Igneous rocks are sparsely distributed throughout the Middle Devonian–Permian Maritimes Basin (Fyffe and Barr 1986; Dunning et al. 2002). They appear to decrease in volume through time from: (i) Middle to Late Devonian tholeiite, minor alkali basalt, bimodal gabbro and A-type to evolved I-type granite; (ii) Lower Carboniferous bimodal igneous rocks, and most sparsely, (iii) Upper Carboniferous bimodal igneous rocks (New Brunswick Department of Natural Resources 2010a). The felsic igneous rocks, as well as the interbedded lithic sandstone, conglomerate, and feldspathic sandstone of the Cumberland Hill Formation, the topic of this paper, were considered part of the Late Wesphalian Pictou Group, and consequently associated with the latter episode of magmatism based on stratigraphic relationships (Fyffe and Barr 1986). However, recent mapping indicated that these volcanic rocks lie within the Visean to early Namurian Mabou Group, and therefore are associated with the middle episode of volcanic activity (St. Peter 1997; New Brunswick Department of Natural Resources 2010a). A mid-Visean U-Pb zircon date of 335 ± 2 Ma substantiates this interpretation (New Brunswick Department of Natural Resources 2010a). Hence the felsic volcanic rocks in the Cumberland Hill Formation appear to have approximately the same age as several alkali mafic units in New Brunswick (e.g., the Hardwood Ridge, Royal Road, and Queenstown basalts), and the tholeiitic-alkali mafic rocks of the Cap aux Diables Formation in the Magdalen Islands (Barr et al. 1985; Fyffe and Barr 1986; LaFleche et al. 1998; New Brunswick Department of Natural Resources 2010a).

The mafic rocks have continental within-plate characteristics (Barr et al. 1985), and HIMU-OIB Pb-isotopic signatures that have been related either to a mantle plume (Pe-Piper and Piper 1998) or to decompression melting below a pull-apart rift (LaFleche et al. 1998). However, the plume model (Murphy et al. 1999; Keppie and Krogh 1999) appears inconsistent with U-Pb chronology of Devonian to early Carboniferous volcanic rocks within the southern Magdalen Basin (Dunning et al. 2002).

In this paper, we selected key samples from the Cumberland Hill Formation in order to petrologically and geochemically characterize the rocks and compare the rhyolite and trachyte to similar rocks elsewhere in the Maritimes Basin. This work is critical when assessing uranium potential in the Cumberland Hill Formation, as similar rocks in the Maritimes Basin have been identified as hosts of economic uranium mineralization (e.g., Dahlkamp 1993; Plant et al. 1999; Cuney 2009; Nash 2010).

Late Devonian to Early Permian rocks of New Brunswick comprise the western part of the Maritimes Basin of Atlantic Canada. The Maritimes Basin is a successor basin within which there are a number of sub-basins separated by fault-bounded ridges of basement (Fig. 1a). In New Brunswick, these rocks occur within deep depositional centres or subbasins, or on shallowly buried or partially exposed basement uplifts and platforms (e.g., St. Peter and Johnson 2009).

The volcanic rocks of the Cumberland Hill Formation crop out as several inliers and are unconformably overlain by the Upper Carboniferous Pictou Group on the New Brunswick Platform (Fig. 1b). The outcrops are likely related to a single volcanic centre partially hidden beneath Pictou Group rocks, because the area incorporating the inliers clearly defines a circular area of high magnetic response on aeromagnetic maps (Thomas and Kiss 2005).

Fyffe and Barr (1986) examined some of the felsic volcanic rocks in the vicinity of Cumberland Hill, including lava flows and tuffs in a rhyolitic unit and underlying trachytic unit. Subsequently, St. Peter (1997) delineated the distribution of these units in more detail, depicting the trachyte as mostly underlying, but in part interdigitating with, the rhyolite (Fig.2). There is also one massive intrusive porphyry plug with several smaller dykes/pipes, interpreted by Thomas and Kiss (2005) as volcanic feeder dykes, based on their bulls-eye positive magnetic anomalies (50–375 nT) within the magnetically anomalous area. The exact thickness of these units is unknown as they are poorly exposed; however, both trachyte and rhyolite units are believed to be tens of metres thick (St. Peter 1997).

