Christiansen, E.H., and Keith, J.D., 1996, Trace element systematics in silicic magmas: A metallogenic perspective: Geological Association of Canada Short Course Notes Volume 12, Trace Element Geochem- istry of Volcanic Rocks: 24-26 May 1996, Winnipeg, Manitoba Canada, p. 115-151. 116 RH. Christiansen and J.D. Keith

Consequently, the composition of silicic magma reflects its tectonic setting, but in a secondary and sometimes inconsistent fashion.

Magma Sonrces The "source" of a magma is more a hypothetical construct to aid the understanding and modelling of magma evolution than a physical reality. Nonetheless, many magmas appear to derive much of their mass from a distinctive reservoir and, herein, we envision the history of a magma as beginning with partial melting in a single reservoir. Almost immediately, the magma begins to communicate with its environment, exchanging heat and mass as a non-isolated open system; we consider these contributions to the magma in subsequent sections. To describe a magma source completely, we need to know its bulk composition, its mineralogy, the fugacities of the volatile species, and the values of the relevant partition coefficients. Each of these is complexly dependent on the others and on the prevailing temperature and pressure. The source of a magma places important constraints on the composition of any partial melt that is derived from it. At equilibrium, the activities of all of the components must be equal in the melt and coexisting solids. Thus, the degree of Al and Si saturation, Fe2+/Fe3+ ratios, various volatile fugacities, etc. are initially controlled by the composition of the source. In a primary melt, the mineral compositions are the same in the source residues and the magma. Moreover, simple partition coefficients can be used to describe the concentration ratios of each trace element in the meLt to that in the solid at equilibrium. Ratios of incompatible elements (e.g., RblNb) are similar in the source and in the magma. The types and abundance of mineral phases that remain in the source are reflected in compatible trace element depletions in the derived magma. On the other hand, enrichments of a specific trace element in the melt may indicate that a phase was eliminated in the partial melting reaction. Source composition can even control the extent of melting, as described below. In each of these ways, silicic magmas are images of their sources. However, subsequent processes greatly degrade this image.

Partial Melting Processes The generation of magma involves partial melting of solids in the source. The extent and type of melting are important for the elemental composition of the resulting magma. To describe trace element systematics in magmas, it is useful to define the extent of melting as F (the weight fraction of melt present in a partially molten system) and the bulk partition coefficient as D (the ratio of the concentration of a trace element in the assemblage of crystals to its concentration in the coexisting melt). Because trace elements have different affinities for different minerals, the bulk partition. coefficient for each element depends on the types and proportions of the minerals that are in equilibrium with the melt. Compatible elements have D > 1. Sr, Ba and Eu all partition strongly into feldspars in .silicic melts and dre usually compatible elements. Rb, Li and Nb partition only weakly into the principal minerals found in silicic magmas and are usually incompatible elements, with D < 1. The main variable in the partial melting process is the extent, or degree, of melting. Whatever the type of melting process, low degrees (small F) of melting produce melts enriched in incompatible elements and depleted in compatible elements, compared to the source rock.. Larger degrees of melting cause incompatible elements to drop in concentration and compatible elements to increase 117 until at F = 1 (100% melting) the melt has the same composition as the source (Fig. 1). The extent of partial melting is controlled by the heat flux (either conductive or advective), the magnitude of decompression, the amount of fluxing volatiles introduced, or the proportions of hydrous minerals for dehydration melting. Regarding this latter point, a crustal protolith with abundant hydrous minerals may produce a magma with low incompatible element concentrations because a larger proportion of the protolith may melt. In contrast, a nearly dry protolith, even one that has experienced a previous episode of melt extraction, may produce magmas that are more enriched in incompatible elements by virtue of a small melt fraction. For example, decomposition of 5% biotite in a nearly anhydrous high-grade metamorphic rock produces about 5% partial melt (Burnham, 1979; Clemens, 1984) that is comparatively enriched in incompatible trace elements (Fig. 2). To understaild how different melting processes also affect trace element concentrations in magmas, consider the difference in compatible trace element concentrations between batch partial melting and fractional melting, two end-member models used to describe partial melting processes. Partial melting that occurs in accord with the assumptions of batch partiru melting produces melts that are not extremely. depleted in the compatible elements, which are limited to concentrations proportional to lID (Hanson, 1978). For example, if the D for Sr is 5, the lowest concentration (at very small F) that can occur in a single batch of partial melt is 1/5 of the initial concentration in the solid source (Fig. 1). If a series of melts is related by varying degrees' of batch partial melting of a single source, then Sr would vary by no more than a factor of 5 from the most enriched to the most depleted melt. Most volcanic and intrusive suites thought to be cogenetic violate this limitation, I I • suggesting that variations in the degree of batch partial melting are not the dominant cause of elemental variation in silicic magmas. Of course, this conclusion has no bearing on the importance of batch partial melting in producing magmas that are subsequently modified by contamination or fractional crystallization. Although incompatible trace element concentrations are greatly affected by the extent of melting, their ratios are less changed (Fig.s 1 and 3). As a result, ratios of I' incompatible elements may be examined to deduce the nature of a magma'ssource(s). I : I' Unfortunately, few trace elements are highly incompatible in silicic magmas, because of the great , I I variety of accessory minerals that contain them and because of the high partition coefficients for I cool, highly polymerized silicic magmas (Nash and Crecraft, 1985; Mahood and Hildreth, 1983). Another important partial melting model is fractional melting, in which each bit of magma . escapes from the source as soon as it is produced. For fractional melting, there is no lower limit to the concentration of compatible trace elements in a partial melt (Fig. 1). The initial fractions of melt (low F) are extremely depleted in compatible elements and enriched in incompatible elements. Imagining the physical conditions that would allow the instantaneous separation of each small fraction of melt is difficult. In most cases, a certain amount of melt must be generated before the source becomes permeably. Natural melting processes probably include aspects of batch and fractional melting. . More complicated mociels of partial melting include the contribution of an external fluid or melt to the production of magma (Allegre and Minster, 1978). This type of model may be especially applicable for the initial generation of subduction-zone magmas, where a fluid, derived by dehydration of descending oceanic crust, introduces soluble elements and, simultaneously induces partial melting in the overlying mantle wedge. The resulting magma is a composite of the fluid, melt from the mantle wedge, and solids modified by equilibration with the fluid and the melt (Fig. 4). .. • 118 E.H. Chrisp.ansen and J.D. Keith • 100 r-;::~~=-=;:::;:::;:::::;:=:;-11 100 A I2f Ch P~rtia/ijr:e/t1iJil B Fractional Melting & • Fractional C stallization • CI/Co =1/(O(1-F)+F) • I 10 10 • i I' i i • o o o o • --o --o • 0.1 • 0.1 • • 0.01 0.01 ..• 0.001 1 0.80.6 0.4 0.2 o 0.8 0.6 0.4 0.2 o • F (fraction of melt) - F (fraction of melt) .,• Figure 1 Mathematical models ofpartial melting andfractional crystallization processes can be used to understand the trace element concentrations of magmas. The relationship between the weight fraction of melt (F) and the i • concentration of a trace element in a melt (Cl) relative to its original concentration (Co) is shown. Co is the trace .. i'l I element concentration in the original solid source for .batch partial melting and is the trace element concentration in .. the original magma for fractional crystallization. D is the bulk crystalliiquid partition coefficient. (A) Batch partial' melting, in which the melt is in equilibrium with the solids until it separates. (B) Fractional melting, in which the • melt is continuously removed from a proto lith or source, and fractional crystallization, in which the solids are .. continuously removedfrom a magma, have identical effects on trace element concentrations. .. .., 1000 .. I ~=:~!i!!:~~~U Ymelt ~. j 500 ~(je Sov. .at! 300 ~\(j'(\ e' -E 'O\~ Co 200 <0'1 ~(je Co e.j Sov. = -.0cr:: oo~ ell i .'!..e'« .., 100 I <0\0\\

50 Xo 30 5 10 20 30 50 100 Nb (ppm)

Figure 2 The decomposition of a hydrous mineral such as biotite may control the degree of partial melting of a crustal source; in this case the amount of melt produced is directly proportional to the amount of biotite. For a biotite-poor (5%) source with low concentrations of incompatible elements, points Xo and Xmelt represent the compositions of the original solid source and a 5% partial melt. For a source with 20% biotite and higher concentrations of incompatible elements, points Yo and Ymelt represent the source and a 20% partial melt.. Because of the low degree of melting, incompatible element concentrations are higher in melt Xmelt than in the melt Ymelt, derivedfrom the enriched source. Points on curves represent 10% melt increments. DNb=0.1 and DRb = 0.05 for both sources. 119

80 800 F F B A 0.1 0.1 700 ONb = 0.3 60 600 ___ 0.2 -500 ORb E E C- o. .3;40 .3;400 0.3 .c .c r-- 0:: 0:: 300 0.6 20 200 1.0 100 0.9 Original Composition. --.J Or\ginal Composition 0 0 0.1 0.2 0.5 1 2 5 10 20 0.1 0.2 0.5 . 1 2 5 10 20 Rb/Nb Rb/Nb

Figure 3 Incompatible element ratios are relatively unchanged by variable degrees of (A) batch partial melting or (B) fractional crystallization. Dl{b = 0.3 and was held constant. DRb was vaned as shown. F is the fraction of melt in the system. Except when the partition coefficients are very different (e.g., DRb=0.9), the RbINb ratios change by a factor of2 or less between F=l and F=O.l.

300~------~~---' S-type Granites 100 A-type 30 Granites E a. 10 .3: .c 0:: 3 1 Batch Partial AII""lti",,., 0.3 1.0

0.1 2 3 5 10 20 30 50 100 Nb +Y (ppm)

Figure 4 The effect of mantle metasomatism on incompatible trace element concentrations during partial melting of the mantle (upper curve) compared to melting of normal mantle (lower curve). Rb represents the soluble elements I introduced into the mantle wedge by metasomatism above a subducting slab of oceanic lithosphere and Nb+Y- represent the insoluble high field strength elements. DNb = DY = 0.1 and DRb= 0.05 are the bulk distribution coefficients for the elements and are assumed to be the same for both sources. F is the fraction of melt in the system. Field boundaries are from Pearce et al. (1984). Shaded region shows composition ofI-type rhyolites from Macdonald et al. (1992). Batch partial melting of metasomatized mantle produces magmas with the characteristics of I-type granites. zi

E.H. Christiansen and J.D. Keith

The resulting magmas are enriched in the elements carried by the fluid, here represented by Rb, and do not have trace element patterns like the source before fluid interaction. Restite unmixing, where residual solids are variably separated from the partially molten mass, is 4 important at the site of magma generation and controls the initial composition of the separated ~ magma (crystals plus liquid). After magma segregation, restitic mineral phases may become 4 dissolved at lower pressure or after continued reaction with the melt. However, restite unmixing is C probably not important in the generation of the observed elemental trends in rhyolites and granites, • except some S-types thought to be derived from sedimentary parents, that show abundant • petrographic evidence for residual materials. In general, granitic magmas, especially the mineralized • ones that are highly evolved, have cleared themselves of restite. The eruption of large volumes of 4 t nearly crystal-free rhyolite attests to this. Most of the chemical variation in granitic plutons probably 41 is caused by fractional crystallization rather than variable proportions of minimum melt and restite 41 from the source (e.g., McCarthy and Hasty, 1976; Bateman and Chappell, 1979; Lee and Christiansen, 1983). During pluton solidification, cumulates and melts may be mixed in variable • proportions. • Application of inverse techniques, such as those described for batch partial melting by Minster • and Allegre (1978), to estimate extent of partial melting and source composition, has not generally • been successfUl with silicic igneous rocks. This failure is probably because most silicic magmas • experience extensive fractionation andlor contamination during their histories and because some • crustal melts may entrain substantial amounts of residual solids and cumulus phases. • All these simple models assume Henry's law behaviour of trace elements and complete • equilibration of melt and coexisting residual solids. Because of the relatively low temperatures and • highly polymerized rhyolitic melts produced by crustal melting, diffusion is too slow to remove • compositional gradients before melting occurs. Consequently, disequilibrium melting may be more • common in the crust than in the mantle. Bea (1995) questioned the applicability of simple • mathematical models to the partial melting of crustal rocks. He contends that the effective partition • co~fficients for trace elemet:lts in the major phases approach one at the low temperatures • characteristic of crustal melting. This is so because the melting rate may exceed the diffusion rate • for many trace elements. For minerals with diameters of 2 mm and trace element diffusion rates of • about 10-16 cm 2s-1, it takes ).6 m.y. to achieve equilibrium between melt and residual crystals. If • I : this happens, then silicic melts would be more direct images of their sources than generally • ',il imagined; the concentration of a trace element in the melt would be simple function of the • , ,'I composition of the minerals that participated in the melting reaction. Another cause of non-Henrian • behaviour of trace elements during crustal melting arises because some trace elements are essential • structural components of accessory phases. The concentrations of such elements in melts are ..• controlled by the solubility of minerals, not by simple Henry's law behaviour. The best example of .. this is Zr in zircon, but it probably applies to U and REE (rare-earth elements) in monazite, and Y and REE in xenotime. Disequilibrium processes may also disturb the Sr and N d isotopic ratios of •.. crustally derived magma, so that the melt's isotopic composition no longer reflects that of the ~ source. This may happen if Sm and Nd are largely held in residual accessory phases with steep ~ REE patterns, such as monazite, that do not equilibrate with the magma before it separates .. (Creaser, 1995). Alternatively, if magma produced is caused by the breakdown of hydrous ~ minerals, the decomposition of biotite could release highly radiogenic Sr to a small melt fraction. =.-~ = 121

These problems probably only apply to magmas produced by very small degrees of melting.

