James D. Miller Precambrian Research Center Department of Geological Sciences University of Minnesota Duluth

 Geochemical Analyses for Exploration  The Problem with Cumulates  Major Element Chemistry  Trace Element Chemistry  Mineral Chemistry  Assay Data for Cu-Ni-PGE Mineralized Intrusions

PURPOSE OF GEOCHEMICAL ANALYSES OF MLI ROCKS IN EXPLORATION (IN ORDER OF IMPORT)  ESTABLISH GRADE OF ORE DEPOSIT  EVALUATE THE POTENTIAL FOR MINERALIZATION S  EVALUATE THE COMPOSITION OF THE PARENTAL MAGMA (SOURCE OF METALS) AND POSSIBLE CONTAMINANTS (COMMONLY THE SOURCE OF S)  EVALUATE THE CRYSTALLIZATION AND DIFFERENTIATION HISTORY OF THE MAGMA

XRF+ICP-MS Full digestion >$60/smpl ICP-MS/AES Part. digestion $20-25/smpl Fire Assay $20-30/smpl No Si 2009 Acme Analytical Lab Brochure

The Classic View of Cumulate Rocks is that they are Mixtures of Primocrysts and a Liquid Component Primocrysts are: Enriched in high-T solid solution components (Mg in mafic phases, Ca in plagioclase) Enriched in compatible trace elements (e.g. Ni in Ol, Sr in Pl) Liquid component is: Enriched in low-T solid solution components (Fe in mafic phases, Na,K in plagioclase) Enriched in incompatible minor and trace elements

The concentration (X) of any element (a) in a cumulate rock (WR) is dependent on: The relative proportions of primocrysts (PC) and the liquid component (LC) The relative proportions of primocrysts (PC) and the liquid component (LC) The compositions of those components The compositions of those components X a (WR) = %PC 1 * X a (PC 1 ) + %PC 2 * X a (PC 2 ) %LC* X a (LC)

What parts of this mass balance can we know? X a (WR) = %PC 1 * X a (PC 1 ) + %PC 2 * X a (PC 2 ) %LC* X a (LC) Modes of Primocrysts? Problem is that cumulus phases continue to crystallize post-cumulus rims Plagioclase – possible if zoning preserved, but painstaking Olivine and pyroxene –not precisely, zoning lost due to subsolidus re-equilibration Oxide - ????

What parts of this mass balance can we know? X a (WR) = %PC 1 * X a (PC 1 ) + %PC 2 * X a (PC 2 ) %LC* X a (LC) Compositions of Primocrysts? Problematic because most primocrysts are solid solutions phases Plagioclase – possible, but cumulus cores can be very complexly zoned Olivine and Pyroxene – Ease of re-equilibration leads to “trapped liquid shift” Oxides – easily re-equilibrated and oxy-exsolved

What parts of this mass balance can we know? X a (WR) = %PC 1 * X a (PC 1 ) + %PC 2 * X a (PC 2 ) %LC* X a (LC) Compositions of the “Trapped” Liquid Component.... and might this be representative of the Parent Magma? Problem is... the amount of liquid in the cumulate changes over time due to compaction driven by bouyancy and crystal accumulation, and.... the composition of the liquid in the cumulate changes over time due to fractional crystallization of the intercumulus liquid. – McBirney 2005 From Tegner et al., 2009

DO NOT PLOT CUMULATES ON PLOTS INTENDED FOR MAGMA COMPOSITIONS!!! Primocrystic Plagioclase Liquid Component fractionation

Dunite Cumulate (O) adcumulate mesocumulate orthocumulate porphyriticliquid olivine adcumulate mesocumulate orthocumulate porphyritic liquid plactioclase + olivine Troctolite Cumulate (PO) Fo 77 olivine, An 76 plagioclase, mg# 50 liquid Fo 84 olivine, mg# 60 liquid

Major and Minor Element chemistry can serve as approximate proxies for abundances of primocryst phases e.g. Al – Plagioclase Mg – Augite + Olivine Ti – Ilmenite and Ti-magnetite From Joslin (2004)

Major and Minor Element chemistry (that includes accurate SiO 2 ) can be used to calculate CIPW Norms. Helpful for metamorphosed / altered cumulates (assuming a closed system) Helpful for calculating average An content (Ca/(Ca+Na+K)) of complexly zoned plagioclase

Like major elements, the absolute concentration of a trace element in a cumulate rock is typically dependent on the relative proportions AND compositions of the primocrysts and the liquid component. Primocrysts Liquid Bio? Ap? FC

Compatibility – degree to which an element prefers to partition into the solid over the liquid phase. Kd (i) 1 – Mineral-Liquid Partition Coefficient for element i in mineral 1 Kd (i) 1 = C (i) mineral 1 / C (i) liquid (C (i) - concentration of element i in wt. %) Kd (i) 1 > 1 – Compatible, Kd (i) 1 < 1 – Incompatible D (i) – Bulk Rock Partition Coefficient for element i D (i) = x 1 Kd (i) 1 + x 2 Kd (i) 2 + x 3 Kd (i) (x 1 – proportion of mineral 1)