The rhyolite occurs as aphanitic to porphyritic flows and ignimbritic tuffs with flow banding and microporphyritic to porphyritic textures (Fig. 3a). Flow-aligned phenocrysts of feldspar (mainly anorthoclase; Fyffe and Barr 1986; 0.1–0.5 mm in length) and quartz (0.3–1 mm) are enclosed in a quartzofeldspathic groundmass with rare green clinopyroxene (hedenbergite; Fyffe and Barr 1986). The groundmass is composed of quartz, feldspar, devitrified glass, minor magnetite, ilmenite, pyrite, rare galena, fluorite and accessory minerals similar to those in trachyte samples (i.e., microlites of zircon, apatite, monazite, and allanite). The rhyolite is also extensively altered, although to a lesser degree than the trachyte, with most primary magmatic phases replaced by a secondary assemblage (i.e., carbonate, sericite, albite, and chlorite).

The trachyte occurs in two varieties, porphyritic and aphanitic, the latter of which is dominant. The porphyritic trachyte contains phenocrysts of feldspar (sanidine or anorthoclase) up to 2 mm in size, rare clinopyroxene, and plagioclase (Fig. 3b). The groundmass is largely composed of devitrified glass with phenocrysts of feldspar, clinopyroxene, and quartz, and rare ilmenite, hematite, sphalerite, and galena. Accessory minerals (fine-grained microlitic zircon, apatite, monazite, and allanite) were identified and confirmed using a scanning electron microscope including backscattered electron imaging, energy-dispersive X-ray spectroscopy and X-ray element mapping. The aphanitic variety displays a typical trachytic texture with flow-aligned feldspars (0.1–0.5 mm in length) enclosed in a fine-grained matrix similar to that of the porphyritic type. Both varieties are pervasively altered, with only relicts of primary minerals present, and secondary minerals dominated by calcite, sericite, chlorite, albite, and hematite alteration.

Fifteen samples were collected from the Cumberland Hill Formation, crushed and pulped in a soft iron swing mill, then analyzed for major elements using glass discs by X-ray fluorescence (XRF) spectrometry in the Nova Scotia Regional Geochemical Centre at Saint Mary’s University, Halifax. Samples were subsequently analyzed for trace element compositions using a Perkin Elmer Optima 3000 inductively coupled plasma-mass spectrometer in the Activation Laboratories in Ancaster, Ontario. Prior to analysis, samples were digested in nitric acid after lithium metaborate fusion to ensure the complete dissolution of refractory accessory minerals. Precision and accuracy of the XRF data are reported by Dostal et al. (1994). The analytical error of the trace element determinations is 2–10%.

According to the Na₂O + K₂O vs SiO₂ classification of LeMaitre et al. (1989; Fig. 4a) and the Zr/TiO₂ vs SiO₂ diagram (Fig. 4b) of Winchester and Floyd (1977), the rocks can be subdivided into two groups, trachyte and peralkaline rhyolite, analogous to comendite/pantellerite fields according to Winchester and Floyd (1977). Most rocks are peralkaline with a high agpaitic index (AI = mole (Na+K)/Al; Shand 1951) and acmite in their CIPW normative mineralogy. Considering loss on ignition-free analyses (Table 1), the trachyte on average (n = 7, 1σ) has moderate SiO₂ (65 ± 2.48 wt. %) accompanied by high total alkalis (9.5 ± 1.81 wt. %), low CaO (2.5 ± 1.21 wt. %), high FeO^t/MgO (ca. 25) and elevated concentrations of high field strength elements (HFSE) and other incompatible elements (e.g., Zr + Nb + Ce + Y = 1334 ± 409.70; Whalen et al. 1987). The average of the rhyolitic rocks (n = 8, 1σ) has higher SiO₂ (75.0 ± 0.66 wt. %), equivalent total alkali concentrations (9.0 ± 0.29 wt. %), very low CaO (0.23 ± 0.08 wt. %), high FeO^t/MgO (ca. 140), and high HFSE and other incompatible elements (e.g., Zr + Nb + Ce + Y = 3126 ± 161.75; Whalen et al. 1987).