Fractional Crystallization The compositional evolution of most magmas is driven largely by cooling and fractional crystallization. The extent of crystallization of a magma chamber is an important control on the trace element concentrations of the residual melt and the solids. The effect of fractional crystallization, with or without complete removal of crystals, can be modelled using the Rayleigh fractionation equation (e.g., Allegre and Minster, 1978; Fig. 1). If high quality analyses of cogenetic igneous rocks are available, inverse modelling techniques (Allegre et aI., 1977) can be used to deduce the nature of the fractionating mineral assemblage and extent of frac~ionation; reasonable values for indiVidual mineral/liquid partition coefficients must be used. The important role of accessory phases in controlling the abundances of many trace elements makes the inverse approach more difficult for silicic magmas than for mafic magmas, but has yielded reasonable results (e.g., Christiansen et aI., 1984). Obviously, rocks derived from highly fractionated magmas are compositionally quite different from those magmas that have experienced only modest amounts of fractional crystallization. Although concentrations of ore elements in silicic magmas or cumulates are seldom, if ever, of ore grade, significant enrichments may occur that set the stage for subsequent ore-generating processes. The extent of fractional crystallization is especially critical for those elements that have either very high or very low bulk partition coefficients. Compatible elements with high partition coefficients become sequestered in solid phases where they may become subsequently remobilized, depending on the nature of the host mineral and its reactions with later fluids. If fractionation proceeds extensively, magmatic concentrations of compatible elements may drop below the critical level needed for the generation of ore (Candela, 1989). Copper behaves as a compatible element in most felsic magmas, largely because its partitions strongly into the small proportion of immiscible sulphides that eventually fractionate. Consequently, porphyry Cu deposits are typically associated with less fractionated igneous rocks where the Cu concentrations remained relatively high. Copper deposits are rarely found in association with highly fractionated leucogranites, which have inherently low concentrations of Cu. On the other hand, incompatible elements are concentrated in residual melts, locally setting the stage for subsequent processes to enrich the element further. The ,I' behaviour of Be exemplifies that of an incompatible element. Be deposits_ are typically associated with highly evolved granites, rhyolites, and pegmatites that had extensive histories of fractionation to concentrate Be. In highly evolved rhyolitic magmas, the effect of fractional crystallization is most obvious in trace element rather than major element variations. Fractional crystallization drives a silicic magma toward the composition of a minimum melt in the granite system. In high-Si02 rhyolite magma, the / coexisting solids (subequal proportions of plagioclase, sanidine and quartz) and the melt have nearly the same major element composition and large amounts of fractionation can occur without large changes in major element compositions. However, trace elements continue to change in accord with Rayleigh fractionation. For example, the zonation in the Bishop Tuff in eastern California spans a 2% change in Si02, but compatible and incompatible element abundances vary by several hundred percent (Hildreth, 1979). These changes are consistent with fractional crystallization (Christiansen, 1983; Michael, 1983), but have been attributed to other poorly • I ' 122 RH. Christiansen and J.D. Keith • i •~ constrained processes such as volatile complexing, "thermo gravitational diffusion," or "complex interactions of petrogenetic processes" (Hildreth, 1979; Macdonald et aI., 1992). The fact that trace ~ element enrichment and depletion patterns in rhyolites and trachytes correspond to those produced '" by fractionation of their mineral assemblages, including distinctive accessory minerals, argues •~ persuasively for the comparatively simple process of crystal and liquid separation (Wolff and ~ Storey, 1984). •~ Mixing and Assimilation .., , I I Once a batch of magma is produced, it immediately begins to interact with its surroundings in a long process of trying to come to chemical, thennal, and mechanical equilibrium. Compositional '"~ changes in the magma result from open-system behaviour, or contamination. This process becomes ~ especially important in the crustal entrapment zones above subduction zones where mantle-derived magmas stall because of relatively high densities, partially crystallize to induce wall-rock melting, ..,~ and then mix with the crustal magma while being reintruded by fresh draughts of mantle-derived ..., magma.. Contamination continues to be important when a magma batch achieves buoyancy and moves out of its source region and into totally different rock types in the overlying crust and/or .. mantle. •~ Obviously, the composition of the contaminants is important to the final composition of the .. mixture, but the particular type of contaminating process plays an equally important role. The magma may assimilate wall rocks by bulk digestion if a block falls into the magma. For example, Johnson et aI. (1990) found Proterozoic zircons ina tuff erupted from the Mo-mineralized Questa magmatic system of northern New Mexico. Apparently, these were derived from nearly complete dissolution of Proterozoic wall rocks. Commonly, assimilation of wall rocks forces the magma to crystallize. The effects of simultaneous assimilation and fractional crystallization (AFC; DePaolo, 1981a) are contrasted with simple fractional crystallization in Figure 5. Moreover, a magma may encounter other bodies of magma, from the same or different sources and, in its attempt to mix with these extraneous magmas, large compositional variations may be created. If mixing proceeds tb completion, the final composition depends on the mixing proportions. Commonly, mixing is also accompanied by crystallization as new phases fractionate, once they are stabilized by the changing compositional and thermal regime created by mixing. On a smaller scale, the original magma, itself a heat source that can induce the production of more magma by partial melting, may become contaminated by partial melts drawn from the wall rock. The composition of these contaminating magmas is governed by appropriate partition coefficients, wall rock compositions, and the degree to which equilibrium is achieved. More subtle magma - wallrock interaction may occur by diffusive exchange driven by compositional and thermal gradients. Some speculate that shallow magma chambers also can be contaminated by hydrothennal fluids during caldera collapse or resurgence (Hildreth et at., 1991). These! processes of contamination occur throughout the history of all magmas, from the moment of initial production, through rise and emplacement as a pluton or eruption as .volcanic rock, and are partially responsible for obscuring the connection between igneous rocks and their original sources. ---~ .. -...... -.

100 300.------~_r.------. 123 A I AFC r=O.21 B _ Fractional .::. 10 250 Crystallization .:.; Assimlation ...•.. & FC .! Contaminant

1 _200 j, ~ E 0 Co ~ .e ~,~ C3 .0 0.1 ~ 150 DNb=O.10 ~~ DNb=O.25 DRb=O.25 DRb=O.10

0.01 100

50L----L--~~~~~----~~-J 0.001 1 0.8 0.6 0.4 0.2 0 1 F (fraction of melt remaining)

Figure 5 Fractional crystallization.and assimilation combined with fractional crystallization (AFC; DePaolo, 1981a) are contrasted in.these diagrams. The rate of assimilation/rate of crystallization is the same in both diagrams (r=O.2). (A) Cl is the trace element concentration in the evolving liquid and Co is its original composition. The numbers on the curves show the ratio between the trace element concentration in the assimilated material and the original magma. Partition coefficients (D) are shown in boxes. Thin lines are for simple fractional crystallization. (B) The extent to which incompatible trace element ratios are changed by assimilation and fractional crystallization depends on the composition of the contaminant (here assumed to be a pelitic sediment) and upon the partition coefficients (DNb and DRb) for the trace elements. Dots on lines represent 10% increments ofcrystallization. In the case shown here, little change in incompatible element trace element ratios is evident.

Volatile Fugacities Volatiles play an important role in the evolution and eventual mineralization of silicic magmas. Magmatic volatile fugacities and fugacity ratios depend on source compositions, fractionation histories, and mixing and assimil.ation. However, the effect of each volatile is so diverse that they are worth considering separately. The principal volatile components of interest are H20, HF, HCl, C02 and the various S species. The fugacities of these species largely reflect those in their sources (see Carmichael, 1991, for a discussion centred onj02), but various evolutionary processes in open magmatic systems may change the values (pichavant and Hammouda, 1995). Dramatic and consistent variations in j02 between the various types of silicic magmas have been identified. Proportions of mantle-derived magma (with widely varying j02) and the dramatically reduced (C-containing sedimentary rocks) and oxidized (sedimentary and altered igneous rocks that reacted with the atmosphere) magmas derived from the continental crust determine the prevailing f02 of silicic. magma. Dehydration of oceanic crust in subduction zones also releases an oxidized fluid that passes this characteristic on to magmas formed in this environment. Munoz (1984) has also shown that halogen fugacities vary systematically in silicic igneous rocks and associated mineral deposits. , i Silicic rocks related to Cli mineralization have higher j HClIjHF than those associated with Mo or =• Sn-W -Be. Moreover, volatile fugacities control the stability of many phases that are important for I.,. ... ore element concentration or dispersion. For example, the solubility of sulphur as an immiscible ... sulphide melt strongly controls the Cu concentrations of evolving silicic magmas. If magma E becomes sulphide saturated, Cu partitions strongly into these blebs and is commonly removed from .-t. the magma by fractionation. Hence, porphyry Cu deposits are rarely developed from highly E E

124 E.H. Christiansen and J.D. Keith

evolved rhyolitic magmas, but may be related to their more mafic precursors. Another good example of the role of volatile fugacity controlling mineral stability and trace element evolution lies in the effect of f02 on the oxidation state and consequently, magmatic and hydrothermal behaviour of Sn. In magmas with 102 below the QFM (quartz-fayalite-magnetite) oxygen buffer, the compatibility of Sn in mafic silicates and oxides is apparently lower and Sn becomes strongly enriched in the residual liquid by fractional crystallization processes (Ishihara, 1977). Sn+2 is apparently less compatible in minerals than Sn+4, which predominates at high j02. Moreover, low j02 increases the solubility of Sn in hydrothermal fluids derived from, or equilibrating with, reduced granites (Eugster, 1986). In contrast, high f02 produces high Mo+6 /Mo+4 and lowers the partition coefficients for Mo. This probably explains why Mo and Sn deposits are both related to highly evolved magmas, but do not commonly occur as ore deposits related to the same magma (Keith et al., 1993). ) Fluid saturation can be treated as a phase separation as well. Most silicic magmas are probably fluid-saturated during much of their evolution as the result of the extremely low solubility of CO2. The effects of the migration and partitioning of trace elements into this kind of fluid are probably minor. More dramatic changes may occur in magmas that reach water saturation. Some elements preferentially partition into an aqueous fluid (CI, the alkalis and alkaline earths, Pb, Zn, Cu) compared with others that prefer the melt (F and high field strength elements--those with high charge to radius ratios like Nb, Ta, Zr, and Ti). The mobility and redeposition of these elements significantly alter the patterns produced by crystal-melt equilibria and must be carefully avoided to ascertain magmatic patterns. For example, high Flel ratios (>5) in magmas may be produced by the separation of a Cl-rich aqueous fluid rather than by distinctive sources or melt evolution paths.

Tectonic Setting . Although there are broad correlations between tectonic setting and the geochemistry of magmas, in detail the composition of a silicic magma is not simply nor directly tied to the tectonic setting in wQ.ich it was formed, differentiated, and emplaced (e.g., Twist and Harmer, 1987). Rather, the composition is the product of its partial melting history, source(s), crystallization and contamination processes, and prevailing volatile fugacities. Consequently, the trace element signatures of granites do not uniquely identify the~r tectonic setting, but point directly to their melting and crystallization histories. It is this latter inference that is most important for the association of ore with rhyolitic magmas. We emphasize the compositional characteristics of granites, rather than their tectonic settings, in this paper.~ We use trace element characteristics and so-called tectonic discrimination diagrams to help identify magma sources and evolutionary processes. However, to the extent that these controls on magma chemistry are linked to plate tectonic setting, tectonic discrimination is; meaningful and the tectonic settings of the principal types of silicic magmas are discussed belo~. For example, silicic igneous rocks from mid-ocean ridges are distinct from those associated with oceanic plumes, compositions, mineral assemblages, and volatile fugacities are all distinctive. This is apparently the result of homogeneity in the processes and magma sources available in these two distinctive tectonic settings. Magmas formed in subduction zones appear to have consistently highj02,jHCI, and distinctive trace element patterns. In oceanic island arcs, uniformity in source materials, magma producing reactions, and ascent paths leads to uniformity in island-arc igneous rocks. As a result, the associated ores are consistent and F=:======-:==~--.--~.-~-.- ..-

125

predictable. In contrast, where subduction-generated magmas interact with compositionally heterogeneous continental crust, they take on elemental and volatile characteristics that reshape their compositions and ore-generating capacities. Subduction-zone magmas may become heavily contaminated with sedimentary and other continental materials along continental margins. This typically changes the magma composition to become more peraluminous and less oxidized; two factors that radically change the assemblage of fractionating accessory phases and the partition coefficients of trace (including ore) elements. Likewise, the heterogeneity of continental crust and its variable contributions to all types of magma (plume-, rift-, subduction-, or collision-generated) that pass through it or that are produced in it, imparts dramatic variations to all continental magma associations and complicates the association of tectonics and magma composition and contributes to the formation of certain metallogenic provinces.

SILICIC MAGMA TYPES A useful framework for the discussion of the trace element compositions of silicic magmas is provided by the compositional criteria proposed by Chappell, White, and co-workers in a series of papers beginning in the 1970's (e.g., Chappell and White, 1974, 1992; White and Chappell, 1983; Collins et al. 1982; Whalen et al., 1987). Although the origins of these compositional types are still controversial, the application of the criteria for distinguishing various kinds of granites and rhyolites is more clear cut. Ishihara (1977) identified distinctive ilmenite- and magnetite-series granites and emphasized the importance of oxygen fugacity in the trace element character and ore associations of granitic magmas. The consistent identification of distinctive granite types in specific tectonic settings led to the idea that granite compositions and tectonic settings were genetically related. In his paper, "Granite type and tectonic environment,:' Pitcher (1982) espoused this connection and extended the subdivision of subduction-related granites into several groups. Pearce et al. (1984) used trace elements to distinguish various types of granites, and divided granitic rocks into groups that are similar to those of Chappell and White. However, Pearce et al. (1984) distinguished the granite types as products of their tectonic setting; rather than as compositional types that reflect their source components and magmatic processes. In this review, we discuss the compositional characteristics. and ore associations of four principal types of rhyolitic/granitic magma as follows, OR (ocean ridge)-, I-, S-, and A-type (Table 1). The letters representing the last three categories were intended by Chappell and White to describe the crustal source of each type of magma in a rather strict sense. I-types from igneous source materials, S-types from sedimentary materials, and A-types from anhydrous, melt-depleted rocks. The letters are retained because of their familiarity, but the connotation of strictly representing a crustal granite source is abandoned here. The various groups ~e simply compositional types of granites that reflect their protracted evolutionary histories, including, the extent and style of partial melting, their sources, degree of fractionation and/or contamination, and inherited volatile fugacities. Because these compositional controls are in turn affected by the plate tectonic setting, the general tectonic setting of each type and some problems involved when comparing tectonic setting and granite composition are discussed.

