Acid rock drainage issues in Nova Scotia, in particular in the Halifax Group, were broadly researched in the late 1980s and 1990s (e.g., Pasava et al. 1995; Feetham et al. 1997; Fox et al. 1997; Zentilli and Fox 1997, and references within) but have seen limited research since that time. One of the earliest and most extensively studied sites associated with ARD in HRM involved the construction and subsequent expansion(s) of the Robert Stanfield International Airport (previously referred to as the Halifax International Airport). Here a direct correlation was demonstrated between construction activity in the Halifax Group and significant fish kills in the nearby Shubenacadie River between 1960 and 1976 (Pasava et al. 1995; Worgan 1987). Remediation and treatment costs have averaged one million dollars annually at the airport, and treatment of surface waters continues (Hicks 2003).

Construction on Highway 107 east of Dartmouth near West Petpeswick Lake during the late 1980s exposed a continuous section of Halifax Group slate which resulted in the production of ARD and the degradation of water quality and aquatic habitat. Subsequently, the acid-producing outcrop was capped with shotcrete and a small treatment facility was constructed (Fox et al. 1997). Other historic ARD issues that have arisen in HRM and elsewhere in Nova Scotia also were outlined in Fox et al. (1997).

These incidents led the Nova Scotia Department of Environment to instate the "Sulphide Bearing Material Disposal Regulations" which require testing for ARD prior to ground disturbance (Nova Scotia Environment Act 1995). However, within HRM evidence suggests that current development in Fairview, Hammonds Plains, and Middle Sackville continues to have ARD problems. The local residents in these areas who are on well-water have water soſteners and filters to combat the resulting rusty stain on fixtures, discoloured laundry, and sulphur odour (Goodwin 2004). In addition, past and on-going infrastructure development such as the interchange of highways 102 and 103 as well as road building and the development of retail commercial sites in the Bayers Lake Industrial Park continue to be sources of natural ARD and acidification of local lakes (Kerekes et al., 1984; Rajaratnam 2009). More recently, construction was halted during the building of the new Dartmouth bus terminal and Bayers Lake underpass as pyritic slate was encountered on these sites, dramatically increasing project costs (Talpin 2011a, b; Stephenson 2011).

The portable X-5000 XRF provides rapid and simultaneous analyses of several elements with minimal sample preparation. To assess the reliability of analyses using the portable XRF the results are compared with the ACME results on the same samples. In addition, two internal reference samples (metasandstone and slate from the Goldenville and Halifax groups, respectively) were routinely analyzed to evaluate the portable XRF’s performance. This check aids in estimation of the precision and accuracy and hence ensures meaningful interpretation of the data. Scatter plots for selected elements (S, Ca, and As) from ACME and the portable XRF results were constructed and compared to the total S analyzed by BC-RIT (Fig. 3). Correlation coefficients (r-values) on each pair were also calculated. The results show excellent reproducibility (r > 0.95). Comparing the results from ACME to the portable XRF data yielded correlation coefficients that range from 0.88 to 0.98 for K2O, TiO₂, MnO, Fe₂O₃^T, Ba, Cu, Nb, Pb, Rb, Sr, V, Y, Zn, and Zr pairs (not shown). There is no significant discrepancy in the data between the ACME and portable XRF and hence the portable XRF results can be used with a significant degree of confidence.

In rocks of the Goldenville and Halifax groups away from the contact metamorphic effects of the South Mountain Batholith in the HRM area, regional greenschist facies (chlorite zone) metamorphic minerals and textures are preserved. Here the original clay-sized material in the slate as well as the matrix in metasandstone and metasiltstone has been recrystallized to a mixture of very fine-grained sericite, epidote, and chlo-rite. Quartz is the most abundant detrital component in the metasandstone, with lesser amounts of feldspar and minor or no lithic fragments (Fig. 4). Most feldspar grains are sub-rounded and consist of albitic plagioclase. It is rarely twinned and partially altered to sericite, epidote, or calcite. Potassium feldspar is rare. The metasandstone has between 18 and 38% matrix and is classified (using the system of Boggs, 2001, shown in Fig. 4a) as feldspathic wacke (Fig. 4b). Detrital muscovite is present in many samples and ranges in abundance up to 5%. Trace amounts of zircon, titanite, and tourmaline also occur. Lithic fragments are rare and are dominantly quartzite.