The major and trace element signatures of the volcanic rocks of the Cumberland Hill Formation are interpreted to vary predominantly due to the degree of fractionation. Conversely, a heterogeneous magma source could also contribute to the variations exhibited. In order to determine the fractionation trends, Nb was utilized as a differentiation index following White et al. (2006; Fig. 5). With increasing concentration of Nb, the trachyte samples display positive correlation with Al₂O₃, and negative correlation with TiO₂. The reduction in TiO₂ concentration can most likely be attributed to crystallization of Fe-Ti oxides. Rhyolite samples, in contrast, have relatively constant Al₂O₃ (Fig. 5a) and TiO₂ (Fig. 5b). Trace elements also display different fractionation trends in trachyte and rhyolite samples. Trachyte samples show marked decreases in Ba with increasing Nb (Fig. 5c), reflecting the crystallization of feldspars, as well as an increase in incompatible trace elements including Zr, La, and Ga (Fig. 5d, e, f). However, the abundances of these trace elements in the rhyolite remain virtually unchanged with increasing Nb. The contrast in fractional crystallization trends between trachyte and rhyolite is highlighted by the CaO-K₂O-Na₂O ternary diagram (Fig. 6), which illustrates that trachyte samples plot in a trend away from the CaO apex, whereas rhyolite sample variation is parallel to the K₂O-Na₂O join. This difference is most readily explained by the crystallization of Ca-bearing plagioclase and clinopyroxene in the trachyte magma, and alkali feldspar crystallization in the rhyolite magma.

The chondrite-normalized rare earth element (REE) patterns for each suite (Fig. 7a) are strongly enriched in light REE (LREE). The rhyolite has higher LREE contents with La_n values of ca. 440–510 compared to the trachyte with La_n ca. 270–430. The trachyte shows steeper LREE to heavy rare earth element (HREE) profiles, with La_n/Yb_n ratios ca. 8–11 compared to ca. 4–6 in rhyolite. A similar relationship is observed for La_n/Sm_n, which in the trachyte is ca. 2.6–3.4 and in the rhyolite is ca. 2.2–2.6. The shape of the REE patterns in the trachyte does not change significantly, although the absolute abundances are escalating with the increasing degree of differentiation. The REE patterns of the rhyolite are characterized by a pronounced negative Eu anomaly (average Eu/Eu* = 0.08), consistent with low-pressure fractional crystallization of alkali and/or plagioclase feldspar. Primitive mantle-normalized trace element patterns in both rock types (Fig. 7b) are fairly similar. Both are highly fractionated and peak at Th-Nb; however, the rhyolite displays strong, negative Ba, Sr, and moderately negative Eu anomalies, whereas the trachyte contains only slightly negative anomalies in these elements. The Nb/Ta ratios of the trachyte (14.27 ± 0.47) and rhyolite (13.91 ± 2.48) are the same, but slightly lower than that of the mantle (typically ~ 17.5; Sun and McDonough 1989).

