1. Introduction
Alkaline and carbonatitic rocks are an important petrological record of anorogenic magmatism within cratons (Ernst & Bell, Reference Ernst and Bell2010), and hence these rocks provide invaluable information for deciphering the geodynamic evolution of these ancient continental blocks (Tappe et al. Reference Tappe, Foley, Stracke, Romer, Kjarsgaard, Heaman and Joyce2007). Taking the example of the North American continent, it is characterized by numerous anorogenic igneous suites subsequent to cratonization at ~1.8 Ga (Windley, Reference Windley1993; Ernst & Bleeker, Reference Ernst and Bleeker2010). Among these, alkaline rock and carbonatite complexes occurring to the north of Lake Superior represent an important stage of Midcontinent rifting during Mesoproterozoic time (Rukhlov & Bell, Reference Rukhlov and Bell2010). The Midcontinent Rift has received much attention as it is one of the best-preserved Proterozoic intra-continental rifts, and potentially holds important clues as to the nature of Precambrian continental rifting (Van Schmus & Hinze, Reference Van Schmus and Hinze1985; Palmer & Davis, Reference Palmer and Davis1987; Heaman et al. Reference Heaman, Easton, Hart, Hollings, Macdonald and Smyk2007; Hollings, Smyk & Cousens, Reference Hollings, Smyk and Cousens2012). Previous investigations indicated that the magmatism is diverse in lithology and composition, and includes the main stage of tholeiitic eruptions, lamprophyre, mafic–ultramafic intrusions (for example, the Duluth and Abitibi dykes), alkaline rock and carbonatite intrusions, rhyolite and related granophyre, etc. (Paces & Miller, Reference Paces and Miller1993; Heaman et al. Reference Heaman, Easton, Hart, Hollings, Macdonald and Smyk2007; Vervoort et al. Reference Vervoort, Wirth, Kennedy, Sandland and Harpp2007; Hollings et al. Reference Hollings, Richardson, Creaser and Franklin2007; Hollings, Smyk & Cousens, Reference Hollings, Smyk and Cousens2012; Bright et al. Reference Bright, Amato, Denyszyn and Ernst2014). Previously, most studies have suggested that a plume (Keweenawan) was responsible for the rifting as a large volume (>106 km3) of magma developed over a short period of ~20 Ma (Burke & Dewey, Reference Burke and Dewey1973; Ernst & Bell, Reference Ernst and Bell2010; Ernst & Bleeker, Reference Ernst and Bleeker2010). According to Heaman et al. (Reference Heaman, Easton, Hart, Hollings, Macdonald and Smyk2007), the magmatism was initiated with lamprophyre and mafic magmatism (1150–1130 Ma), followed by mafic–ultramafic intrusions (1115–1110 Ma), the main rift magmatism (1108–1094 Ma) and synchronous alkaline–tholeiitic magmatism (1110–1100 Ma).
Within the Superior Province of the Canadian Shield, numerous Proterozoic alkaline rock and carbonatite complexes have been identified (Sage, Reference Sage1983, Reference Sage1987). One of these complexes, the Prairie Lake alkaline rock – carbonatite complex is situated northwest of Marathon, Ontario (49°02′N, 86°43′W), approximately 26 km from the shores of Lake Superior (Fig. 1). The complex has a surface area of 8.8 km2 (Fig. 2) and consists predominantly of foidolitic and carbonatitic rocks emplaced in Archaean gneisses of the Superior craton along the Trans-Superior Tectonic Zone (Sage, Reference Sage1987).
Distribution of the alkaline rock and carbonatite complexes within the Canadian part of the Midcontinent Rift. The spatial range of the proposed Keweenawan plume is also shown (modified from Ernst & Bell, Reference Ernst and Bell2010; Rukhlov & Bell, Reference Rukhlov and Bell2010).
Geology of the Prairie Lake complex.
Previous geochronological studies have suggested that the Prairie Lake complex might have been emplaced between 1020 and 1200 Ma (see discussion in Section 5 below). These investigations were not able to determine whether the different types of rocks in the complex were synchronously emplaced or not. In this study, we firstly present U–Pb geochronological data for baddeleyite and other related minerals to constrain accurately and precisely the time of emplacement of this complex, and then use Sr–Nd–Hf isotopic and trace-element data to consider the relationships between, and genesis of, the silicate and carbonatitic rocks of the complex.
2. Geological setting
The Midcontinent Rift is one of the most spectacular geological features of the North American continent, and extends from central Kansas, through Lake Superior and further east to Michigan (Fig. 1). It has been proposed that it represents an aborted or failed rift resulting from an upwelling mantle plume during Mesoproterozoic time at ~1.1 Ga (Burke & Dewey, Reference Burke and Dewey1973). The magmatism associated with the incipient rifting has been termed the Keweenawan Large Igneous Province (Ernst & Bell, Reference Ernst and Bell2010). The rift is characterized by a large volume (~1–2×106 km3) of igneous material (Cannon, Reference Cannon1992; Merino et al. Reference Merino, Keller, Stein and Stein2013), which includes tholeiitic volcanic rocks and related rhyolite and granophyre (Keweenawan Supergroup), mafic–ultramafic sills and intrusions, represented by the layered Duluth Complex in Minnesota (Paces & Miller, Reference Paces and Miller1993), and alkaline rock and carbonatite intrusions in Canada (Fig. 1).
Within the Archaean rocks of the Quetico Subprovince of the Superior Province, numerous alkaline and carbonatitic rocks are distributed along the Trans-Superior Tectonic Zone and Kapuskasing Structural Zone, respectively (Fig. 1). Located within the former zone, four alkaline rock – carbonatite complexes (Coldwell, Chipman Lake, Killala Lake and Prairie Lake) have been recognized (Fig. 1). The Coldwell complex is one of the largest alkaline complexes in the world with an area of ~490 km2, and is composed of subalkaline and alkaline gabbro, nepheline syenite, diverse contaminated syenites, ferroaugite syenite and A-type quartz syenites (Mitchell & Platt, Reference Mitchell and Platt1978; Mitchell, Platt & Cheadle, Reference Mitchell, Platt and Cheadle1983; Mitchell et al. Reference Mitchell, Platt, Lukosius-Sanders, Artist-Downey and Moogk-Pickard1993) with an emplacement age of ~1108 Ma (Heaman & Machado, Reference Heaman and Machado1992). The Killala complex consists of troctolite, gabbro, monzonite, syenite, nepheline monzonite–syenite and mafic–syenitic dykes, but lacks quartz syenites. Neither the Coldwell nor the Killala Lake complexes contain carbonatites. The Chipman Lake complex at the northern end of the Trans-Superior lineament consists of abundant carbonatitic dykes (Platt & Woolley, Reference Platt and Woolley1990).
2.a. Prairie Lake complex: geology
The limited outcrop of this small intrusion coupled with the significant modal heterogeneity of the rocks on a metre (or less) scale makes it difficult to construct a detailed geological outcrop map of the complex. Examination of drill cores also indicates significant complexity as particular petrographically defined units cannot be correlated even between closely sited drill holes. Regardless of these problems it is possible to recognize several broad petrographic units. On the basis of recent exploration activities (R. H. Mitchell, unpub. data; Purtich, Armstrong & Yassa, Reference Purtich, Armstrong and Yassa2010), a schematic map is provided to show the distribution of these units (Fig. 2). They are described below in approximate order of intrusion.