..

127

I -Type Silicic Magmas Rhyolites and granites generated at subduction zones are diverse, but most have a broad range of similar features, which together distinguish I-type silicic magmas from other types (Table 1). Some of the most important compositional features ofI-type granites and rhyolites are their high ratios of large ion lithophile elements to high field strength elements (Pearce and Cann, 1973) and high fugacities of 02, H20, Hel, and oxidized sulphurous species. In mafic rocks, high BaJNb ratios are employed to examine the relative depletion of high field strength elements as compared with other equally incompatible elements. In evolved felsic magmas v:here sanidine and biotite are stable, Ba behaves as a strongly compatible element and the BaJNb ratios drop. In such silicic magmas, Rb/Nb and Th/Nb ratios give better indications of this depletion in the source. These differences are especially clear when I-type rhyolitic magmas are compared with those from the ocean basins, including oceanic plumes and mid-oceanic ridges (Fig. 6). Rb/Nb ratios should be source related because they do not change dramatically as a result of closed system fractionation or variations in degree of partial melting (Fig. 3). However, contamination with upper continental crustal materials may increase this ratio in open magma systems (Fig. 5). Because of low Nb (+Y) concentrations, I-type rhyolites and granites fall in the volcanic arc field of Pearce et al. (1984), which we have relabeled here to emphasize compositional rather than tectonic characteristics (Fig. 7). As expected from these characteristics, I-type magmas have deep negative Nb anomalies on chondrite-normalized trace element diagrams (Fig. 8). Most I-type magmas have highj02 (several log units above QFM; Ewart, 1979) and consequently high Mg/Fe and Fe20~eO ratios in mafic silicates and bulk rocks (Fig. 9). Many are calc-alkalic in the classification of Miyashiro (1974), . which is based on this parameter, but some range to lower MglFe ratios and are.tholeiitic. Early fractionation of magnetite, because of high j02, and other phase relations caused by high jH20 drive the characteristic calc-alkalic fractionation trend marked by Si02 enrichment and Fe-depletion. This process has operated over time to produce Earth's unique continental crust. High water fugacities are shown by the presence of hydrous minerals in relatively mafic rocks. High jHCVjHF is shown by biotite and amphibole compositions (Munoz, 1984; Ague and Brimhall, 1988) and high F/CI ratios in melt inclusions (Lowenstem, 1995; Fig. 10). Volcanic and many plutonic associations are unimodal with broadly continuous trends on Si02 variation diagrams. Among I-type magmatic series, intermediate compositions, andesite or granodiorite, are commonly the most abundant rock types. These simple characteristics are complicated by the extent of fractionation and by the nature of the lithospheric contaminants. Consequently, wide ranges of compositions appear that correlate with the type of overriding plate- whether it includes continental or oceanic crust, thick sedimentary basins or granulite facies metamorphic terranes, ancient metasomatized mantle or young depleted lithospheric mantle. The key features of I-type magmas are tied to the evolution of oceanic crust. Mter its formation at an oceanic'ridge, the basaltic crust becomes hydrated, oxidized, and mineralized as ocean water is pumped through it in large hydrothermal convection cells. Sediment accumulates above the igneous crust. As this altered oceanic crust sub ducts and dehydrates, oxidized aqueous fluids carry soluble elements into the overlying mantle. This creates magmas (and the continental crust derived from it) that are oxidized and have high abundances of the water soluble elements (e.g., Cl, S, Cu, Ba, Rb and·K) compared with oceanic basalts. The characteristic Nb-Ta-Ti depletion of subduction-related I-type magmas is probably the result of residual titanates in dehydrating (or melting) oceanic crust -----~------~

128 RH. Christiansen and J.D. Keith

(e.g., Ryerson and Watson, 1987). The released fluid is depleted in elements stored in relatively insoluble titanates like rutile, a common mineral in eclogites. Alternatively, partial melting of the basaltic slab may produce a low temperature siliceous melt that is rutile-saturated. The rutile is left behind when the magma rises into the overlying mantle wedge. There it reacts with peridotite to produce the more common mafic I-type magmas, or rarely erupts as adakite (Yogodzinki et ai., 1995). 1,200 , • • • Figure 6 RbINb ratios of various types -O~ I-type 1,000 o • e of silicic igneous rocks. Data are from ,1 Macdonald et al. (1992), the authors' ~i • S-type '" analyses of topaz rhyolites from the 800 0 • western United States, and sources cited in 0 OR-type E text for discussed localities. RhlNh ratios ,Q. §D are lowest in rhyolites from ocean ridges S 600 • o 0 1 ~ .0 • A-type (OR-type) and highest in S- and I-types. 0:: 1- • 0 Subalkalic A-type' rhyolites from the 400 ,." Peralkaline continents have slightly higher RbINb 8 <> ratios than other A-types, including 0 Plume peralkalic rhyolites and those from oceanic 200 0, plumes.

0.1 0.2 0.5 1 2 5 10 20 50 Rb/Nb(ppm)

1000 1000 S-type S-Type Granite A-Type Granite 300 100 • • • Bingham (Cu) E 100, • • Pine Grove (Mo) E "'- "'- .s • Macusani .s - 10 ~~Q,eJ? 'T:~7>rjIJ ~P~;'~9. 3 --i'?fo~ 2 5 10 20 50 100 200 500 1000 2 5 10 20 50 100 200 500 1000 Nb+Y(ppm} Nb+Y(ppm}

Figure 7 (A) Left: compositions of selected igneous rock suites that are related to mineralization plotted on Pearce et al. (1984) tectonic discrimination diagram. The field names have been changed to emphaSize compositional, rather than, tectonic characteristics. Datafrom sources cited in text. (B) Right: the effect o/various processes and source composiiions on trace element trends anli positions on this diagram.

Sedimentary materials may contribute to the genesis of subduction zone magmas in two ways. First, oceanic sediment is subducted along with the oceanic crust in many subduction zones. Cosmogenic lOBe is produced in the atmosphere and becomes enriched in oceanic sediment but decays away rapidly with a half-life of 1.5 m.y. Measurable activities of lOBe (Morris and Tera, 129

1989) in modern arc lavas demonstrate that substantial amounts of young subducted sediment contribute to I -type magmas fonned in subduction zones. Sedimentary materials in the continental crust may also contribute to I-type magmas during assimilation of wall rock by mantle-derived magma. DePaolo (1981b) and Ague and Brimhall (1988) show the effect of this kind of crustal interaction for many plutons in California, including the Sierra Nevada and Peninsular Ranges batholiths. Assimilation of graphitic pelites created strongly reduced ilmenite-series granites in the western portions of these batholiths. Assimilation of more oxidized crustal materials occurred to

,... ..Ie ..• ,... Figure 8 Normalized trace element patterns for the principal types of silicic igneous rocks. The strongly compatible elements, Sr, P, and Ti, have been omitted. Normalizing factors are mostly cho1ulritic abundances and taken from Thompson (1984). Data from Macdonald et al. (1992) and sources cited in text. The pattern ofan intraplate alkalic • basalt is shown for reference with a dashed line. ..• ss:::i0 - ~ I I' 130 E.H. Christiansen and J.D. Keith

5

2 I-type ;? 1 0 • S-type ! '.• Figure 9 Fe2031FeO ratios 0 0.5 for silicic igneous rocks of

the east and is evidenced by increases in incompatible element abundances and jHPljH20 and jHPljHCl as indicated by biotite compositions; here, magmatic j02 was not strongly affected. In many continental magmatic arcs, progressive assimilation of crustal sediments by mantle-derived magma created S-type magmas that may be indistinguishable from those derived purely by crustal anatexis. Figure 10 F and CI abundances in various types of glassy rhyolites and glass 0.3 inclusions in phenocrysts. Data S-type are from obsidian data of • ...... Macdonald et al. (1992), Bailey ~0 0.1 A-type .J1980), 'Christiansen et al. ! 0.· (1986) and glass inclusion data. C3 Peralkaline summarized in Lowenstern () 0.03 (1995). B-Bingham (Cu) lava Plume o flows, Utah; PG - Pine Grove (Mo) tuff, Utah, glass 0.01 inclusions. 0.005 0.01 0.03 0.1 0.3 1 3 5 F (wt%) Tectonic Setting I-type silicic magmas are found overwhelmingly at convergent plate margins, including oceanic island arcs and continental magmatic arcs. As a group, I-type rhyolites and granites are largely the )~. / same as those identified by Pearce et al. (1984) as volcanic-arc granites. We have not separately distinguished sub-types of these granites (e.g., Pitcher, 1982; Macdonald et al., 1992) in this review because they share so many common characteristics. The sub-types probably represent variations in the amount of contamination by continental materials.

Example of I-type Magmatism at the Bingham Porphyry Copper Deposit. The felsic volcanic rocks associated with the Bingham porphyry copper deposit are examples of 131

strongly mineralized, subduction-related magmas. Bingham, Utah, is one of the largest known porphyry Cu deposits with reserves plus past production of more than 2,000 Mt (Babcock et al., 1995). Mo, Au, Ag, Pb and Zn have also been recovered. The deposit lies in the northeastern Basin and Range province and formed far inland in a wide continental magmatic arc that cut across Archean and Proterozoic basement buried by miogeoclinal sedimentary rocks. The mineralized stock lies in the central part of an east-tilted, basin-and-range fault block; this extension and tilting occurred during the late Cenozoic and are unrelated to the generation of the ore deposit. Middle Tertiary .volcanic rocks, exposed east of the mineralized intrusion, are cogenetic with the Cu deposit (Keith et at., 1995). The volcanic rocks consist of an older (40-37 Ma) and a younger (35-34 Ma) series. The older series of rocks contains a diverse assemblage of mafic alkalic lavas (biotite-bearing melanephelinites) to high-potassium, I-type lava flows, breccias and lahars that include shoshonites, latites and dacites (K. Waite, unpublished analyses). The younger series of breccias and lava domes has similar major element compositions, but includes more silicic lavas including high-silica rhyolite. The Bingham copper deposit developed around a shallowly emplaced quartz monzonite sto~k that, based on composition and age, appears to represent a subvolcanic chamber for part of the older series of volcanic rocks. The younger series is apparently unrelated to mineralization and must have been erupted from vents only slightly farther to the east. The compositions of the fresh igneous rocks at Bingham show that they have most of the characteristics of typical volcanic arc rocks, including high large ion lithophile abundances (K, Ba, Rb), high BaINb and Rb/Nb ratios and high MglFe ratios because of high magmatic f02. The lavas and fresh plutonic rocks are metaluminous (molar Al203/(CaO+Na20+K20) < 1.0, but molar Al203/(Na20+K20) > 1.0). In most respects, they are like continental arc rocks worldwide. The mineralized rocks are not highly fractionated, consistent with their association with Cu mineralization; Cu is a highly compatible element Glassy rocks that could have preserved magmatic F/CI ratios are rare in the older volcanic series at Bingham, but whole-rock data have F/CI ratios that are typically between 1 and 3 (Fig. 10). These ratios are lower than those found in glass from sub alkalic A-type rhyolites, such as topaz rhyolites. Volcanic glasses from other subduction-related felsic magmas extend to even lower FICl ratios. Loss of easily mobilized CI after eruption could account for the higher F/CI ratios in the Bingham lavas when compared with glassy I-type rocks. For example, biotites from the hydrothermally altered rocks at Bingham show that the orthomagmatic deposit formed at highjHClIjH20 and low jHFijHCI (Munoz, 1984). The characteristics of the mineralized series of lavas at Bingham. can be largely, if somewhat generally, explained as the result of oceanic slab dehydration in a subduction zone, rise of the buoyant fluid into the overlying mantle inducing partial melting to form diverse mafic magmas that : I rose into the crust, assimilating large proportions of the heterogeneous continental 'crust, fractionating, and experieI}cing repeated magma recharge with diverse magma types limiting the extent of fractionation. Magma mixing seems to have been especially important for the generation of the ore deposits at Binghain. Alkalic mafic magmas appeared to have mixed with more silicic I-type magmas to produce potassic intermediate magmas with high Cu, Cr and Ni contents. These mixed magmas erupted as shoshonites and latites whose parental magma chambers crystallized to form quartz monzonite that became Cu (Mo, Au)-mineralized. The elemental effect of mixing is most clearly seen in plots of highly compatible elements like Cr and Ni against an incompatible element like Rb (Fig. 11). The alkalic magmas, presumably derived from small degree partial melts of t t

