Mobility of palladium in the surface environment: data from a regional lake sediment survey in northwestern Ontario

Transcription

1 Mobility of palladium in the surface environment: data from a regional lake sediment survey in northwestern Ontario Eion M. Cameron 1 & Kéiko H. Hattori 2 1 Eion Cameron Geochemical Inc., 865 Spruce Ridge Road, Carp, Ontario, Canada, K0A 1L0 ( eioncam@attglobal.net) 2 Department of Earth Sciences, University of Ottawa, Ottawa, Canada, K1N 6N5 ( khattori@uottawa.ca) ABSTRACT: This paper describes part of a project to examine geochemical methods in exploration for Pd deposits in northwestern Ontario. Palladium is one of the platinum group elements (PGE) that is mobile in dilute surface waters as hydroxy complexes, a feature that can facilitate its use in exploration. Earlier studies around the Lac des Iles Pd mine have shown that Pd is widely dispersed in solution even from sulphide-poor mineralization (Hattori & Cameron 2004). The region that includes Lac des Iles has been covered by a lake sediment survey by the Ontario Geological Survey (Dyer & Russell 2002). These lake sediment data provide a further means of better understanding the geochemistry of Pd in the surface environment and its application to exploration for PGE deposits. Factor analysis of the lake sediment data reveals three associations: Factor 1, an association of Cr, Co, Ni, Pt and Pd; Factor 2, an association of Pd, S and As; and Factor 3, an association of Cu, Ag and Au. High scores for Factor 1 extend for 10 km down-ice from the mineralized mafic rocks of Archean age at Lac des Iles and reflect the dispersion of clastic material by ice sheets, supplemented by hydromorphic dispersion. High scores for Factor 2 are much more widespread and are spatially associated with Quaternary glaciofluvial deposits, notably eskers. A plot of Pd alone in lake sediments throughout the area largely reflects this association with eskers. The elemental assemblage of Factor 2 originates from the leaching of fragments of Nipigon diabase (Pd) and Sibley sedimentary rocks (S, As), both of Proterozoic age, contained within the eskers. The high permeability of the eskers permits soluble elements to be leached. All three elements, Pd, S and As, form neutral and anionic complexes and are not adsorbed by negatively-charged iron oxyhydroxide colloids and coatings, which restrict the dispersion of elements that dissolve as cations, such as Cu, Ni and Co. Factor 3, Cu, Ag and Au, occurs in areas of Proterozoic bedrock. This paper is part of the study: Geochemical Detection of Disseminated Palladium Mineralization in North-Western Ontario commenced in 2002 and funded by the Ontario Mineral Exploration Technology program. Our initial field studies focussed on the Lac des Iles area, where sulphide-poor, Pd-rich platinum-group elements (PGE) mineralization of Archean age is mined. Palladium is the only member of the PGE that is significantly mobile in dilute surface waters. It forms hydroxyl complexes, Pd(OH) 2, Pd(OH) 3 and Pd(OH) 4 2. This mobility favours the use of Pd in geochemical exploration for PGE mineralization. Nevertheless, it was not known if the Pd would be mobilized during weathering of the Lac des Iles-type ore, because of the paucity of associated sulphide. The results of the field and laboratory studies show that Pd is, indeed, mobilized from the ore and is carried in solution to be fixed with organic material in swamps and in lake sediments (Hattori & Cameron 2004). The region that includes the Lac des Iles deposit has been covered by a high-density survey of lake sediments and waters (Fig. 1) carried out by the Ontario Geological Survey Geochemistry: Exploration, Environment, Analysis, Vol , pp (Dyer & Russell 2002). The availability of these highquality data provides an additional means for understanding the geochemistry of Pd in the surface environment. This paper utilizes these data, combined with follow-up visits to anomalous sites, to evaluate factors controlling the distribution of Pd. The area covered by the lake sediment survey is shown in Figure 1. The western half is mainly underlain by Archean rocks. These comprise a variety of metasedimentary and metavolcanic rocks intruded by granitic rocks and by less abundant mafic and ultramafic rocks. The area includes parts of the Wabigoon and Quetico subprovinces of the Superior province. The eastern part of the lake sediment survey area has rocks of Proterozoic age: the Sibley Group of sedimentary rocks, 1537 Ma (Sutcliffe 1991), and Nipigon diabase sills, 1109 Ma (Sutcliffe 1991). The lake sediment data considered in this paper are confined to that sector between West and West. This sector contains the Lac des Iles Pd mine and a number of showings of Pd mineralization in mafic and ultramafic intrusions of Archean age /03/$ AEG/ Geological Society of London