Calcite (CaCO₃), and to a lesser degree dolomite [Ca, Mg(CO₃)2], occur in nodules and as minor cement in metasandstone of the Taylors Head and Bluestone formations. Calcite nodules and laminations also occur within the slate of the Beaverbank and Bluestone formations, and rhodochrosite (MnCO₃) has been documented in units equivalent to the Beaverbank formation (O’Beirne-Ryan 1996). Ankerite [Ca(Fe, Mg, Mn)(CO₃)₂] is not common but has been noted in metasandstone units within gold districts of the Goldenville Group (MacDonald 1998).

Magnetite and minor ilmenite are the main magnetic minerals in the metasandstone and metasiltstone beds (King 1997; Pelley 2007). They form small (0.1–0.2 mm), inclusion-rich to inclusion poor, euhedral crystals in the matrix and are meta-morphic in origin; however, the more magnetic beds contain magnetite up to 2 mm in diameter, which is likely partly detrital in origin (Pelley 2007).

Cubic pyrite (1 to 2 cm in diameter) is typically rare in metasandstone of the Taylors Head formation but more abundant in the slaty rocks and metasandstone of the Cunard formation and locally is present in the Bluestone formation at lower metamorphic grades. Based on petrographic evidence the pyrite predates the regional foliation (e.g., Hicks 1996).

Pyrrhotite is not common in metasandstone beds in the Taylors Head formation but does occur locally with magnetite in the minor interbedded metasiltstone and slate. In the Beaverbank formation pyrrhotite typically occurs in the cores of spessartine garnet grains (e.g., Fox et al. 1997; White 2010a). Pyrrhotite, along with pyrite, is common in the Cunard formation (e.g., Pasava et al. 1995) and locally present in the Bluestone formation. It can occur as relatively inclusion-free porphyroblasts but is typically inclusion-rich, flattened parallel to foliation, and elongate parallel to the intersection lineation and regional fold axis, indicating it grew during regional chlorite-grade metamorphism that was synchronous with regional deformation.

As metamorphic grade increases toward the contact with the South Mountain Batholith, the matrix minerals in the metasandstone beds are comprised of coarser grained white mica and decussate biotite and ultimately poikiloblastic cordierite at the contact. In these less aluminous metasandstone beds, andalusite and sillimanite are rare. In the pelitic units of the Taylors Head and Beaverbank formations the matrix material also becomes coarser grained and, like in the metasand-stone, decussate biotite appears. Closer to the contact with the South Mountain Batholith, biotite content increases and ovoid cordierite appears; it has typically been weathered out leaving tiny voids in the rock. Plagioclase compositions become increasingly Ca-rich with higher An contents (up to An₅₀) at the contact with the South Mountain Batholith (Mahoney 1996; White 2003; Clarke et al. 2009).

Jamieson et al. (2005 a, b) documented a change in the opaque mineralogy with increasing metamorphic grade where rutile and pyrite in the outer aureole are replaced by ilmenite and pyrrhotite closer to the contact. As noted by Clarke et al. (2009), pyrrhotite is the dominant sulphide in the contact aureole adjacent to the SMB, forming up to 15 modal % in some layers in the Halifax Group (Betts-Robertson 1998). Like in the lower grade metamorphic rocks, pyrrhotite remains flattened and elongate parallel to the relict foliation and intersection lineation. The pyrrhotite is typically anhedral and contains abundant inclusions of quartz, sheet silicates, Fe-Ti oxides, and other sulphide minerals that define inclusion trails parallel to the relict foliation in the matrix (i.e., Fox et al. 1997; Clarke et al. 2009). In contrast, pyrrhotite in the metasandstone beds of the Cunard formation is cubic (up to 1 cm in diameter) because it has pseudomorphed pyrite (Clarke et al. 2009).

In the shear zone at the contact between the SMB and country rocks, the intersection lineation becomes steeper and forms a mineral lineation defined by elongate quartz+biotite ± cordierite aggregates and elongate calc-silicate nodules in the metasandstone or boudinaged andalusite and elongate pyrrhotite in the pelitic rocks (Culshaw and Bhatnagar 2001; White et al. 2008). In contrast to the Taylors Head formation, the main magnetic mineral is pyrrhotite in the Beaverbank, Cunard, and Bluestone formations. Other minor sulphide minerals associated with pyrrhotite, typically as inclusions, include chalcopyrite, sphalerite, arsenopyrite, and galena. Although pyrrhotite is the dominant sulphide in some layers (Fox et al. 1997: Clarke et al. 2009), no samples collected in the study contain over 5 modal % total sulphide minerals.