132 E.H. Chrispansen and J.D. Keith •41 41 metasomatized Precambrian mantle lithosphere, added Cr, Cu, Ni, S, H20, Nb, K, Ba, and 41 perhaps Cl and C02 to the more fractionated magmas. Alkalic lavas at Bingham have the highest Cu 41 contents of any of the nearly 300 fresh mafic to intermediate composition lavas from the northern 41 Basin and Range province (Barr, 1993). Here, it seems that Cu, Cl, and S may have come from 41 ancient subduction-generated materials stored in veins or dykes in the lithospheric mantle formed in • the Precambrian and returned to much younger subduction-generated magmas of Cenozoic age. • Dehydration of oceanic crust in the Tertiary may have also contributed. t t t Ores Associated with I-type Magmas t The most important metallic ores associated with I-type igneous rock suites are Cu, Pb, Zn, Au, and Ag. Skarn deposits of Ware also an important association with I-type plutons on the • continents. Continental I-types are also enriched in Mo compared with those in·island arcs and • granodiorite-type Mo deposits (Mutschler et al., 1981) are commonly developed in I-type plutons. • Moreover, porphyry Cu deposits on the continents, such as Bingham, have substantial amounts of •t Mo. Climax-type Mo deposits are associated with siliceous, incompatible element and F-rich, oxidize!i magmas with some characteristics of A-type granites (White et al., 1981), including high .. • F, U, Be, and Rb. In fact, compositions of rhyolites associated with Climax-type Mo deposits in • the western United States (Pine Grove, Utah, and Henderson, Colorado) plot in the A-type field on • •(I Figure 11 Cr and Rb concentrations in volcanic 1400 rocks associated with the • • Bingham porphyry copper • 1200 • Bingham Utah deposit, Utah (K. Waite, • • •• • unpublished data). The o Younger Series • younger volcanic series is 1000 • f~ til Older Series unrelated to mineralization and • Mixing • ! \ evolved principally by crystal • Minette fractionation, fractionation • .E' 800 Trends~\ • Shoshonite a. trends are lines with ticks. • a. !. \ The older volcanic series is • ---"- 0 600 f ,. contemporaneous with • Fractional f+. \ mineralization and appears to Crystallization f+. \ • 400 have resulted from mixing Trend between Cu-rich mafic alkalic • magmas .( minette) and a • 200 fractionating sequence ofmore • ;~~ normal I-type magmas. • Mixing produced shoshonites oL-~--~~~~~~~~~~~~~~ • .0 50 100 150 200 250 and latites that are enriched in • compatible trace elements. Rb (ppm) • • Pearce diagrams (Fig. 7). The distinction between evolved A-types and I-types is difficult because • extensive fractionation produces common features in both types of rhyolites (see Pearce et al., • 1984). However, the highf02 (and the attendant mineralogy), spatial and temporal association with • normal calc-alkalic rhyolites, and low FICl ratios (Fig. 10) in melt inclusions from the • Mo-mineralized Pine Grove magma system (Lowenstern, 1995; Keith and Shanks, 1988), suggest • •t • 133

that these magmas may be highly fractionated I-type magmas, rather than A-type magmas. Blevin and Chappell (1992) and Blevin et al. (1995) persuasively show the relationships between I-type plutons and Cu, Pb, Zn, Mo, W, and Au are the result of source, oxygen fugacity, and extent of fractionation. For example, source control is illustrated by the observation that I-types are much more strongly associated with Cu-Au-Mo than with Sn, which is associated with S-types. In this review, the sources (and contaminants) of most I-type magma series are taken to be mixtures of metasomatic fluid derived from a subducting slab (high Cl, some metals, jHzO, and fOz), mantle-derived magmas (high Cu, Ni, Au), and continental crust (probable source ofMo and Pb). Varying proportions of these metal sources control the potential ore deposits that can form from a magma. High magmatic concentrations of HzO and Cl help subduction zone magmas produce ore-laden hydrothermal fluids. The role of magmatic fractionation can be seen in the association of '" Cu deposits with the least fractionated, most oxidized rocks, the association of W deposits with .. moderately fractionated rocks, and Mo and Sn with the most fractionated rocks (Fig. 12) . .. Apparently, Cu is depleted by fractional crystallization processes and W, Mo, and Sn are enriched. .- During the crustal residence of subduction zone magmas, there are great opportunities for hybridization or miXing with repeated injections of mantle-derived magma, forestalling magma •.. evolution by fractional crystallization and keeping compatible element concentrations (e.g., Cu) ,. high enough for mineralization and reducing the opportunity for development of incompatible • element (e.g., Mo) ores. Finally, the strong enrichment of Mo in oxidized I-types and lack of • enrichment of Sn is consistent with the inferred effect of oxygen fugacity on the behaviour of these .. elements in fractionating magmas . • S-Type Silicic Magmas • S-type granites occur as isolated plutons or in batholithic belts. Some plutons in both associations • have wide contact metamorphic aureoles and were characteristically emplaced at relatively great • depths. Strongly peraluminous rhyolites of Cenozoic age are rare. Our understanding of the " • compositional characteristics of this important group of magmas is limited to studies of granites and • rare, mostly altered volcanic rocks (e.g., Clemens and Wall, 1984; Zeck, 1970). The glassy • volcanic rocks from the Macusani region of Peru (Noble, 1984; Pichavant et al., 1988a, b) are • exceptions and are used here to represent some characteristics of fractionated S-type magmas. .1 • The single most important elemental characteristic of S-type silicic magmas is that they are • strongly peraluminous (molar Alz0 3/(CaO+NazO+KzO) >1.1). This characteristic is revealed in •III their distinctive mineral assemblages including cordierite, andalusite, muscovite, or garnet and as • accessories tourmaline, ilmenite, monazite, apatite and zircon (Clarke, 1981; Zen, 1988). Biotites • are typically more Fe-rich and aluminous than those of 1- or A-type magmas. S-type silicic magmas • are characteristically enriched in K (and Rb) and depleted in Na compared with I-types of similar • SiOz content (Chappell and White, 1992). The depletion ofNa explains why they are peraluminous • and is thought to be inherited from Na-depleted shales or greywackes in their sources. During • weathering at Earth's surface, Na (as well as Ca and Sr) is extracted from solids and transported to • seawater. At least some K and Rb is stored in clay minerals in shales and other sediments. • Consequently, sedimentary rocks commonly have higher KINa and ~blNa than their unweathered • protoliths (Taylor and McLennan, 1985). When these same sedimentary rocks participate in magma • genesis, as magma sources or contaminants, these characteristics are passed on to the derivative • • 4 4 4 E.H. Christiansen and J.D. Keith 4 4 8r------~ , • 6 I- ,• E a. , a. ~4 •411 t 2 • 66 68 70 72 74 76 78 • Si02(wt%) • ~r------~ • o • 16 (]]) o • o • 9 0 • o • 8 <> <> • <> • 4 o • •o • 66 68 70 72 74 76 78 Si02(wt%) • 100 r------. • I-type o • o • 80 • S-type • • • E 60 A-type • .9; o .c • a.. Peralkaline 40 <> •.. Plume o • 20 .. .0 .- 66 68 70 72 74· 76 78 Si02 (wt"Io)

Figure 12 Concentrations of various ore elements as a junction of Si02 content for various types of rhyolitic obsidians (Macdonald et al., 1992). The highly evolved Macusani glass is the only S-type obsidian shown and its ~ ~ concentrations of Sn (250 ppm) and W (76 ppm) are too high and Mo content (0.3 ppm) too low to be shown. A-type rhyolites are generally the most,enriched in the incompatible ore elements, except for Mo, where evolved I-type rhyolites are just as enriched, anq the reduced Macusani glass most depleted. =~

~ =-' 135

magmas. Less soluble elements like Nb and Y, are not strongly mobilized by weathering processes, but they may become enriched as residual elements in mature shales. Since most continental rocks have a heritage that traces their origins to a subduction zone, most immature sedimentary sequences also have the hallmarks of arc magmas, namely high Ba!Nb, Rb/Nb, and LalNb ratios. Consequently, in the discriminant diagrams of Pearce et aI. (1984), S-type granites are distinguished by high Rb compared with I-type granites. Nb and Y may be enriched compared with I-types, by the residual processes noted above and by partial melting and fractionation processes (Fig. 7). The aluminous compositions of these magmas also explain other compositional features. For example, many S-type magmas are strongly depleted in REE, Zr, c:md Th compared to comparably siliceous I-types. These features are probably not source related but are the result of the low solubility of accessory phases, such as zircon, monazite and thorite in peraluminous melts. Depletion in these trace elements is probably a two-stage process, occurring during partial melting r' events that leave accessories in the residue and continuing during fractional crystallization of the •.­ magma. In highly evolved, peraluminous magmas, xenotime and columbite lave low solubilities ,. and may fractionate to further increase the already high Rb/(Nb+Y) ratio and, in extreme cases, driving the residual melt composition to the left on Pearce diagrams (Fig. 7). Such a process may • explain the relatively low Nb and Y concentrations of Macusani rhyolites. On the other hand, apatite is highly soluble in peraluminous siliceous magmas; this controls the concentration of P20S in the •i. " melt (Pichavant et aI., 1992). For example, Chappell and White (1992) have shown that P20S I' decreases during the fractionation of I-types much more rapidly than in S-types, which may even I • show P enrichment (Fig. 13) . ..,

.. ,1'1 :, I •.. 0.25 ,I' I-type Ii ~ • • • 0.2 • S-type 0 • • •• • • OR-~pe • ,. • • • •• • ~0.15 • • • A-type• • • 0 • ! ••• • ~ LO 0 • • • N • • • • Cl.. 0.1 • • -• • • • -.~ • • • • ••0 •• -•• - • 0.05 • •• • • • • 0 • 0 100 200 300 400 500 600 • Sr (ppm) • Figure 13 Phosphorous contents of S-type granites and rhyolites are significantly higher than those of other • granite types because of greater solubility ofphosphate minerals in peraluminous magma. Data from Chappell et • al.(1991), Macdonald et al. (1992), Pollard et al. (1995), Pichavant et al. (1988a), and Clemens and Wall (1984). • •.. ~; .. II II II • I 136 E.H. Christiansen and J.D. Keith 'i .,• : High concentrations of B and the presence of tounnaline are particularly good indicators of •til S-type silicic magmas. B is strongly concentrated in sediments (approximately 100 ppm) compared til to the mantle (0.1 ppm; Ryan and Langmuir, 1993). Apparently this characteristic is passed on to metasedimentary rocks and to magmas derived from them. Highly evolved A-type rhyolites have • much lower B and tourmaline is rare in their intrusive equivalents. The A-type Spor Mountain • vitrophyres of western Utah, for example, have less than 100 ppm B and whereas peraluminous •tI! glasses from Macusani have over 2000 ppm (Pichavant et aI., 1988a), but both have comparable .. enrichments of Rb and Be, elements with similar partition coefficients. • Another important characteristic of S-type granites and rhyolites is low f02. Many of these .. granites lack magnetite and ilmenite is the only Fe-Ti oxide. This fact was recognized by Ishihara (1977)~ but, as he points out, not all ilmenite-series granites are S-types, many are contaminated and reduced I-types. Low oxygen fugacities can also be inferred from the low Fe203/FeO ratios (Fig. 9; Blevin and Chappell, 1992) and high Fe/Mg in mafic silicates. The compositions of coexisting minerals in the S-type Macusani volcanic rocks of Peru also show that they crystallized at low f02, below QFM (Pichavant et aI., 1988b ). Isotopically S-types have the characteristics of the crust from which they are largely derived, including high 8180 (reflecting enrichments during a weathering cycle; Table 1), and generally high Sr and low Nd isotope ratios (Chappell and White, 1992). However, Sr and Nd isotopic compositions are sensitive to the age of the source, not just the parent/daughter ratios. Consequently, S-types derived from young sedimentary rocks do not have high Sr isotope ratios. The S-types granites of the New England fold belt of eastern Australia illustrate this relationship (Shaw and Flood, 1981).

Tectonic Settings S-type granites are found in several settings that are generally related to plate convergence. Perhaps the best constrained are the S-type granites that clearly formed because of the collision of India and Asia. Moreover, many S-types are synkinematic and foliated. These facts led to the notion that S-type granites are syn-collisiorial granites, as in the geochemical classification of Pearce et al. (1984). However, abundant S-types are also found in other plate convergent settings, where they formed as parts ofI-type suites. The S-type granites may slightly pre- or post-date the other granitic rocks in a given area. Most of these associations formed far from the convergent plate margin (pitcher, 1982) and in regions underlain by thick sequences of metasedimentary rocks including metamorphosed shales, greywackes, and sandstones. Some members of the strongly peraluminous granites of the western Cordillera of North America are examples of this setting (Miller and Bradfish, 1980; Lee et al., 1981) and eastern Asia (Takahashi et aI., 1980), although some muscovite:..bearing plutons are more appropriately seen as strongly contaminated I-types because they lack all of the characteristics of S-type magmas. Pitcher (1982) separates these tectonic settings into a Hercynotype (collisional) and Andinotype (subduction without large collisions). In the latter, S-types form a much smaller proportion of the granites. The cause of the melting that produces S-type silicic magmas varies by tectonic setting. Those that form above still active subduction zones probably form as a result of melting induced by mafic mantle-derived magmas, with which many undoubtedly mixed and hybridized both before and after silicic magma was produced. Other S-types lack a clear association with such mafic magmas and -~ -.. ------~------

137

probably formed as a result of volatiles flowing from a zone of metamorphic dehydration to higher levels of the crust (Fyfe and Kerrich, 1985), as a result of heating following tectonic thickening of continental crust during collision (pitcher, 1982), or as a result of decompression following tectonic unroofing of collisional orogenic belts (Hodges, 1990).