2 300 E.M. Cameron & K.H. Hattori Fig. 1. Region in northwestern Ontario covered by lake sediment survey of Dyer & Russell (2002). The area discussed in this paper lies between and W, and N. GEOLOGICAL SETTINGS There are two sources for Pd geochemical anomalies in the survey area. The source of most economic significance is a suite of late Archean mafic-ultramafic rocks (Fig. 2) forming a circular pattern, c. 30 km in diameter, with aeromagnetic anomalies, the largest of which is the Lac des Iles complex. These intrude tonalitic and volcano-sedimentary rocks of the Lac des Iles greenstone belt. The Lac des Iles Complex hosts the Lac des Iles Pd mine of North American Palladium Ltd., which has been in operation since 1993, with a current reserve of 159 million tonnes grading at 1.55 g/t Pd, 0.17 g/t Pt, 0.12 g/t Au, 0.06% Cu, and 0.05% Ni at a Pd cut-off grade of 0.7 g/t (Lavigne & Michaud 2001). Other intrusions hosting mineralization with greater than 2 ppm Pd include the Legris Lake, Buck Lake, Tib Lake, Demars Lake, Wakinoo Lake and Towle Lake complexes (Fig. 2). Palladium mineralization is restricted to gabbroic rocks that show textural complexity, such as magma mingling and brecciation (Sutcliffe et al. 1989; Lavigne & Michaud 2001; Pettigrew & Hattori 2002; Hinchey et al. 2003). Homogeneous gabbros and ultramafic rocks are largely barren, such as voluminous ultramafic rocks in the northern part of the Lac des Iles Complex, hornblende gabbro in the southern part of the Complex, and leucogabbro in the centre of the Legris Lake Complex. Mineralization in the area is characterized by low contents of sulphide minerals, usually less than 3 vol. %. Sulphides are mainly pyrrhotite and chalcopyrite with microscopic grains of pentlandite. Pyrite is not common in mineralized zones, but is locally abundant in rocks that have undergone late alteration. Sulphides form aggregates less than 1 cm in size and disseminated in silicate minerals without apparent veining and patterns. Accordingly, surface exposure of the mineralization does not produce gossans, but minor stains of Fe oxides and hydroxides are locally observed. The second source of Pd geochemical anomalies is the Nipigon diabase and related mafic and ultramafic rocks of Proterozoic age. The Nipigon diabase is part of mafic magmatism associated with the Midcontinental Rift at 1.1 Ga. Like other rift-related magmatic episodes, mafic rocks associated with the Midcontinental Rift contain high PGE and host PGE mineralization. They include the Duluth Complex, Crystal Lake Gabbro, Coldwell Complex and Seagull Lake intrusions. As shown in Figure 1, Proterozoic diabase and related intrusions are mainly exposed in the eastern part of the total lake sediment survey area. Quaternary deposits have a substantial influence on the distribution of geochemical anomalies. Much of the area shown in Figure 3 has a cover of till, which may range from thin and discontinuous to tens of metres in thickness. There are also glaciofluvial sediments, including prominent esker complexes. The distribution of the till and glaciofluvial sediments reflects a predominant ice direction from the NE to the SW. There are organic deposits (peat) scattered throughout the area. Topographic relief is low, usually less than 100 m for the area shown in Figure 3. NATURE OF LAKE SEDIMENTS This discussion is summarized from Cameron (1994). The principal sedimentary environments in lakes are deltaic, nearshore, and deep. Coarser particles, sand or silt, carried in stream and river water are deposited in deltas where these waters enter the lake. Near-shore sediments, derived by erosion of the shoreline and shallow bottom, are also relatively coarse, since

3 Palladium mobiility in the surface environment 301 Fig. 2. Geology of the area surrounding the Lac des Iles mine. Occurrences of PGE mineralization associated with mafic/ultramafic intrusions are shown by triangle. the action of waves and currents removes fine particles to deep water. It is the sediment of deep basins, commonly called gyttja, that is sampled for geochemical exploration purposes. This comprises clay- or fine silt-sized mineral and rock particles brought to the lake in suspension in stream waters, together with chemical precipitates produced within the lake by organisms and by inorganic reactions. Gyttja is a gel that becomes increasingly watery towards the sediment/water interface. Thomas (1969) found a good correlation between the clay and organic carbon content of sediment from deep basins. In lakes from the southern Canadian Shield, gyttja may be up to 5 m thick. There are three principal sources for the material forming lake sediment. The allogenic fraction is particulates transported by streams, as airborne dust, in rain, or derived by shore erosion. The mineralogical composition of this fraction largely reflects that of local bedrock or overburden and this applies to the clay-sized material, as well as to the coarser fractions. The endogenic fraction of sediments is formed within the water column, and includes biological remains. Whereas the distribution of the allogenic fraction is determined by physical factors in the drainage system, the endogenic fraction is chemically and biologically controlled. Despite this fundamental difference in origin, there are strong relationships between the allogenic and endogenic