The Quarry Lake granodiorite (Fig. 2) is medium- to coarse-grained, equigranular to megacrystic, with alkali feldspar and less abundant plagioclase megacrysts. It has abundant metasedimentary xenoliths of the Goldenville and Halifax groups. Quartz forms anhedral to subhedral grains in the ground-mass but in places forms subhedral megacrysts. Plagioclase is typically in the groundmass as subhedral crystals displaying normal to oscillatory zoning, and ranging in composition from An 5–35, whereas alkali feldspar occurs as euhedral perthitic megacrysts (e.g., MacDonald 2001). Biotite is mainly subhedral to euhedral in the groundmass and forms up to 25 modal percent. Muscovite, cordierite, and garnet occur in trace amounts. The monzogranite and leucomonzogranite units to the southwest (Fig. 1) display textures similar to the granodio-rite but contain more muscovite and cordierite and less biotite (MacDonald 2001).

Pyrrhotite and chalcopyrite are the main sulphide minerals in the Quarry Lake granodiorite, together with trace amounts of sphalerite, galena, arsenopyrite, and molybdenite which typically occur as inclusions in the pyrrhotite (e.g., Clarke et al. 2009). Although the pyrrhotite may retain a similar flattened morphology as in the adjacent country rock, it is inclusion-free, small (<2 mm in size), and in some samples appears to be interstitial. The overall modal abundance of sulphide minerals in the batholith is less than 2%, considerably less than in the adjacent country rocks but can be several volume % of the granite at the contact (Clarke et al. 2009). No sample of granitoid rock collected during the present study contains over 1% sulphide minerals.

A detailed description of sulphide mineral textures and chemistry in the contact aureole and adjacent granitoid rocks of the South Mountain Batholith was provided by Clarke et al. (2009).

Major element oxide and trace element data can provide useful information on ARD prediction. Because the ARD behaviour of a rock is depended on its mineral assemblage, the effects of changes in mineralogical composition on acid-rock generation and buffering reactions must be considered when predicting ARD. Because the portable XRF does not measure elements lighter than Na, data from ACME is used for SiO₂, Al₂O₃, K₂O, Na₂O, and Fe₂O₃^T plots (Table 1a, Fig. 5).

Variations in SiO₂ content and SiO₂/Al₂O₃ ratio can be linked to the relative proportions of quartz, feldspar, and matrix minerals in the samples. White and Barr (2010) demonstrated this relationship for metasedimentary rocks in the Goldenville and Halifax groups in southwestern Nova Scotia. In the present study, petrographic data also were compared to their chemical compositions (Table 1a) and the results indicate that the boundary between sandstone and siltstone/mudstone is at 65% wt.% SiO₂ with a corresponding SiO₂/Al₂O₃ ratio of 3.5 (Fig. 5a), identical to the results of White and Barr (2010) from elsewhere in Meguma terrane. Samples with more than 65% SiO₂ when plotted on the chemical discrimination diagram of Herron (1988) are wacke to litharenite, whereas those with less than 65% SiO₂ are mainly shale to Fe-rich shale (Fig. 5b). On the chemical classification of sandstone diagram of Pettijohn et al. (1987), samples with more than 65% SiO₂ plot mainly in the litharenite field (Fig. 5c).

Metasandstone, siltstone, slate, and calc-silicate nodules were analyzed in each formation and plotted according to strati-graphic position against several chemical parameters (Figs. 6 and 7). The Taylors Head formation is typically low in CaO and MnO (Fig. 6a, b) and there is no difference between metasand-stone and metasiltstone/slate. However, metasandstone samples with a carbonate matrix/cement are slightly higher in CaO with calc-silicate nodules having up to 23 wt. % CaO and no significant MnO (Fig. 6a, b). Fe₂O₃^T tends to be higher in the metasiltstone and slate samples than in metasandstone samples (Appendix A–D). The metasiltstone-rich Beaverbank formation is higher in CaO and MnO with the metasandstone having the lowest concentrations and the metasiltstone and slate slightly higher (Appendix A, B). The coticule nodules and layers are significantly higher in MnO, CaO, and Fe₂O₃^T (Fig. 6a, b, c) but that correlation is not 1:1 as some metasiltstone samples with high CaO are low in MnO and Fe₂O₃^T. In contrast, all samples in the overlying Cunard formation display low CaO and MnO but Fe₂O₃^T remains high. Metasandstone and metasiltstone samples from the Bluestone formation are somewhat similar in CaO, MnO, and Fe₂O₃^T to those from the underlying Cunard formation; however, the calc-silicate nodules are significantly higher in CaO and only slightly higher in MnO and Fe₂O₃^T (Fig. 6a, b).