2.a.1. Biotite pyroxenite and calcite carbonatite-(I)
Biotite pyroxenites are ultramafic coarse-grained, very friable rocks occurring principally in the NE sector of the complex (Fig. 2). The biotite pyroxenites are cut by many veins of very coarse-grained calcite carbonatite, here termed carbonatite-I. These veins and schlieren range in size from irregular segregations a few centimetres in length to large veins up to 1 m in width. This unit at its eastern extremity appears to have been intruded by carbonatite-(I) containing clasts of melanocratic ijolite. Carbonatite-(I) is a very friable coarse rock consisting of large interlocking calcite crystals with rare apatite present as aggregates of radial prisms. These rocks contain very little apatite or magnetite.
2.a.2. Ijolite series rocks
Much of the central region of the complex is occupied by diverse ijolite series rocks (Fig. 2). These are modally heterogeneous on a metre scale. Varieties present include: ijolite (pyroxene+nepheline), calcite mela-ijolite (silicocarbonatite or hollaite), mela-ijolite (pyroxene > nepheline with biotite and andradite–schorlomitic garnet) and wollastonite ijolite. Ijolite and mela-ijolite compose the bulk of this series. Several texturally distinct varieties can be recognized, although their relative disposition is unknown. The ijolitic rocks are notable in that they also contain accessory zirconolite, marianoite–wohlerite solid solutions and Zr-bearing titanite. Ijolite breccias are characteristic of the eastern parts of this unit.
A large area of wollastonite ijolite exposed in the southeastern portion of this unit has been shown by recent exploration trenching to be a megaxenolith incorporated in ijolitic breccia. Smaller clasts of wollastonite ijolite can be found in other parts of this breccia. The unit is clearly older than many of the other ijolites.
2.a.3. Potassic syenitic rocks
Mela-syenites, characterized by the presence of Na-poor potassium feldspar, nepheline and pyroxene, occur in the central and southern parts of the complex and are intimately mixed with ijolite-suite rocks (Fig. 2). The poor exposure prevents determination of their disposition but they are considered to be younger than the ijolites. Some of the malignites appear to be modally gradational into ijolite-suite rocks. The presence of a single non-perthitic feldspar plus nepheline indicates that these rocks are malignites (Mitchell & Platt, Reference Mitchell and Platt1979) and not melanocratic nepheline syenites (sensu lato). Characteristic accessory minerals are marianoite–wohlerite and Zr- and Nb-bearing titanite.
2.a.4. Heterogeneous carbonatite-(II)
This extensive complex unit forms a group of rocks extending from the northwest of the complex along the western margin to the SE margin of the complex (Fig. 2). The unit is characterized by great modal and textural diversity. Typically, the rocks are banded (or layered) on a centimetre to millimetre scale. Banding results from wide modal variations in the content of calcite, apatite, mica, olivine and magnetite. Fine-grained rocks are ‘interbedded’ with coarse-grained rocks. Large (up to 10 cm), rounded, coarse-grained apatite–mica-poor calcite carbonatite clasts can be found in the finer grained apatite carbonatites.
Modal layering in many places appears to be parallel to the margins of the complex. However, within any given area, the strike of the banding can vary significantly, and can range from near vertical to horizontal over a short distance. Extrapolation of the strike of layering from one area to another is not possible.
Carbonatite-(II), having more apatite and magnetite relative to carbonatite-I, contains numerous rounded clasts (centimetre to decimetre size) of ultramafic rocks derived from the deeper, and earlier-formed, parts of the complex. These clasts vary widely in mode and texture, and range from coarse-grained ijolite and biotite pyroxenite to micro-mela-ijolite and alnoitic lamprophyre. Some xenoliths of calcite ijolite contain flow-aligned prisms of calcite. The distribution of clasts is erratic, and clasts do not appear to be more abundant adjacent to the margin of the complex. Carbonatite-(II) also contains clasts of ‘phoscorite’ and ‘disaggregated phoscorite’. Phoscorites, an imprecisely defined (Krasnova et al. Reference Krasnova, Petrov, Balaganskaya, Garcia, Moutte, Zaitsev, Wall, Wall and Zaitsev2004) but commonly used term, are considered in this work to be cumulate rocks consisting of magnetite+apatite+olivine and/or diopside with perovskite and/or pyrochlore.
Although there are continuous modal gradations, several varieties of carbonatite can be recognized as composing the heterogeneous carbonatite-(II) suite. These include: (a) coarse-grained apatite-poor calcite carbonatite; (b) olivine–apatite–magnetite calcite carbonatite with disaggregated phoscorite; (c) phlogopite–tetraferriphlogopite–magnetite–perovskite calcite carbonatite.
The modally variable olivine–apatite–magnetite calcite carbonatite composes most of the suite. Intrusive relationships between these diverse carbonatites have not been established and their distribution within the carbonatite-(II) suite is unknown. It is possible that the phlogopite perovskite-bearing carbonatites are younger than the olivine–apatite–magnetite carbonatites as the micas with respect to their composition are ‘more evolved’ than phlogopite in the latter rocks.
The carbonatite-(II) suite rocks contain a wide variety of minerals. Many of the oxide minerals and olivine are intergrown with other phases, and in particular with magnetite, and represent transported disaggregated cumulates.
The predominant minerals of carbonatite-(II) are: olivine (Fo88–74), calcite (with 1–2 wt% SrO), fluorapatite (rare earth element (REE) and Sr-poor; <1 wt%), Ti-magnetite (3–12 wt% TiO2), green pleochroic phlogopite–barian phlogopite and red pleochroic phlogopite–tetraferriphlogopite. Accessory minerals (1–5 vol.%) include Na–Ca-pyrochlore, U–Ta-bearing pyrochlore, Pb-pyrochlore, niobian zirconolite, perovskite–latrappite–loparite solid solution, baddeleyite and pyrite. Trace minerals (<1 vol.%) are thorianite, calzirtite, pyrrhotite, galena, cobaltian pentlandite, chalcopyrite, ancylite-(Ce), rhabdophane-(Ce) and Mn-ilmenite.
2.a.5. Dolomitic carbonatite
This unit is found only in a small area at the northern margin of the complex. It is bounded in the west by carbonatite-(II) and biotite pyroxenite in the east. Temporal relationships with these units are not established. This carbonatite appears to lack apatite and mica, but does contain REE-fluorocarbonates. It is considered to be a late-stage hydrothermal REE-enriched rock.
The Prairie Lake complex has been considered to be economically significant for uranium, niobium and tantalum deposits, owing to the occurrence of pyrochlore–betafite, wohlerite, perovskite and calzirtite in the carbonatite and ijolite, with apatite and/or wollastonite as potential by-products (Mariano, Reference Mariano1979; Mariano & Roeder, Reference Mariano and Roeder1989; Purtich, Armstrong & Yassa, Reference Purtich, Armstrong and Yassa2010).