fractions. For example, the correlation noted above between organic matter and clay content is determined by similar settling rates. The Fe and Mn oxides that are precipitated in the water column commonly coat finegrained allogenic silicates. These oxide coatings, and organic coatings, provide important sites for the adsorption and co-precipitation of trace elements in solution. The complex nature of the relationship between the endogenic fraction and fine-grained allogenic material makes it hazardous to postulate, for example, that an element is organically-bound simply because its distribution correlates with that of the organic fraction. The authigenic fraction is formed within the sediment by diagenetic reactions. LAKE SEDIMENT SURVEY DATA Sampling and analytical methods are described by Dyer & Russell (2002). Samples of gyttja were collected by gravity corer from the float of a helicopter. To avoid anthropogenic influences and surface diagenetic effects, only material from >20 cm below the sediment surface was sampled. In some cases the endogenic fraction contains CaCO 3. The ph and conductivity of the water at each site was measured. Sediment samples were dried at <40 C, then disaggregated for 10 seconds in ceramic and sieved to minus 80 mesh (<177 µm). A variety of analytical methods were used. For the elements discussed here, one of the following three methods was involved: (1) 5 to 10 g of sample were analysed for Au, Pd and Pt by fire assay (FA) followed by ICP-MS measurement; (2) 0.5 g of sample was decomposed by nitric acid - aqua regia, the resulting solution being analysed by ICP-MS and ICP-ES; (3) 10 g of sample were pressed into a briquette and analysed by instrumental neutron activation analysis (INAA). Every 10 th sample was an analytical duplicate to measure precision and every 20 th sample was a standard. More than 50 elements were determined. For the factor analysis we have selected the 10 elements that are most closely associated with Pd mineralization. For the area shown in

4 302 E.M. Cameron & K.H. Hattori Fig. 3. The portion of the lake sediment survey area by Dyer & Russell (2002) between and discussed in this paper, showing (a) the occurrence of mafic and ultramafic rocks; (b) Quaternary glaciofluvial deposits, from Mollard (1979a, b), arrows indicating ice direction; (c) lake sediment sampling sites. Boxes in map (c): A: Lac des Iles area, Figures 7, 8 and 9; B: Towle Lake area, Figure 10; C: Eleph Lake area, Figures 12 and 13; D: Keni Lake area, Figure 14; E: Dog Lake area, Figure 11. Table 1. Summary data for ten elements selected for factor analysis from 675 lake sediment samples from the area shown in Figure 3. Precision estimates are from Dyer & Russell (2002). Element Analytical Method Detection Limit Precision Median Range Au (ppb) FA-ICP-MS <1 <1 to 35 Pd (ppb) FA-ICP-MS <0.3 to 47 Pt (ppb) FA-ICP-MS <0.3 to 6.6 Cr (ppm) ICP-ES <4to174 Cu (ppm) ICP-ES to433 Ni (ppm) ICP-ES to 133 S (ppm) ICP-ES to Ag (ppm) ICP-MS <0.02 to 0.43 Co (ppm) ICP-MS to 57 As (ppm) INAA <0.5 to 51 Figure 3, there are 675 samples with reported values for all of these elements, including Pd, Pt and Au by FA-ICP-MS. Summary data for the 10 elements in the 675 samples are shown in Table 1. The detection limits and estimates of precision are as reported by Dyer & Russell (2002), who also provide analyses of standard samples. FACTOR ANALYSIS Element distributions in nature are usually a mixture derived from several components. Factor analysis provides a way of separating and identifying the distributions associated with individual components. Figure 4 shows frequency distribution plots for Pd and Pt in the 675 samples of lake sediments using logarithmic axes, plots that demonstrate that both elements have a lognormal distribution. The other elements listed in Table 1 have similar lognormal distributions. Thus prior to factor analysis, all data were logarithmically transformed. A principal components matrix was extracted from the data and the three factors with eigenvalues greater than 1 were rotated using the Varimax procedure. In Table 2 Factor Matrix A is shown which comprises the 10 elements most closely related to

5 Palladium mobiility in the surface environment 303 Fig. 4. Frequency distribution plots for Pt and Pd in 675 samples of lake sediments. Values below the detection limit are set at half the detection limit. PGE mineralization. Factor Matrix B (Table 3) comprises the same 10 elements plus the elements Fe, Mn and Ca. The numbers comprising the factor matrices are termed loadings. The square of any loading represents the proportion of the total variance of that element that is accounted for by the given factor. Thus a loading of 0.5 represents 25% of the total variance of that element and 0.9 represents 81% of the total variance, whereas a loading of 0.1 represents only 1% of the Table 2. Factor Matrix A. Factor analysis of 675 lake sediment samples for the ten elements most closely associated with PGE mineralization. This table shows the Varimax rotated matrix for the 3 principal component factors with eigenvalues great than 1. All