Total weight percent sulphur also varies with stratigraphic position and rock type, with median values of 0.04 in the Taylors Head formation, 0.32 in the Beaverbank formation, 1.29 in the Cunard formation, 0.17 in the Bluestone formation, and 0.01 in the granitoid rocks of the South Mountain Batholith (Fig. 7a; Appendix A–E). The higher values in the Beaverbank, Cunard, and Bluestone formations are attributed to abundant pyrrhotite. Some of the highest values (> 5 wt. %) are in hornfelsic samples from around the South Mountain Batholith. Typically there is a positive correlation between total sulphur and total metals (Cu + Pb + Zn) (e.g., Hammarstrom et al. 2003) but only the Cunard formation displays this trend (Fig. 7b). However, Pb, Zn, Cu, and As show no significant differences with stratigraphic position and all of the anomalous samples are located in the contact aureole of the South Mountain Batholith (Fig. 7c, d, e, f).

The chemical change at the contact between the Taylors Head and Beaverbank formations marks a dramatic and rapid change in the original seawater chemistry where conditions were favourable for deposition of Ca, Mn, Fe, and S. Under anoxic conditions Mn-oxide served as the oxidant, thereby leading to the eventual precipitation of manganese carbonate near the sediment-water contact (Graves and Zentilli 1988). These conditions are also favourable for sulphide mineral precipitation. The Beaverbank formation contains some of the highest total carbon values compared to the other formations (Table 1a).

The Beaverbank formation and equivalent Moshers Island formation elsewhere in the Meguma terrane are also enriched in Pb, Zn, Cu, and As (Graves and Zentilli 1988); however, the Taylors Head, Cunard, and Bluestone formations are also locally enriched in these elements (e.g., White 2010b). In the HRM area this enrichment appears to be associated with contact metamorphic effects from the South Mountain Batholith as indicated by the increased presence of chalcopyrite, galena, sphalerite, and arsenopyrite in the contact aureole (Clarke et al. 2009).

In Nova Scotia, the Sulphide Bearing Material Disposal Regulations state that rocks having sulphide sulphur contents equal to or greater than 0.4 weight % must be considered as hazardous ARD material. The portable XRF results, like the results from Acme Analytical Laboratories, are presented as total weight % sulphur which, based on the data in Table 2, is about 17% higher than total sulphide sulphur, except in samples from the Cunard formation where the ratio is almost 1:1 (Table 2). Although assigning all sulphur to sulphides is perhaps overly conservative as it ignores contributions from non-acid generating phases (e.g., sulphates and organic matter), the total weight % sulphur and total sulphide sulphur can be considered interchangeable; hence, a rock with total sulphur content equal to or greater than 0.47 can be considered hazardous (Fig. 7a). However, this parameter does not take into account the possible neutralizing capacities of minerals in the rock (Weber et al. 2005) or the varying reactivities of different sulphide minerals.

The most important acid neutralizing minerals in the Gold-enville and Halifax groups are calcite and dolomite and, to a much lesser extent, rhodochrosite and ankerite. The neutralizing power of feldspar varies considerably, depending on its composition. Altered (i.e., sericitized) feldspar has much less neutralizing power than its unaltered or carbonatized equivalents. Potassium-rich feldspar (orthoclase/microcline) and Na-rich plagioclase (albite) have less neutralizing power than the more Ca-rich plagioclase (anorthite). The metasandstone in the lower grade regional metamorphic rocks typically contains albitic plagioclase; however, in the contact metamorphic aureole plagioclase (10–40 modal %) is typically more Ca-rich (oligoclase to andesine) which adds to the neutralizing power. Minerals containing Mg (biotite and chlorite) also contribute to buffering ARD, but play a minor role (Weber et al. 2005). In the Goldenville and Halifax groups it is likely that the abundance of plagioclase and its alteration products, along with carbonate minerals, largely determine the neutralizing power in high sulphur samples. In comparison, the granitoid rocks of the SMB have total sulphur less than 0.20 weight % and pose no apparent ARD risk (Fig. 8; Appendix E).

