Fig. 4), if measured ⁴⁰Ar is corrected using the trapped contribution ((⁴⁰Ar/³⁶Ar)_t = 0.146), obtained from the y-intercept on the (⁴⁰Ar/³⁶Ar)_m vs. (³⁹Ar/³⁶Ar)_m diagram. In comparison, the age of Plag#14 is similar to that of other fragments (Fig. 4) when we assume (⁴⁰Ar/³⁶Ar)_t = 0. On the contrary, the ⁴⁰Ar/³⁹Ar age for Plag#10 is ~5.0 Ga (assuming (⁴⁰Ar/³⁶Ar)_t = 0.146) and ~5.3 Ga (assuming (⁴⁰Ar/³⁶Ar)_t =0), and both approaches yielded geologically meaningless ages.

Both these plagioclase-rich lithic fragments (i.e., Plag#10 and Plag#14) have the lowest (⁴⁰Ar/³⁶Ar)_m and (³⁹Ar/³⁶Ar)_m (Fig. 3). Assuming non-radiogenic trapped (⁴⁰Ar/³⁶Ar)_t = 0.143, the percentage of radiogenic ⁴⁰Ar (⁴⁰Ar*) released from Plag#10 and Plag#14 is approximately 87 and 76 %, respectively, while the rest of the samples yielded >94 % ⁴⁰Ar* (Table 2). Plag#10 and Plag#14 have higher ³⁶Ar_m compared to that of other fragments; however, the ⁴⁰Ar/³⁹Ar ages do not appear to be a function of the abundance of measured ³⁶Ar (Fig. 6). As we have assumed the same value of non-radiogenic trapped ⁴⁰Ar/³⁶Ar (i.e., 0.143), the correction for trapped contribution in measured ⁴⁰Ar is higher for Plag#10 and Plag#14, owing to their higher abundance of ³⁶Ar. Overcorrection for (⁴⁰Ar)_m could explain the lowest ⁴⁰Ar/³⁹Ar age for Plag#14 (i.e., 3.96 Ga). It is possible that Plag#14 fragment may have a lower or no trapped (⁴⁰Ar/³⁶Ar)_t component. On the other hand, the highest apparent ⁴⁰Ar/³⁹Ar age for Plag#10 can be due to the higher amount of non-radiogenic trapped ⁴⁰Ar/³⁶Ar. The ⁴⁰Ar^*/³⁹Ar_K ratio for the Plag#10 fragment is the highest (i.e., ~578) when compared to that of other plagioclase-rich lithic fragments including Plag#14 (ranging from 303 to 391) (Table 2). We conclude that Plag#10 sample contains parentless ⁴⁰Ar. This can explain the geologically meaningless apparent age (i.e., an age of a sample cannot be older than the age of solar system). The Plag#10 fragment also has the highest CRE age (i.e., 24 to 110 Ma, depending on the choice of production rate; Table 2), indicating longer residence on the lunar surface. We propose that during residence on the lunar surface, ⁴⁰Ar that was present in the lunar atmosphere was ionized and implanted in the Plag#10 fragment by solar wind, resulting in entrapment of excess or parentless ⁴⁰Ar.

Fig. 6

No correlation exists between ³⁶Ar concentration and ⁴⁰Ar/³⁹Ar ages of plagioclase-rich lithic fragments. a ⁴⁰Ar/³⁹Ar ages of individual plagioclase-rich lithic fragments compared with the abundance of ³⁶Ar. b Comparison excluding the Plag#10 fragment. A poor correlation is observed between the ages and ³⁶Ar abundances of the fragments. The ⁴⁰Ar/³⁹Ar ages are determined after correcting measured ⁴⁰Ar for non-radiogenic trapped contribution. The numbers beside the symbols represent the sample names (e.g., 10 represents sample Plag#10)