data were logarithmically transformed prior to analysis. Shown in bold are loadings of 0.50 or greater, which represent 25% or more of the total variance of the element. Element Factor 1 Factor 2 Factor 3 Co Cr Ni Pt Pd S As Cu Ag Au Table 3. Factor Matrix B. Factor analysis of 675 lake sediment samples for the ten elements most closely associated with PGE mineralization plus the elements Fe, Mn and Ca. This table shows the Varimax-rotated matrix for the 3 principal component factors with eigenvalues great than 1. All data were logarithmically transformed prior to analysis. Shown in bold are loadings of 0.50 or greater, which represent 25% or more of the total variance of the element. Element Factor 1 Factor 2 Factor 3 Fe Co Mn Cr Ni Pt S As Ca Pd Cu Ag Au total variance. Loadings of 0.5 or more are shown in bold in Table 2 and such loadings identify the elements that principally comprise each factor. The result of the factor analysis is relatively simple. For Factor Matrix A, Factor 1 is comprised mainly of the elements Co, Cr, Ni, Pt and Pd; Factor 2, Pd, S and As; and Factor 3, Cu, Ag and Au. In Factor Matrix B, the elements Fe and Mn are added to Factor 1 and Ca to Factor 2. Factor scores were computed from Factor Matrix A. For each sample this involves multiplying a vector of standard scores for the N elements by a matrix of factor score coefficients. Standard scores are normalized, with a mean of 0 and standard deviation of 1, and were computed using logarithmically-transformed data. Factor scores should preserve this lognormal distribution; the scores for each of Factors 1 to 3 have means of 0 and standard deviations of 1. Histogram plots for each set of factor scores (Fig. 5) show that they approximate a normal distribution, with moderate right-skewing for Factor 2. Figure 6 presents contour plots for the upper (anomalous) portion of the distribution for each set of factor scores. Factor 1 (Co, Cr, Ni, Pt, Pd) The principal area of high scores for this factor is southwest (down-ice) from the Lac des Iles mafic-ultramafic complex, including the Lac des Iles Pd mine. There is also a small area of high values at Towle Lake, near an inflection in a linear belt of mafic/ultramafic intrusions, which include the Stocker and Powder Hill PGE prospects. The third area of high values is around the north end of Dog Lake, in an area identified during reconnaissance mapping as comprised of granitic rocks (Ontario Geological Survey 1964). Smaller areas shown in Figure 6 with high scores for this factor have not been investigated. Figure 7 shows a larger-scale contour plot of Factor 1 scores in the area around Lac des Iles; Figures 8 and 9 show the bedrock and Quaternary geology of the same area. High values for Factor 1 extend over Lac des Iles and SW from Lac des Iles. The mafic/ultramafic complex at Lac des Iles comprises three domains (Lavigne & Michaud 2001). The largest is mainly ultramafic rocks centred on Lac des Iles. The lake is shallow, with many islands, and mapping on these islands indicates that much of the lake is underlain by ultramafic rocks. The second domain (Fig. 8) is the Mine Block intrusion of chaotic gabbroic rocks immediately south of the lake, which contains all known mineralization, such as the Roby, Twilight, Baker and Creek zones. The third domain is homogenous hornblende gabbro

6 304 E.M. Cameron & K.H. Hattori Fig. 5. Histogram plots of factor scores for the three factors shown in Factor Matrix A (Table 2). Fig. 6. Contour plots of factors 1, 2 and 3 together with outlines of the geological units most closely related to the factors. Quaternary geology from Mollard (1979a, b). forming the barren Camp Lake intrusion, located southwest of the lake of that name. The highest scores for Factor 1 are (a) within Lac des Iles, (b) SW from the southern part of the lake, in part along the drainage channel from the lake (Riviere Lac des Iles), and (c) from Shorty Lake, adjacent to the Baker Zone. The Factor 1 anomaly appears to reflect the rocks and mineralization from the Mine Block intrusion and the ultramafic northern domain. The barren Camp Lake intrusion shows no dispersion train. The Factor 1 distribution in the Lac des Iles area is interpreted to represent clastic dispersion of mafic material from the source rocks to the southwest by the ice sheets. Following the retreat of the ice sheet, fine, clay-sized material from this dispersal train was carried into the lakes and settled as an allogenic component of the gyttja. There was also weathering and leaching of elements from the tills on land, with these elements carried to the lakes to be precipitated as the endogenic component of the gyttja. Some of the elements comprising this factor, such as Ni and Co, are relatively mobile in the Shield environment, whereas others, such as Pt, are not. Searcy (2001) examined till samples from the Roby Zone at the Lac des Iles mine to 2.5 km down-ice (240, see Fig. 7) and found distinctive

