SHRIMP zircon age constraints on Mesoarchean crustal development in the Vredefort dome, central Kaapvaal Craton, South Africa
Published:January 01, 2006
- PDF LinkChapter PDF
Richard A. Armstrong, Cristiano Lana, Wolf Uwe Reimold, Roger L. Gibson, 2006. "SHRIMP zircon age constraints on Mesoarchean crustal development in the Vredefort dome, central Kaapvaal Craton, South Africa", Processes on the Early Earth, Wolf Uwe Reimold, Roger L. Gibson
Download citation file:
SHRIMP (sensitive high-resolution ion microprobe) U-Pb zircon dating of the Archean basement rocks in the core of the Vredefort dome, South Africa, is integrated with results of lithological and structural mapping and geochemical data to assess the tectonic evolution of the central parts of the Kaapvaal Craton. Recent tectonic models have envisaged the rocks in the core of the dome as part of a much larger, >3200 Ma, Archean cratonic shield (Kaapvaal Shield) that attained stability during partial melting of its early crust and emplacement of voluminous, Mesoarchean, crustally derived granitoid bodies between 3200 and 3100 Ma. Our studies have indicated, however, that the oldest sialic rocks (tonalite-trondhjemite-granodiorite—TTG) in the core of the dome were only formed between 3120 and 3100 Ma, probably in an island arc setting, and that they were accreted onto the craton shortly thereafter. Tectonic accretion of the TTG rocks led to crustal thickening, partial melting, and emplacement of 3100–3080 Ma granites and granodiorites at midcrustal levels under high-grade metamorphic conditions, slightly later than the main pulse of granitoid magmatism in the Kaapvaal Shield. This event was followed almost immediately by collapse of the thickened crust, leading to rapid exhumation of the midcrustal rocks prior to the deposition of rift-related volcanic and sedimentary supracrustal strata at ca. 3074 Ma. The Mesoarchean tectonomagmatic activity in the core of the dome was terminated with emplacement of 3068 Ma aplite dikes.
While the Kaapvaal Craton of southern Africa (Fig. 1) does not contain the oldest rocks on Earth, it contains one of the most comprehensive Archean geological records from ca. 3.6 Ga onward (Hunter, 1974; de Wit et al., 1992; Kröner and Layer, 1992; Kröner et al., 1996; Poujol et al., 2003). In recent years, the volume of geochronological data for the craton has grown considerably and has allowed further refinement of geodynamic models for its evolution (see reviews in Poujol et al., 2003; Eglington and Armstrong, 2004). In their original synthesis, de Wit et al. (1992) proposed a two-stage evolution for the craton, with the initial cratonic shield (Kaapvaal Shield) having grown through collisional tectonics and vertical magmatic accretion between 3600 and 3200 Ma, after which lateral accretion along the margins of the Shield dominated.
The results of more recent field mapping and geochemical studies and U-Pb single zircon and monazite dating are consistent with the de Wit et al. (1992) model for the eastern parts of the craton (Armstrong et al., 1990; Kamo and Davis, 1994; de Ronde and de Wit, 1994; Lowe, 1994, 1999; Kröner et al., 1996), with the geochronological data having allowed the identification of distinct microcontinental blocks within the Barberton region (Fig. 1) (e.g., Lowe, 1994, 1999). Between 3200 and 3100 Ma, however, local-scale tectonomagmatic processes appear to have been replaced by regional-scale partial melting of the preexisting >3200 Ma tonalite-trondhjemite-granodiorite (TTG) suites and emplacement of high-level granite-granodiorite-monzogranite (GGM) batholiths throughout the Kaapvaal Shield (Robb et al., 1991; de Wit et al., 1992; de Wit, 1998; Moser et al., 2001; Poujol et al., 2003). This regional anatectic/magmatic event, the last major episode of vertical differentiation in the Kaapvaal Shield, has been interpreted by some workers as having been responsible for the establishment of a thick, stable lithosphere that was capable of supporting the large Neoarchean to Paleoproterozoic volcano-sedimentary basins that developed subsequently on the craton (Robb et al., 1991; de Wit et al., 1992; Kamo and Davis, 1994; de Wit, 1998; Moser et al., 2001; Poujol and Anhaeusser, 2001; Poujol et al., 2003). Although the exact cause of the regional GGM event remains elusive, several workers have suggested that it incorporated upper mantle melting and substantial reconfiguration of the subcrustal lithosphere beneath the craton (Moser et al., 2001; Flowers et al., 2003; Prevec et al., 2004).
In the central parts of the Kaapvaal Craton, Hart et al. (1999), Moser et al. (2001), Flowers et al. (2003), and Poujol et al. (2003) have suggested a similar Mesoarchean evolutionary history in the Johannesburg and Vredefort domes to that proposed by de Wit et al. (1992), with a >3.2 Ga TTG–greenstone crust that experienced high-grade metamorphism and GGM magmatism at ca. 3.1 Ga. In this paper, we present new SHRIMP (sensitive high-resolution ion microprobe) zircon U-Pb geochronological results from the Vredefort dome that indicate that emplacement of the oldest sialic rocks (TTG suite) in the core of the dome occurred only after the assembly of the Kaapvaal Shield at 3200 Ma and that the GGM event was also younger than the one that affected the Shield. This necessitates a revised geodynamic model for the evolution of these rocks.
The ∼80-km-wide Vredefort dome, located ∼120 km southwest of Johannesburg, is the eroded remnant of the central uplift of the originally 250–300-km-diameter, 2023 ± 4 Ma (Kamo et al., 1996) Vredefort impact structure (for more details, refer to review by Gibson and Reimold, 2001). The dome consists of an ∼40-km-wide core of Mesoarchean crystalline basement, which is surrounded by a 15–20-km-wide collar of Neoarchean to Paleoproterozoic supracrustal strata (Fig. 2). These supracrustal rocks, which range in age from 3.07 to ca. 2.1 Ga (Armstrong et al., 1991), belong to (from the oldest to the youngest) the Dominion Group and the Witwatersrand, Ventersdorp, and Transvaal supergroups.