The abundances of U and Th in the felsic rocks of the Cumberland Hill Formation are high and variable. Rhyolite display significantly higher contents of U (14.0 ± 4.0 ppm, n = 8), and Th (33.0 ± 3.1 ppm, n = 8) compared to trachyte (U = 2.4 ± 2.4 ppm, n = 7; Th = 12 ± 1.4 ppm, n = 7). The large range in U may be related to devitrification and alteration of the rhyolite groundmass. The rhyolite samples also have a higher U/ Th ratio (0.44 ± 0.13, n = 8) than trachyte (0.32 ± 0.09, n = 7), which is comparable to the whole rock or melt inclusion data from non-mineralized peralkaline rhyolite from various U districts in the world (Cuney and Kyser 2009). The increase of U/Th ratio in the rhyolite compared to trachyte may reflect monazite fractionation. Based on the variability of the U and Th concentrations in the rhyolite and trachyte samples, we interpret that remobilization of these elements has occurred. This remobilization was likely caused by devitrification of the groundmass because zircon is modally insignificant, and not the principal host of these elements based on the lack of correlation of radioactive trace elements with Zr (see Dupuy and Dostal 1983).

Overall the geochemical and mineralogical similarities of the spatially and temporally associated trachyte and rhyolite imply that they are genetically related. However, discontinuous and/or contrasting variation trends and significant differences in incompatible trace element ratios indicate that the rhyolite was not derived from the trachyte via continuous fractional crystallization, although fractional crystallization may be responsible for the variations within the individual suites. The evolution of the trachyte in the melt phase involved crystallization of feldspars, clinopyroxene, and Fe-Ti oxides, whereas the evolution of the rhyolite was dominated by crystallization and fractionation of alkali feldspar.

An alternative model to continuous fractional crystallization for the origin of the rhyolite is assimilation-fractional crystallization. However, the low and overlapping Th/Ta ratio of the trachyte and rhyolite (Fig. 8a), a sensitive indicator of crustal contamination (Gorton and Schandl 2000), demonstrates that assimilation-fractional crystallization and crustal contamination processes cannot solely account for derivation of the rhyolite from the trachyte. In addition, Pe-Piper and Piper (1998) reported an ε_Nd value of +3, with an assumed age of ca. 310 Ma from the Cumberland Hill rhyolite. This relatively high value implies that significant assimilation of crustal material is not likely to have taken place.

K/Rb ratios in trachyte samples are much higher than in rhyolite samples (Fig. 9a, b), which suggest that the evolution of the rhyolite may have been affected by fluids (Dostal and Chatterjee 1995). Again, this mechanism cannot explain the origin of the rhyolite, but rather indicates that fluids played an important role during the evolution of these rocks and perhaps even led to the mobilization of U (Cuney and Kyser 2009; Nash 2010) in the rhyolite (Fig. 9b).

A comparison with similar rocks elsewhere (e.g., Pecerrillo et al. 2003) suggests that the Cumberland Hill trachyte and rhyolite could represent various stages of fractional crystallization in an evolving alkali basaltic magma. The absence of a negative Nb anomaly suggests that the parent magmas were not derived from subcontinental lithospheric mantle, but more plausibly originated from asthenospheric mantle. This interpretation is supported by the rhyolite ε_Nd value of +3 (Pe-Piper and Piper 1998), which implies upward migration of enriched asthenospheric magma into a juvenile lithosphere.

LaFleche et al. (1998) and Pe-Piper and Piper (1998) described alkali basaltic rocks of comparable age from other parts of the Maritimes Basin. The presence of alkali gabbroic rocks of a similar age (Johnson 2008) in the area suggests that the felsic rocks of the Cumberland Hill Formation could be derived from such parent magma by fractional crystallization. In fact, peralkaline felsic volcanic complexes are typically associated with a shallow-seated alkali intrusions ranging in composition from gabbro to highly fractionated granitic rocks. Although geochemical data are insufficient from the mafic alkali rocks in the vicinity of the Cumberland Hill Formation, chemical analyses of alkali basaltic rocks reported by LaFleche et al. (1998) from the Magdalen Islands indicate that similar basalts could have been a parent to trachyte and rhyolite. Trachyte of similar compositions have been documented to evolve from alkali basalt by extensive fractional crystallization (Peccerillo et al. 2003) and peralkaline rhyolite could have evolved from such magma by further fractional crystallization in a subvolcanic magma chamber. A likely scenario for the Cumberland Hill felsic rocks is that they are the products of fractional crystallization of alkali basaltic magma, and that the trachyte and rhyolite represent separate magma pulses with the latter being younger having undergone more extensive fractionation prior to emplacement.