Results yielded total fusion ⁴⁰Ar/³⁹Ar ages that are interpreted as minimum crystallization ages of the plagioclase-rich lithic fragments. If samples have been partially outgassed due to heating following crystallization (e.g., due to impacts on lunar surface), this would result in loss of radiogenic ⁴⁰Ar from the lithic fragments, therefore lowering the ⁴⁰Ar/³⁹Ar ages. On the other hand, implantation of unknown amount of parentless ⁴⁰Ar from lunar atmosphere through SW would result in higher ⁴⁰Ar/³⁹Ar ages. Therefore, it is difficult to characterize the amount of loss of ⁴⁰Ar and/or implantation of ⁴⁰Ar in single-millimeter fragments. Although we were unable to resolve possible details of the thermal history following crystallization, results are comparable to an integrated ⁴⁰Ar/³⁹Ar age for step heat experiments, assuming no loss of ⁴⁰Ar or implantation of ⁴⁰Ar, and thus provide minimum ages of crystallization (e.g., Lister and Baldwin 1996). However, it is highly possible that lunar regolith could have experienced complicated history of ⁴⁰Ar degassing and implantation as discussed below.

Discussion

During the early lunar evolution (>4 Ga), a greater amount of ⁴⁰K was available to decay to ⁴⁰Ar in the lunar crust and this radiogenic ⁴⁰Ar escaped to lunar atmosphere (e.g., Eugster et al. 2001). A fraction of the escaped ⁴⁰Ar was ionized and accelerated by solar wind. Eventually, a small fraction of ⁴⁰Ar (5–8.5 %; Manka and Michel 1971) was implanted by the solar wind into the lunar surface. The ⁴⁰Ar/³⁶Ar ratio of trapped Ar in lunar regolith has decreased throughout lunar history, roughly in phase with the decreasing ⁴⁰K; however, the initial ⁴⁰Ar/³⁶Ar composition of argon trapped in the regolith is uncertain and dependent on the ambient environment at the time of crystallization. This is especially relevant if impact ejecta are the source of regolith. On the other hand, the lunar regolith experiences gardening processes and impact events by micrometeorites that may have a variable effect dependent on regolith size. For example, grains that spent more time on the lunar surface are susceptible to impacts by micrometeorites unlike grains at a few centimeters depth. The gardening process can randomly excavate small grains multiple times. These processes affect the lunar regolith and therefore the trapped ⁴⁰Ar/³⁶Ar composition of regolith material.

What is the trapped non-radiogenic ⁴⁰Ar/³⁶Ar composition in lunar regolith?

The selection of trapped non-radiogenic ⁴⁰Ar/³⁶Ar is critical for the determination of ⁴⁰Ar/³⁹Ar ages. However, in the case of total fusion experiments and step heat experiments where isochron plot does not yield a meaningful y-intercept (i.e., 67215 clast; Norman et al. 2003), it is difficult to determine the trapped non-radiogenic ⁴⁰Ar/³⁶Ar composition. For lunar regolith, the trapped ⁴⁰Ar/³⁶Ar composition has a range of values. For example, orange glass from 74001/2 core drilled near the Shorty crater at the Apollo 17 landing site has ⁴⁰Ar/³⁶Ar composition ranging from 8 to 10 (Eugster et al. 1980). Two Apollo 14 breccia samples (14301 and 14318) have trapped ⁴⁰Ar/³⁶Ar composition ranging between 10 and 14 (Megrue 1973; Reynolds et al. 1974; Bernatowicz et al. 1980). ⁴⁰Ar/³⁶Ar values for breccia samples from Apollo 16 regolith lie in the range of <1 to 12 (McKay et al. 1986). In the present study, the measured ⁴⁰Ar/³⁶Ar ratio in plagioclase-rich lithic fragments ranges from 0.62 to 15.5, which is similar to the ratios reported.