3. Analytical methods
In this study, baddeleyite, perovskite, apatite, titanite and calcite were separated, hand-picked, mounted in epoxy resin and polished until the centres of the grains were exposed. Before any microprobe and isotopic analysis, back-scattered electron (BSE) images were obtained using a LEO1450VP scanning electron microscope (SEM), in order to assess internal compositional variation and textures, and identify potential target sites for compositional, U–Pb and Sr–Nd–Hf analyses. All analyses were conducted at the Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing.
3.a. Major- and trace-element analyses
Major-element compositions of the minerals were obtained using a JEOL–JAX8100 electron microprobe with a 15 kV accelerating potential and a 12 nA beam current. Counting times were 20 s. Total Fe is expressed as FeOT. The analytical uncertainties are within 2% for those major elements with concentrations >5%, but ~10–20% for other elements owing to their low concentrations.
Trace-element, including REE, compositions were obtained using an Agilent 7500a Q–ICP–MS (quadruple inductively coupled plasma mass spectrometer), which is equipped with a 193 nm excimer ArF laser ablation system (GeoLas Plus). The elemental concentrations were calculated using GLITTER 4.0 and calibrated using internal standards of 43Ca for apatite, calcite, perovskite and titanite, and 90Zr for baddeleyite. NIST610 was used as an external reference material during analyses (Griffin et al. Reference Griffin, Powell, Pearson, O'Reilly and Sylvester2008).
3.b. In situ U–Pb analyses of baddeleyite by SIMS
Baddeleyite U–Pb analyses were performed using a CAMECA 1280 ion microprobe located at the Institute of Geology and Geophysics in Beijing. A detailed description of the analytical procedure can be found in Li et al. (Reference Li, Li, Liu, Tang, Yang and Zhu2010). In contrast to other U-bearing minerals, such as zircon, monazite and titanite, in situ secondary ion mass spectrometry (SIMS) U–Pb age determination of baddeleyite is difficult owing to crystal orientation effects which bias the 206Pb/238U ratio (Wingate & Compston, Reference Wingate and Compston2000). Therefore, the oxygen flooding technique was applied during this study as it has been shown that this technique can significantly reduce the orientation effects (Li et al. Reference Li, Li, Liu, Tang, Yang and Zhu2010). During analysis, Pb/U ratios were calibrated with a power law relationship between Pb/U and UO2/U relative to a Phalaborwa baddeleyite standard which has a U–Pb age of 2059.6 Ma (Heaman, Reference Heaman2009). The measured Pb isotopic compositions were corrected for common Pb using non-radiogenic 204Pb measured. Small corrections are insensitive to the choice of common Pb composition, and hence an average of the present-day crustal composition (Stacey & Kramers, Reference Stacey and Kramers1975) is applied for the common Pb assuming that this originates mainly from surface contamination introduced during sample preparation. Uncertainties of individual analyses are reported at a 1σ level, and mean ages for pooled Pb/Pb analyses are quoted with a 95% confidence interval using the Isoplot/Ex v. 3.0 program (Ludwig, Reference Ludwig2003).
3.c. In situ Sr–Nd–Hf isotopic analyses by laser ablation
The in situ Sr–Nd–Hf isotopic analyses were conducted using a Neptune MC-ICP-MS (multi-collector inductively coupled plasma mass spectrometer) instrument. Detailed analytical protocols are given by Yang et al. (Reference Yang, Sun, Xie, Fan and Wu2008, Reference Yang, Wu, Wilde, Liu, Zhang, Xie and Yang2009) and Wu et al. (Reference Wu, Yang, Xie, Yang and Xu2006); only a brief summary is given here.
The Sr isotopic data were acquired in static, multi-collector mode with low resolution using eight Faraday collectors and the mass configuration array from 83Kr to 88Sr, monitoring Kr and Rb (Yang et al. Reference Yang, Wu, Wilde, Liu, Zhang, Xie and Yang2009). The key point of in situ Sr isotopic data acquisition is corrections for interfering elements. During this study, these effects were accounted for in the order of Kr, Yb2+, Er2+ and Rb (Yang et al. Reference Yang, Wu, Wilde, Liu, Zhang, Xie and Yang2009), but interferences from Fe dioxides, Ga and Zn oxides, 176Lu2+ and 176Hf2+ are not considered owing to their very low signals during analyses. In order to avoid potential matrix-mismatch effects, the in-house standards of LAP and SAP (apatite), NW-1 and Hainan coral (calcite), and AFK (perovskite) were used for the external correction of 87Sr/86Sr ratios. The 87Sr/86Sr ratios of these standards obtained by using thermal ionization mass spectrometry (TIMS) solution techniques are 0.71138±2 (LAP), 0.72652±1 (SAP), 0.70250±2 (NW-1), 0.70918±2 (Hainan coral) and 0.70335±4 (AFK), respectively (Wu et al. Reference Wu, Arzamastsev, Mitchell, Li, Sun, Yang and Wang2013). However, 87Rb/86Sr ratios were not corrected as this ratio is extremely low for the analysed apatite, calcite and perovskite, and hence has no effect on the initial 87Sr/86Sr ratio of the samples investigated.
The in situ Nd isotope analytical technique is similar to that for the Sr isotope analysis as described above. The most important difficulty needed to be overcome during analysis is the interference of 144Sm on 144Nd, since our previous study shows that the influence of 142Ce on 142Nd is insignificant (Yang et al. Reference Yang, Sun, Xie, Fan and Wu2008). For the correction of 144Sm interference on 144Nd, a method proposed by McFarlane & McCulloch (Reference McFarlane and McCulloch2007) was applied during this study. Similarly, the in-house standards of AP1 and AP2 (apatite), and AFK (perovskite) were used for the external corrections of 147Sm/144Nd and 143Nd/144Nd in order to overcome the potential matrix-mismatch effects. The reported 147Sm/144Nd and 143Nd/144Nd ratios by TIMS analysis are 0.0825 and 0.511352±24 for AP1, 0.0764 and 0.510007±30 for AP2, and 0.0659 and 0.512609±27 for AFK, respectively (Wu et al. Reference Wu, Arzamastsev, Mitchell, Li, Sun, Yang and Wang2013; Yang et al. Reference Yang, Wu, Yang, Chew, Xie, Chu, Zhang and Huang2014).
In situ Hf isotopic analysis of baddeleyite is much easier than that for Sr and Nd isotopes. The isobaric interference of 176Lu on 176Hf is not significant owing to the extremely low 176Lu/177Hf of the minerals investigated (normally < 0.001). However, interference of 176Yb on 176Hf is not negligible. During this analysis, the mean 173Yb/172Yb ratio of an individual spot was used to calculate the fractionation coefficient (βYb), and then this coefficient was used to calculate the interference of 176Yb on 176Hf (Wu et al. Reference Wu, Yang, Xie, Yang and Xu2006). Zircon standards 91500 and GJ-1 were used for external calibration. During analytical sessions, the 176Hf/177Hf value obtained for 91500 was 0.282302±36 (2SD, n = 37), which is similar to the TIMS value of 0.282305 (Wu et al. Reference Wu, Yang, Xie, Yang and Xu2006). As an unknown sample during analyses, the GJ-1 zircon gives a 176Hf/177Hf ratio of 0.282004±7 (2σ, n = 20), agreeing well with the recommended value (0.282000±5) from the literature (Morel et al. Reference Morel, Nebel, Nebel-Jacobsen, Miller and Vroon2008). However, in order to avoid the potential matrix effect between zircon and baddeleyite, the Phalaborwa baddeleyite (Phala-1) was analysed during analyses and yielded a 176Hf/177Hf ratio of 0.281200±16 (2σ, n = 7), which is slightly lower than the previously reported values by TIMS (0.281229±11) and laser ablation (0.281238±12) analyses (Wu et al. Reference Wu, Yang, Xie, Yang and Xu2006, Reference Wu, Yang, Li, Mitchell, Dawson, Brand and Yuhara2011). Therefore, a small correction of 30 ppm was applied for the analysed baddeleyite.