The core of the Vredefort dome exposes a complex amphibolite- to granulite-facies migmatite terrane, consisting dominantly of stromatic migmatites and syntectonic granitoids and volumetrically minor mafic granulites, amphibolites, and nonmigmatitic ortho- and paragneisses (Fig. 2) (Stepto, 1979, 1990; Lana et al., 2003a, 2004; Lana, 2004). The migmatites, restitic gneisses, and granitoids record a complex Archean tectonometamorphic history involving four ductile deformation events (D1 to D4) and widespread melting (Lana et al., 2003a, 2003b, 2003c). Published U-Pb single zircon and monazite data from various rock types in the central parts of the core of the dome indicate that the high-grade metamorphism occurred ∼3.1 b.y. ago (Kamo et al., 1996; Hart et al., 1999; Moser et al., 2001; Flowers et al., 2003). Analyses of two multigrain fractions of shocked and highly cracked zircon from a postmetamorphic dolerite dike gave discordant, minimum 207Pb/206Pb ages of ca. 2.57 Ga, data used by Hart et al. (1999) to indicate the minimum age for the Archean metamorphic and deformation events. Recently, Moser et al. (2001) and Flowers et al. (2003) obtained U-Pb single zircon ages between 3.13 and 3.08 Ga for several synmetamorphic dikes of granite, tonalite, and charnockite. The results presented here form part of an integrated field, geochemical, and geochronological study that has produced the first coherent lithological and structural map of the core of the dome (Lana et al., 2003a, 2004; Fig. 2). They have allowed the identification of four main periods of tectono-magmatic activity in these rocks.
The oldest event recorded in the dome involved submarine to subaerial ultramafic-mafic volcanic activity with eruption of likely >3.4 Ga komatiites, komatiitic basalts, and tholeiites. This was followed by or interspersed with deposition of pelites, graywackes, and banded ironstones (Stepto, 1979, 1990; Hart et al., 1981, 1990; Minnitt et al., 1994; Lana et al., 2003b, Lana, 2004). The komatiitic lavas are characterized by massive and pillowed basalt lava flows displaying vesicles, variolites, and spinifex textures. The lavas alternate with thin layers of chert and banded iron formation (Minnitt et al., 1994; Lana et al., 2003b). Field, petrological, and geochemical data from these sequences indicate that the mafic and ultramafic rocks might have a similar origin to the komatiites and komatiitic basalts in the Johannesburg dome and the Barberton Greenstone belt (Lana et al., 2003b). The best outcrops in the Vredefort dome occur in an inlier in the southeastern core (Fig. 2).
Over much of the core of the dome, the volcano-sedimentary sequence forms volumetrically minor rafts within large, heterogeneous tonalite-trondhjemite-granodiorite intrusive bodies (Stepto, 1990; Flowers et al., 2003; Lana et al., 2004). The earliest recognizable intrusion is a biotite-rich trondhjemitic gneiss (sample ABBG of this study), which is found as xenoliths within the main trondhjemitic gneiss (sample ABG of this study) and which contains evidence of an early S1 fabric that has been largely transposed by D2 (Lana et al., 2004; Lana, 2004). Both xenoliths and host gneisses are found in scattered outcrops in the northeastern sector of the core of the dome (Fig. 2). These rocks were intruded by a porphyritic granodiorite in the western parts of the core of the dome (Fig. 2) (Lana et al., 2004). Emplacement of this granodiorite was accompanied by significant deformation, involving the development of a dominant subhorizontal S2 fabric in the trondhjemitic gneisses. S2 has, however, been rotated to the vertical in the outer parts of the core during the impact-induced doming event (Fig. 2; Lana et al., 2003c).
Emplacement of leucosomes and granitoids with a wide range of compositions occurred during the peak amphibolite- to granulite-facies metamorphic event in the TTG and supracrustal rocks (Lana et al., 2004; Lana, 2004). This was accompanied by the development of northwest-trending peak-metamorphic S3 high-strain zones characterized by both coaxial and non-coaxial strains (Fig. 2) during a northeast-southwest–directed compressional event (Lana et al., 2003a, 2004; Lana, 2004). Field and geochemical evidence (Lana et al., 2004) indicates that the leucosomes that intruded the upper levels of the migmatite (amphibolite-facies zone) are compositionally similar to highly evolved GGM rocks that are found elsewhere in the craton as upper crustal sheeted dike and batholithic granitic bodies (Anhaeusser, 1999; Poujol and Anhaeusser, 2001; Poujol et al., 2003). Leucosomes in the granulite-facies zone range from enderbitic to charnockitic in composition (Lana et al., 2004; Lana, 2004). The charnockites have similar compositions to, and probably share a common source with, the granitic leuco-somes in the amphibolite-facies zone (Lana et al., 2004). The enderbites are relatively more mafic and are characterized by significant amounts of xenoliths of mafic granulite (e.g., remnants of greenstones; Lana et al., 2004; Lana, 2004).
Collapse of the thickened crust along northeast-trending, subhorizontal, extensional D4 mylonitic shear zones followed immediately after the attainment of peak metamorphic conditions (Colliston and Reimold, 1990; Lana et al., 2003b). The collapse led to juxtaposition of the greenschist-facies greenstone sequence in the southeastern sector of the dome with granulite-facies migmatites in the central parts of the dome (Lana et al., 2003b) and may have been accompanied by emplacement of several aplite dikes that intrude into or crosscut minor mylonitic shear zones (Lana et al., 2004; Lana, 2004).
SAMPLING AND METHODOLOGY
Zircons from the nonmigmatitic gneisses, migmatites, and granites exposed in the core of the Vredefort dome were separated for SHRIMP U-Pb dating. The host trondhjemitic gneiss (ABG) and the xenoliths of biotite-rich gneiss (ABBG) were collected in the northeastern sector of the dome (Fig. 2), whereas the syn-D2 porphyritic granodiorite (POR) and trond-hjemitic melanosome (VALG) originate from the western sector of the dome (Fig. 2). Samples of the enderbitic leucosome (LES) and magnetite-rich granodiorite (GRADIS), all representing partial melts formed during D3 (Lana et al., 2003a, 2004), were collected in the amphibolite-granulite facies transition zone, some 5 km northeast of Vredefort town (Fig. 2). One sample of an undeformed aplite dike (APL), which crosscuts S4 and which should, thus, provide a minimum age for the polyphase tectonometamorphic events in the dome, was collected in the Kudu dimension stone quarry ∼3 km north of Parys (Fig. 2).