Comparison of results with previously reported isochron ages

The ⁴⁰Ar/³⁹Ar total fusion ages determined for the lithic fragments are the first ages ever determined for the trench soil collected at station 8 using any isotopic study. However, an approximately 0.5-m-diameter boulder also collected at station 8 (~50 m from the trench, Fig. 1), at the base of the Sculptured Hills, has been extensively studied. Jackson et al. (1975) concluded that the coarse-grained, cumulus-textured boulder formed at a depth of 8–30 km in the lunar crust and was excavated by a large basin-forming impact event. The boulder originated via impact generated ballistic transport to the present site and cannot be related to the geology of the Apollo 17 landing area or the Sculptured Hills. Samples 78235, 78236, 78238, 78255, and 78256 are derived from the same boulder. Figure 7 gives the compilation of ages derived by various isotopic dating methods. The ages range from 4.00 ± 0.09 Ga (metamorphic age for 78238 using Rb-Sr study (Edmunson et al. 2009) to ~4.43 ± 0.05 Ga (crystallization age based on Sm-Nd for most retentive phases (Nyquist et al. 1981)). The Rb-Sr metamorphic age is concordant with an ⁴⁰Ar/³⁹Ar age of 4.11 ± 0.02 for 78235/78236 (Aeschlimann et al. 1982). As can be seen in Fig. 7, the ⁴⁰Ar/³⁹Ar total fusion ages of plagioclase-rich lithic fragments (except Plag#10) fall within the age range observed for the station 8 boulder 78238. However, it should be noted that there is no apparent genetic relation between the lithic fragments and the boulder based on the origin of the boulder 78238 inferred by Jackson et al. (1975).

Fig. 7

Comparison of ⁴⁰Ar/³⁹Ar ages of plagioclase-rich lithic fragments with other studies of relevant lunar rocks. Ages for rocks from station 8, Apollo 17, and Mg-suite highland rocks compared with the plagioclase-rich lithic fragments. The numbers beside the symbols represent the sample names (e.g., 14 represents sample Plag#14). Data from: 1 Edmunson et al. (2009); 2 Aeschilimann et al. (1982); 3 Nyquist et al. (1981); 4 Fernandes et al. (2013); 5 Carlson and Lugmair (1981a); 6 Carlson and Lugmair (1981b); 7 Shih et al. (1993)

The crystallization ages of Mg-suite rock samples are also compared with the ⁴⁰Ar/³⁹Ar ages of the plagioclase-rich lithic fragments (Fig. 7). Sample 78155, also collected at station 8 (~20 m from the trench at the rim of Bowen-Apollo crater, Fig. 1), a thermally annealed polymict breccia of anorthositic norite composition (Bickel 1977), has a ⁴⁰Ar/³⁹Ar crystallization age of 4.19 ± 0.04 Ga (Fernandes et al. 2013), within the range observed for the plagioclase-rich lithic fragments reported here (Fig. 7). Delano and Bence (1977) reported two ~4-mm plagioclase-rich lithic fragments belonging to the Mg-suite that had been collected ~5 km from station 8 (station 3; regolith 73260) at the Apollo 17 site. These two lithic fragments, both belonging to the Mg-suite, had ⁴⁰Ar/³⁹Ar ages of ~4.2 Ga obtained by stepwise heating. A comparison of ⁴⁰Ar/³⁹Ar ages of plagioclase-rich lithic fragments with the crystallization age ranges observed for other lithologic components reveal that the lithic fragments overlap the age range observed for polymict breccia (anorthositic norite and noritic) and Mg-suite rocks, suggesting these lithologic components as possible origins for the lithic fragments as discussed in “Origin of the plagioclase-rich fragments” section below.

