Electric lithosphere of the Slave craton Alan G. Jones Geological Survey of Canada, 615 Booth Street, Ottawa, Ontario K1A 0E9, Canada Ian J. Ferguson Geological Sciences, University of Manitoba, Winnipeg, Manitoba R3T 2N2, Canada Alan D. Chave Deep Submergence Laboratory, Department of Applied Ocean Physics and Engineering, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543, USA Rob L. Evans Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543, USA Gary W. McNeice* Phoenix Geophysics Ltd., 3781 Victoria Park Avenue, Scarborough, Ontario M1W 3K5, Canada ABSTRACT to determine structures within the subcontinental lithospheric mantle The Archean Slave craton in northwestern Canada is an ideal of the craton and the topology of the lithosphere-asthenosphere bound- natural laboratory for investigating lithosphere formation and evo- ary. Serendipitously, we have discovered an anomalous upper mantle lution, and has become an international focus of broad geoscienti®c conducting region that is collocated both with the central Slave Eocene investigation following the discovery of economic diamondiferous diamondiferous kimberlite ®eld and with a geochemically de®ned ul- kimberlite pipes. Three deep-probing magnetotelluric surveys have tradepleted harzburgitic layer. recently been carried out on the craton using novel acquisition procedures. The magnetotelluric responses reveal an unexpected and remarkable anomaly in electrical conductivity, collocated with the kimberlite ®eld that is modeled as a spatially con®ned upper mantle region of low resistivity (,30 V´m) at depths of 80±1001 km, and is interpreted to be due to dissolved hydrogen or carbon in graphite form. This geophysically anomalous upper mantle region is also spatially coincident with a geochemically de®ned ultradeplet- ed harzburgitic layer. The tectonic processes that emplaced this structure are possibly related to the lithospheric subduction and trapping of overlying oceanic mantle at 2630±2620 Ma. Keywords: Slave Province, magnetotelluric surveys, electromagnetic, Archean, electrical conductivity, plate tectonics. INTRODUCTION The Slave craton (Fig. 1) in the northwestern part of the Canadian shield is a small Archean geologic province, ;600 3 400 km in ex- posed extent, that hosts the oldest dated rocks on Earth, the 4.027 Ga Acasta gneisses (Stern and Bleeker, 1998). It has become the focus of intense geoscienti®c scrutiny following the discovery of diamonds in 1991 (Fipke et al., 1995) and the opening of North America's ®rst producing diamond mine, Ekati, in 1998 (Rylatt and Popplewell, 1999). Excellent outcrop and existing geoscienti®c data make the Slave craton an unparalleled natural laboratory for testing competing theories of Early Archean lithospheric assembly, growth, and subsequent evolution. Models for assembly processes include repeated cycles of differ- entiation and collisional thickening (Jordan, 1988), collision of island arcs comprising depleted material (Ashwal and Burke, 1989), buoyant subduction and imbrication by lithospheric-scale stacks (Helmstaedt and Schulze, 1989), and basal accretion by cooling asthenospheric ma- terial (Thompson et al., 1996). These models imply different physical characteristics that are measurable using geophysical methods; in par- ticular, deep-probing electromagnetic methods are well suited for de- ®ning lithospheric structures (Jones, 1999). Geoscienti®c studies of the Slave craton and its relationship to the Proterozoic terranes to the west have been undertaken as part of Lithoprobe's SNORCLE transect investigations (Slave Northern Cor- dillera Lithospheric Evolution transect; Clowes, 1997). In addition, a wealth of geologic data for the craton has been accumulated as part of ongoing Geological Survey of Canada activities and scienti®c studies of mantle samples provided to academia by industry. The primary objectives of the electromagnetic experiments were Figure 1. Slave craton and locations of magnetotelluric sites. Open circlesÐ1996 all-weather road sites. Open squaresÐ1998 and 1999 winter road sites. StarsÐ1998±1999 (white) and 1999±2000 (gray) *Present address: Geosystem Canada Ltd., 927 Raftsman Lane, Ottawa, lake-bottom sites. Models were derived for pro®les A-B and A-C. Ontario K1C 2V3, Canada. Also shown is north-trending Pb isotope line. YKÐYellowknife. q 2001 Geological Society of America. For permission to copy, contact Copyright Clearance Center at www.copyright.com or (978) 750-8400. Geology; May 2001; v. 29; no. 5; p. 423±426; 4 ®gures. 