The potential for acid generation is dependent on the ratio of acid-producing potential (APP) to acid-consuming ability (ACA). Following the legally binding British Columbia Research Initial Test the neutralizing power of 23 samples has been calculated (Table 2) and plotted (Fig. 9a). A 1:1 ratio (acid/ base) is plotted and assumes that the tested samples that plot below this ratio are net acid consumers (e.g., Schoeffer and Clawson 1996). However, Lawrence and Scheske (1997) noted that different mineral groups have different relative reactivities (i.e., carbonate minerals react faster than feldspar minerals), and hence the control on neutralization varies with mineral-ogy and a more suitable value for the neutralizing potential is 0.54. Because it is difficult to predict the short- and long-term acid generation potential of a rock unit a "zone of uncertainty" was defined between the 1:1 and 1:2 gradient lines (Fig. 9a, b) (Bell and Bullock 1996; Campbell et al. 2001).

Total S wt% versus CaO wt% can be considered equivalent to APP/ACA in a theoretical sense (e.g., Downing and Made-isky 1997) and it has been shown that data from the portable XRF is comparable to Acme Analytical Laboratories and British Columbia Research Initial Test results (Fig. 3). Hence, a plot of total S wt. % versus CaO wt. % for data collected from the portable XRF can be used with a significant degree of confidence as a first approximation for acid generating potential of the rock.

Figure 9b shows a wide distribution of sample populations which indicates that three units are potentially acid producers (Bluestone, Cunard, and Beaverbank formations), and two units are non-acid producing (Taylors Head formation and South Mountain Batholith). About 97% of the samples collected from the Cunard formation are potential acid producers with an APP/ACA and S/CaO ratios of >1.0 (Fig, 9a, b) and a high median total sulphur content of 1.31 %. About 3% of the samples are non-acid producers. In comparison, 26 % of the samples in the stratigraphically higher Bluestone formation and 24 % of the samples from the underlying Beaverbank formation are potential acid producers with average total sulphur of 0.30% and 0.90%, respectively. However, about half (50–55%) of the samples from these two formations lie in the zone of uncertainty and hence can be acid producers under the right conditions. The high sulphur percentage and acid producing potential in the Bluestone, Cunard, and Beaverbank formations is consistent with major basin-wide stagnation in which anoxic conditions existed during the depositional history of these formations (Waldron 1987; Graves and Zentilli 1988). The high proportion of acid-producing rocks in the Cunard formation can be directly linked to the low acid-buffering CaO content (median = 0.09 wt. %) compared to the Bluestone and Beaverbank formations (median = 0.39 and 1.29 wt. %, respectively) (Appendix A–C).

About 99% of the samples from the Taylors Head formation are non-acid producing (Fig. 9b) and have a low total sulphur content (median = 0.04 wt. %) (Appendix D). One sample (D12-W09-101) that is a potential acid producer has low total sulphur (0.023 wt. %) and low CaO (0.01 wt. %) and is one of two samples that is classified as quartz arenite (Fig. 4b) and hence is not likely to produce acid due to the high quartz content. The other sample classified as quartz arenite (D12-W07-015) is a spotted metasiltstone which contains 0.86 wt. % sulphur (Appendix D). It is from a thin (< 20 cm thick) bed within a massive metasandstone bed and, although this sample has the potential to produce acid, the effects would likely be diluted by the surrounding non-acid producing metasandstone.

About 95% of the granitoid samples from the South Mountain Batholith are non-acid producing (Fig. 9b) and have the lowest average total sulphur content (0.017 wt %) of all the units sampled (Appendix E). One sample (CW-270B) plots in the zone of uncertainty and has a total sulphur content of 0.859 wt. %. It has an elevated sulphur percentage compared to the other granitoid samples because it is a pyrite-bearing aplite-pegmatite dyke cutting the South Mountain Batholith. Because this is a minor unit the acid-producing potential is considered as low.

A useful application of susceptibility measurements is to determine the presence of sulphide-bearing minerals in out-crop. High magnetic susceptibility measurements are indicative of the presence of ferromagnetic and paramagnetic minerals (e.g., King 1997; Howells and Fox 1998; Fitzgerald and Goodwin 2005). In the Taylors Head formation the high magnetic susceptibility measurements are due to the presence of magnetite and to a lesser degree ilmenite. In contrast, the high readings in the Beaverbank, Cunard, and Bluestone formations are due to the presence of pyrite in the lower metamorphic grade rocks and pyrrhotite in the higher metamorphic grade contact aureole of the South Mountain Batholith (Fig. 10). The susceptibility data aid in delineating problematic, potentially acid-producing units before excavation takes place. However, detailed petrographic analysis is required to determine the character of the magnetic mineral. As shown in Figure 10, the metasandstone in the Taylors Head formation yields high magnetic susceptibility measurements but based on other data this rock type is low risk in terms of producing acid rock drainage.