Mineral separates were prepared from 7 to 10 kg rock samples that were crushed and pulverized at the School of Geosciences, University of the Witwatersrand, Johannesburg. Zircons were separated using a Wilfley table, heavy liquids (bromoform and methylene iodide), and a Frantz isodynamic separator. SHRIMP II and SHRIMP RG U-Th-Pb analyses were performed at the Research School of Earth Sciences at the Australian National University, Canberra. Zircons were handpicked under a binocular microscope and mounted in epoxy, together with the Research School of Earth Sciences zircon standards FC1 and SL13. All grains were then polished to half their thickness to expose internal structures. Preparation for the SHRIMP analyses included extensive use of transmitted and reflected light microphotography, together with SEM (scanning electron microscope) cathodoluminescence imaging, to decipher the growth complexities in the zircon grains and to target the best areas for in situ analysis. For the zircon age calibration, the Pb/U ratios were normalized relative to a value of 0.1859 for the 206Pb*/238U ratio of the FC1 reference zircons, equivalent to an age of 1099 Ma (Paces and Miller, 1993). Uranium and thorium concentrations were determined relative to the SL13 standard. SHRIMP analyses comprised 6 repeated scans through the species 196Zr2O, 204Pb (common Pb), background 206Pb, 207Pb, 208Pb, 238U, 248ThO, and 254UO.
The data were reduced in a manner similar to that described by Williams (1998, and references therein), using the SQUID Excel Macro of Ludwig (2000a). Uncertainties in the Pb/U calibration from the various analytical sessions are included in the computed ages (e.g., the concordia ages) by SQUID but are not included in the individual analyses as reported in the data tables (Tables 1–7; uncertainties in the calibrations are, however, given in the tables). Corrections for common Pb were made using the appropriate model values of Stacey and Kramers (1975).
Uncertainties given for individual analyses (ratios and ages) are at the 1σlevel unless stated otherwise; however, uncertainties in the calculated weighted mean, intercept, or concordia ages are reported at 95% confidence limits. Concordia plots and age calculations were carried out using Isoplot/Ex (Ludwig, 2000b) or SQUID.
A sample of the dominant trondhjemitic gneiss (ABG) in the dome contains light brown, generally subhedral zircons. Many grains are highly fractured, with some radial cracks linked to volume expansion in areas surrounding (metamict) high-U zones (e.g., Lee and Tromp, 1995) and also display sets of parallel, near-planar fractures as a result of the 2.02 Ga impact (see also Kamo et al., 1996; Gibson et al., 1997). Resorption features and reworking of terminations indicate metamorphic modification. The zircons have the oscillatory zoning typical of crystallization in felsic magma. Many grains also preserve core structures (Fig. 3A). These are best identified through cathodoluminescence imaging, but the overall poor (dark) luminescence of these grains makes unraveling the more subtle internal structures difficult and sometimes impossible. For the analysis of these complex zircons from this and other samples, SHRIMP spots were targeted in the cores, the zoned magmatic grains, and, where spatial resolution allowed, in the reworked or metamorphic rims. Least altered areas were targeted, but given the overall poor state of preservation of the zircons from these rocks, this was not always possible.
The SHRIMP U-Pb analyses of zircons from the trondhjemitic gneiss (ABG) are largely discordant, with the data scattering about a discordia trend that appears to be dominated by Mesoproterozoic Pb loss (Fig. 4; Table 1), although the scatter of data does not necessarily indicate a unique trend. Even the more concordant points show a range in apparent 207Pb/206Pb dates (Table 1), which can be interpreted either as real differences in ages (i.e., a heterogeneous population of magmatic and inherited zircons) or the result of early Pb loss. The latter could result from metamorphism shortly after crystallization of the trondhjemite protolith or from the Paleoproterozoic impact event. Given these complexities, the determination of an emplacement age is difficult and subjective. The favored interpretation is to calculate an upper intercept age of 3113.8 ± 5.5 Ma (MSWD [mean square of weighted deviates] = 0.91; probability = 0.50) from the nine most concordant points of the zoned, magmatic population (Fig. 4). Although grain cores were targeted for dating in anticipation that these would be significantly older, this is not necessarily the case, and in most instances these cores are not discernibly older than
the magmatic population (e.g., analyses 6.1, 6.2, and 9.3; Table 1), although discordance and early Pb loss could obscure this relationship. Where possible, analyses of small overgrowths of what appear to be metamorphic zircon or recrystallized magmatic growth zones were analyzed, and although these few analyses do hint at an age only slightly younger than that calculated for the magmatic phase, they are too discordant and scattered to provide any accurate dates (Fig. 4). One analysis (7.1) within error of concordia has a 207Pb/206Pb age of 3093 ± 10 Ma (2σ).
The zircons separated from a biotite-rich gneiss (ABBG) xenolith are brown, clear to translucent, anhedral, highly fractured, and show signs of resorption as well as shock deformation. Concentric, oscillatory, magmatic zoning is generally better preserved in the lower-U cores of individual grains, which grade into more highly altered and higher-U margins (Fig. 3B). The population as a whole is highly discordant and follows much the same pattern as that described for the zircons from the host trondhjemitic gneiss (ABG). Rare small tips or overgrowths of very high-U, low-Th/U zircon (e.g., analyses 13.1 and 15.1; Table 2) suggest growth of a later metamorphic phase, but these analyses are, unfortunately, extremely discordant, and no useful geochronological information could be obtained from them apart from minimum 207Pb/206Pb ages. A minimum age of 2120 ± 24 Ma (1σ) calculated for analysis 13.1 does, however, seem to indicate that this metamorphic growth predated the Vredefort impact event (dated at 2023 ± 4 Ma by Kamo et al., 1996).