Examples of S-type Rhyolites and Granites The granites associated with the large W skarns of western Canada, MacTung and CanTung, are S-type granites, and contain biotite, muscovite, and gamet (van Middelaar and Keith, 1990). , • I MacTung reportedly contains 63 m.t. of ore with 0.95% W03 and CanTung contains over 6 m.t. of I ore at 1.6% W03 (Einaudi et ai., 1981). Granites that caused both deposits lack magnetite and have Fe-rich biotites. Rb concentrations are moderately high, indicating extended fractionation and potential enrichment in W set the stage for ore generation. CanTung granites are more fractionated

~ than those at MacTung and are related to the higher grade W deposit. Nb and other high field ,. strength elements, especially Zr, have low concentrations and the granites plotin the S-type field of ..r Figure 7. Biotite compositions reveal that these plutons crystallized and became mineralized at ~ relatively highjHFifHCI compared with I-type granites (van Middelaar and Keith, 1990). The Sn-mineralized Phuket Supersuite of southwest Thailand displays many characteristics of ,.• S-type granites and illustrates some processes that are important in the' generation and mineralization ,. 86 of such granites. The evolved Sn-mineralized plutons of the suite have high 87Sr/ Sr ratios ..,.r (>0.710), peraluminous compositions (molar Ah03/(CaO+Na20+K20) = 1.05-1.17), 10wNa/K ratios, low 102 with ilmenite and rutile as the characteristic Fe-Ti oxides, and gamet, monazite and ~ tourmaline. Some granites are also topaz-bearing. The trace element characteristics of these granites .. are like those of other S-type granites with high Rb/(Nb+Y) (Fig. 7), high B (20 to 1300 ppm), !it low Zr (15 to 90 ppm), low REE concentrations, and high P205 (about 0.1 wt% in evolved rocks ill with about 1000 ppm Rb). These arethe characteristics of a fractionated magma with substantial •[ proportions of a weathere~ sedimentary protolith. Such are cool peraluminous magmas with :low • solubilities of monazite and zircon and high apatite solubility (pichavant et ai., 1992). Nonetheless, II• Pollard et al. (1995) conclude that the granites in this association resulted from extensive crystal .. fractionation of I-type magmas and were not derived from a metasedimentary protolith. As Pollard et al. (1995) argue, high 87Sr/86Sr ratios may be derived from meta-igneous rocks, but it is difficult •.. to understand how such a protolith could produce strongly peraluminous magma with the observed .. phase stabilities, fractionation trends, high B and P, very high RblNb ratios, and a f02 low enough •.. to preclude magnetite crystallization and induce Sn incompatibility. Without the inclusion of sedimentary materials, closed system fractionation of an I-type silicic magma, produces residual •.. magmas that are metaluminous, low in B, P205, and Sn, and relatively oxidized. Unevolved and .. unmineralized portions of the supersuite are oxidized I-type granites. Consequently, the • compositional characteristics of the mineralized plutons of the Phuket supersuite probably • devdoped as sedimentary materials were incorporated in mantle-derived I-type magmas. The • Phuket granites are probably typical of many suites that evolve from subduction-related I-type • magmas to S-type magmas as high proportions of weathered, carbon-bearing sedimentary rocks are • assimilated. • • • ..• 138 E.H. Chri*ansen and J.D. Keith •.. Ores Associated with S-type Magmas Many important ore deposits are associated with S-type granites. The most characteristic are Sn and • W. The Sn deposits of southeastern Asia are associated with S-type granites (Beckinsale et al., • 1979). And as noted above, W deposits in the Canadian Cordillera are associated with • peraluminous muscovite-garnet granites with the trace element characteristics of other S-type • granites. Be, Li, and Ta are important in phosphate-rich pegmatites derived from strongly • peraluminous magmas, such as the Greenbushes pegmatite, western Australia, which contains half .. of the world's Ta resources (Partington et al., 1995). .. . The principal controls on the ore associations of S-type granites include the modest enrichments .. of these ore elements in the weathered continental crustal materials from which the granites are • domimintly composed. Most other granite types contain significant fractions of mantle-derived .. dilutants to these generally incompatible elements. Perhaps the most important control for the .­.. abundance of Sn deposits, however, is the low 102 and not any substantial Sn enrichment in the ... source materials (Lehmann, 1982). This condition probably reflects the incorporation of organic carbon derived from carbonaceous sedimentary materials. Sn behaves as an incompatible element .. and becomes strongly enriched during fractionation of reduced magmas in contrast to strongly .. III! fractionated but uncontaminated I-type granites. Magmatic fractionation must be an important .., control on the enrichment of W in S-type granites, though the incompatibility of W is not as .. sensitive to extreme oxidation state as either Sn or Mo (Keith et al., 1993). Another important ., control is the peraluminous character of the magma which increases the solubility of phosphorous . (pichavant et al., 1992), allowing phosphate-rich rare element pegmatites to evolve by fractional • crystallization or very small degrees of partial melting. Low Nb/Ta ratios may be achieved by • extended fractionation, since niobium-rich minerals begin to crystallize in low-temperature .-• peraluminous magmas. Unfortunately, there are few peraluminous obsidians to compare with the .. A- and I-type obsidians in Figure 12. The extremely high conc;entrations of Sn (250 ppm) and W ~ (76 ppm) in the Macusani glass (Macdonald et al., 1992) are probably not representative of most • S-type granitic magmas. On the other hand Mo contents (0.3 ppm) were much lower in the .. fractionated but reduced Macusani magma as compared to fractionated and oxidized I-types, which •." I can contain as much as 8 ppm Mo (Fig. 12). •~ A· Type Silicic Magmas ..• Of all the silicic magma types considered here, A-type granites and rhyolites may be the most A diverse and, perhaps, the least understood. The has variously connoted alkalic (as in peralkalic), . l anorogenic (as in rifting- or plume- as opposed to subduction-related), or even anhydrous, although = each of these descriptive terms is questionable if applied universally. First, a major category of these rockS is metaluminous or marginally peraluminous, not peralkalic (molar Al20yf(Na20+K20) ! " = < 1), although all have high alkali contents compared with 1- or S-types. Second, the use of the • j term "extensional orogeny" makes it difficult to call silicic magmas "anorogenic," if they form = during episodes of crustal extension and rifting; the meaning is clear when used in the petrologic ~ context to contrast these rocks from "orogenic" andesites and rhyolites like those found above = ~ subduction zones. Third, a large group of these rhyolites and granites are found in orogenic belts and formed shortly after the cessation of contractional orogenies. These post-orogenic rocks are difficult to categorize. Finally, highly fractionated A-type magmas are water-rich, although it is ..~ ..~ ~ 139

probably important that their sources were relatively water-poor. Although we categorize rhyolites and granites of this class as peralkalic or sub alkalic (metaluminous to slightly peraluminous), both classes share some important compositional features, including high Si02, Na20+K20, Fe/Mg, and low Al203, CaO, and P205. Most A-type magmatic rocks have Si02 contents greater than 68%. Compared with granites of similar Si02 content, common trace element characteristics are low Rb/Nb ratios (Fig. 6), the absence of deep Nb-Ta anomalies on chondrite-normalized trace element diagrams (Fig. 8), enrichments of many high-field strength elements (Nb, Ta, W, Y and especially Zr and Hf in the peralkalic varieties), enrichments in Be, Zn, 9a, Sn, Pb, Th, U and the halogens (high F/CI in aluminous types and lower F/CI in peralkalic types, Fig. 10), and characteristically lo~ concentrations of compatible elements (Mg, Ca, Ti, Sr, Eu, Ba) (e.g., Collins et al., 1982; Anderson, 1983; Whalen et al., 1987; Ramo and Haapala, 1995; Macdonald et al., 1992). Typical A-type silicic rocks also have low Y/Nb and high GalAI ratios. A wide range off02 values has been recorded for A-types, from near QFM to several log units above that (e.g., Anderson and Bender, 1989). In the Macdonald et al. (1992) compilation of obsidian analyses, Fe203/FeO ratios of nonperalkalic A-type rhyolites are comparable to those of ,. I-types (Fig. 9), although Fe/Mg ratios are consistently high. Peralkalic rhyolites probably have high Fe3+/Fe2+ ratios because of the effect of alkalis in stabilizing Fe3+ in silicate melts and not •t­ because of an inherently highf02 (Kress and Carmichael, 1991). Volatile fugacities in peralkalic ,. rhyolites are poorly constrained because the mineral assemblages are largely anhydrous and Fe-Ti Ie oxides are not stable in many peralkalic melts. However, the f02 in most peralkalic rhyolites is ,. probably near QFM, as judged from Fe20~eO ratios and by the common appearance of fayalite '-r and quartz. Absolute, water concentrations in peralkalic rhyolites are controversial. Kovalenko et al. (1994), from Pantelleria, and Webster et al. (1993), from Fantale, Ethiopia, have reported as -­ much as 4 wt% H20 in melt inclusions included in quartz in peralkalic volcanic rocks. However, •.. Lowenstern and Mahood (1991) found less than 2% H20 in their glass inclusions from Pantelleria. • Few hydrous minerals are stable in peralkalic melts; amphibole is one of the few and in sub alkalic • magmas amphibole is stabilized at higher dissolved water contents than biotite. Most A-type • rhyolites and granites have high concentrations of fluorine. For example, Precambrian A-type .. granites from the southwestern United States have significantly higher concentrations of F than any Ie of the Phanerozoic granite types from that region (Christiansen and Lee, 1986). An important .. characteristic of peralkalic magma is a low F/CI ratio. In an absolute sense, many peralkalic silicic ..• magmas have higher concentrations of F than many sub alkalic magmas, but they are consistently more enriched in Cl and consequently have low FICI ratios (Fig. 10), sometimes as low as those in I .. I .. glassy I-type magmas but IIl;ore generally like those of mafic magmas derived from mantle. plumes. • High FICI ratios are not m~aningful in granites or in most volcanic rocks, because halogens, like .. water, have few mineralogic sites in fully crystalline rocks. Granites and crystalline rhyolites .. generally have FICI ratios that are higher than the magma from which they crystallized because F is lilt more compatible in mineral structures than Cl. Fluorine is hosted by fluorite, micas, apatite, topaz, .. and in other silicates, where it may replace 0, in greater quantities than Cl. Consequently, useful .. comparisons of the halogen contents of silicic magmas can only be made with glassy volcanic ..• rocks, and even these may have lost halogens during eruptive degassing (Fig. 10). Glass inclusion ..• III """t t 140 E.H. Christiansen and J.D. Keith • studies may provide better estimates of the magmatic concentrations of these mobile elements; those • done so far agree with bulk rock studies and show that sub alkalic magmas have FICI ratios that • exceed 1 (Webster and Duffield, 1991; X.Y. Zhang and E.H. Christiansen, unpublished). • Many of the sub alkalic group of A-type granites become enriched in heavy REE (HREE) relative • to light REE (LREE) during fractional crystallization. In most I-type magmas LREEIHREE ratios • increase during fractionation because of the smaller partition coefficients for the slightly larger • LREE. Consequently, when this. trend was first observed (Hildreth, 1979), liquid-state • differentiation processes were invoked, such as thermogravitational diffusion or the Soret effect. • These processes have been shown to be unimportant in natural silicic magmas. In A-type rhyolite • magma, the principal control on changing LREE IHREE ratios is the accessory mineral assemblage, • which commonly includes several minerals with strong affinities for LREE, such as allanite, • chevkinite, xenotirne, huttonite, or monazite. I• Many features of A-type granites and rhyolites could result from extreme fractionation. Note the , overlap of many incompatible element concentrations on Figure 12. Thus,-the question naturally arises, are these rhyolites simply fractionated I-types or S-types? Sometimes these disti:nctions • between A-types and these other granite types are clear, but ambiguities are more common. Whalen • I et al. (1987) conclude that high GalAl ratios can be used to distinguish fractionated A-type granites • from the other types. Most A-type granites have 1000xGalAl > 2.6, although some fractionated S­ • and I-types may exceed this discriminating value. They also use Zr+Ce+Y > 250 ppm as a • dis9rirninating parameter. Although this feature may be characteristic of fractionated alkalic A-types, • the sub alkalic types have low solubilities for the accessory phases (zircon, allanite, monazite and • xenotime) that control the concentrations of these phases and Zr+Ce+Y commonly drops during • fractionation. We have found that low RbINb ratios (less than 7) are a simpler way to distinguish • most A-type granites from I- and S-types (Keith et al., 1993) and this ratio is less sensitive to the • effects of fractionation or contamination (Fig.s 3 and 5). However, niobates form in· some highly • fractionated A-type granites and may raise RbINb ratios to higher values. For example, highly • evolved rhyolites with topaz and complex Nb oxides such as the Honeycomb Hills rhyolites of • western Utah (Congdon and Nash, 1991) and some ongonites from the Mongolian type locality • (Kovalenko and Kovalenko, 1976) have Rb/Nb ratios that exceed 15. High FICI ratios and low j02 • may also be helpful in distinguishing some A-type granites and rhyolites from I-types. Highly • fractionated S-type magmas may become F-rich and crystallize topaz, but they can usually be • distinguished from topaz-bearing A-types by much higher P20S and B contents that are manifest in • abundant phosphate minerals and tourmaline in S-type leucogranites and pegmatites. • • Tectonic Settings • A-type rhyolites and granite;s are found in various tectonic settings. Such silicic magmas are • characteristic of those erupted above mantle plumes, both on the continents (the Snake River Plain • and Yellowstone Plateau) and in the ocean basins (Iceland). However, the most common • association of A-type silicic magmas on the continents is with rifting, where they are the silicic • end-members of bimodal volcanic associations. This association is especially common in the • western United States where lithospheric extension since about 20 Ma has been accompanied by the • eruption of bimodal suites of basalt and rhyolite, with rare intermediate magma (Christiansen and • Lipman, 1972). Peralkalic rhyolites and sub alkalic ,rhyolites and granites with topaz are the • •II • 141

characteristic products. The tectonic relations of the Proterozoic granites of A-type affinities that extend in a wide belt from southwestern North America to the Baltic Shield remain controversial, although good evidence for crustal thinning, extension, and mafic dyke intrusion has been described for the rapakivi granites of southern Finland (Ramo and Haapala, 1995). The tectonic association of "post-orogenic" A-types is also problematic. For example, the late Precambrian granites of the Arabian shield are A-types and formed during the [mal stages or immediately after the collisional accretion of the Pan-African Orogeny (Stoesser and Camp, 1985). The A-types of southeast Australia appear to have formed in a similar late orogenic association, but here evidence of bimodal magmatism and at least local extension is found (Collins et at., 1982). The main item these settings have in common is a lack of input of mafic magma from a subduction zone. Instead, the mafic magmas are the result of processes related to lithospheric extension andlor mantle plumes.

