Abstract Magmatic sulfide deposits fall into two major groups when considered on the basis of the value of their contained metals, one group in which Ni, and, to a lesser extent, Cu, are the most valuable products and a second in which the PGE are the most important. The first group includes komatiite- (both Archean and Paleoproterozoic), flood basalt-, ferropicrite-, and anorthosite complex-related deposits, a miscellaneous group related to high Mg basalts, Sudbury, which is the only example related to a meteorite impact melt, and a group of hitherto uneconomic deposits related to Ural-Alaskan–type intrusions. PGE deposits are mostly related to large intrusions comprising both an early MgO- and SiO 2 -rich magma and a later Al 2 O 3 -rich, tholeiitic magma, although several other intrusive types contain PGE in lesser, mostly uneconomic quantities. Most Ni-rich deposits occur in rocks ranging from the Late Archean to the Mesozoic. PGE deposits tend to predominate in Late Archean to Paleoproterozoic intrusions, although the limited number of occurrences casts doubt on the statistical validity of this observation. A number of key events mark the development of a magmatic sulfide deposit, partial melting of the mantle, ascent into the crust, development of sulfide immisciblity as a result of crustal interaction, ascent of magma + sulfides to higher crustal levels, concentration of the sulfides, their enrichment through interaction with fresh magma (not always the case), cooling and crystallization. Factors governing this development include (1) the solubility of sulfur in silicate melts and how this varies as a function of partial mantle melting and subsequent fractional crystallization, (2) the partitioning of chalcophile metals between sulfide and silicate liquids, and how the results of this vary during mantle melting and subsequent crystallization and sulfide immiscibility (degree of melting and crystallization, R factor and subsequent enrichment), (3) how effectively the sulfides become concentrated and the factors controlling this, and (4) processes that occur during the cooling of the sulfide liquid that govern aspects of exploration and mineral beneficiation. These topics are discussed first in general terms and then with specific reference to deposits at Noril’sk, Kambalda, and Voisey's Bay. With regard to Voisey's Bay, quantitative modeling is consistent with the very low PGE concentrations in this deposit being the result of some sulfide having been left behind in the mantle during partial melting. Both the Noril'sk and Voisey's Bay deposits are shown to be economic because of subsequent upgrading of the ores, which are located in magma conduits, through interaction with fresh, sulfide-unsaturated magma passing along the conduits.

Abstract Magmatic sulfide deposits fall into two major groups when considered on the basis of the value of their contained metals, one group in which Ni, and, to a lesser extent, Cu, are the most valuable products and a second in which the PGE are the most important. The first group includes komatiite- (both Archean and Paleoproterozoic), flood basalt-, ferropicrite-, and anorthosite complex-related deposits, a miscellaneous group related to high Mg basalts, Sudbury, which is the only example related to a meteorite impact melt, and a group of hitherto uneconomic deposits related to Ural-Alaskan–type intrusions. PGE deposits are mostly related to large intrusions comprising both an early MgO- and SiO 2 -rich magma and a later Al 2 O 3 -rich, tholeiitic magma, although several other intrusive types contain PGE in lesser, mostly uneconomic quantities. Most Ni-rich deposits occur in rocks ranging from the Late Archean to the Mesozoic. PGE deposits tend to predominate in Late Archean to Paleoproterozoic intrusions, although the limited number of occurrences casts doubt on the statistical validity of this observation. A number of key events mark the development of a magmatic sulfide deposit, partial melting of the mantle, ascent into the crust, development of sulfide immisciblity as a result of crustal interaction, ascent of magma + sulfides to higher crustal levels, concentration of the sulfides, their enrichment through interaction with fresh magma (not always the case), cooling and crystallization. Factors governing this development include (1) the solubility of sulfur in silicate melts and how this varies as a function of partial mantle melting and subsequent fractional crystallization, (2) the partitioning of chalcophile metals between sulfide and silicate liquids, and how the results of this vary during