7 Palladium mobiility in the surface environment 305 Fig. 7. Contour plot of scores for Factor 1 in the vicinity of Lac des Iles. Contours at intervals of 0.2 from 1.2 to 2.0. The shores of Lac des Iles and its drainage are shown by shadowed lines. Large arrow at 240 indicates ice direction from the Roby Zone, shown by a circle, which is currently being mined. Small arrows indicate drainage directions. Rectangular symbols are lake sediment sampling sites. Eastings and northings in metres. Cr-rich andradite garnets dispersed at 240 from the Roby Zone. We examined the till-covered area SW of Lac des Iles along the Riviere Lac des Iles. C-horizon till samples contain elevated concentrations of Pd, Ni and Cr. And the presence of numerous boulders of gabbroic rocks and clinopyroxenite similar to lithologies in the mineralized zones indicates glacial transport to the SW. Of the 675 lake sediment samples from the area studied, the highest score for Factor 1 (+2.4) occurs in Shorty Lake (Fig. 7), immediately adjacent to the undeveloped Baker Zone. The Pd component of this high score is due to hydromorphic movement of Pd from the Baker Zone. Much of the Pd migrating in solution from the Baker Zone is intercepted and fixed in organic material in swamps encircling the lake, the remaining Pd entering the lake. Humus from the swamps have Pd contents up to 105 ppb (Hattori & Cameron 2004) and the lake sediment contains 13.6 ppb Pd (Dyer & Russell 2002). The hydromorphic dispersion of Pd continues south along the drainage from Shorty Lake to Camp Lake, where the sediment contains 14 ppb Pd. However, Camp Lake does not have a high score for Factor 1 because it lies south of the glacial dispersion train (Fig. 7), which would have contributed Co, Cr, Ni and Pt, the other elements comprising Factor 1. The second anomalous area for Factor 1 scores is near Towle Lake (Fig. 10). There is a tight cluster of high values, including scores of 2.1 and 2.2, in samples taken from Towle Lake itself. The cluster occurs at an inflection in direction for the Towle Lake mafic complex. Rocks are poorly exposed in the area. Considering the continuous occurrences of Pd mineralization along the strike of the Towle Lake complex, it is possible that there is additional hidden PGE mineralization in the area. Given the assemblage of elements that form this factor, there also may be undiscovered ultramafic rocks. Scores >1.0 are scarce elsewhere in the area shown in Figure 10, including areas underlain by mafic rocks that are mineralized, such as mafic/ultramafic rocks in the southern part of Demars Lake. There are a number of PGE prospects within the area shown in Figure 10 associated with pyroxenite and leucogabbro. The third area of high scores for Factor 1 is at the north end of Dog Lake (Figs. 6, 11). The area has been interpreted on a reconnaissance scale as composed of granitic rocks (Ontario Geological Survey 1964). High scores are orientated along prominent lineaments striking E NE that crosses Creighton Bay, the northern tip of Dog Lake. Approximately on strike from this feature, near Fisher Lake, west of Dog Lake, and in an area considered to be granitic, drilling in 1970 intersected pyroxenite, sheared serpentinite and minor sulphide with anomalous values for PGE (B. Schnieders, pers comm., 2003). The lineaments may represent a splay off the Quetico fault to the south, which is an east-trending dextral shear zone that forms the boundary between the Wabigoon and Quetico subprovinces. Other unmapped occurrences of ultramafic rocks are known to occur in this area. Factor 2 (Pd, S, As) The distribution of high scores for Factor 2 (Fig. 6) has a strong spatial correlation with Quaternary glaciofluvial deposits, most notably eskers. To ascertain the nature of this correlation we visited two areas with high scores for this factor, Eleph Lake and Keni Lake. The Eleph Lake area (Figs. 12, 13) is dominated by sandy glaciofluvial deposits with a number of boulder-filled eskers with their axes orientated along the southwest direction of ice movement. There are few outcrops of bedrock, which are mainly of granite with minor mafic intrusive rocks. We examined the esker adjacent to Eleph Lake, exposed on the wall of an aggregate pit. Rounded boulders from 5 to 40 cm in diameter comprise much of the esker set in a matrix of pebbles and coarse sand. Boulders of diabase, which are considered to

8 306 E.M. Cameron & K.H. Hattori Fig. 8. Outline geology of the Lac des Iles area. Lac des Iles is a shallow lake with many islands. Mapping on the islands indicates that ultramafic rocks and subsidiary mafic rocks underlie much of the lake (Sutcliffe et al. 1989). Roby Zone is currently being mined. Geology modified from Lavigne & Michaud (2001). Eastings and northings in metres. be Nipigon diabase based on the characteristic textures and mineralogy, comprise more than half of the boulders. There are also boulders of reddish sandstone, recognised to be from the Sibley Group, and granite boulders. The closest outcrops of Sibley Group sedimentary rocks are 30 km to the east of Eleph Lake, i.e., up-ice flow. Many boulders of diabase have been thoroughly weathered and are disintegrated. We collected the least weathered diabase boulders, which were