-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 G10 DI CPX G10D G1 G3 DI CR G4 G12 Mg-ILM G4D G3D G5 G5D CPX G9 G0 CR Crs-ILM G11 PC3 G10 DI CPX G10D G1 G3 DI CR G4 G12 Mg-ILM G4D G3D G5 G5D CPX G9 G0 CR Crs-ILM G11 Kimberli c and Diamondiferous Crustal Multivariate Analysis of Kimberlite Indicator Mineral Trends across Canada Brett Ferguson 1 , Bruce Eglington 1 University of Saskatchewan 1 Fig. 1. Fig. 2. Fig. 3. Fig. 4. Fig. 5. Fig. 6. Fig. 7. Fig. 8. TiO 2 -MgO Kimberlitic Ilmenite Discrimination Plot (Wyatt et al., 2004) showing Mg-rich ilmenites sourced from kimberlite pipes Cr 2 O 3 -MgO Chromite Diamond Inclusion and Intergrowth Plot (Fipke et al., 1995) Cr 2 O 3 -CaO Clinopyroxene Diamond Inclusion Field Plot (Fipke et al., 1998) Sample Locations of microprobed indicator minerals (n=122,913) shown in blue shown with Archean cratons PCA scaled loadings showing negative (red) correlation between more favourable kimberlitic and diamondiferous minerals versus positive (blue) crustal correlation PC3-PC1 analysis showing vectored trends. Positive and negative groupings can be seen along PC3 axis between kimberlitic and crustal components High-Potential (High scoring) PC3 samples displayed with remainder of sample locations showing cluster following kimberlite trend PCA Scaled Loadings Showing Diamondiferous Kimberlite Variables The Grϋer classificaon uses a variety of ranges of major oxides, as well as incorporang a Calcium Intercept (CA_INT) where; IF CaO ≤ 3.375 + 0.25*Cr 2 O 3 THEN CA_INT= 13.5*CaO/ (Cr 2 O 3 + 13.5) IF NOT CA_INT= CaO – 0.25*Cr 2 O 3 As well as a Magnesium number (MGNUM) where; MGNUM= (MgO/40.3)/ (MgO/40.3 + FeOt/71.85) Incorporang the ranges, an example of the classificaon of G10; Classified as G10 where; Cr 2 O 3 [wt.%]: ≥ 1.0 to < 22.0 CA_INT [wt.%]: 0 to < 3.375 MGNUM [wt.%]: ≥0.75 to < 0.95 The suffix “D” (Diamond-facies) may also be added to minerals that show a strong composional and P-T associaon with diamond in the G10, G3, G4 and G5 classificaons where; G10D = Cr 2 O 3 ≥ 5.0 + 0.94*CaO, or Cr 2 O 3 < 5.0 + 0.94*CaO and MnO< 0.36 (wt.%) G3D, G4D, G5D = Na 2 O > 0.07 (wt.%) (Grϋer et al., 2004) Garnet: Garnets were classified using the updated Grϋer classificaon scheme ( Grϋer et al., 2004) which show composional variaons in Cr, Ca, Mg, Mn, Fe and Ti. These, in turn, reflect chemical, physical and lithological environments that can occur together with diamond (Grϋer et al., 2004). The microprobe analysis of major oxides (in wt.%) was used to categorize garnets as the following (Listed in order of classificaon, this being of significant importance due to slight overlap between classificaons); G1- Low-Cr megacrysts G11- High-TiO 2 peridoc G10- Harzburgic G9- Lherzolic G12- Wehrlic G5- Pyroxenic, websteric G4- Pyroxenic, websteric G3- Eclogic G0- Unclassified All classificaon diagrams and mulvariate analyses were performed using the ioGAS soſtware package. Principal Component Analysis: Principal component analysis (PCA) was used in this study to analyze correlaons between different variables (classificaons) of indicator minerals (i.e. G10, DI chromite, Mg-ilmenite, etc.) in the goal of finding a ‘kimberlite sourced and potenally diamondiferous’ trend unassociated to a more crustal sourced trend. PCA uses orthogonal transformaon to convert correlated variables (clustered data) into a set of linearly uncorrelated values (principle components). In order to perform PCA at the per sample level, each sample locaon (n= 122,913, shown in Fig. 4.) was broken up into raos of percent of mineral per sample (n-mineral class/total mineral) for each mineral analyzed (i.e. n-G10/Total Garnet = %G10). Results: The raos were analyzed in the PCA (per sample) resulng in two major trends (shown in Fig. 6.) along the principle component 3 (PC3) axis The kimberlic and diamondiferous is negavely loaded on PC3 while the crustal trend is posive (Fig. 5.). PC3 scores for samples provides a single variable (from mulple) which was used to discriminate ‘high-potenal’ samples with high negave scores correlang to kimberlic trends (Fig. 8.). Combining mulple variables into a single PC3 score decreased sample locaons dramacally (n= 23,769), with clusters of high scoring samples becoming main areas of interest in terms of exploraon potenal. Summary: Properly classifying indicator minerals into sub-categories of higher and lower potenal for coinciding with diamond formaon and transport is an essenal part to any diamond exploraon program. Using PCA on the raos of these variables to create one new kimberlic and diamondiferous variable dramacally decreases the number of samples to focus on as well as eliminates the need to layer mulple indicator minerals on one map, which oſten makes for more difficult interpretaon. Even in areas where microprobe data is low in relaon to total grains (i.e. NWT), by using PCA scores basic trends can be seen correlang to high - potenal areas making this a powerful tool in any exploraon program. Introducon: The goal of this study was to reduce the dimensionality of a Kimberlite Indicator Mineral database in regard to microprobe data from grains across Canada, so as to idenfy prospecve diamond districts. Principle component analysis (PCA) was applied for this purpose. Approximately 200,000+ indicator minerals were compiled and categorized into sub-classifications, some