4. Analytical results
4.a. Mineral compositions
In this study, 8 carbonatite, 25 ijolite, 4 syenite and 1 gneissic xenolith samples were selected for mineral separation and further geochemical analysis. Only apatite, calcite, baddeleyite, titanite and perovskite were selected for compositional and isotopic analyses. Mineral compositions are listed in the online Supplementary Material available at http://journals.cambridge.org/geo and the REE distribution patterns are shown in Figure 3. Apatite occurs in all the rocks, and does not show any significant difference in composition (53.7–56.5 wt% CaO, 39.1–43.0 wt% P2O5, 0.1–1.3 wt% SiO2). In terms of trace elements, apatite is characterized by high Sr (3760–7610 ppm) and low Ba (10–36 ppm) contents (Fig. 4a, b). The chondrite-normalized REE distribution pattern is steep, with only the apatite from the county rock (X-1) showing significant depletions in light REEs (LREEs) (Fig. 3a).
Rare earth elemental distribution patterns of (a) apatite; (b) calcite; (c) titanite and perovskite; and (d) baddeleyite. Data sources: Phalaborwa baddeleyite – Wu et al. (Reference Wu, Yang, Li, Mitchell, Dawson, Brand and Yuhara2011); Catalao carbonatite baddeleyite – unpublished data (F. Y. Wu).
(a) Ba v. Sr and (b) Zr v. Nb variations between apatite and calcite, and (c) trace elemental ratios of calcite and apatite from different rock types.
Calcite is the major mineral of carbonatite-(II) (hereafter referred to simply as carbonatite), but is also common in the calcite ijolite and syenites. As shown in Figure 3b, calcite from different samples shows variable REE distribution patterns, but no distinct differences can be identified between the different rock types. Most calcites are characterized by a steep REE distribution pattern with high LREE and low heavy REE (HREE) abundances. Compared to apatite, calcite has higher Sr (6590–89690 ppm) and Ba (201–7394 ppm) (Fig. 4a, b), but lower Nb, Hf, Pb, Th and U concentrations (Fig. 4c).
Titanite from the ijolite contains ~29 wt% SiO2, 31.2–33.6 wt% TiO2 and 26.1–26.5 wt% CaO with minor Al2O3 (0.3–0.7 wt%) and FeO (1.5–2.0 wt%). With respect to trace elements, titanite has high Nb (5909–22485 ppm), Ta (171–1241 ppm), Zr (4978–18465 ppm) and Sr (721–1135 ppm) with low Rb (0.1–0.9 ppm). Titanite has lower contents of REEs, and greater depletion of La and Ce than apatite (Fig. 3c). The perovskite is mostly euhedral with a grain size of ~200 μm. The perovskite exhibits a variable composition. Perovskite from the ijolite (NE3-1, NE5-1, WT2 and WT8) has 52.5–54.6 wt% TiO2 and 36.8–38.1 wt% CaO with minor FeOT (1.1–1.5 wt%) and Na2O (0.4–0.5 wt%). However, perovskite from the carbonatite (NW-1) has a higher concentration of Nb2O5 (~16 wt%) with TiO2 of 37.2–38.2 wt%, CaO of 30.8–32.2 wt% and minor FeO (3.6–3.8 wt%) and Na2O (2.9–3.6 wt%). With respect to trace elements, the perovskites are characterized by high concentrations of Sr (2003–3764 ppm) and Ta (369–791 ppm), but low Zr (139–390 ppm) and Hf (3.4–8.9 ppm) and minor Rb (0.1–0.7 ppm). The REE distribution patterns show extreme LREE enrichment with La abundances of about 13000–23000 times higher than those of chondrites (Fig. 3c).
Baddeleyite occurs in both the carbonatite and ijolite and contains 94.9–98.6 wt% ZrO2 with minor HfO2 (0.4–0.6 wt%). The baddeleyites contain significant amounts of Nb (3938–11035 ppm) and Ta (285–2880 ppm). In terms of REE distribution patterns, the baddeleyite is characterized by depletion of LREEs and enrichment of HREEs with significant positive Ce anomalies (Fig. 3d). Baddeleyites from the ijolite (P10L and MT25) have lower REE concentrations than those from the carbonatite (NW-4, P10A, PL97 and PL110). Compared with the Phalaborwa complex (Wu et al. Reference Wu, Yang, Li, Mitchell, Dawson, Brand and Yuhara2011), the baddeleyite in the Prairie Lake complex has much higher REE concentrations, especially for the HREEs, but contains much lower LREEs than in the niobium-rich Catalao carbonatite in Brazil (Fig. 3d).
4.b. Baddeleyite U–Pb ages
Six samples of baddeleyite were extracted from the carbonatite and ijolite, but only four were selected for U–Pb age determination (Table 1). These baddeleyites have a grain size of ~100 μm. The U concentrations are 228–2400 ppm for NW-4, 178–2210 ppm for P10A, 0.4–130 ppm for MT25 (mostly < 5 ppm) and 125–2049 ppm for P10L, respectively. For the carbonatite, the obtained 207Pb–206Pb weighted average ages are 1157.2±2.3 Ma for NW-4 and 1158.2±3.8 Ma for P10A (Fig. 5a, b). Baddeleyite from the ijolite (MT25) has a much lower U content, making age determination difficult, although a 207Pb–206Pb weighted average age of 1170±18 Ma was obtained (Fig. 5c). However, baddeleyite from an ijolite clast in sample P10L within the carbonatite yielded an age of 1163.6±3.6 Ma (Fig. 5d).
Baddeleyite U–Pb isotopic data by SIMS analysis
The crossed data are not included in age calculation owing to their low U and Th abundances, and hence large errors in isotopic ratio.
Baddeleyite U–Pb Concordia diagrams for the Prairie Lake complex. (a) NW-4 (carbonatite); (b) P10A (carbonatite); (c) MT25 (ijolite); and (d) P10L (ijolitic clast within carbonatite).