analysed for Pd, Pt and Au (Table 4). The contents of Pd and Pt in these boulders and in an outcrop of diabase north of Lac des Iles are substantially higher compared to mantle peridotites and much higher than oceanic basalts (Table 4). The second area investigated with high scores for Factor 2 is around Keni Lake (Fig. 14). A small, unnamed lake, Lake 553, with no stream inflows, drains into Keni Lake through a narrow swampy path. The sediment sample from Lake 553 has the highest score for Factor 2 of any sample in the entire area shown in Figure 6. The sediment sample from Keni Lake also has a high score for this factor. Other samples collected nearby have lower scores. Another feature of sample 553 is the high content of Ca as carbonate: essentially this lake sediment is a marl. On visiting the site we noted a large esker containing numerous bounders of Nipigon diabase ends at the south shore of Lake 553. Several springs emerge at the base of the esker and flow directly into the lake; the esker appears to be the principal source of water for the lake. We sampled the spring water and analyses for selected elements are shown in Table 5; for comparison, three samples of lake water from the Lac des Iles area and a sample from Eleph Lake are shown. The most obvious features of the waters from the spring at the base of the esker are the high Ca and low Cu content, features that it shares with the water from Eleph Lake, the sediment from which also has a high score for Factor 2. The data and setting of lakes suggest that the high Pd content of the sediment in Lake 553 and Keni Lake derive from the spring water emerging from the esker. Factor 2 is interpreted to represent elements, Pd, S, As plus Ca that have arrived in the lake in solution and have precipitated to form an endogenic component of the gyttja Factor 3 (Cu, Ag, Au) The samples with the highest scores for this factor are located northeast of Lac des Iles (Fig. 3), in terrain where the lakes are entirely surrounded by Nipigon diabase. As shown in Figure 1, only a small part of the area discussed in this paper between W and W is underlain by Nipigon diabase, whereas to the east, much of the area is of Nipigon diabase or related mafic and ultramafic rocks. In this latter area there are distinct zones within diabase terrain where lake sediments are anomalous in Cu, Ag and Au. The anomalies for these elements in Proterozoic terrane are not surprising considering numerous occurrences of past producing mines of Ag and Cu. Silver veins are mostly associated with diabase dykes and sills and Cu with minor Au deposits are associated with volcanic rocks and felsic intrusions (Sutcliffe 1991).

9 Palladium mobiility in the surface environment 307 Fig. 9. Quaternary geology of the Lac des Iles area modified from Mollard (1979a). Arrows indicate direction of ice flow. Eastings and northings in metres. Fig. 10. Contour plot of scores for Factor 1 in the vicinity of Towle Lake. Contours increase in increments of 0.2 from 1.0 to 2.0. Sites with scores of 1.0 or greater outside the contoured area are also shown. The patterned area is the Towle Lake mafic/ultramafic intrusion. Its location is interpreted from an aeromagnetic map compiled by Platinum Group Metals Ltd. Stars indicate occurrences of PGE mineralization and rectangular symbols are lake sediment sampling sites. Arrow indicates predominant ice direction within the illustrated area. Quaternary geology is mainly thin till over bedrock with peat in depressions (Mollard 1979a). Eastings and northings in metres. DISCUSSION In the Canadian Shield most lake sediments are mixtures of elements that have been transported in solution and in clastic form (Cameron 1994). Glacial action dispersed rock fragments down-ice. When the ice receded, fine-grained clastic materials from till and glaciofluvial deposits were carried to the lakes to form sediment. Only the finest, clay-sized particles, carried in suspension, were able to reach the centre of the lakes and deposited to become a component of gyttja. Chemical weathering of glacial deposits occurred on land and mobile elements were carried in solution to the lakes and precipitated. The Factor 1 anomaly, SW of Lac des Iles, and comprising the elements, Co, Cr, Ni, Pt and Pd, is an example of this two-stage process: dispersion by ice, followed by transport by water in suspension and solution to the lakes. Studies by Searcy (2001) and Hattori & Cameron (2004) over and near the Baker Zone PGE mineralization provide a method for evaluating the relative importance of clastic vs. hydromorphic processes in the transport of specific elements to the lakes. Searcy provided data