Implications for lunar surface processes based on CRE ages

The CRE ages of the lithic fragments represent integrated durations of exposure on the surface of the Moon and reveal an exposure and a burial history of the regolith samples. The variation in CRE ages (from ~1 to 24 Ma (based on ³⁸Ar production rate for surface exposure) or ~5 to 110 Ma (based on average ³⁸Ar production rate for 1 to 6 cm depth on lunar surface) indicates that individual plagioclase-rich lithic fragments have not remained stationary and instead have experienced movement within the lunar regolith for durations of 1 Ma. On the Moon, regolith from a depth of 9 mm is thought to overturn at least once in approximately 10 Ma (Hörz et al. 1991). The estimated CRE range, determined for the plagioclase-rich lithic fragments is consistent with previous suggested regolith overturn rates. The CRE ages for rock samples also collected at station 8 sample range from 17 to 292 Ma (Drozd et al. 1977; Hudgins et al. 2008; Fernandes et al. 2013). The CRE ages observed for some plagioclase-rich lithic fragments are similar to previously reported CRE age of ~17 to 21 Ma for the 78155 samples (Hudgins et al. 2008; Fernandes et al. 2013). Trench regolith from station 8 contained mature samples based on carbon content and maturity index (McKay et al. 1974). The deepest sample from trench (78421) is a mature regolith containing a high abundance of agglutinates (68 %). Based on the low albedo data (Apollo Lunar Geology Investigation Team 1973) and inferred regolith maturity on the basis of fossil track data (Goswami and Lal 1974), it was argued that a large fraction of the regolith at the Apollo 17 site was exposed and reworked at the lunar surface for significant periods of time. Since the plagioclase-rich lithic fragments were derived from the regolith at the base of Sculptured Hills, it is possible that the regolith at station 8 may have been affected by downslope movement of materials from higher elevations. The observed range for CRE ages, either ~1 to 24 Ma or ~5 to 110 Ma, for eleven plagioclase-rich lithic fragments suggests that there is a heterogeneous distribution of submillimeter-sized fragments with a range of cosmic ray exposure ages in the top layer of lunar regolith at station 8.

Origin of the plagioclase-rich fragments

The lunar regolith generally spans a large range of chemical compositions and the variations result from different proportions of major lithologic components that originate from mare basalt and highland rocks. Korotev and Kremster (1992) presented a mass-balance model to account for the compositional variation observed in the regolith samples collected during the Apollo 17 mission. Korotev and Kremster (1992) concluded that various proportions of three chemical components of mare origin (high-Ti and very-low-Ti basalts, pyrocloastic orange glass) and three components of highland origin (impact-melt noritic breccia, anorthositic norite breccia, Mg-suite) can account for most of the compositional variation observed for Apollo 17 regolith. A mixing model, based on major and trace-element compositions, reveals that the trench regolith sample collected at station 8 contains 25.8 % of HT basalt, 2.2 % of VLT basalt, 12 % of orange glass, 15.7 % of impact-melt NB, 35.7 % of AN breccia, 4.62 % of troctolite Mg-suite, and 3.1 % of norite Mg-suite (Table 3, Korotev and Kremser 1992). With the exception of orange glass, these components contain different proportions of plagioclase grains (Table 3). We use mole percent anorthite (i.e., CaAl₂Si₂O₈) and Fe/(Fe + Mg) in plagioclase-rich lithic fragments to further constrain the origin of the lithic fragments. In Fig. 8, we compare An(%) and Fe/(Fe + Mg) values of plagioclase-rich lithic fragments (Table 1) with that of possible protoliths listed in Table 3. Based on this comparison, we conclude that none of the plagioclase-rich lithic fragments are inferred to be derived from HT mare basalt (Fig. 8a). This comparison also suggests multiple possible origins for a given plagioclase-rich lithic fragments due to overlapping values of An(%) and Fe/(Fe + Mg) for the lithologic components. We observe that the An(%) and Fe/(Fe + Mg) values of the Plag#1 lithic fragment do not overlap with that of any lithologic component. However, impact-melt NB could possibly contribute Plag#1 fragment. At least four lithic fragments may have originated from very-low-Ti mare basalt, while most of the plagioclase-rich lithic fragments may have originated from the highland region. We further constrain the lithologic origin of the plagioclase-rich lithic fragments based on the range of crystallization ages reported for different lithologic components (Table 3). The crystallization ages for highland polymict rocks (i.e., AN and NB) range from 3.75 ± 0.11 Ga to 4.16 ± 0.04 Ga (Stöffler et al. 2006). These overlaps with the younger age range are obtained for the plagioclase-rich lithic fragments (Fig. 7). Based on crystallization ages, it can also be argued that no lithic fragment originated from the VLT mare basalt, and most of the fragments may have been derived from the Mg-suite rocks. Simon et al. (1981) observed that mare volcanics and highland-derived fragments were present in similar proportions in Apollo 17 regolith sample 78221, the deepest part of the trench regolith collected at station 8. Similarly, mixing models based on major and trace-element composition indicate that the top layer of trench regolith at the station 8 (i.e., 78481) is composed of 40–50 % of mare and 52–59 % of highlands (Rhodes et al. 1974; Korotev and Kremser 1992). A recent study suggests that pure anorthosite is widely distributed over the entire Moon (Nagaoka et al. 2014). Contrary to thier suggestion, our results indicate that the fragments collected at the base of the Sculptured Hills are mostly derived from the Mg-suite rocks. However, it should be noted that the plagioclase-rich fragments analyzed in this work were selectively sampled and therefore should not be expected to yield similar results to previous studies based on bulk compositional data (Rhodes et al. 1974; Simon et al. 1981; Korotev and Kremser 1992). Due to the known lithologic complexity of the Sculptured Hills (Robinson and Jolliff 2002), and the plagioclase-rich fragments targeted for analysis, it is not surprising that the eleven plagioclase-rich lithic fragments were strongly biased against mare basaltic fragments.