4.c. Apatite U–Pb age
Apart from baddeleyite, apatites from the carbonatite NW-1 and NW-4 samples were selected for U–Pb analyses by TIMS and in situ techniques. Apatite NW-1 is characterized by high U concentrations of 61–168 ppm (with an average value of 115 ppm) constrained by laser ablation analyses (online Supplementary Material available at http://journals.cambridge.org/geo), making it a potential standard for apatite SIMS age determination (Li et al. Reference Li, Li, Wu, Yin, Ye, Liu, Tang and Zhang2012). Ten analyses on this sample by TIMS yielded a 206Pb–238U age of 1168.3±4.5 Ma, a 207Pb–235U age of 1162.1±3.3 Ma and a 207Pb–206Pb age of 1155.3±5.8 Ma, showing a slight discordance (Table 2; Fig. 6a). Using the SIMS technique, the calculated 207Pb–206Pb age is 1167±40 Ma (Fig. 6b). Using NW-1 as a calibration standard for the laser ablation analyses, the NW-4 apatite gives an intercept age of 1166±28 Ma on the Tera-Wasserburg diagram (Fig. 6c). If the 207Pb correction method is applied, the obtained weighted mean 206Pb–238U age is 1153±24 Ma (Fig. 6c), identical to the ages obtained above from baddeleyite within errors. It is noted that the composition of common lead is critical to calculation of the above ages obtained by both SIMS and laser ablation techniques. In our work, the model of Stacey & Kramers (Reference Stacey and Kramers1975) at 1160 Ma was applied. Fortunately, our apatites have low common lead, and the data cluster proximal to the Concordia curve (Fig. 6). In this case, the calculated age is not significantly controlled by the uncertainty of common lead composition.
TIMS U–Pb isotopic data of the NW-1 apatite
Analysed by Kevin R. Chamberlain (University of Wyoming, UWYO), and Yuri Amelin (Australian National University, ANU).
Sample NW-1 C was not included in age calculation owing to its significant discordance.
U–Pb diagrams for apatite from the Prairie Lake complex. (a) NW-1 (carbonatite, TIMS data); (b) NW-1 (carbonatite, SIMS data, the plots in grey are excluded for age calculation); (c) NW-4 (carbonatite, laser ablation data).
4.d. Sr–Nd–Hf isotopic compositions
4.d.1 Apatite
Apatite has been extracted from all the samples investigated, and analysed for its Sr and Nd isotopic compositions using an in situ laser ablation technique (Table 3). All of the apatites have low 87Rb/86Sr ratios (0.00001–0.00013) with 87Sr/86Sr ratios ranging from 0.70246±3 to 0.70261±2 (Table 3; Fig. 7a). In terms of rock type, 87Rb/86Sr and 87Sr/86Sr ratios are: 0.00002–0.00013 and 0.70251±2 to 0.70261±2 for carbonatite; 0.00001–0.00012 and 0.70246±3 to 0.70260±2 for ijolite; and 0.00001–0.00002 and 0.70253±2 to 0.70260±3 for syenite, respectively. The weighted average 87Sr/86Sr ratios are 0.70251±2 for carbonatite (2σ, n = 8), 0.70253±2 for ijolite (2σ, n = 25) and 0.70256±5 for syenite (2σ, n = 4), respectively (Fig. 7a). These data indicate that there is no significant variation among the different rocks within the complex, and the average value of 0.70253±1 for all samples (2σ, n = 37) can be considered as the best estimation of the initial Sr isotopic composition of apatite for the Prairie Lake complex.
Sr–Nd isotopic data of minerals from the Prairie Lake complex
Sr isotopic variations of (a) apatite, (b) calcite and (c) perovskite.
The 147Sm/144Nd ratios and 143Nd/144Nd isotopic compositions of apatite range from 0.0690 to 0.1094 and from 0.511806±51 to 0.512142±10, respectively (Table 3; Fig. 8a). For the individual rocks within the complex, the 147Sm/144Nd and 143Nd/144Nd ratios of the carbonatite range from 0.1006 to 0.1094 and from 0.512080±10 to 0.512142±10, respectively. The calculated εNd(t)1160 values are from +3.1±0.2 (PL97) to +3.7±0.1 (NW-1), with an average value of +3.44±0.20 (2σ, n = 8) (Fig. 8a). Apatite from the ijolite shows larger variations of 147Sm/144Nd from 0.0690 to 0.1064 and 143Nd/144Nd from 0.511806±51 to 0.512141±20, with εNd(t)1160 values of +2.6±0.5 and +4.1±0.2 and an average value of +3.55±0.17 (2σ, n = 25) (Fig. 8a). For the syenite in the complex, apatites have 147Sm/144Nd values of 0.0738 to 0.0848 and a 143Nd/144Nd ratio of 0.511883±10 to 0.511984±18, with εNd(t)1160 values of +3.1±0.3 to +4.0±0.2 and an average value of +3.61±0.49 (2σ, n = 4) (Fig. 8a). Therefore, the εNd(t)1160 values of apatite from the carbonatite (+3.44±0.20), ijolite (+3.55±0.17) and syenite (+3.61±0.49) are identical within analytical uncertainties, and the average value of +3.52±0.11 (n = 37) for all the samples represents the best estimate of the εNd(t)1160 value for the complex.
Nd isotopic variations of (a) apatite and (b) titanite and perovskite.
In addition, it is noted that the apatite from the gneissic xenolith has a much lower 87Sr/86Sr ratio of 0.70170±11 compared to those from the carbonatite, ijolite and syenite. Its 147Sm/144Nd and 143Nd/144Nd ratios are 0.0968 and 0.510924±63, respectively, with an εNd(t)1160 value of −19.1±0.6 and a depleted mantle model age of 2.88 Ga (Table 3; Fig. 8a), indicating its derivation from the Archaean country rocks.
4.d.2 Calcite
Calcite has low REE contents; thus, only Sr isotopic compositions were determined (Table 3). Comparatively, calcite has even lower 87Rb/86Sr ratios than apatite (Fig. 7b). Of the eight samples of calcite extracted from the carbonatite, the 87Rb/86Sr and 87Sr/86Sr ratios are 0.00000–0.00006 and 0.70250±2 to 0.70261±2 with an average 87Sr/86Sr ratio of 0.70253±3 (2σ, n = 8). Of the 23 samples of calcite from the ijolite, their 87Rb/86Sr and 87Sr/86Sr ratios are 0.00000–0.00008 and 0.70245±4 to 0.70262±2, with an average 87Sr/86Sr ratio of 0.70254±2 (2σ, n = 23). The calcite samples from the syenite also have uniform 87Rb/86Sr (< 0.00002) and 87Sr/86Sr (from 0.70255±3 to 0.70259±1) ratios, with an average 87Sr/86Sr ratio of 0.70258±3 (2σ, n = 4). Therefore, calcites from the carbonatite (0.70253±3), ijolite (0.70254±2) and syenite (0.70258±3) have identical Sr isotopic compositions, and an average of 0.70255±1 (2σ, n = 35). These values are comparable to those of apatite from the carbonatite (0.70251±2), ijolite (0.70253±2) and syenite (0.70256±5).
4.d.3 Perovskite
Five perovskite samples from the carbonatite and ijolite were selected for Sr–Nd isotopic analysis (Table 3). Of these samples, only two perovskites (NW-1 and WT8) were adequate for Rb–Sr isotope study, and give 87Rb/86Sr ratios of 0.0001 to 0.0038 and 87Sr/86Sr ratios of 0.70232±7 to 0.70273±11 (Table 3; Fig. 7c). The weighted 87Sr/86Sr ratios are 0.70263±5 (n = 8) for NW-1 and 0.70253±3 (n = 16) for WT8. In terms of Nd isotopic data (Fig. 8b), perovskites from the five samples show similar isotopic variations to those of apatite, with 147Sm/144Nd, 143Nd/144Nd and εNd(t)1160 values ranging from 0.0573 to 0.1088, 0.51176±4 to 0.51216±9, and +1.9±1.4 to +4.9±1.0, respectively. Their average εNd(t)1160 value is +3.37±0.14 (n = 66), comparable to that of apatite.