Examples of A-type Rhyolites and Granites The rhyolites of the Spor Mountain Formation exemplify the sub alkalic group of A-type silicic magmas. The Spor Mountain Formation consists of topaz-bearing rhyolite erupted about 21 Ma (Lindsey, 1982) and marks the initiation of extension in the Basin and Range province in western Utah. Hosted within a tuff and lava flow complex, is North .America' s largest deposit of Be. Davis (1984) estimates that the total resource is at least several million tons of bertrandite ore that averages about 0.7% BeO. Nearby, a topaz-bearing granite of the same age also contains topaz and magmatic (?) beryL The rhyolite vitrophyres are enriched in Be (about 60 ppm) and F (as much as 2% in glass inclusions preserved in quartz phenocrysts and 1.2% in vitrophyres) and consequently devitrified rocks commonly have topaz as a vapour phase mineral. FICI ratios in the glasses are characteristically high (Fig. 10). Magmatic phases include smoky quartz, ~anidine, oligoclase, Fe-rich biotite and accessory zircon, monazite, thorite, uraninite and complex oxides of Fe-Ti-Mn-W-Nb-Ta. The magmatic.fO.z was apparently low in the Spor Mountain rhyolite (near QFM), but other aluminous A-type rhyolites were more oxidized and crystallized magnetite and titanite. On Pearce discrimination diagrams (Fig. 7), these rhyolites plot within the A-type granite field, but as a group Cenozoic topaz rhyolites from the region straddle the A-type and S-type II boundary. Isotopically the Be-mineralized rhyolite and granite have low 8180 (6 to 10 per mil), relatively low initial Sr isotope ratios (0.706), and high ENd (-6). This latter observation is

important because S-type, and even some I-type, granites in the same region have ENd as low as ;11 -18 (Farmer and DePaolo, 1983) and interacted extensively with the early Proterozoic or Archean

crust that underlies the region. A large component of mantle-derived basalt must be present in the i!1 magma, probably because of basalt intrusion in the crust followed by partial melting, although this interpretation is non-unique. Extensional tectonics and attendant mafic magmatism, intimate hybridization of the lower (?) continental crust, coupled with low degrees of partial melting «10%) and an extended fractionation history, explains most of the isotopic and trace element characteristics of sub alkalic A-type granites in general. The peralkalic A-type rhyolites have even higher concentrations.ofhigh field strength elements like Zr, Hf, Nb and REE than the subalkalic varieties (Fig. 8). These enrichments are partially the result of source enrichments because peralkalic silicic magmas generally arise from mafic magmas derived from mantle unaffected by subduction zone processes. As noted above, the oxidized, hydrous, silicic (?) magmas generated above subduction zones are depleted in Nb, as compared • 142 E.H. Christiansen and J.D. Keith • with other incompatible elements and consequently have negative anomalies on • chondrite-normalized trace element diagrams. The enrichment of high field strength elements, • especially Zr and Hf, also results from the high solubility of many accessory minerals in peralkalic • magmas (Watson and Harrison, 1983; Montel, 1986). For example, zircon solubility is increased • by 4 to 10 times in peralkalic rhyolite as compared with typical metalurninous rhyolite at the same • temperature and water content. This effect is obvious when Zr, with its variable behaviour, is • compared with an element like Rb that is incompatible in both peralkalic and sub alkalic silicic • magmas (Fig. 14). Likewise, monazite, apatite, allanite, tho rite and columbite-tantalite either are • unstable or have high solubilities in typical peralkalic magmas (e.g., Montel, 1986). Consequently, • LREE, Th, Nb, and Ta (trace elements accommodated in these accessory minerals) all behave as • strongly incompatible elements, in contrast to their moderately incompatible or even compatible • behaviour in many sub alkalic A-type magmas. Most peralkalic rhyolites show evidence for • extensive fractional crystallization. Ca, Mg and P are strongly depleted, almost to the point of being .. trace elements in some strongly peralkalic magmas. Feldspar-compatible trace elements always have .. extremely low concentrations. .. The compositions of peralkalic rhyolites from the McDermitt Caldera, northern Nevada (Rytuba • and McKee, 1984), are shown in Figures 7 and 8. The Miocene McDermitt caldera complex formed • contemporaneously with the onset ·of Basin and Range extension in the region. RiftIng may have .. been related to the initial rise of the plume speculated to presently lie beneath the Yellowstone .. Plateau (Zoback et at., 1994). Most of the silicic magmas are mildly peralkalic ash-flow tuffs and, .. lava domes and their compositions fall in the A-type granite field on Pearce discrimination diagrams .. because of their high concentrations of both Y and Nb (Fig. 7): However, the Rb concentrations ~ are generally lower than in the aluminous A-types with comparable Y + Nb concentrations. FICI .~.. ratios in glassy peralkalic samples are lower than in the aluminous varieties, and more like those found in plume-related rhyolites and basalts (Fig. 10) Chondrite-normalized trace element patterns ~ illustrate the lack of Nb and Zr depletions (Fig. 8), that reflect the high solubilities of the controlling .~• . accessory minerals and a probable mantle derivation from non-orogenic magmas coupled with little· continental contamination. Mantle-like Sr and Nd isotope ratios are consistent with this suggestion ~ . (Tegtmeyer and Farmer, 1990). Extensive fractionation of this mantle-derived magma apparently ~ caused the enrichments of incompatible elements. Barium, Sr and Eu are strongly depleted by extensive feldspar fractionation from less siliceous parental magmas. The McDermitt caldera complex is highly mineralized with deposits of Hg, Sb, Cs, Li and U. As a result of extensive =~ fractionation, the peralkalic tuffs had high contents of these elements and were the source rocks .. from which these metals were leached by hydrothermal systems late in the development of the .. caldera (Rytuba and McKee, 1984). Peralkalic A-types generally have lower U and Th than the .. ~ subcilkalic A-type rhyolites and granites of the continents and, in this regard, are similar to those

found on ocean islands. - I Silicic igneous rocks found on ocean islands related to mantle plumes, though rare, have many = characteristics of A-type granites found on the continents including high concentrations of high field strength elements, including especially high Nb and Ta, and volcanic series that include peralkalic rhyolites. The compositions of these oceanic silicic rocks plot in the within-plate granite field of Pearce et at. (1984) and the A-type field Whalen et at. (1987). They are nonetheless distinctive from A-types found on the continents. A typical feature of these rhyolites is lower concentrations of large 143

1400 Figure 14 Rb and Zr Rhyolite Obsidians 1200 concentrations in peraluminous • • Peraluminous compared with peralkaline • rhyolites. Apparently, Rb is an 1000 • • Others incompatible element in • • peralkaline and peraluminous silicic magmas, whereas Zr is an 800 Ec. •• incompatible element in .::, -- ... peralkaline magmas, but is a ~ I, 0::: 600 i • compatible element in peraluminous magmas in which Zr concentrations (as low as 50 400 ppm) are limited by the insolubility andfractionation of 200 zircon. Modified from , Christiansen et al. (1984) by addition ofdata from Macdonald .. I etal.(1992). 200 400' 600 800 1000 Zr (ppm) ion lithophile elements, like Cs and Rb, than other siliCIC magma type, excepting the rare granites found at normal mid-ocean ridges. Consequently, these rhyolites occupy a distinctive part of the RblNb-Rb diagram shown in Figure 6, with nearly chondritic Rb/Nb ratios. They are more enriched in Rb than OR-type silicic magmas and have higher RblNb ratios. Other distinctive . I features of oceanic plume-type rhyolites are their high REE (including Eu), Zr, Hf, Sn and Zn contents in both peralkalic and sub alkalic rocks and low Pb, Li, Rb, an~ Cs. , , Rhyolites from Iceland and the Galapagos islands typify the oceanic plume association. These , ' rhyolites are all metaluminous or peralkalic. Their characteristics are illustrated on the Pearce discrimination diagrams (Fig. 7) and on chondrite-normalized trace element diagrams (Fig. 8). F/Cl ratios in melt inclusions and obsidians from metaluminous silicic rocks of oceanic islands are about 1 or higher. Peralkalic rocks from these oceanic islands have lower FICI ratios of about 1 to 0.5, ,. typical of the more mafic rocks on oceanic islands (Schilling et al., 1980) consistent with a direct lie link between the basalts and rhyolites by fractional crystallization. In fact, most studies of these rhyolites emphasize the role of fractional crystallization from basaltic parents through trachytic • intermediate compositions. The low concentrations of large ion lithophile element, compared with • continental A-types (and even many 1- and S-types), are likely the result of the absence of ..• continental crustal contamination of subduction zone fluids. • Sources of A-type Magmas • These two varieties of A-type silicic magma probably have fundamentally different proportions of .. I • source materials (e.g., Christiansen et al., 1986; Eby, 1992). The peralkalic varieties almost .. certainly are dominated by mantle materials. This is supported by their commonly low initial Sr .. isotope ratios and high Nd isotope ratios (e.g., Kovalenko et al., 1995; Tegtmeyer and Farmer, .. 1990). From the perspective of trace element concentrations, it would be difficult to produce a lilt ie silicic magma with no Zr depletion by partial melting any type of crustal material. The characteristic lilt lack of Zr anomalies in peralkalic rhyolites is strong evidence that they evolved from mafic magmas with little crustal input. Likewise, the absence of negative Nb anomalies in these rocks argues not .. \. ..• p

144 E.H. Christiansen and J.D. Keith

only against the involvement of contemporaneous subduction-derived magma, but also against the incorporation of continental materials in generaL Moreover, the low FICI ratios characteristic of peralkalic rhyolites are also consistent with a mantle origin. Continental crustal materials generally have high FICI ratios because of the absence of mineralogical sites for Cl in the common hydrous minerals. Biotite, amphibole, and apatite in crustal rocks are almost always F-rich. This is particularly true of high-grade metamorphic rocks such as those proposed to be the source of A-type granites (Filippov et al., 1974). Many sub alkalic A-types probably evolved from magmas that included large proportions of crustal material (e.g., Collins et al., 1982; Christiansen et al., 1988; Clemens et al., 1986; Anderson and Bender, 1989; Whalen et al., 1987; Ramo and Haapala, 1995). This is suggested by their Nd and Sr isotopic compositions, substantial depletions of Zr, and high FICI ratios. However, the trace element patterns, similar incompatible element ratios in silicic and adjacent mafic magmas, and isotopic compositions argue that partial melting occurred in a zone where abundant mafic magmas lodged and hybridized with the crust before these small degree melts were produced and extracted (e.g., Hildreth et al., 1991). Unless melt extraction from, or metamorphic dehydration of crustal protoliths can produce both low RblNb and RblSr ratios and high SmlNd ratios, a significant mantle input is suggested for most continental A-types. The fact that these same features also characterize A-type silicic magmas in the ocean basins is consistent with a mantle heritage. The extensional tectonic setting of many of these magma systems allowed small melt fractions to segregate easily into dykes and did not promote extensive mixing between crustally derived melts and the contemporaneous basalts, as is characteristic of I-type magma series generated in subduction zones in which intermediate composition magmas are common. One of the most important processes that creates the trace element characteristics of sub alkalic A-type magmas is a small degree of melting of a hybridized crustal protolith. Melt fractions of less than 15% have been invoked in several studies as leading to large initial enrichments in incompatible trace elements (e.g., Christiansen et al., 1988). Assuming batch partial melting, only m.oderate depletions of compatible elements are produced and consequently primitive A-type melts may have high Ba for example. Small melt fractions may be the natural result of melting being induced by the decomposition of small amounts of biotite or amphibole in a nearly hybridized with volatile-poor mafic magII}a or in a high-grade metamorphic protolith. Subsequent strong fractionation in closed or open systems appears to be required to create the observed depletions in many A-type rhyolites and granites. An extensional tectonic environment is probably critical for the segregation of small volumes of melt from their protoliths. In such an environment, dykes could rapidly fill with newly formed magma and be drained out of a crustal source region heated by basalt intrusion. These melts appear to rise without substantial mixing with the mantle-derived magmatic heat sources because of the presence of extension-related crustal fracture systems and, in some cases, relatively low flux of mantle-derived magma. In compression environments, the accumulation of a relatively large volume of buoyant magma may be necessary to overcome the strength of the rock and induce rise and separation. Before this can occur, new mafic magmas may be added to~ the melting region. Apparently in supra-subduction zone settings, magma mixing is common and suppresses the chemical effects of fractionation. ~e~======~~~~~~======~====~~======~' •t I • 145 •I Ores Associated with A-type Granites and Rhyolites Ore deposits directly related to peralkalic silicic magmas are sparse, but include a variety of • elements that become concentrated by fractional crystallization in peralkalic magmas but lack • insoluble mineralogical hosts. Notable ore deposits include, Hg-Sb-Cs-Li-U deposits in the • peralkalic McDermitt caldera of Nevada (Rytuba and McKee, 1984), epithermal Au deposits are •I associated with sub alkalic rhyolites in Oregon and peralkalic rhyolite and trachtye intrusions in the I Black Hills of South Dakota, Be-Zr-Nb-REE mineralization is associated with peralkalic granite and • syenite at Thor Lake, Northwest Territories, and in peralkalic pegmatites of Khaldan-Buregtey, • Mongolia (Kovalenko et al., 1995). • Ores related to sub alkalic A-types include Be, U, Li, Sn, and Mo. Small Sn deposits are found • in topaz rhyolites in western North America (Huspeni et al., 1984; Eggleston, 1987) and larger • deposits are found associated with A-type granites in Jiangxi province, China (lowP topaz granites • of Zhu, 1990) and Nigeria. Both kinds of A-type granite are found in Nigeria, but the biotite • granites are more mineralized, here with Sn and lesser Nb-U-Zn-W (Bowden and FO.nnaird, 1978; • Imeokparia, 1982). It is important to note that A-type rhyolites are generally no more enriched in • Mo than the oxidized I-types; this contrasts with distinctly higher Be, Sn, Zn, U, etc. in the A-types • (Fig. 12) and is consistent with the control of Mo incompatibility by high f02. Be ores and gem • red-beryl (Keith et al., 1994) are extracted from separate topaz rhyolite dome complexes in western • Utah that are also related to small deposits ofU and Li. All of these ore elements are incompatible • elements that become strongly enriched by fractional crystallization and low degrees of partial • melting. Moreover, this suite of elements has moderately higher concentrations in the continental • crust, which appears to play an important role as source or contaminant in these non-peralkaline • silicic magmas. • The suite of enriched elements in A-type magmas then is a function of source (nonorogenic • mantle for peralkaline varieties versus a similar mantle plus crust for sub alkaline varieties), small • degrees of melting, coupled with extensive fractionation in whichj02 and peralkalinity control that • nature of the fractionating phases. Peralkaline series can become dramatically enriched in Zr, Nb, • U, and Th because minerals rich in these high field strength elements are much more soluble in • peralkaline than sub alkaline magmas. During fractionation of magmas with a continental crustal i • component, reduced magma series can become enriched in Sn and oxidized variants in Mo. All will I , I • become enriched in Be, U, Li and other incompatible elements. The apparent lack of continuous •It mafic magma recharge into A-type magma chambers may allow fractionation to dominate the It diluting effect of magma mixing that is so common in I-type magma associations. High F concentrations in peralkaline and sub alkaline varieties also assist in lowering the magma's so1.idus • temperature (Manning, 1981) and viscosity (Baker and Vaillancourt, 1995) to allow for more •It effective crystal-melt separation and enhance the enrichment of incompatible elements. Peralkalic It melts have even lower viscosities because of the depo1ymerizing affect of the alkalis (Baker and It Vaillancourt, 1995). It It It OR-type Silicic Magmas It Rhyolites and granites found at ocean ridges are rare and small in volume, but have distinctive t features that contrast with other types. Consequently they are informative about the sourceS and t processes that are important for silicic magmas in generaL Most of these silicic rocks are