10 308 E.M. Cameron & K.H. Hattori Fig. 11. Plot of scores for Factor 1 in the vicinity of Dog Lake. Solid squares indicate lake sediment sampling site. Those sites with scores >0.0 are labelled. On the reconnaissance geology map (Ontario Geological Survey 1964) the whole area is interpreted to be granitic. However, there is an area of high aeromagnetic values (shaded) and the symbol UM marks the position where serpentinite and pyroxenite containing sulphides and anomalous PGE have been intersected during drilling. The interrupted lines indicate prominent lineaments that may represent a splay off the Quetico fault. Eastings and northings in metres. Fig. 12. Contour plot of Factor 2 scores of the area near Eleph Lake. Rectangular symbols are lake sediment sampling sites. Eskers indicated by herringbone pattern. Eastings and northings in metres. on C-horizon soils, which are comminuted fragments of the mineralization and host rocks. Hattori & Cameron provide data on humus from swamps that surround Shorty Lake beside the Baker Zone. The humus acts as a partial chemical trap for metals liberated from bedrock and soils in solution, which migrate down-slope to the swamps and lakes. A comparison of the C-horizon soils with the swamp humus (Fig. 15) shows that Pd contents are much higher in the humus of the swamps bordering the lake than the C-horizon soils, whereas Ni, Cr, Cu, and Cr are substantially lower. A similar plot shows that Co is lower in humus. These comparisons suggest that for the elements comprising Factor 1, transport in solution was more important for Pd, but less important for Cr, Ni, Co and Pt. By contrast, the elements comprising Factor 2, Pd, S, As, were derived by leaching of rock fragments in eskers and other glaciofluvial deposits and carried to the lakes in solution. Nipigon diabase boulders and fragments in the eskers provide a source for leaching Pd (Table 4). Although the Nipigon diabase is also enriched in Pt, and Pt may also form hydroxy complexes (Wood et al. 1992), the solubility of Pt is much lower than that of Pd and thus its mobility is restricted compared to Pd. Sibley sediments contain evaporitic rocks with abundant sulphate minerals. Lake sediment surveys in the eastern part of the area shown in Figure 1 show prominent As anomalies over areas of Sibley sedimentary rocks, indicating that fragments of Sibley rocks provide a source for As as well as S. Eskers form ridges and are permeable. Precipitation passes freely through the esker to the water table, then to the nearby lakes and streams. Palladium, S and As migrate either as neutral complexes or as anions, e.g., Pd(OH) 2, Pd(OH) 4 2, SO 4 2,

11 Palladium mobiility in the surface environment 309 Fig. 13. Quaternary geology of the area near Eleph Lake. Modified from Mollard (1979a). Eastings and northings in metres. Table 4. Contents of Pd, Pt and Au by FA-ICP-MS in samples of diabase boulders from the Eleph Lake esker and from a Nipigon diabase sill north of Lac des Iles. Also concentration of precious metals in mantle peridotites (Guillot et al. 2000) and average oceanic basalts (Crocket 2002). Coordinates given here and elsewhere in the paper conform to NAD 27. Sample Description Easting, m Northing, m Pd ppb Pt ppb Au ppb R Diabase boulder, Eleph esker R Diabase boulder, Eleph esker R Nipigon diabase outcrop Mantle peridotites Mid-oceanic ridge basalts H 3 AsO 3,H 2 AsO 4, HAsO 4 2. Many trace metals dissolve as cations (e.g. Pb 2+,Cu 2+,Ni 2+,Co 2+,Zn 2+ ), which become less soluble as the ph increases. At the near-neutral ph of most groundwaters, the solubility of trace-metal cations is limited by precipitation or adsorption to hydrous metal oxides, clay or organic matter, whereas oxyanions become less strongly adsorbed as the ph increases. Thus as water of near-neutral ph passes through the esker, cations tend to be retained, whereas anions and neutral species continue. Arsenic is a special case, since it can occur in two valency states in groundwater, As(III) and As(V). The former has high solubility as H 3 AsO 3. If groundwater containing H 3 AsO 3 enters a lake and the water become oxygenated, the As(V) species H 2 AsO 4, HAsO 4 2 are formed, which have a greater tendency to be adsorbed. On reaching swamps and lakes, trace metals are incorporated into organic matter or are otherwise precipitated. When sulphate is reduced to sulphides in such environments, Pd and As may be co-precipitated. CONCLUSIONS The high mobility of Pd in the surface environment confers an advantage to geochemical methods of exploration for PGE deposits. Palladium can migrate in solution to be fixed in swamp humus along drainage (Hattori & Cameron 2004) or lake sediments, permitting lower sampling densities for exploration than might be required for a less mobile element. However, this mobility has also led to the formation of lake sediment anomalies that are unrelated to mineralization: the result of the leaching of large volumes of permeable eskers that contain Pd-bearing rock fragments. High absolute values for Pd in the survey area (Fig. 16) are related mainly to the distribution of glaciofluvial deposits. Of the known occurrences of PGE mineralization in the area, only that at Lac des Iles shows as a distinct anomaly with clusters of values above 8 ppb Pd. Factor analysis of lake sediment data permits the discrimination of Pd leached from eskers from that originating from the PGE