4.d.4 Titanite
Titanite was extracted only from four samples and only Sm–Nd isotopic data were obtained. The 147Sm/144Nd and 143Nd/144Nd isotopic ratios range from 0.1084 to 0.1475, and 0.512139±24 to 0.512448±45, respectively, with εNd(t)1160 values from +3.3±0.4 to +3.5±0.3 (Table 3; Figs 8c). The calculated average εNd(t)1160 value is +3.39±0.15 (n = 62), identical to that of apatite and perovskite.
4.d.5 Baddeleyite
Six samples of baddeleyite from the carbonatite and ijolite were analysed for Hf isotopic compositions (Table 4). All have 176Lu/177Hf ratios less than 0.0008, except those of P97, which have much higher 176Lu/177Hf ratios (Fig. 9). The average 176Lu/177Hf, 176Hf/177Hf and εHf(t)1160 values are from 0.0002 to 0.0010, 0.282173±8 to 0.282207±13, and +4.32±0.32 to +5.25±0.46, respectively (Table 4), with an average εHf(t)1160 value of +4.56±0.47 (n = 6). However, if the data are separated according to rock type, the carbonatite and ijolite have εHf(t)1160 values of +4.53±0.64 (n = 4) and +4.79±0.57 (n = 2), consistent within uncertainties.
Lu–Hf isotopic data of baddeleyite from the Prairie Lake complex
Hf isotopic variations of baddeleyite. Data for the Duluth zircon are from Woodhead & Hergt (Reference Woodhead and Hergt2005), Kemp et al. (Reference Kemp, Foster, Schersten, Whitehouse, Darling and Storey2009), Kimura, Tani & Chang (Reference Kimura, Tani and Chang2012) and Fisher, Vervoort & Dufrane (Reference Fisher, Vervoort and Dufrane2014).
5. Discussion
5.a. Emplacement age of the Prairie Lake complex
The Prairie Lake complex intruded Archaean granitic gneisses, and is considered to have been emplaced during the Proterozoic Era. The first ages of 1164 and 1059 Ma were determined for biotite in the carbonatite by the K–Ar method (Gittins, Macintyre & York, Reference Gittins, Macintyre and York1967). Subsequently, isochron ages of 1030±60 Ma (Bottriell, Reference Bottriell1975), 1033±59 Ma (Bell & Blenkinsop, Reference Bell, Blenkinsop and Pye1980) and 1030±70 Ma (Bell et al. Reference Bell, Blenkinsop, Cole and Menagh1982) were determined by Rb–Sr methods. In a subsequent study, S. J. Pollock (unpub. MSc thesis, Carleton Univ., 1987) obtained ages of 1165±30 Ma (Rb–Sr isochron for whole-rock ijolite–carbonatite), 1135±15 Ma (Rb–Sr isochron of biotite), 1130±10 Ma (Rb–Sr isochron of whole-rock ijolite–carbonatite and biotite) and 1200±40 Ma (Sm–Nd isochron of whole-rock ijolite–carbonatite plus separated apatite, calcite and garnet). Given the potential late-stage hydrothermal alteration, these authors suggested that the age of 1165±30 Ma could be the best estimate for the emplacement time of the complex. Bell & Blenkinsop (Reference Bell, Blenkinsop and Bell1989) reported a whole-rock Rb–Sr age of 1023±74 Ma for the carbonatites, and Kwon, Tilton & Grüenenfelder (Reference Kwon, Tilton, Grüenenfelder and Bell1989) reported a Pb–Pb age of 1155±36 Ma for calcite in the carbonatite. Note that the isochron ages have very large errors, and mineral ages are required to constrain the emplacement time of the complex.
Using the SIMS technique, Sano et al. (Reference Sano, Oyama, Terada and Hidaka1999) obtained a U–Pb age with a large error of 1156±45 Ma for apatite from ijolite. Recently, Rukhlov & Bell (Reference Rukhlov and Bell2010) using TIMS analysis for baddeleyite and zircon from phoscorite, obtained an age of 1164±4 Ma. Similarly, Wu et al. (Reference Wu, Yang, Mitchell, Bellatreccia, Li and Zhao2010) reported a Pb–Pb age of 1159±5 Ma for calzirtite from the carbonatite. These data indicate that the complex was most probably emplaced at ~1160 Ma. Nevertheless, it remained unclear as to whether the different rocks of the complex were emplaced synchronously since these ages were obtained from different laboratories using different methods.
In this study, two samples of baddeleyite from the carbonatite gave 207Pb–206Pb ages of 1157±2 (NW-4) and 1158±4 (P10A) Ma. A TIMS analysis of the NW-1 apatite yielded a slightly discordant age around 1160 Ma, which is verified by SIMS analysis. In addition, laser ablation of calzirtite in NW-1 gives a 207Pb–206Pb age of 1170±11 Ma, which is identical within analytical error to the 207Pb–206Pb age of 1159±5 Ma given by SIMS analysis (Wu et al. Reference Wu, Yang, Mitchell, Bellatreccia, Li and Zhao2010). Laser ablation of the NW-4 apatite gives an intercept age of 1166±28 Ma. If all these ages are considered together, a mean age of 1158±3 Ma is obtained, which is identical to the TIMS age of baddeleyite/zircon by Rukhlov & Bell (Reference Rukhlov and Bell2010).
For the ijolitic rocks, baddeleyite from MT25 gives an age of 1169±22 Ma. The large error is owing to the extremely low concentration of uranium (less than 5 ppm), and hence low signal during analysis. Fortunately, an ijolitic clast (P10L) within the carbonatite gives a more precise age of 1164±4 Ma, which is identical to those obtained from the carbonatite. Therefore, it can be concluded that the Prairie Lake complex was emplaced at ~1160 Ma.
5.b. Emplacement age relative to Midcontinent Rift magmatism
Midcontinent rifting was accompanied by numerous stages of magmatism. According to Heaman et al. (Reference Heaman, Easton, Hart, Hollings, Macdonald and Smyk2007), four stages of magmatism can be identified. Stage 1 (1150–1130 Ma) magmatism, interpreted as the earliest manifestation of Midcontinent rifting, is represented by numerous lamprophyre dykes and minor felsic volcanic rocks, which are temporally related to the well-known Abitibi dyke swarm (1141 Ma, Ernst & Buchan, Reference Ernst, Buchan, Mahoney and Coffin1997; Heaman et al. Reference Heaman, Easton, Hart, Hollings, Macdonald and Smyk2007). Stage 2 (1115–1105 Ma), the onset of the Midcontinent rifting, is represented by diverse ultramafic intrusions, basaltic sills and flows, rhyolites and alkaline rocks. Stage 3 (1100–1094 Ma) is the main period of Midcontinent rifting, and is represented by mafic intrusions (Duluth Complex) and basalt flows, with minor amounts of alkaline rocks (Lackner Lake carbonatite). Stage 4 (<1094 Ma) represents the waning stage of the Midcontinent rifting, and is expressed by porphyry and basaltic dykes.