The discordant analyses tend to scatter along a splayed discordia generated by the effects of multistage Pb loss, and as a result, calculation of an age or ages is limited to consideration of the few more concordant points from grains 1, 4, and 17 (Table 2). These grains are useful in that the cathodoluminescence images show clear core/rim structures (Figs. 3C and 3D), and multiple analyses within the different domains on two of the grains (an analysis of the rim of grain 17 was discordant) provide both the least discordant data from the whole data set and also suggest a significant age difference between the two generations of zircon. A weighted mean 207Pb/206Pb age of 3113.8 ± 4.8 Ma is calculated from the four least discordant core analyses (1.3, 1.5, 4.2, and 17.1). Similarly, the weighted mean 207Pb/206Pb age of 3077.1 ± 7.3 Ma from three analyses (1.1, 1.4, and 4.1) sited within the dark rims or embayments suggests metamorphic growth at this time. The apparent age of the cores is indistinguishable from that calculated for the host trondhjemite gneiss (ABG), and the dates calculated from the limited data sets for the rims of both samples are also similar. However, the scarcity of concordant or near-concordant analyses does not allow the effects of early Pb loss to be satisfactorily evaluated (Fig. 5), so that these results must nevertheless be considered minimum age estimates.
Syn-D2 Porphyritic Granodiorite
Zircons in the porphyritic granodiorite (POR) are light brown and clear, and although most grains are subhedral with evidence of metamorphic rounding and resorption, some original euhedral forms are preserved. All zircons are fractured, with the most intense fracturing related to shock deformation. Some grains are also traversed by fluid inclusion tracks. Cathodoluminescence imagery shows that all grains have cores of various shapes, sizes, and states of preservation (e.g., Fig. 3E). Cores generally comprise concentric, oscillatory-zoned zircon surrounded by more strongly zoned growth with higher U contents. In some cases the zoning in these central areas is truncated by the later growth, but the growth zones can also appear to be continuous, although with changing compositions. Most grains are mantled or embayed by very thin overgrowths of zircon that are often difficult to discern, even with cathodoluminescence imaging (CL) imaging.
Many of the data points plot on or near the concordia but with some spread between ca. 3130 Ma and ca. 3080 Ma, and this is clearly not a single age population. The data in Figure 6 are plotted according to the zircon characterization determined from the CL images. Analyses of the cores (defined here as the central sector-zoned zircon and the surrounding dark-CL growth) and analyses of the small overgrowths representing the youngest phase of growth are plotted as open and filled ellipses, respectively. The ten more concordant data of these overgrowths combine to give a weighted mean 207Pb/206Pb age of 3105 ± 5 Ma (MSWD = 1.5; probability = 0.14). Although these younger overgrowths are volumetrically relatively small, they are interpreted to give the best estimate of the age of this porphyritic granodiorite. Many of the analyses of cores are concordant and show a small spread in ages about a mean of ca. 3125 Ma (Fig. 6 inset). These inherited cores provide an indication of the age of the preexisting crust in the region.
The zircons from the trondhjemitic melanosome (VALG) are highly variable in shape and internal structure. The absence of shock deformation features in these grains may reflect lower peak shock pressures in rocks further from the center of the dome than the other samples or heterogeneous local shock conditions (Gibson, 2002; Gibson et al., 2002; Gibson and Reimold, 2005). Oscillatory compositional zoning is present in at least parts of many grains, which suggests a magmatic origin. Most grains have substantial cores, many of which are structurally discordant with respect to the overgrowths of zircon (Fig. 3F), and are interpreted as inherited zircons. Overall, the analyses show a great range in discordance, but most data do follow a discordia with an upper intercept date of 3111 ± 14 Ma (Fig. 7) but with some significant scatter (MSWD = 15; probability = 0.00). If the more discordant analyses plus those analyses clearly sited in older, inherited cores are ignored, the data are regressed to give a more well-constrained upper intercept age of 3106.5 ± 8.4 Ma (n = 15, MSWD = 0.88, probability = 0.57; Fig. 7 inset). The calculated lower intercept indicates Mesoproterozoic Pb loss is dominant. The apparently inherited magmatic cores analyzed (8.1, 9.1, 10.1, and 14.1; Fig. 7; Table 4) give ages ranging from not much greater than the host (3107 Ma) to the oldest concordant age of 3154 ± 7 Ma. Two analyses of a rim on a rounded grain (analyses 6.1 and 6.2; Table 4) appear to provide evidence for at least a minor component of zircon growth during the Paleoproterozoic, with an upper intercept age calculated on these two points alone suggesting a date of ca. 2088 Ma. This is somewhat older than the age of the Vredefort impact event, but it is possible that the analyses incorporated a small component of an older core.
Zircons in the enderbitic leucosome (LES) collected some 5 km northeast of Vredefort town (Fig. 2) are brown, generally opaque, anhedral, up to 500 μm in length, strongly resorbed or rounded, and strongly fractured. Internally, they are also extremely complex, with virtually all grains comprising an inherited core and a broad, inconspicuously zoned overgrowth interpreted to represent the magmatic zircon from the enderbitic leucosome (Figs. 3G and 3H). Rare embayments of dark-CL zircon invade some cores and appear to predate the overgrowths (analysis 14.1; Fig. 3H; Table 5). The extremely low Th/U ratio of 0.01 measured for this zircon indicates a metamorphic origin (Williams and Claesson, 1987).
The U-Pb analyses of the cores show variable discordance, and even the concordant analyses show a small spread of apparent 207Pb/206Pb ages (Table 5), possibly a consequence of early Pb loss, but equally plausibly reflecting real heterogeneity in this inherited component (Fig. 8). A weighted mean 207Pb/206Pb age for the most concordant group of 7 core analyses gives 3410 ± 14 Ma (MSWD = 2.2), but with a low probability-of-fit (0.043) reflecting the scatter of data (Fig. 8). Similarly, the data for the overgrowths give a weighted mean 207Pb/206Pb age of 3094.5 ± 9.1 Ma (MSWD = 1.8; probability = 0.087). A single analysis of a dark-CL embayment (analysis 14.1; Fig. 3; Table 5) gives a concordant, intermediate age of 3200.1 ± 5.7 Ma, the significance of which is uncertain (Fig. 8). However, it could record a hitherto unrecognized metamorphic event in the basement. This enderbite contains abundant mafic granulite xenoliths, and it is these rocks that presumably were the source for the 3410 Ma inherited cores.