mantle melting and subsequent crystallization and sulfide immiscibility (degree of melting and crystallization, R factor and subsequent enrichment), (3) how effectively the sulfides become concentrated and the factors controlling this, and (4) processes that occur during the cooling of the sulfide liquid that govern aspects of exploration and mineral beneficiation. These topics are discussed first in general terms and then with specific reference to deposits at Noril’sk, Kambalda, and Voisey's Bay. With regard to Voisey's Bay, quantitative modeling is consistent with the very low PGE concentrations in this deposit being the result of some sulfide having been left behind in the mantle during partial melting. Both the Noril'sk and Voisey's Bay deposits are shown to be economic because of subsequent upgrading of the

Duke Island, with an area of about 60 sq. mi., is situated on the North American Pacific Coast at latitude 55° N. The ultramafic complex exposed on the island is the southernmost of some thirty distinctive ultramafic bodies of Early Cretaceous age distributed along the 350-mi. length of southeastern Alaska. A similar belt of corresponding bodies occurs in the Ural Mountain; there are at least two complexes of the same type in British Columbia, a possible occurrence in California, and two in Venezuela. The oldest rocks on Duke Island are clastic sedimentary and volcanic rocks of probable Late Triassic or Jurassic age. These are metamorphosed to greenschist- and amphibolite-facies grades and intruded by gabbroic and granitic rocks. The ultramafic rocks underlie two main areas totalling 9 sq. mi., and about a dozen minor areas. They are emplaced in gabbro that is intensely amphibolitized 1 around their edges and permeated by highly aluminous hornblende-anorthite (An 90–96) pegmatite derived from the ultramafic complex. The unaltered gabbroic rocks have tholeiitic affinities and are largely cumulates. They comprise two differentiation series, one ranging from picrite through olivine gabbro to hypersthene gabbro, and the other ranging from norite to olivine ferronorite. Olivine has reacted to orthopyroxene in both series, and hornblende is a late phase in all rock types. Some of the hornblende appears to be primary (postcumulus), but most is metamorphic or metasomatic in origin. The main ultramafic rock types, in order of abundance, are olivine clinopyroxenite, hornblende-magnetite clinopyroxenite, peridotite (wehrlite), dunite, and hornblendite. In many places the olivine-bearing units show well-developed rhythmic layering featuring pronounced grain-size sorting and remarkable sedimentation structures. It is evident, therefore, that the rocks are mainly cumulates, and that the layers were deposited by magmatic currents, which at times must have closely resembled turbidity currents. Hornblende-magnetite clinopyroxenite shows similar layering at one place, with a few of the layers being graded in modal content of magnetite. Another kind of layering, termed “inch-scale layering,” common to olivine clinopyroxenite, is defined by vague alternations of granular olivine and pyroxene plus more massive units of perpendicularly oriented pegmatitic pyroxene and small discontinuous bands of dunite, all with thicknesses of about 1 in. This layering apparently formed by recrystallization and partial replacement of weak stratification developed by crystal sedimentation. Petrographically, the dunite is mainly a cumulate of olivine and minor chromite; peridotite is largely a cumulate of olivine, clinopyroxene, and some chromite; olivine clinopyroxenite is a cumulate of clinopyroxene and subordinate olivine; hornblende-magnetite clinopyroxenite is mostly a cumulate of pyroxene and subordinate magnetite. Discrete postcumulus minerals are clinopyroxene and minor hornblende in the olivine-chromite cumulates, hornblende and magnetite in the olivine-clinopyroxene cumulates, and abundant hornblende in the clinopyroxene-magnetite cumulates. All rock types are virtually devoid of orthopyroxene and plagioclase. The cumulus crystals have generally undergone some recrystallization during the postcumulus stage and in places are intensely recrystallized. Replacement of pyroxene by olivine has locally produced irregular bodies of secondary dunite up to several hundred feet on a side, and veins and pods of coarse pyroxene are common to the olivine-bearing rocks. The abundance of hornblende as a postcumulus phase demonstrates that the trapped intercumulus magma was rich in H 2 O and compositionally different from the settled minerals. The hornblende-anorthite pegmatite