From Rollinson (1993) Compatible Incompatible Bulk Rock Partition Coefficient of Ce,Yb, and Ni for Crystallization of: 1) Troctolite (70% Pl, 30% Ol) D (Ce) = x Pl Kd (Ce) Pl + x Ol Kd (Ce) Ol =.7* *.007 = D (Yb) = x Pl Kd (Yb) Pl + x Ol Kd (Yb) Ol =.7* *.065 = D (Ni) = x Pl Kd (Ni) Pl + x Ol Kd (Ni) Ol =.7* *25= 7.5 2) Olivine Gabbro (63% Pl, 12% Ol, 25% Cpx) D (Ce) = x Pl Kd (Ce) Pl + x Ol Kd (Ce) Ol + x Cpx Kd (Ce) Cpx =.63* * *.09 = D (Yb) = x Pl Kd (Yb) Pl + x Ol Kd (Yb) Ol + x Cpx Kd (Yb) Cpx =.63* * *.09 = D (Ni) = x Pl Kd (Ni) Pl + x Ol Kd (Ni) Ol + x Cpx Kd (Ni) Cpx =.63* * *8 = 5 LREE HREE

F (fraction of liquid remaining) Rayleigh Distillation: C L /C o = F (D-1) Conclusions: Fractional crystallization of mafic magmas gradually increases the concentrations of similarly incompatible elements, but has a minimal effect on their ratios; and strongly decreases the concentrations of compatible elements F (fraction of liquid remaining) C L /C o TroctoliteOlivine Gabbro

From Rollinson (1993) Fractional crystallization increases the REE abundance, but has a neglible effect on the REE pattern Since incompatible elements are 2-3 orders of magnitude greater in abundance than primocrysts, the REE pattern of cumulates (especially orthocumulates and adcumulates) will approximate that of their parental magmas and the magma source Fractional crystallization of olivine from a komatiitic melt From Jirsa and Miller (2006)

Spidergrams Tectonic Discrimination Diagrams rock/chondrite Increasing incompatibility From Bedard (2001) Negative anomalyPositive anomaly

From Miller (2004) Stratigraphic variations in the compositions of solid- solution cumulus minerals generally reflect the progressive differentiation of the parental magma and the occurrence of recharge event... but not exactly.

PLAGIOCLASE Zoning is preserved and records a history of cumulus and postcumulus crystallization White (2009) Strategy 1: Compare only An of cumulus cores... Problem: Cores are commonly complexly zoned Strategy 2: Calculate An from CIPW norm Problem: Integrates cumulus and postcumulus components

OLIVINE AND PYROXENE Zoning is NOT preserved and thus integrates cumulus and postcumulus compositions Overcoming the Trapped Liquid Shift Solution: Compare only like types of cumulates (ortho, meso, ad) Problem: Evaluating the type of cumulate is qualitative

Evaluating the Trapped Liquid Shift Gradual increase in incompatible elements; can assume nearly constant over limited stratigraphic thickness Magma Recharge POcf cumulate From Miller (2006) TLS

Evaluating the Trapped Liquid Shift POcf cumulate From Miller (2006) TLS Postcumulus mineral abundance are general proxies for amount of trapped liquid

Evaluating the Trapped Liquid Shift POcf cumulate From Miller (2006) TLS Assuming that well foliated cumulates have lower porosity – i.e. lower volume of trapped liquid From Meurer & Boudreau (1997)

Mineral chemistry also allows the estimation of the magma composition in equilibrum with that mineral A procedure for calculating the equilibrium distribution of trace elements among the minerals of cumulate rocks, and the concentration of trace elements in the coexisting liquids I Jean H. Bedard Chemical Geology 118 ( 1994 ) K D = (X FeO ol /X FeO liq )*(X MgO liq/ X MgO ol ) = 0.3 (Roedder and Emslie, 1970) which translates in determining the mg# of the liquid as: mg# liq = 100 / (3.333(FeO/MgO) ol + 1) This assumes no trapped liquid shift. Therefore, one should apply this only to adcumulates

Precious Metals Zone (PMZ) Meters above Cu-Au break Sonju Lake Intrusion Variation in the Cu/Pd is one of the best monitors of sulfide saturation in magmatic systems, but need high precision Pd analyses (<2 ppb DT) From Joslin (2004) From Miller (2004) Greenwood Lake Intrusion

D sulf/sil ~ D sulf/sil ~10 2 Pd is several orders of magnitude more compatible in sulfide melt relative to Cu

after Barnes and others, 1987 R = X sil /X sulf

Pd (ppb) Cu/Pd From Joslin (2004)From Jirsa & Miller (2006)

Quadrant w/ Best Potential Quadrant w/ Ore Grade Quadrant w/ No Potential Quadrant w/ Potential at Depth Indicates duration since initial saturation

From Jirsa & Miller (2006)