Syn-D3 Magnetite-Rich Granodiorite
Zircons in this granodiorite (GRADIS) are euhedral to anhedral, with pronounced metamorphic rounding of pyramidal terminations. As with many other samples from the Vredefort dome, the zircons are highly fractured, with some radial cracks linked to volume expansion in areas surrounding (metamict) high-U zones. Many grains show typical (e.g., Kamo et al., 1996; Gibson et al., 1997) near-planar shock fracture patterns. Cathodoluminescence imaging shows almost all zircons comprise a central, relatively strongly zoned “core,” which is mantled by weaker, broad zones of zircon growth. Characterizing the central zones as exotic cores or simply as an early phase of magmatic growth is difficult as in some cases the change from central to outer areas is abrupt and in other instances it is transitional (Fig. 3I). Elevated Th/U ratios in both types would be consistent with a magmatic origin for both structural types, and it is assumed they have a common origin. A number of the zircons have very thin bright-CL rims or spots within the outer parts of the grains that appear to be metamorphic overgrowths plus reworking of the rims. Only one of these bright-CL rims (on Grain #15; see Fig. 3I) was thick enough to access with the SHRIMP, and it has an extremely low U content of ∼18 ppm and a significantly younger apparent age of about ca. 2769 Ma.
Overall, the SHRIMP analyses show a complex Pb loss and age pattern (Fig. 9). Some data cluster along the concordia, giving a concordia age of 3083.7 ± 8.6 Ma from the 14 analyses shown in Figure 9. Many other points are highly discordant and follow a multibranched pattern of Pb loss.
A cautionary note here is that if the analyses are treated separately according to their petrographic characterization as “core” or “overgrowth” (as interpreted from the CL images), two separate 207Pb/206Pb ages can be determined from the concordant analyses. The cores give a 207Pb/206Pb age of 3092 ± 6.2 Ma (n = 7; MSWD = 0.62; probability = 0.72) and the rims an age of 3074 ± 10 Ma (n = 7; MSWD = 1.3; probability = 0.25). It is hardly possible to know whether this data treatment is valid or not, given the apparent complications of early Pb loss and the difficulty in characterizing these structures unambiguously. It is possible that the rims have preferentially suffered the early Pb loss discussed above or that the analyses of the rims included the small bright-CL reworked zones, thus producing a mixed, apparently younger, date. If the latter scenario is correct, then the date of 3092 ± 6.2 Ma cited above would be the most accurate estimate of the time of crystallization of this granodiorite.
The more discordant data trend toward a Mesoproterozoic lower intercept. A number of analyses (11.2, 15.1, and 15.2; Fig. 9; Table 6) plot along a seemingly different, shallow, discordia trend and clearly show the effects of either the 2.02 Ga impact or some intermediate but early metamorphic event (Fig. 9). The latter scenario appears to be more probable given the difference between the observed Pb-loss trend and the theoretical 2.02 Ga discordia shown in Figure 9.
Zircons extracted from aplites, pegmatites, and other highly evolved rocks can have unusual compositions (e.g., extreme U contents and discordance), which makes them difficult to date accurately. Inheritance is also a common problem. The zircons from the undeformed aplite sample (APL) analyzed in this study are anhedral or fragmental, with very rare euhedral forms. Many show dense shock fracture patterns. The dominant zircon type has a platey form and is thus generally fragmented (presumably in the sample crushing) and has some internal zoning (but this is generally weak and not continuous) and high-U margins. A number of large (>250 μm), anhedral platey fragments that are relatively clear and uncomplicated are also present and probably represent broken fragments of the magmatic zircon described above. These unusual grains are interpreted to represent the magmatic population of this aplite. A small number of possibly inherited cores were found.
The U-Pb analyses are generally discordant, and it is not possible to define a single Pb-loss trend (Fig. 10). Apart from analysis 14.2, which was of an inherited core, all analyses were sited within zircons interpreted to represent the magmatic population. The only concordant or near-concordant data came from analyses within two zircons from the clear, anhedral population. A weighted mean 207Pb/206Pb age calculated from three analyses on grains #3 and #4 gives 3068.3 ± 6.3 Ma (MSWD = 0.046, probability = 0.96). Repeat analyses on different areas of grain #4 (analyses 4.1–4.3; Table 7), however, did not reproduce this age, and it is clear that this particular grain must either be zoned in terms of age or, more likely, must have suffered early Pb loss. The single analysis of a core (#14.2) was discordant and gave a minimum 207Pb/206Pb age of 3251 ± 34 Ma (1σ) (Table 7).
If our interpretation of these zircons as magmatic zircons that crystallized in the aplite is correct, then this date provides a firm age constraint on the aplite. This is important as it provides a lower age constraint on the timing of deformation of the basement granitoids, which must have been complete by ca. 3068 Ma.
This study highlights a number of factors that can complicate how U-Pb data are deciphered and dates are determined from metamorphic terranes that have been subjected to multiple igneous and/or metamorphic events. The first is the very complex nature of almost all the zircon populations. Even with detailed microphotography and cathodoluminescence imaging, it was difficult to definitively categorize many of the zircon types (e.g., identify whether apparent cores were older, inherited grains or simply an early stage of crystallization in continuously evolving conditions). Some cores that were clearly inherited (e.g., fragments of grains surrounded by overgrowths of magmatic zircon) gave apparent ages that are sometimes indistinguishable from those of the overgrowths, despite using the small-scale sampling of the SHRIMP technique (at the 20–30 μm scale used in this study).
A second problem is that many of the zircons display sets of parallel, near-planar fractures as a result of the 2.02 Ga impact (Kamo et al., 1996; Gibson et al., 1997). This, as well as the highly altered and/or metamict nature of many of the grains (Reimold et al., 2002), sometimes made selection of sites for analysis difficult but critical, as these altered areas have invariably suffered severe Pb loss. Apart from this more recent Pb loss, the zircons from the basement lithologies in the core of the dome appear to have lost Pb shortly after their crystallization and during one or more Proterozoic events. Results of SHRIMP U-Pb analysis of Archean rocks from the western parts of the craton (e.g., Poujol et al., 2002) and those presented here suggest that early Pb loss seems to be a common feature regionally. The early Pb loss in the basement rocks of the Vredefort dome must have occurred within a few million years of crystallization of the parent rocks and could, plausibly, be related to polyphase intrusions of TTG magma into the crust and to the ensuing high-grade metamorphic conditions as well as to the formation of syntectonic granites and granodiorites emplaced during the metamorphism.