423 Figure 3. Resistivity model of winter road magnetotelluric data pro- jected along north-south pro®le A-B (Figure 1). LdGÐLac de Gras. Given this rotational invariance, we can obtain qualitative infor- mation about lateral variation in Earth conductivity from response maps. Figure 2 shows the arithmetically averaged phases (Berdichev- sky and Dmitriev, 1976) for a period of 300 s from the all-weather and winter road sites, plus phases from preliminary processing of data from the ®rst year of lake-bottom deployments. The phases show a Figure 2. Map of averaged magnetotelluric phases at 300 s period distinct high of almost 808 at sites in the Lac de Gras region. Phases based on data from land sites (small dots) and preliminary pro- decrease away from this region along all three roads; 68 decrease to cessing of 1998±1999 lake-bottom sites (large named dots). YKÐ the north toward the Jericho kimberlite pipe (J), 68 decrease to the Yellowknife; JÐJericho pipe; KÐKennady lake pipes. On Slave cra- southeast toward the Kennady kimberlite pipes (K), and 108 decrease ton at 300 s period, electromagnetic waves are penetrating 100±150 km into lithosphere. VEÐvertical exaggeration. to the southwest toward Yellowknife (YK). The phases from the lake- bottom data (lighter shaded regions in Fig. 2) are consistent with those from the winter road sites, i.e., high phases in the center of the craton ELECTROMAGNETIC EXPERIMENTS, DATA PROCESSING, and low phases to the north and south. These lake-bottom phases sug- AND OBSERVATIONS gest that the phase maximum extends west of the north-trending Pb Three electromagnetic experiments using the magnetotelluric tech- isotope boundary (Thorpe et al., 1992) to at least Big Lake. Veri®cation nique (Jones, 1999) have been conducted on the Slave craton since of this westerly extension requires complete analysis of the lake-bottom 1996. The ®rst survey in fall 1996 comprised conventional broadband data, including those only now becoming available. (periods of 1024±104 s) acquisition every 10 km along the only all- The winter road magnetotelluric responses along pro®les A-B and weather road on the craton. This pro®le, located in the southwest corner A-C were modeled by using a two-dimensional inversion code that of the Slave craton, extends east-west ;100 km on either side of Yel- simultaneously ®ts the data and minimizes structure (Rodi and Mackie, lowknife (Fig. 1). 2001). The dominant feature of the best-®tting resistivity model ob- Subsequently, broadband data acquisition took place along ice tained for pro®le AB (Fig. 3) is the high-conductivity upper mantle roads over frozen lakes (Fig. 1) during March 1998, 1999, and 2000 region (r,30 V´m) at a depth of ;80 km beneath the Lac de Gras by using a novel deployment procedure. At each site, ®ve electrodes region. Along the winter roads, the anomaly does not exist north of were lowered to the lake bottom, through holes drilled through the ice, ;65.258N or south of ;63.758N. The resistivity model for pro®le A- to record the electric ®elds; the magnetometers were installed on the C (not shown) also images an anomaly at ;80 km beneath Lac de nearby shores to avoid noise caused by ice movement. Gras that is absent to the north (beneath Contwoyto Lake) and south The third experiment involved deploying sea¯oor magnetotelluric (beneath Kennady Lake). Sensitivity studies show that the anomaly instrumentation (Petitt et al., 1994) into lakes around the craton from must be at least 15 km thick for an internal resistivity of ;15 V´m. ¯oat planes. The 10 instruments were deployed at 19 locations (Fig. The attenuating effects of the anomaly make imaging the resistivity 1), with a year's recording at each location. structure beneath the anomaly impossibleÐdeeper imaging requires The magnetotelluric data processing code used to estimate the longer periods than in the present data set and will be addressed with magnetotelluric tensor response functions was a multiremote reference the lake-bottom data. variant of the Jones-JoÈdicke scheme (method 6 in Jones et al., 1989). A plan view of the central Slave mantle conductor mapped by our Subsequently, the response functions were analyzed for electric-®eld winter road data is shown in Figure 4A (dark gray). The high phase distortions caused by local, near-surface inhomogeneities and to deter- value observed at Big Lake suggests that the conductor extends to the mine the appropriate geoelectric strike direction (McNeice and Jones, west (light gray) across the Pb isotope boundary. This boundary in- 2001). The responses showed weak dependence on geoelectric strike, dicates that the crust of the western Slave province is fundamentally requiring conductivity structure to vary slowly with lateral distance in different from that of the eastern Slave, but the mantle-conductivity both the crust and mantle. anomaly may not respect this crustal boundary. 424 GEOLOGY, May 2001 Figure 4. Spatial locations of geophysical and geo- chemical mantle anoma- lies in center of Slave cra- ton. Also shown are locations of magnetotel- luric sites (dotsÐwinter road; starsÐlake bottom), Pb isotope line, mantle-zo- nation boundaries of GruÈt- ter et al. (1999), and im- portant kimberlite pipes on craton. DBÐDrybones; CrLÐCross Lake; SÐ Snap Lake; CLÐCamsell Lake; KÐKennady Lake; A44ÐDiavik's pipe A44 on Lac de Gras; PLÐPoint Lake; PÐPanda pipe at Ekati mine; TÐTorrie; RÐ Ranch; JÐJericho. A: Spatial location of discov- ered central Slave mantle conductor and its possi- ble extension to west to account for high phase observed at Big Lake. B: Spatial location of discovered ultradepleted harzburgitic layer from Grif®n et al.