12 310 E.M. Cameron & K.H. Hattori Fig. 14. Keni Lake area, showing selected data from lake sediment samples and ph of lake waters by Dyer & Russell (2002). Geology from Kaye (1965). Eastings and northings in metres. Table 5. Analyses of selected elements in waters from lakes and from a spring at the base of the esker that flows into Lake 553 near Keni Lake. Shorty Lake is adjacent to the Baker Zone Pd mineralization near Lac des Iles. Location Easting, m Northing, m Ca ppm Mg ppm Na ppm S ppm Cu ppm ph Lac des Iles < Lac des Iles Shorty Lake Eleph Lake < Esker Spring. Lake < Fig. 15. Box and whisker plots comparing Pd, Pt, Ni, Cu and Cr in C-horizon soils over and peripheral to the Baker Zone (Searcy 2001) and adjacent samples of swamp humus around Shorty Lake (Hattori & Cameron 2004). For C-horizon samples N = 46, and humus samples, N = 26. mineralization at Lac des Iles. Palladium related to eskers forms an element assemblage with As and S, the result of leaching of fragments of Nipigon diabase and Sibley Group sedimentary rocks contained in the eskers. At Lac des Iles Pd is associated with Co, Cr, Ni and Pt. This assemblage is the result of glacial dispersion of PGE-mineralized mafic-ultramafic rocks, followed by the transport of these elements from glacial sediments to nearby lakes in solution or as clay-sized solids in

13 Palladium mobiility in the surface environment 311 staking of the strongest anomalies. As shown by the Pd data for this area, the strongest anomalies may not be most closely related to mineralization. The project was funded by the Ontario Mineral Exploration Technologies Program. Maurice J. Lavigne of North American Palladium, Neil Pettigrew of Avalon Venture Ltd., Bernie Schnieders of Northern Development of Mines in Thunder Bay all generously provided assistance for our field work. We are most grateful to Richard Dyer and Cam Baker of Ontario Geological Survey for providing helpful information and the collection of lake sediments from Shorty Lake. Richard Dyer kindly read the manuscript and offered helpful comment. Fig. 16. Palladium contents in lake sediments with outline of Quaternary glaciofluvial deposits. suspension. Data from lake sediment surveys carried out by provincial and federal governments in Canada are generally distributed by timed release. Often this results in the immediate REFERENCES CAMERON, E.M Lake sediment sampling in mineral exploration. In: HALE, M. & PLANT, J.A. (eds) Drainage Geochemistry. Handbook of Exploration Geochemistry, 6, CROCKET, J.N Platinum-group element geochemistry of mafic and ultramafic rocks. In: CABRI, L.J. (ed.) The Geology, Geochemistry and Mineral Beneficiation of Platinum-Group Elements. Special Volume, 54. Canadian Institute of Mining, Metallurgy and petroleum, DYER, R.D.& RUSSELL, D.F Lac des Iles-Black Sturgeon River Area Lake Sediment Survey: Operation Treasure Hunt. Open File Report Ontario Geological Survey. GUILLOT, S., HATTORI, K.H. & DE SIGOYER, J Mantle wedge serpentinization and exhumation of eclogites: Insights from eastern Ladakh, NW Himalaya. Geology, 28, HATTORI, K.& CAMERON, E.M Using the high mobility of palladium in the surface media in exploration for platinum-group element deposits: Evidence from the Lac des Iles region, north-western Ontario. Economic Geology, in press. HINCHEY, J., HATTORI, K.H. & LAVIGNE, M.J Geology and timing of palladium mineralization in the South Roby Zone of the Lac des Iles deposit, Northwestern Ontario. Ontario Geological Survey Open File Report 6107, 16 pp. KAYE, L Geology of the Eayrs - Starnes Lakes Area, West Part, District of Thunder Bay. Map Scale 1: Ontario Geological Survey. LAVIGNE, M.J. & MICHAUD, M.J Geology of North American Palladium Ltd s Roby Zone Deposit, Lac des Iles: CIM. Exploration and Mining Geology, 10, MOLLARD, D.G. 1979a. Northern Ontario Engineering Geology Terrain Study. Data Base Map, Heaven Lake. Ontario Geological Survey, Map Scale, 1(100), 000. MOLLARD, D.G. 1979b. Northern Ontario Engineering Geology Terrain Study, Data Base Map, Kaministikwia. Map Scale 1: Ontario Geological Survey. ONTARIO GEOLOGICAL SURVEY Atikokan-Lakehead sheet, Map Geological Compilation Series. PETTIGREW, N.T. & HATTORI, K.H Palladium-copper-rich PGE mineralization in the Legris Lake mafic ultramafic complex, western Superior Province of Canada. Transaction, Institute of Mining and Metallurgy. Vol, 111, B46 B57. SEARCY, C.A Preliminary Data Results of Drift Exploration for Platinum Group Elements, Northwestern Ontario. Open File Report Ontario Geological Survey. SUTCLIFFE, R.H Proterozoic geology of the Lake Superior area; in Geology of Ontario. Ontario Geological Survey, Special Volume, 4, SUTCLIFFE, R.H., SWEENY, J.M. & EDGAR, A.D The Lac des Iles Complex, Ontario; petrology and platinum-group-elements mineralization in an Archean mafic intrusion. Canadian Journal of Earth Sciences, 26, THOMAS, R.L A note on the relationship of grain size, clay content, quartz and organic carbonin some Lake Erie and Lake Ontario sediments. Journal of Sedimentary Petrology, 39, WOOD, S.A., MOUNTAIN, B.W.& PUJING, P The aqueous geochemistry of platinum, palladium and gold; recent experimental constraints and a re-evaluation of theoretical predictions. The Canadian Mineralogist, 30,