Pb loss during the Proterozoic has previously been related to the ca. 2020 Ma impact event and the ca. 1100–1300 Ma Kibaran/Grenville orogeny (Kamo et al., 1996). The impact event has been dated at 2023 ± 4 Ma based on a concordant age obtained from unshocked zircons from impact-related pseudotachylitic breccia and Vredefort Granophyre (impact melt rock; Kamo et al., 1996). Kamo et al. (1996) observed that the Pb loss seems to be more prominent in populations of highly shocked zircons. Our SHRIMP analyses do not indicate significant Pb loss during the impact event (or shortly thereafter). This may be partly because the samples analyzed here did not contain significant volumes of pseudotachylitic breccia and, therefore, have had less chance of thermally induced Pb loss associated with the impact event, and also because severely shocked zircons were generally avoided where possible. The dominant Pb loss indicated by this study seems to have occurred during the Mesoproterozoic, possibly coincident with the Kibaran-Namaquan orogeny, and also associated with intrusion in the dome of several laterally extensive gabbroic sheets of the same age (Reimold et al., 2000). These sheets also seem to have induced significant hydrothermal alteration and resetting of the Ar isotopic system not only in the core of the Vredefort dome (Reimold et al., 1990; Spray et al., 1995), but also throughout the Witwatersrand basin (e.g., Reimold et al., 1990; Trieloff et al., 1994; Friese et al., 2003). An important observation from this study is that even within a population of zircons from a single sample, there is not necessarily a single Pb-loss pattern, and certainly the Pb-loss pattern for the region is variable. Clearly, an appreciation for the appropriate Pb-loss model is needed for interpretation of intercept ages that are calculated from the Archean basement complex from discordant data and especially from multigrain analyses or small data sets. In the case of the Vredefort dome, several previous studies (e.g., Kamo et al., 1996; Moser, 1997; Hart et al., 1999) have made use of results calculated from upper intercept dates assuming Pb loss to be exclusively associated with the impact event at 2.02 Ma (i.e., the assumed lower intercept age). In some instances this might be appropriate (e.g., for highly shock-metamorphosed zircon), but in other cases this can lead to artificially high apparent upper intercept dates. Significantly, the Archean rocks in the western parts of the craton also record evidence of Pb loss during the Kibaran/Namaquan orogeny (Poujol et al., 2002), although these samples were derived from an area substantially closer to the locus of geological activity at this time along the southern and western edges of the craton.
Timing of Magmatic and Deformation Events in the Core of the Vredefort Dome
Previous models for the evolution of the rocks exposed in the core of the Vredefort dome (e.g., Moser et al., 2001; Flowers et al., 2003) have favored a two-stage process in which an initial, >3.3 Ga, TTG component underwent partial melting between 3.13 and 3.08 Ga to produce a suite of trondhjemitic gneisses and tonalites, granites, and quartz syenites. They have suggested that partial melting of the older crust occurred during a 3.107 Ga (Hart et al., 1999) high-grade metamorphic event and correlated this with the craton-wide magmatic event that led to the formation of the voluminous granodiorite-granite-monzogranite (GGM) suites in the upper crust of the Kaapvaal Shield (see review in Belcher and Kisters, this volume). In these models, little attention has been given to the deformation structures found in the rocks in the dome or their tectonic significance, although large-scale tectonic inferences have been made.
As part of their model, Moser et al. (2001) maintained that at least 40% of the lower crust supposedly exposed in the central core of the Vredefort dome melted to form a 10-km-thick granitic layer that is now exposed in the outer part of the core of the dome. However, new structural and geochemical data (Lana et al., 2003a, 2004) have indicated that the core of the dome, while being compositionally heterogeneous on an outcrop scale due to its migmatitic and polydeformed character, shows no definitive gross kilometer-scale layering consistent with such a model. Lana et al. (2004) also indicated that the section exposed in the dome comprises the same TTG suite, with only metamorphic grade increasing toward the center, and that voluminous granitic/charnockitic rocks intruded throughout the core of the dome during peak-metamorphic conditions. The SHRIMP data presented here provide further refinements of the geological model proposed by Lana et al. (2004), with 3 magmatic suites and associated tectonic features identifiable.
The first suite of sialic rocks is represented by the trondhjemitic gneisses, including the melanosomes in the amphibolite-facies zone (Lana et al., 2004), which intruded the greenstone sequence before or during D1 and the porphyritic granodiorite that was emplaced during the D2 deformation event (Lana et al., 2003a, 2003c, 2004). Both the xenolith ABBG and its host, the trondhjemitic gneiss ABG, gave ages of ca. 3114 Ma. These age data are indistinguishable from the 3106 ± 8 Ma ages for trondhjemitic melanosomes in the migmatites (sample VALG) and support geochemical interpretations that link the melanosomes to the nonmigmatized TTG suite (Lana et al., 2004). The 3105 ± 5 Ma age obtained for zircons extracted from the porphyritic granodiorite (POR) also shows that its emplacement occurred at this time or only slightly later than the trondhjemitic gneisses. Thus, the data suggest multiple intrusions of TTG magma within a very restricted period around 3.11 Ga. The ages of D1 and D2 can also be similarly bracketed as both events occurred during emplacement of the TTG granitoids. An approximate age of 3105 Ma is suggested for D2, given that this event occurred during the emplacement of the porphyritic granitoid.