of which reflect their potenal co-derivaon and transport with diamonds via kimberlites. The study looked at raos between these classificaons (per sample) in the theory that the more ‘kimberlic’ and ‘diamondiferous’ indicator minerals would occur together more frequently compared to the more crustal indicator minerals. PCA facilitated idenficaon of a new variable based on a combinaon of mineral categories, creang an easier, more robust, way of selecng higher potenal targets in lieu of looking at individual mineral categories. Indicator minerals assessed in this study were garnet, chromite, ilmenite and cr-diopside (clinopyroxene). Methods: The kimberlite indicator minerals analyzed in this study include garnet, ilmenite, chromite and clinopyroxene. These are considered to be some of the more favorable mantle sourced minerals associated with diamond. Classifying these indicator minerals into various diamond potenal sub- categories is key in aiding with any exploraon program in the search for diamond-bearing kimberlite The following methods were used in classifying the aforemenoned indicator minerals of focus: High-potential PC3 samples (n=23,769) displyed in pink, Archean cratons displayed in blue Map Interpretaon: Analyzing the high-scoring PC3 samples (Fig. 8.), several promising clusters (see aached fi gures below) are displayed with addional detail. Two trends in northern Alberta and one in northeastern Manitoba that cluster well in the PC3 variable show high abundances of diamond facies garnets, G10 Harzburgic garnet, Mg-ilmenite (kimberlic), DI chromite as well as DI clinopyroxenes. This applicaon can also be used in areas of high density clusters with many high-scoring samples. The NWT exhibits this, and using a larger scale map to look at the spaal distribuon of high-scoring samples compared to the remainder (Fig. 7.) shows a definive cluster of high - scores following the same trend as known kimberlites. Note the correlaon between the high-scoring samples with Archean craton. Typically diamond-bearing kimberlites are found on-craton, which is promising to see majority of high-potenal samples plot on- craton or near craton margins. References: Grüer, H.S., Gurney, J.J., Menzies, A.H. and Winter, F. (2004): An updated classificaon scheme for mantle-derived garnet, for use by diamond explorers; Lithos, 77, pages 841-857. Fipke, C.E., Gurney, J.J. and Moore (1995): Diamond exploraon techniques emphasising indicator mineral geochemistry and Canadian examples; Geological Survey of Canada, Bullen 423, 86 pages. Wya, B.A., Baumgartner, M., Ancar, E. and Grüer, H. (2004): Composional classificaon of "kimberlic" and "non- kimberlic" ilmenite; Lithos, 77, pages 819- 840. Simandl, G.J., Ferbey, T., Levson, V.M., Robinson, N.D., Lane, R., Smith, R., Demchuk, T.E., Raudsepp, I.M., and Hickin, A.S. (2005): Kimberlite and Diamond Indicator Minerals in Northeast Brish Columbia, Canada - A Reconnaissance Survey, Brish Columbia Ministry of Energy, Mines Petroleum Resources GeoFile 2005-25, 25 pages. Harvey, S.E, Kjarsgaard, B.A., and Kelley, L.I. (2001): Kimberlites of central Saskatchewan: Compilaon and significance of indicator mineral geochemistry with respect to diamond potenal; in Summary of Invesgaons 2001, Volume 2, Saskatchewan Geological Survey, Sask. Energy Mines, Misc. Rep. 2001-4.2. Acknowledgements: This study would not have been possible without generous contribuons of me and data from the following people: Barre Elliot, Doug Irwin and Kelly Pierce (Northwest Territories Geological Survey), Colin Card (Saskatchewan Geological Survey), Shawn Harvey (Cameco) and Levi Kalinsky (Cameco). Ilmenite: Ilmenite is an important indicator of kimberlite due to its Mg-rich species (picroilmenite) oſten being sourced from kimberlite pipes. The MgO-TiO 2 Kimberlic Ilmenite Discriminaon Plot (Fig. 1.) pictured leſt was used to discriminate between ilmenites sourced from kimberlite (shown in red) versus those having formed in a crustal origin (shown in blue) (Wya et al., 2004). The ‘undefined’ region prevents any overlap between those of kimberlic or crustal sources. Chromite: 98% of worldwide chromite inclusions in diamond fall into the High-Cr, moderate to high MgO diamond inclusion and diamond intergrowth fields in Cr 2 O 3 -MgO plots (Harvey et al., 2001; Fipke et al., 1995). This is an important pathfinder to peridote sourced diamonds. The same Cr 2 O 3 - MgO plot (Fig. 2.) was applied in classifying chromites for this study, with chromites that fall into the diamond inclusion and diamond intergrowth field being displayed in red on the Cr 2 O 3 vs MgO in Chromites plot. Clinopyroxene: Clinopyroxene (cr-diopside) found in associaon with other mantle-sourced minerals can be used as an indicator for kimberlite, and also provide proximity to source as clinopyroxenes typically are destroyed aſter relavely short transport. Cr 2 O 3 -CaO plots represenng the composions of CPX found as inclusions in diamonds (Simandl et al., 2005; Fipke et al., 1998) can also show potenal for coinciding with diamond in the diamond stability field. The plot displayed below (Fig. 3.) shows CPX in this study that fall into the diamond inclusion field, highlighted in pink. Idenfying diamond districts, one can then apply geological interpretaons, such as glacial transport, to find the source of indicator minerals.