The alkaline rock and carbonatite complexes within the rift area are mostly considered to belong to stage 2 magmatism (Heaman et al. Reference Heaman, Easton, Hart, Hollings, Macdonald and Smyk2007). However, the available geochronological data indicate that the alkaline magmatism extends over a significant time interval (Heaman & Machado, Reference Heaman and Machado1992; Heaman et al. Reference Heaman, Easton, Hart, Hollings, Macdonald and Smyk2007; Rukkhlov & Bell, Reference Bell and Simonetti2010; and this study). For the complexes within the Kapuskasing Structural Zone, the complexes include Firesand River (1139.7±2.5 Ma), Valentine Township (1114.7±1.1 Ma), Nemogosenda (1105.4±2.6 Ma) and Lackner Lake (1100.6±1.5 Ma). Within the Trans-Superior Tectonic Zone, only Prairie Lake (1158±3 Ma) and Coldwell (1107±2 to 1109±8 Ma) have reliable age determinations. In addition to complexes within the two main tectonic zones, carbonatite complexes are found at Big Beaver House (1093.0±1.7 Ma) and Schryburt Lake (1083.5±2.9 Ma) to the northwest of these zones (Fig. 1). Thus, the ages of the alkaline rock and carbonatite magmatism range from 1158 to 1084 Ma, indicating that such magmatism occurred during all stages of the Midcontinent rifting.
The range of age determinations can be interpreted as either reflecting magmatism associated with passive continental rifting or that associated with the initial stages of plume-induced rifting. If these complexes were related to mantle plume activity, then the lifetime of this plume must be as long as >70 Ma (Heaman et al. Reference Heaman, Easton, Hart, Hollings, Macdonald and Smyk2007; Hollings, Smyk & Cousens, Reference Hollings, Smyk and Cousens2012). It is also noted, from these age data, that the Prairie Lake complex is the oldest alkaline–carbonatite complex associated with the Midcontinent rifting. Detailed discussion of the merits of passive versus active rifting in the genesis of the Midcontinent Rift is well beyond the scope of this work. Our data have no direct bearing on whether the emplacement of the Prairie Lake complex was related to crustal extension and passive rifting or an upwelling mantle plume. Our data merely indicate that the earliest manifestation of magmatism in this region was at 1160 Ma. If this was related to plume activity then the plume activity would extend from 1160 to 1094 Ma. In support of a plume-related origin, Sasada et al. (Reference Sasada, Hiyagon, Bell and Ebihara1997) found that apatite from Prairie Lake has an extremely high 136Xe/130Xe ratio and excess 129Xe, and suggested that it might be related to less-degassed primordial mantle, which is comparable to a proposed plume model (Bell & Simonetti, Reference Bell and Simonetti2010; Ernst & Bell, Reference Ernst and Bell2010).
It has been proposed that the Prairie Lake complex is spatially and age-related to the nearby large alkaline complexes, and that these complexes represent differentiated segments of magma that travelled on different paths, but that have been generated from the same magma source (Gittins, Macintyre & York, Reference Gittins, Macintyre and York1967). However, from the geochronological data presented here, the Prairie Lake complex was emplaced much earlier than the Coldwell and Killala complexes. Other petrological studies indicate that these A-type granitoid complexes such as the Coldwell complex (Mitchell et al. Reference Mitchell, Platt, Lukosius-Sanders, Artist-Downey and Moogk-Pickard1993), lacking carbonatite, are derived by partial melting of metasomatized lower crust (Martin, Reference Martin2006). Therefore, those complexes are not genetically related to the Prairie Lake carbonatite complex.
5.c. Aspects of the petrogenesis of the Prairie Lake complex
It is not an objective of this paper to discuss the petrogenesis of the Prairie Lake complex as this is beyond the scope of this work. However, our new isotopic data do have a bearing on this problem and are commented on below.
The petrogenesis of alkaline rock and carbonatite complexes remains one of the major unresolved problems in petrology (Bell, Reference Bell1998; Harmer & Gittins, Reference Harmer and Gittins1998; Harmer, Reference Harmer1999). It is commonly suggested that nepheline-bearing rocks and carbonatites can be related by fractional crystallization or liquid immiscibility to a common parental magma. Typically, evidence for immiscibility is circumstantial and the hypothesis is based primarily on extrapolation of laboratory studies of synthetic haplocarbonatites (Lee & Wyllie, Reference Lee and Wyllie1998; Kjarsgaard, Reference Kjarsgaard1998; Brooker & Kjarsgaard, Reference Brooker and Kjarsgaard2011; Martin et al. Reference Martin, Schmidt, Mattsson and Guenther2013), and/or experimental geochemical studies (Jones et al. Reference Jones, Walker, Picket, Murrel and Beate1995; Veksler et al. Reference Veksler, Petibon, Jenner, Dorfman and Dingwell1998, Reference Veksler, Dorfman, Dulski, Kamenetsky, Danyushevsky, Jeffries and Dingwell2012) rather than direct petrographic or geological evidence for the process. For example, conjugate liquids as represented by melt inclusions have not been described from ijolite–carbonatite complexes, although conjugate immiscible melt pairs have been observed for wollastonite nephelinite and natrocarbonatite (Mitchell, Reference Mitchell2009; Mitchell & Dawson, Reference Mitchell and Dawson2012).
The major observations stemming from this work are that: (1) All of the rock units occurring at Prairie Lake were emplaced contemporaneously within the resolution of the age determinations at ~1160 Ma; (2) All the rocks have effectively identical initial Sr and Nd isotopic compositions, indicating derivation from a common parental magma. To our knowledge, the Prairie Lake complex has the most homogeneous isotopic composition of any carbonatitic complex. Our data certainly do not support the contentions of Gittins & Harmer (Reference Gittins and Harmer2003) that the relationships between silicate and carbonatite rocks within alkaline complexes is spatial rather then genetic; (3) None of the rocks exhibit isotopic evidence for significant contamination by crustal material.
None of these isotopic data provide conclusive evidence for any particular petrogenetic scheme and must be considered in conjunction with geological evidence. However, geological observations suggest that the Prairie Lake plutonic and hypabyssal rocks formed within a continuously replenished, continuously fractionating magma chamber beneath a nephelinitic volcano. Consequently, there are several distinct episodes of ijolite and carbonatite formation in addition to cumulate formation and disaggregation. There is petrographic evidence for fractional crystallization and magma mixing in many of the ijolites and carbonatites but no evidence for liquid immiscibility. That the Prairie Lake rocks appear to have crystallized from a very homogeneous magma is either a consequence of these mixing processes or derivation from a homogeneous source. The details of the petrogenesis will be presented in a subsequent paper.