Silica-oversaturated peralkaline felsic volcanic rocks, particularly rhyolite, are commonly enriched in U and represent a potential source of uranium for volcanic-related hydrothermal U deposits, such as the Streltsovka caldera (Transbaikalia, Russia: Ischukova 1997; Chabiron et al. 2001, 2003; Cuney and Kyser 2009; Nash 2010), sediment-hosted U deposits in siliciclastic sedimentary basins, such as the U deposits in the Tim Mersoi basin in Niger (Forbes et al. 1984; Forbes 1989; Plant et al. 1999; Pagel et al. 2005; International Atomic Energy Agency 2009), and the New Horton mineralization (U-Cu-Ag-Au) within the Carboniferous basin (Boyd 1978; McLeod and Smith 2010; New Brunswick Department of Natural Resources 2010b).

The crystallization of U-bearing accessory minerals in highly depolymerized peralkaline melts is typically suppressed (Cuney 2009; Cuney and Kyser 2009) leading to progressive enrichment of U in residual melts during fractional crystallization and entrapment of U in the glassy groundmass. Subsequently U can be leached during the alteration or devitrification of the glass (Nash 2010). Cuney (2009) and Cuney and Kyser (2009) inferred that ignimbritic tuffs, in which U is hosted in the glass, are the most favourable type of volcanic rock; such rhyolitic pyroclastic rocks are abundant in the Cumberland Hill Formation.

Rhyolite of the Cumberland Hill Formation has relatively high abundances of U compared to associated trachyte and could be a source of U mineralization. The U concentrations in these rhyolite samples are elevated compared to similar felsic and mafic Late Tournaisian–Early Visean deposits of the Maritimes Basin (Magdalen Islands, 0.4–1.9 ppm; Central New Brunswick, 0.8–1.9 ppm and the Cobequid Highland dykes, 0.2 ppm; Pe-Piper and Piper 1998). The abundances of U in the rhyolite are comparable to the non-mineralized U concentrations of the peralkaline rhyolite of the Streltsovka caldera (Fig. 7b), the largest, volcanic-related, U ore field in the world (Cuney 2009). Despite similar U concentration, it is important to consider where U is hosted in rhyolite. When U is hosted in zircon or other resistant accessory minerals it cannot be readily remobilized. Based on current geochemical work, it appears U is present within the glassy groundmass of the rhyolite in the Cumberland Hill Formation and can therefore be released and re-deposited and thus could be a source of U within U- (Cu-V-Ag) deposits within local parts of the Carboniferous basin.

Felsic volcanic rocks from the Cumberland Hill Formation consist of peralkaline rhyolite and trachyte. Both contain elevated concentrations of incompatible trace elements, with the rhyolite more enriched and fractionated than the trachyte. Although both rock types appear to be genetically related, differences in their geochemical signatures imply the rhyolite was not derived from the trachyte by continuous fractional crystallization. Examination of the Th/Ta ratio also indicated crustal contamination cannot account for the observed compositional contrast. A preferred model is that the trachyte and rhyolite originated from the same or highly similar parent magma, and represent separate magma pulses with the latter being younger, having undergone more extensive fractionation prior to emplacement. The absence of a negative Nb anomaly suggests the parent magma was not derived from subcontinental lithospheric mantle, but more plausibly originated from an asthenospheric mantle source. Lastly, Cumberland Hill rhyolite contains elevated U concentrations, with U hosted within the glassy groundmass. Since U can be leached and re-deposited, the rhyolite could be a source of sediment-hosted U mineralization within the basin.