Table 3 Lithologic components that contributed to the trench regolith sample collected at station 8, Apollo 17

Fig. 8

Comparison of Fe/(Fe + Mg) and An(%) for plagioclase-rich lithic fragments with lunar lithologies. Fe/(Fe + Mg) and An(%) for plagioclase-rich lithic fragments (filled circles) compared with mare basalt and highland rocks. The hatched area in all the panels represent the range of values (Table 3) observed for different lithologic components. a Numbers beside the circles represent sample names (e.g., 1 Plag#1). No lithic fragments overlap the range of values reported for high-Ti basalt. b Plag#24, Plag#14, Plag#22, and Plag#25 have compositions similar to VLT basalt. c Except Plag#1, all the lithic fragments overlap the range of composition observed for AN component. d Plag#24, Plag#14, Plag#3, Plag#9, and Plag#8 have compositions similar to that of NB component. e The compositions of Plag#23, Plag#22, Plag#25, Plag#10, Plag#3, and Plag#21 are within the range observed for troctolite component of Mg-suite rocks. f Three plagioclase-rich lithic fragments (i.e., Plag#3, Plag#6, and Plag#9) have compositions within the range observed for the norite component of Mg-suite rocks. Source of data for different lithologic compositions is given in Table 3

Conclusions

We present the first ⁴⁰Ar/³⁹Ar ages on plagioclase-rich lithic fragments separated from trench regolith collected at station 8 during the Apollo 17 mission. The apparent ⁴⁰Ar/³⁹Ar ages for plagioclase-rich lithic fragments in general range from 4.0 to 4.4 Ga assuming a trapped ⁴⁰Ar/³⁶Ar of 0.146 (Fig. 3), the y-intercept of the regression between (⁴⁰Ar/³⁶Ar)_m and (³⁹Ar/³⁶Ar)_m. One sample (Plag#10) contains excess ⁴⁰Ar interpreted to result from implantation of parentless ⁴⁰Ar together with solar wind while exposed on the lunar surface. The CRE duration of Plag#10 is the longest among all the grains analyzed suggesting that this grain likely spent a longer duration on the lunar surface, most likely before 4 Ga and hence higher ⁴⁰Ar was likely implanted with the solar wind. The variation in CRE ages (from ~1 to 24 Ma) in plagioclase-rich lithic fragments indicates that individual fragments have experienced recent movement within the lunar regolith during this time likely due to downslope movement at the base of the Sculptured Hills. Based on the chemical mixing model Korotev and Kremser (1992), chemical compositions, and ⁴⁰Ar/³⁹Ar ages, we argue that most of the plagioclase-rich lithic fragments analyzed in this study originated from Mg-suite of highland rocks. No plagioclase-rich fragments were derived from the mare region. The chemical and argon isotopic records preserved in Apollo 17 station 8 plagioclase-rich lithic fragments illustrate the wealth of information obtainable from submillimeter-sized samples returned from planetary objects including the Moon, Mars, and asteroids.