5.d. Isotopic heterogeneity of the mantle source of the Midcontinent magmatism
It was determined in this study that the ijolite, syenite and carbonatite from the Prairie Lake complex show identical Sr–Nd isotopic compositions (Fig. 10a). The ijolite has an initial 87Sr/86Sr ratio of ~0.70254 and a positive εNd(t)1160 value of +3.5, and the syenite has the above values of ~0.70257 and +3.6. Similarly, the carbonatite has a 87Sr/86Sr ratio of ~0.70252 and εNd(t)1160 value of +3.4. In addition, baddeleyites from the ijolite and carbonatite have identical Hf isotopic compositions (Table 4), and the Nd–Hf isotopic correlation indicates that these rocks have a similar isotopic composition to that of oceanic island basalt (OIB; Fig. 10b). In agreement with Bell & Blenkinsop (Reference Bell and Blenkinsop1987), Rukhlov & Bell (Reference Rukhlov and Bell2010) and (Rukhlov, Bell & Amelin, Reference Rukhlov, Bell, Amelin, Simandl and Neetz2015), we conclude that the silicate and carbonatitic magmas are derived from a weakly depleted mantle source that was formed at 3 Ga ago. Our data fit the Sr and Nd development lines for such a source as given by Rukhlov, Bell & Amelin (Reference Rukhlov, Bell, Amelin, Simandl and Neetz2015).
(a) Sr–Nd and (b) Nd–Hf isotopic variations of the Prairie Lake complex. Sr–Nd literature data for the complex (grey circles in (a)) are from S. J. Pollock (unpub. MSc thesis, Carleton Univ., 1987), and Midcontinent Rift magmatism is from Hollings, Smyk & Cousens (Reference Hollings, Smyk and Cousens2012). The fields of MORB (mid-ocean ridge basalt), OIB (ocean island basalt), global lower crust and sediments in (b) are from Wu et al. (Reference Wu, Yang, Li, Mitchell, Dawson, Brand and Yuhara2011).
Available data indicate that the igneous rocks formed during the Midcontinent rifting have variable isotopic compositions (Paces & Bell, Reference Paces and Bell1989; Nicholson & Shirey, Reference Nicholson and Shirey1990; Lightfoot, Sutcliffe & Doherty, Reference Lightfoot, Sutcliffe and Doherty1991; Shirey et al. Reference Shirey, Klewin, Berg and Carlson1994; Nicholson et al. Reference Nicholson, Shirey, Schulz and Green1997; Hollings et al. Reference Hollings, Richardson, Creaser and Franklin2007; Hollings, Smyk & Cousens, Reference Hollings, Smyk and Cousens2012; Bright et al. Reference Bright, Amato, Denyszyn and Ernst2014). For example, the vast majority of basalt in the rift is quite uniform with εNd(t) values ranging between −2.5 and +3.5. If only primitive olivine tholeiites are considered, then an even more restricted range of −0.3 to +0.8 is observed (Paces & Bell, Reference Paces and Bell1989). Similarly, the younger (~1100–1094 Ma) volcanic rocks also have a positive εNd(t) value of +1 to +4 (Nicholson et al. Reference Nicholson, Shirey, Schulz and Green1997). In contrast, the 1110–1120 Ma mafic sills in the Nipigon Embayment and Thunder Bay to the west of the studied area have negative εNd(t) values of +0.9 to −9.3 with initial 87Sr/86Sr ratios of 0.7032 to 0.7243 (Hollings et al. Reference Hollings, Richardson, Creaser and Franklin2007; Hollings, Smyk & Cousens, Reference Hollings, Smyk and Cousens2012), indicating their derivation from ancient subcontinental lithospheric enriched mantle or contamination by crustal material during their crystallization.
As for the alkaline complexes, Heaman & Machado (Reference Heaman and Machado1992) reported that gabbro from the Coldwell complex has εNd(t) values of −2.6 to +1.6 with initial 87Sr/86Sr ratios of 0.70336 to 0.70349, whereas syenites have the above values of −0.5 to −4.6 with initial 87Sr/86Sr ratios of 0.70361 to 0.70416, slightly different from the Prairie Lake complex. However, Bell & Blenkinsop (Reference Bell and Blenkinsop1987, Reference Bell, Blenkinsop and Bell1989) found a large Sr–Nd isotopic variation for other alkaline–carbonatitic complexes, such as Nemegosenda, Big Heaver House, Firesand River, Clay-Howells, Seabrook and Schryburt Lake. The Nemegosenda and Clay-Howells complexes have initial 87Sr/86Sr ratios of 0.7037 and εNd(t) values of −2.1 to +0.6, and thus are similar to the Coldwell complex, whereas other complexes have 87Sr/86Sr ratios of 0.7024 to 0.7026 and εNd(t) values of +2.4 to +3.6, similar to the Prairie Lake complex. This variable isotopic composition among the different kinds of rock suggests that the mantle magmatism had a complex relationship between the mantle and crust. Alternatively, these rocks derived from a heterogeneous mantle source, as proposed by Kwon, Tilton & Grüenenfelder (Reference Kwon, Tilton, Grüenenfelder and Bell1989) and Tilton & Kwon (Reference Tilton and Kwon1990). Although no isotopic data are available for the Killala and Chipman Lake complexes, it could be concluded that at least some alkaline–carbonatite complexes in the area (Prairie Lake, Big Heaver House, Firesand River, Seabrook and Schryburt Lake) have the most depleted mantle composition among the various rocks formed during the Midcontinent rifting.
6. Conclusions
Comprehensive age determinations and Sr–Nd–Hf isotopic analyses of baddeleyite, apatite, calcite, titanite and perovskite from the Prairie Lake complex in Ontario, Canada, lead to the following conclusions:
-
(1) Baddeleyite from the carbonatite and ijolite, and apatite from the carbonatite yield identical U–Pb ages of ~1160 Ma, indicating that the different phases of the complex were emplaced contemporaneously;
-
(2) Sr–Nd–Hf isotopic analyses of apatite, calcite, titanite, perovskite and baddeleyite indicate that the different rock types of the complex do not show any significant isotopic variation, indicating that the silicate and carbonate rocks are co-genetic and originate from a single magma type by simple crystal fractionation. The low 87Sr/86Sr ratio of ~0.70254 and positive εNd(t)1160 and εHf(t)1160 values of ~+3.5 and +4.6 suggest that the complex was derived from a weakly depleted mantle;
-
(3) The U–Pb data have reinforced the previous conclusion that the Prairie Lake complex is one of the earliest manifestations of Midcontinent magmatism, and this complex was not formed contemporaneously with other alkaline complexes in the area (Coldwell, Killala).
Acknowledgements
Yang Li, Liang-Liang Zhang and Wei-Qiang Ji are thanked for their assistance during sample preparation and analyses. Kevin R. Chamberlain and Yuri Amelin helped to run the TIMS analyses for the NW-1 apatite. Peter Hollings and Shannon Zurevinski read the early draft and provided invaluable comments on the characters of the Midcontinent Rift and carbonatite petrogenesis. Constructive reviews by Keith Bell, Shannon Zurevinski, Tony Simonetti and anonymous reviewers greatly improved the manuscript. This work was supported by the Natural Science Foundation of China (Grant 41130313). Roger Mitchell's work on alkaline rocks and carbonatites is supported by the Natural Sciences and Engineering Research Council of Canada, Almaz Petrology and Lakehead University. Paul Jones of Nuinsco Resources is thanked for access to their exploration properties at Prairie Lake.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0016756815001120.