The second suite of sialic rocks, represented by the evolved granitic (charnockitic) leucosomes and granitic to granodioritic bodies, intruded during a northeast-southwest compressional event under low-P, high-T metamorphic conditions. The age of this tectonomagmatic event is indicated by zircons in the magnetite-rich granodiorite (GRADIS), which yielded ages of ca. 3090 Ma. The enderbitic leucosomes (LES), which host a large volume of mafic granulite xenoliths (Kamo et al., 1996; Lana et al., 2004; Lana 2004), yielded two distinct ages of 3410 ± 14 Ma and 3095 ± 9 Ma. The 3095 ± 9 Ma age obtained from zircon rims may confirm the fact that most of the leuco-somes and syntectonic granitoids crystallized between 3100 and 3080 Ma. This age interval is broadly consistent with the ages of granitic and quartz syenite intrusions in the dome dated by Moser et al. (2001) and Flowers et al. (2003). The SHRIMP results from the GRADIS and LES samples also suggest that peak metamorphic conditions were reached between 3100 and 3080 Ma, which is within error of or slightly younger than the 3107 ± 9 Ma age previously proposed by Hart et al. (1999) based on a single concordant U-Pb monazite age. The ca. 3410 Ma age obtained from zircon cores in the enderbite is broadly consistent with the 3400–3500 Ma Rb-Sr and Th-Pb whole-rock ages previously reported by Hart et al. (1981) and with inherited zircons from elsewhere in the central part of the Kaap-vaal Craton (Armstrong, et al., 1991). These age data from zircon cores confirm previous interpretations that the oldest greenstone rocks in the core of the dome are similar in age to those exposed in the central and eastern parts of the craton (Hart et al., 1981; Hart et al., 1999; Lana et al., 2003b) and that the enderbites were derived by partial melting of greenstone remnants (Lana et al., 2004).
Although the TTG rocks and the syntectonic granitoids and leucosomes were emplaced during distinct deformation events, it is worth noting that the differences in age between these rocks are within uncertainty. The age results and the geochemical data for these lithologies (see Lana et al., 2004, for discussion) do not exclude the possibility that the TTG suite and the more granitic leucosomes and granitoids are part of a single, continuous, magmatic episode in which the leucosomes represent more fractionated melts from a parental trondhjemitic magma (e.g., Condie and Hunter, 1976).
The youngest suite of rocks analyzed is represented by the volumetrically minor aplite dikes. These dikes intruded parallel to subparallel to the D4 mylonitic shear zones and were not deformed by the D4 mylonitic deformation. The 3068 ± 6 Ma age obtained for the aplite sample is identical to the age established for the deposition of the Dominion Group volcano-sedimentary sequence (3074 ± 6 Ma; Armstrong et al., 1991; Robb et al., 1991) that unconformably overlies the crystalline basement rocks in the core of the dome. It has important implications for the tectonic evolution of the central parts of the Kaapvaal Craton (see below).
Existing models for the Archean crustal evolution in the Kaapvaal Craton suggest formation of a >3.2 Ga cratonic nucleus (the Kaapvaal Shield) during repeated magmatic accretion and tectonic amalgamation of small island arc terranes (de Wit et al., 1992; de Ronde and de Wit, 1994; Kamo and Davis, 1994; Lowe, 1994, 1999; Poujol et al., 2003). New geochronological results from the central parts of the craton (Poujol and Anhaeusser, 2001) suggest that the Kaapvaal Shield extended at least as far west as the Johannesburg dome (Fig. 1) where tonalitic and trondhjemitic gneisses older than 3.2 Ga and GGM rocks ranging in age from 3.11 and 3.13 Ga are exposed (Poujol et al., 2003). In contrast, the crystalline rocks in the northern parts of the craton (Murchison, Pietersburg, and Giyani greenstone belts; Fig. 1) have been interpreted as the product of a series of juvenile volcanic arcs that were successively accreted along a convergent plate boundary between 3.1 and 2.8 Ga (Poujol et al., 2003). More recently, 3.3–3.2 Ga inherited zircons and Nd model ages for gneisses in the Kraaipan belt exposed in the western parts of the craton (Fig. 1) have been used by Schmitz et al. (2004) to suggest that the craton comprises two distinct fragments—a younger Kimberley Block in the west and an older Witwatersrand Block in the east—that collided at ca. 2.9 Ga along a north-south–trending suture immediately east of the Kraaipan-Amalia belts (Fig. 1).
Integral to the models of Poujol et al. (2003) and Schmitz et al. (2004) is that the assembly of the Kaapvaal Shield (or Witwatersrand Block of Schmitz et al., 2004), which also included the crust now exposed in the Vredefort dome, occurred prior to 3.2 Ga and that it was subsequently affected by the regional-scale partial melting of the 3.2 Ga old TTG rocks and emplacement of kilometer-scale GGM suite batholiths at >3.1 Ga. This “final granite bloom” (Bleeker, 2002) removed minimum melt fractions, fluids, and heat-producing elements from the lower crust, cooling it and increasing its rigidity and likely enhancing its coupling to the mantle lithosphere. However, while field, geochemical, and geochronological data indicate the presence of oceanic crust as old as ca. 3.3–3.4 Ga in the Vredefort dome (Reimold et al., 1988; Hart et al., 1981, 1990, 1999; Tredoux et al., 1999; Lana et al., 2003b; Lana, 2004), our study has found no direct evidence of significant TTG magmatism prior to 3.15 Ga. This indicates that the rocks exposed in the core of the dome are too young to have been part of the (>3.2 Ga) Kaapvaal Shield. The geochemical composition of the TTG suites in the core of the dome (e.g., Lana et al., 2004; Lana, 2004) is also consistent with them having been formed during partial melting of hydrated oceanic crust (e.g., Skjerlie and Johnson, 1993; Singh and Johannes, 1996) rather than from remelting of older TTG crust.