•t t t 146 E.H. Christiansen and J.D. Keith

metaluminous to slightly peraluminous tonalites or quartz diorites and include oceanic plagiogranites, such as those described by Coleman and Peterman (1975). These are sodic granites that typically lack K-feldspar and are found in ophiolite complexes, some of which formed at oceanic ridges. Pearce et al. (1984) call these ocean ridge granites, but compositionally similar rocks are also found in ophiolites that may have originally formed in back arc basins above I. subduction zones, such as the Troodos Massif of Cyprus (Aldiss, 1981). i: ,i The most characteristic features of OR-type granites are strong depletions oflarge ion lithophile ! ' : ! elements like Ba, Rb and K (Fig. 8). This feature must be inherited from the low concentrations of these elements in the source of these ocean-ridge granites, presumably by fractionation from MORE (mid-ocean ridge basalt) derived from the depleted upper mantle or by partial melting of these iii basalts above ridge-centred magma chambers. Another important feature is the lack of a negative' 'I'·, Ii:1 . : II Nb-Ta-Ti anomalies in chondrite-normalized trace element patterns. Consequently, these silicic magmatic rocks have the lowest RblNb ratios (Fig .. 6) of any considered here and occupy a distinctive position on the discriminant diagrams (Fig. 7) of Pearce et al. (1984). No data on the FICI ratios of glassy rocks or melt inclusions from silicic ocean-ridge rocks is available. However, if they are fractionates of mid-ocean ridge basalt, they probably have FICI ratios near 1 to as high as 10 (Schilling et al., 1980; Byers et al., 1986). Basalts and peralkalic rhyolites related to mantle plumes have FICI ratios from 1 to 0.1 (Fig. 10). The same kind of comparison suggests that ocean­ ridge granites also crystallize at low j02. •~ :gxtensive fractionation may lead to enrichments of incompatible elements such as Pb, and perhaps Zn, in the silicic magmas, but some other ore metals (notably Cu) are more enriched in the • associated mafic lavas. Moreover, Pb and Zn are more likely to be enriched in continental magma •CI series. The production of magma rich in ore elements that are incompatible at low j02 is difficult CI because of the depleted nature of the MORE source materials.·Moreover, any enriched magmatic t CI fractionate would be very small in volume. The production of ore by secondary processes acting on t such small fractionates would also be difficult. t t SUMMARY OF METAL ASSOCIATIONS • Many ore deposits considered here are related to compositional controls, that is to high • concentrations of ore elements in magma or to an intensive property of magma that enhances •til transport and eventual concentration (e.g.,j02,jHCI, jH20). We have summarized the principal associations of ore deposits (Fig. 15) with the different granite types on the trace element • classification diagram of Pearce et al. (1984). The systematic position of ore-related silicic magmas • on this diagram re-iterates the role of source composition, magmatic processes (such as partial • I • ,:1 melting, fractional crystaIlizav.0n and contamination), volatile/fugacities and tectonic setting on the concentration of ore metals irt silicic magmas. Ray et al. (1995) produced a similar diagram for the • association of granite types and skarn-related ores from British Columbia. Fe, Au and Cu deposits •.. are typically found associated with unfractionated I-type plutonic rocks. All these elements become • depleted during fractionation of silicic magmas. Moreover, increasing continental crustal • contamination dilutes the magmatic concentrations of these elements as well. Consequently, the .. highest probability of ore development is with the least fractionated and least contaminated magmas. .. More fractionated I-type magmas may produce Cu-Mo and then Mo-Cu deposits. The most .1 • •t.. •~ 147

1000 s-Type Ta Li BeLi Granite Sn w REE Figure 15 Trace element . Ta compositions of granites and 300 W MSn Zr o Nb rhyolites associated with various Hg types of mineralization. .-... Modified after Pearce et al . E 100 U c.. Ag Zn (1984) and Ray et al. (1995)...... c.. Cu .0 0:: 30 Au

Fe 10 I-Type Granite OR-Type· Granite 3 2 5. 10 20 50 100 200 5001000 Nb + Y (ppm)

chemically evolved rocks of this sort are associated with Climax-type Mo deposits that lack Cu • mineralization. High f02 seems to be the most critical factor that makes Mo behave as a highly •.. incompatible element (Tacker and Candela, 1987; Keith etal., 1993). Consequently, highly .. fractionated I-types with A-type affinities may host these ore deposits. In contrast, fractionation of. ,. reduced magma series such as those related to S-type granites with incorporated sedimentary ,. carbon, lead to Sn deposits. Sn is a compatible element in oxidized magmas and probably does not ~ reach the levels necessary to generate ore in oxidized magmas (Ishihara, 1977; Lehmann, 1~82). Tungsten's behaviour is interinediate to that ofMo and Sn, and is enriched in fractionated I-types ..•, that have suffered crustal contamination and in some granites with stronger S-type affinities ( Keith • et al., 1989). A-type magmas are most closely associated with incompatible elements like Be, F, U, • Th, Nb, and Zr, and perhaps Sn as well. The low degrees of crustal melting, contributions from •!Ie distinctive mantle reservoirs represented by mafic magmas from plumes and continental rifts, and high degrees of fractionation of either crustal or mantle-derived magma are the primary causes of •Ie this association with incompatible trace elements. Ocean-ridge granites are rarely associated with mineralization. • The role of magma composition in setting the stage for ore generation must be met with another • set of conditions thai leads to the generation of ore. For example, a Mo-specialized granite may be ..• emplaced at sufficiently great depths that it reaches volatile saturation very late in its history and is .. thus umi.ble to generate a hydrothermal ore deposit. Trace element characteristies may help us .. identify the specialization of a granite, but will do little or nothing to help decipher the depth of • emplacement. Other important geologic conditions that would leave no imprint on the magmatic • concentrations are innumerable. Trace element geochemistry is only one of many facets that must be • considered in exploration for rhyolite- and granite-related deposits. No one parameter can be used • to ascertain the compositional character, tectonic setting, or metallogenic affinity of silicic ritagmas. • • 148 B.H. Christiansen and J.D. Keith

ACKNOWLEDGMENTS As with any review paper, we have relied heavily on the data and ideas of others and apologize to those whose contributions were ignored or misinterpreted. We are grateful for the review of M.G. Best and discussions with many graduate students. We also thank K.Waite for permission to use her unpublished data.

REFERENCES Ague, J.J. and Brimhall, G.H., 1988, Regional variations in bulk chemistry, mineralogy, and the compositions of mafic and accessory minerals in the batholiths of California: Geological Society of America Bulletin, v. 100, p. 891-911. Aldiss, D.T., 1981, Plagiogranites from the ocean crust and ophiolites: Nature, v. 289, p. 577-578. Allegre, C.J. and Minster, J.F., 1978, Quantitative models of trace element behavior in magmatic processes: Earth and Planetary Science Letters, v. 38, p. 1-25. All~gre, C.l., Treuil, M., Minster, J.-F., Minster, B. and Albarede, F., 1977, Systematic use of trace element in igneous process: Part I: Fractional crystallization processes in volcanic suites: Contributions to Mineralogy and Petrology, v. 60, p. 57-75. Anderson, J. L., 1983, Proterozoic anorogenic granite plutonism of North America: Geological Society of America Memoir 161, p. 133-154. Anderson, J.L. and Bender, E. E., 1989, Nature and origin of Proterozoic A-type granitic magmatism in the southwestern United States of America: Lithos, v. 23, p. 19-52. Babcock, R.C., Ballantyne, G.H. and Phillips, C.H., 1995, Summary of the geology of the Bingham District, Utah, in Proceedings of Bootprints Along the Cordillera Symposium-- Porphyry copper deposits from Alaska to Chile: Tucson, Arizona. Bailey, D.K., 1980, Volcanism, Earth degassing and replenished lithosphere mantle: Royal Society of London, Philosophical Transactions, v. A297, p. 309-322. Baker, D.R. and Vaillancourt, 1., 1995, The low viscosities of F + H2O-bearing granitic melts and implications for melt extraction and transport: Earth and Planetary Science Letters, v. 132, p. 199-21l. Barr, D.L. 1993, Time, space, and composition patterns of Middle Cenozoic mafic to intermediate composition lava flows of the Great Basin: M.S. Thesis, Brigham Young University, 68 p. Bateman, P.e. and Chappell, B.W., 1979, Crystallization, fractionation, and solidification of the Tuolumne Intrusive Series, Yosemite National Park, California: Geological Society of America Bulletin, v. 90, p. 465-482. Bea, Fernando, 1995, Controls on the trace element chemistry of crustal melts: U.S. Geological Survey Circular 1129, p. 19-20. Beckinsale, R.D., Suensilpong, S., Nakapadongrat, S. and Walsh, J.N., 1979, Geochronology and geOChemistry of granite magmatism in Thailand in relation to a plate tectonic model: Journal of the Geological society of London, v. 136, p. 529-540. Blevin, P.L. and Chappell, B.W., 1992, The role of magma sources, oxidation states and fractionation in determining the. granite metallogeny of eastern Australia: Transactions of the Royal Society of Edinburgh: Earth Sciences, v. 83, p. 305-316. . Blevin, P.L., Allen, e.M., Bryant, C.J. and Chappell, B.W., 1995, A metallogenic template for eastern Australia based on granite source and evolution: U.S. Geological Survey Circular 1129, p. 25-26. Bowden, P. and Kinnaird, 1.A., 1978, Younger Granites of Nigeria -- a zinc-rich tin province: Transactions of the Institute for Mining and Metallurgy, v. 87,p. 1366-1369. Burnham, C.W., 1979, The importance of volatile constituents, in Yoder H.S.,. ed., The Evolution of the Igneous Rocks - Fiftieth Anniversary Perspectives: Princeton University Press, Princeton, N.J., p. 439-482. Byers, C.D., Garcia, M.O. and Muenow, D.W., 1986, Volatiles in basaltic glasses from the East Pacific Rise at 210N: implications for MORB sources and submarine lava flow morphology: Earth and Planetary Science Letters, v. 79, p. 9-20. Candela, P.A., 1989, Felsic magmas, vdlatiles, and metallogenesis: in Whitney, J.A., and Naldrett, A.J., eds., Ore deposition associated with magmas: Reviews in Economic Geology, v. 4, p. 223-233. Carmichael, 1.S.E., 1991, The redox state of basic and silicic magmas: a reflection of their source regions?: Contributions of Mineralogy and Petrology, v. 106, p. 129-14l. Chappell, B.W., and White, A.J.R., 1974, Two contrasting granite types: Pacific Geology, v. 8, p. 173-174. Chappell, B.W. and White, A.J.R., 1992, 1- and S-type granites in the Lachlan fold belt: Royal Society ?f Edinburgh Transactions, v. 83, p. 1-26. . Chappell, B.W., White, A.J.R. and Williams, 1.S., 1991, A transverse section through granites of the Lachlan fold Belt, Excursion Guide, Second Hutton Symposium on Granites and Related Rocks, Canberra, September 1991: Bureau of Mineral Resources, Geology and Geophysics, Record 1991122, 125 p. Christiansen, E.H., 1983, The Bishop Tuff revisited: Compositional zoning by double-diffusive fractional crystallization: Geological Society of America Abstracts with Programs, v. 15, p. 390. Christiansen, E.H., Burt, D.M. and Sheridan M.F., 1986, The geology of topaz rhyolites from the western United States: Geological Society of America Special Paper 205, 82 p. Christiansen, E.H., Bikun, 1.V., Sheridan, M.F. and Burt, D.M., 1984, Geochemical evolution of topaz rhyolites from the Thomas Range and Spor Mountain, Utah: American Mineralogist, v. 69, p. 223-236. Christiansen, E.H. and Lee, D.E., 1986, Fluorine and chlorine in granitoids from the Basin and Range Province, western United States: Economic Geology, v. 81, 1484-1484. Christiansen, E.H., Stuckless, J.S., Funkhouser, MJ. and Howell, K.A., 1988, Petrogenesis of rare-metal granites from depleted crustal sources--an example from the Cenozoic of western Utah, USA: in Taylor, R.P and Strong, D.F. (ed), Recent advances in the geology of granite-related mineral deposits, Canadian Institute of Mining and Metallurgy Special Volume 39, p. 307-321. 149