apparently represents segregations of this magma fractionated into dikes. In the largest of the two main areas of ultramafic rocks, the hornblende-rich rocks form an almost continuous marginal zone but are particularly abundant around the stratigraphically upper parts of the olivine clinopyroxenite that they enclose. The olivine clinopyroxenite section is at least 5,000 ft. thick and is cut through by a separate, younger intrusion represented mainly by peridotite. Along one side of this intrusion, the pyroxenite was truncated, and countless fragments and blocks of it, many of them 100 to 200 ft. on a side, scaled off into the peridotite as it accumulated. Along the other side, the pyroxenite was folded aside. The hornblende-magnetite clinopyroxenite appears originally to have formed a thick layer over the olivine clinopyroxenite; it was then transposed to the margins by the doming effect of the younger intrusion. About a dozen fragments of coarse white quartz, presumably derived from overlying roof rocks, can be seen as xenoliths in the younger intrusion. The other main area of ultramafic rocks shows essentially the same structural relations. A large unit of dunite and peridotite has accumulated in a separate younger intrusion that pushed up through a section of olivine clinopyroxenite more than 5,000ft. thick. The pyroxenite was crowded into a steeply plunging, coupled anticline and syncline, and small bodies of hornblende-magnetite clinopyroxenite along the margins appear to have slumped there during the folding. Intense penetrative deformation is found in both pyroxenite units and in adjoining amphibolitized gabbro and hornblende-anorthite pegmatite. The structural similarities of the two main areas of ultramafic rocks, together with aeromagnetic and drill-hole information, suggest that the two areas represent a single complex continuous at depth, and it appears that this body extends to the southeast of Duke Island to join with olivine clinopyroxenite forming small islands about 1.5 mi. offshore. There is also the possibility that the younger intrusions in each of the two main areas branched from the same feeder. The evidence of repeated intrusion suggests that the whole complex formed in a subvolcanic magma reservoir. Chemical and mineralogical data from the ultramafic rocks indicate that they crystallized from critically undersaturated (alkaline) ultrabasic magma. The main variations can be produced by fractionation of minerals in the order olivine + minor chromite, clinopyroxene+olivine, and clinopyroxene + magnetite, with gradually increasing crystallization of postcumulus hornblende. The peridotite is a combination of the first two stages generated (in effect) by mixing in the magmatic current system. The intense recrystallization of the olivine clinopyroxene cumulates and their replacement by the dunite are believed to be due to transfer of materials via an aqueous vapor phase filtering through the cumulates. The relationships of the ultramafic and gabbroic rocks, indicating that intrusions of tholeiitic basalt magma were followed, after some extended period, by injections of a primitive alkaline magma, suggest that Duke Island constitutes a plutonic parallel to the eruptive sequence of tholeiitic basalt followed by alkaline magma common both to oceanic volcanic islands and to island-arc regions. The main line of ultramafic bodies in southeastern Alaska is spatially coincident with a discontinuous belt of Early Cretaceous volcanic rocks, and the parental magma (or derivatives thereof) may be represented by augite-rich porphyry (probably ankaramite) and hornblende porphyry found in this belt. Other occurrences of the same type of ultramafic body show similar affinities. The two complexes in British Columbia are associated with volcanic belts in which principal rock types are alkaline augite and hornblende porphyries ranging from ankaramite to augite alkali basalt and hornblende trachybasalt. The Uralian ultramafic. gabbroic, and associated rocks bear strong similarities to those at Duke Island, and nepheline-normative augite porphyry dikes, similar to the volcanic porphyries in British Columbia, are common to some localities. The concentric zoning of the major Alaskan-type ultramafic complexes is attributed to diapiric re-emplacement of rudely stratiform sequences of cumulates precipitated in the general sequence, dunite-olivine clinopyroxenite-magnetite clinopyroxenite. But while some of the diapirism may have been caused by continued rise of magma from depth, as a Duke Island, most of it was probably caused by tectonic compression.