The TTG rocks in the core of the Vredefort dome could have formed either in an intraoceanic environment (i.e., island arc or thrust stack; de Wit, 1998; Lowe, 1994, 1999), in similar fashion to other greenstone-TTG associations in the central and western parts of the craton (e.g., Anhaeusser, 1971, 1973, 1999; de Wit et al., 1992; de Ronde and de Wit, 1994; Kamo and Davis, 1994; Lowe, 1994, 1999; Poujol et al., 2003), or in a magmatic arc along the western margin of the Kaapvaal Shield. The paucity of >3.15 Ga inherited zircons in the TTG rocks in the dome is consistent with an oceanic arc, with a relatively thin crustal substrate (e.g., Mason and McDonald, 1978; Whalen, 1985) rather than a continent-margin magmatic arc environment, in which a thick underlying crust would have facilitated formation of voluminous granitoid batholiths (Windley, 1973, 1982). Whatever the cause, our data suggest that the idea of a monolithic Kaapvaal Shield (extending from Barberton to west of the Vredefort dome) that was created exclusively before 3.2 Ga needs revision and that addition of juvenile crustal material to its western margin might have occurred regularly between 3.2 Ga and the collision of the Kimberley Block. Any such terranes are now largely hidden beneath thick younger supra-crustal sequences (Fig. 1).
A similar distinction may also be drawn between the formation of the GGM rocks in the dome versus those in the Shield. Although there may be some overlap in the ages, the regional “final granite bloom” (Bleeker, 2002) in the Shield appears to be 10–20 m.y. older than the GGM intrusions in the dome (see reviews in Poujol et al., 2003; Belcher and Kisters, this volume). In terms of tectonic environment, the granitoids in the dome are associated with a strong, high-grade, subvertical northwest-trending S3 fabric that, so far, has not been identified in the Johannesburg dome, but which seems to be parallel to subparallel to the regional foliation in the relatively younger granite-greenstone rocks of the Amalia/Kraaipan terrane in the western part of the craton (Poujol and Anhaeusser, 2001, 2003; Hirner, 2003). In contrast, the 3.1 Ga granite batholiths in the Barberton region appear to have been emplaced within regional northeast-trending transpressional strike-slip zones (Belcher and Kisters, this volume).
The emplacement of the 3100–3080 Ma granitoids and the development of the D3 northwest-trending crustal shear zones in the dome indicate that high-grade metamorphic conditions were achieved during a northeast-southwest compressional event (Stevens et al., 1997; Lana et al., 2003a). The structural features and the anticlockwise P-T path derived from the greenstone fragments during the high-grade metamorphic event (Stevens et al., 1997) indicate that D3 was intimately linked to crustal thickening. Accretion of the Vredefort rocks onto the Kaapvaal Shield could have occurred during this thickening of the crust at 3100–3080 Ma, which is less than 10–25 m.y. prior to the outpouring of the 3074 ± 6 Ma (Armstrong et al., 1991) Dominion Group lavas in the central parts of the craton.
The fact that the orogenic climax between 3100 and 3080 Ma is only slightly older than Dominion Group volcanism implies that exhumation of the midcrustal levels exposed in the core of the dome must have been particularly rapid. The most effective way to achieve this would be by extensional unroofing of the high-grade rocks shortly after the D3 tectonic and magmatic thickening (e.g., Dewey, 1988; Wong and Gans, 2003; Kisters et al., 2003). Accommodation of the deformation related to extensional unroofing might have occurred principally along a >500-m-wide northeast-trending S4 mylonitic shear zone exposed along the northern margin of the greenstone inlier in the southeastern sector of the dome (Fig. 2) and several smaller mylonitic shear zones that occur in the northern sector of the core of the dome (Lana et al., 2003b). Geothermobarometry results from peak metamorphic assemblages suggest at least 5 km of vertical displacement along the southeastern D4 shear zone located between the upper amphibolite-facies gneisses and greenschist-facies greenstones (Lana et al., 2003b). Other evidence linking D4 with extensional collapse of thickened crust is the peak to post-peak metamorphic assemblages in the S4 fabric, which indicate that D4 was initiated under high-grade metamorphic conditions and that shearing progressed under conditions of declining temperature (Lana et al., 2003b). These observations are consistent with the tectonic setting for the deposition of the Dominion Group lavas in an extensional-intracontinental basin (e.g., Robb et al., 1991; Jackson, 1992) that developed in the collapsing hanging wall of the evolving extensional detachment.
Results of U-Pb SHRIMP analysis of Archean crystalline basement rocks exposed in the core of the Vredefort dome suggest a comparatively short-lived, three-stage magmatic and tectonic evolution. Sialic crust formation commenced with multiple intrusions of 3120–3100 Ma TTG rocks during D1 and D2. The chemical compositions of the TTG rocks (Lana et al., 2004) together with the geochronological data suggest that they were probably derived from partial melting of hydrated oceanic crust in an island arc setting. The TTG and greenstone sequences were later partially melted and intruded by 3100–3080 Ma leucosomes and other granitoids during upper amphibolite- to granulite-facies regional metamorphism that coincided with northeast-southwest D3 compression. D3 and the high-grade metamorphic event are related here to tectonic collision of the Vredefort island arc terrane with the >3200 Ma Kaapvaal Shield to the east. The final Mesoarchean structural event recorded in the core of the dome was associated with the collapse of the thickened crust along northeast-trending, normal dip-slip, peak-to-retrograde, subhorizontal mylonitic shear zones (D4). The collapse was probably associated with or was shortly followed by the emplacement of 3068 ± 6 Ma aplite dikes. This extension may be linked to the rifting that accompanied the outpouring of the 3074 ± 6 Ma Dominion Group volcanic sequence that marked the start of the development of a succession of large sedimentary and volcanic basins on the craton over the next thousand million years.
We would like to thank Carl Anhaeusser for fruitful discussions on the geological evolution of the central parts of the Kaapvaal Craton. Reviews of earlier drafts of the manuscript by Sandra Kamo, Stephen Prevec, and Christian Koeberl are gratefully acknowledged. RAA would like to thank the staff of the Electron Microscopy Unit (EMU) at the Australian National University (ANU) for access to their excellent scanning electron microscope (SEM) facilities. Financial support from the Research Council of the University of the Witwatersrand is gratefully acknowledged. University of the Witwatersrand Impact Cratering Research Group Contribution No. 74.
Figures & Tables
Processes on the Early Earth
- absolute age
- crustal thickening
- Free State South Africa
- igneous rocks
- ion probe data
- island arcs
- Kaapvaal Craton
- mass spectra
- metamorphic rocks
- partial melting
- plutonic rocks
- SHRIMP data
- South Africa
- Southern Africa
- Vredefort Dome
- zircon group