Christiansen, R.L. and lipman, 1972, Cenozoic volcanism and plate tectonic evolution of the western United States: II. Late Cenozoic: Royal Society of London, Philosophical Transactions, v. A271, 249-284. Clarke, D.B., 1981, The mineralogy of peraluminous granites: A review: Canadian Mineralogist, v. 19, p. 3-17. Clemens, J.D., 1984, Water contents of silicic to intermediate magmas: Lithos, v. 17, p. 273-287. Clemens, J.D., Holloway, J.R., and White, A.J.R., 1986, Origin of an A-type granite: Experimental constraints: American Mineralogist, v. 71, p. 317-324. Clemens, J.D. and Wall, V.I., 1984, Origin and evolution of a peraluminous silicic ignimbrite suite: The Violet Town Volcanics: Contributions to Mineralogy and Petrology, v. 88, p. 354-37l. Coleman, R.G. and Peterman, Z.E., 1975, Oceanic plagiogranite: Journal of Geophysical Research, v. 80, p. 1099-1108. Collins, W.J., Beams, S.D., White, A.J.R. and Chappell, B.W., 1982, Nature and origin of A-type granites with particular reference to southeastern Australia: Contributions to Mineralogy and Petrology, v. 80, p. 189-200. Congdon, R.D., and Nash, W.P., 1991, Eruptive pegmatite magma: Rhyolite of the Honeycomb Hills, Utah: American Mineralogist, v. 76, p. 1261-1278. Creaser, R.A., 1995, Radiogenic isotopes in granitic systems: Studies of melting and miring at the source: U.S. Geological Survey Circular 1129, p. 38-39. Davls, L.J., 1984, Beryllium deposits in the Spor Mountain area, Juab County, Utah: Utah Geological Association Publication 13, p. 173-183. DePaolo, D.J., 1981a, Trace element and isotopic effects of combined wallrock assimilation and fractional crystallization: Earth and Planetary Science Letters, v. 53, p. 189-202. DePaolo, D.J., 1981b, A neodymium and strontium isotopic study of Mesozoic calc-alkaline granitic batholiths of the Sierra Nevada and Peninsular ranges, California: Journal of Geophysical Research, v. 86, p. 10470-10488. Eby, G.N., 1992, Chemical subdivision of the A-type granitoids: Petrogenetic and tectonic implications: Geology, v. 20, p. 641-644. Eggleston, T.L., 1987, The Taylor Creek district, New Mexico: Geology, petrology, and tin deposits: PhD. thesis, Socorro, New Mexico Institute of Mining and Technology, 473 p. Einaudi, M.T., Meinert, L.D. and Newberry, R.J., 1981, Skarn deposits: Economic Geology, v. 75, p. 317-39l. Eugster, H. P., 1986, Minerals in hot water, American Mineralogist, v. 71, p. 655-673. Ewart, A., 1979, A review of the mineralogy and chemistry of Tertiary-Recent dacitic, latitic, rhyolitic, and related volcanic rocks, in. Barker, F., ed. Trondhjemites, dacites, and related rocks: Elsevier, Amsterdam, p. 13-12l. Farmer, G.L. and DePaolo, D.I., 1983, Origin of Mesozoic and Tertiary granite in the western United States and implications fro pre-Mesozoic crustal structure. I. Nd ant! Sr isotopic studies in the geocline of the northern Great Basin: Journal of Geophysical Research, v. 88., p. 3379-340l. Filippov, L.V., Savimova, Y.N., Kapitonova, T.A., and Andreyeva, T.P., 1974, Fluorine in Mg-Fe micas from granitoids of various magmatic formations in the folded belt of central Asia: Geochemistry International, v. 11, p. 185-194. Fyfe, W.S. and Kerrich, R., 1985, Fluids and thrusting: Chemical Geology, v. 49, p. 353-362. Hanson, G.N., 1978, The application of the trace elements to the petrogenesis of igneous rocks of granitic composition: Earth and Planetary Science Letters, v. 38, p. 26-43. Hildreth, W., 1979, The Bishop Tuff: Evidence for the origin of compositional zonation in silicic magma chambers: Geological Society of America Special Paper 180, p. 43-75. Hildreth, W., Halliday, A.N. and Christiansen, R.L., 1991, Isotopic and chemical evidence concerning the genesis and contamination. of basaltic and rhyolitic magma beneath the Yellowstone Plateau volcanic field: Journal of Petrology, v. 32, p. 63-78. Hodges, K.V., 1990, Tectonic denudation and anatectic melting in compressional settings: EOS, v. 71, p. 1618. Huspeni, J.R., Kesler, S.E., Ruiz, J., Tuta, Z., Sutter, J.P. and Jones, L.M., 1984, Petrology and geochemistry ofrhyolites associated with tin mineralization in northern Mexico: Economic Geology, v.79, p. 87-105. '­ Imeokparia, E.G., 1982,Geochemistry and relationships to mineralization of granitic rocks from the Afu Younger Granite Complex, central Nigeria: GeOlogical Magazine, v. 119, p. 39-56. .. Ishihara, S., 1977, The magnetite-series and ilmenite-series granitic rocks: Mining Geology, v. 27, p. 293-305. Johnson, C.M., lipman, P.W.,and Czamanske, 1990, H, 0, Sr, Nd, and Ph isotope geochemistry of the Latir volcanic field and cogenetic •r- intrusions, New Mexico, and relations between evolution of a continental magmatic center and modifications of the lithosphere: Contributions to Mineralogy and Petrology, v. 104, p. 99-124. .. Keith, J.D., Christiansen, E.H., Best, M.G., Barr, D.L., Moore, D.K., Waite, K., Tingey, D.G., 1995, The role of mafic alkaline .. magmatism in formation of porphyry- and vein-type mineralization: Examples from the Bingham and Tintic Mining Districts, Utah, USA: GACIMAC Final Program and Abstracts, v. 20, p. A52. .. Keith, J.D., Christiansen, E.H. and Carten, R. B., 1993, The genesis of giant porphyry molybdenum deposits, in, Whiting, B. H., Mason, R . .. and Hodgson, C.J., eds., Gian~Ote Deposits, Society of Economic Geologists Special Publication 2, p. 285-316. a­ Keith, J. D., Christiansen, E. H. and Tingey, D.G., 1994, Geological and chemical conditions of formation of red beryl, Wah Wah Mountains, Utah: Utah Geol~gical Association Publication 23, p. 155-169 . .. Keith, J.D., van Middelaar, W., Clark, A.H. and Hodgson, C.J., 1989, Granitoid textures, compositions, and volatile fugacities associated 1ft with the formation of tungsten-dominated skarn deposits, in Whitney, J.A., and Naldrett, AJ., eds., Ore deposition associated with r­ magmas: Reviews in Economic Geology, v. 4, p. 235-250. . Keith, J.D., and Shanks, 1988, Chemical evolution and volatile fugacities of the Pine Grove porphyry molybdenum and ash-flow tuff ill system, southwestern Utah: in Taylor, R.P and Strong, D.P. (ed~- Recent advances in the geology of granite-related mineral deposits, Canadian Institute of Mining and Metallurgy Special Volume 39, p. 402-423. .. Kovalenko, V.I. and Kovalenko, N.I., 1976, Ongonites (topaz-bearing quartz keratophyres)--subvolcanic analogue of rare metal li-F lit granites (in Russian): Nauka Press, Moscow, 128 p . .. Kovalenko, V.l. and Kovalenko, N.l., 1984, Problems of the origin, ore-bearing and evolution of rare-metal granitoids: Physics of the •.. • r • 150 E.H. Christiansen and J.D. Keith • Earth and Planetary Interiors, v. 35, p. 51-62. • Kovalenko, V.l., Naumov, V.G., Solovova, I.P., Birnis, A.V., Hervig, R.L. and Boriani, A., 1994, Volatile components, composition, and crystallization conditions of the Pantelleria basalt-pantellerite association magmas, inferred from the melt and fluid inclusion data: • Petrology, v. 2, p. 24-42. • Kovalenko, V.l., Tsaryeva, G.M., Goreglyad, A.V., Yarmolyuk, V.V., Troitsy, V.A., Hervig, R.L. and Farmer, G.L., 1995, The • peralkaline granite-related Khaldzan-Buregtey rare metal (Zr, Nb, RJ;::E) deposit, western Mongolia: Economic Geology, v. 90, p. 530-547. • Kress, V.C. and Carmichael, I.S.E., 1991, The compressibility of silicate liquids containing Fe203-and the effect of composition, • temperature, oxygen fugacity and pressure on their redox states: Contributions to Mineralogy and Petrology, v. 108, p. 82-92. Lee, D.E. and Christiansen, E.H., 1983, The granite problem in the southern Snake Range, Nevada: Contributions to Mineralogy and • Petrology, v.83, p.99-116. • Lee, D.E., Kistler, R.W., Friedman, 1. and Van Loenen, R.E., 1981, Two-mica granites of northeastern Nevada: Journal of Geophysical • Research, v. 86, p. 10607-10616. Lehmann, B., 1982, Metallogeny of tin: Magmatic differentiation versus geochemical heritage: Economic Geology, v. 77, p. 50-59. • Lindsey, D.A., 1982, Tertiary volcanic rocks and uranium in the Thomas Range and northern Drum Mountains, Juab County, Utah: U.S. • Geological Survey Professional Paper 1221, 71 p. LowensteIn, lE. and Mahood, G.A., 1991, New data on magmatic H20 contents of pantellerites, with implications for petrogenesis and • eruptive dynaruics at Pantelleria: Bulletin of Volcanology, v. 54, p. 78-83. • Lowenstern, J.B., 1995, Applications of silicate-melt inclusions to the study of magmatic volatiles: in Thompson, J.F.H. (ed), Magmas, • fluids, and ore deposits, :Mineralogical Association of Canada Short Course Volume 23, p. 71-99. .. Macdonald, R., Smith, R.L. and Thomas, J.E., 1992, Chemistry of the subalkalic silicic obsidians: U.S. Geological Survey Professional Paper 1523, 214 p. .. Mahood, ·G., and Hildreth, W., 1983, Large partition coefficients fro trace elements in high-silica rhyolites: Geochimica et Cosmochimica Acta, v. 47, p. 11-30. • Manning, D.A.C., 1981, The effect of fluorine on liquidus phase relationships in the system Qz-Ab-Or with excess water at 1 kb: .. Contributions to Mineralogy and Petrology, v. 76, p. 206-215. .. McCarthy, T.S. and Hasty, R.A., 1976, Trace element distribution patterns and their relationship to the crystallization of granitic melts. II Geochimica et Cosmochimica Acta, v. 40, p. 1351-1358. :Michael, P.J., 1983, Chemical differentiation of the Bishop Tuff and other high-silica magmas through crystallization processes: Geology, .. v. 11, p. 31-34. .. :Minster, J.P. and Allegre, c.J., 1978, Systematic use of trace elements .in igneous processes: Part III: Inverse problem of batch partial melting in volcanic suites: Contributions to :Mineralogy and Petrology, v. 68, p. 37-52. .. :Miller, C.F., and Bradfish, L.J., 1980, An inner Cordilleran belt of muscovite-bearing plutons: Geology, v. 8, p. 412-416. :Miyashiro, A., 1974, Volcanic rock series in island arcs and active continental margins: American Journal of Science, v. 274, p. 321-355. • Montel, J.-M., 1986, Experimental determination of the solubility of Ce-monazite in Si0 - ~03-K20-Na20 melts at 800eC, 2 .. 2 kbar, under H2O-saturated conditions: Geology, v. 14, 659-662. .. Morris, J.D. and Tera, F., 1989, lOBe and 9Be in mineral separates and whole rocks from volcanic arcs: implication for sediment subduction: Geochimica et Cosmochimica Acta, v. 53, p. 3197-3206. •til Munoz, J.L., 1984, F-OH and a-OH exchange in micas with applications to hydrothermal ore deposits: in Bailey, S.W., ed, Reviews in :Mineralogy, v. 13, p. 469-493. .. Mutschler, P.E., Wright, E.G., Ludington, S. and Abbott, J.T., 1981, Granite molybdenite systems: Economic Geology, v. 76, p. 1981. Nash, W.P. and Crecraft, H.R., 1985, Partition coefficients for trace elements in silicic magmas: Geochimica et Cosmochimica Acta, v. .. 49, p. 2309-2322. , .. Noble, D.C., Vogel, T.A., Peterson, P.S., Landis, G.P., Grant, N.K., Jezek, P. and McKee, E.H., 1984, Rare-element-enriched S-type .­ as,h-flow tuffs containing phenocrysts of muscovite, andalusite, and sillimanite, southeastern Peru: Geology, v. 12, p. 35-39. ! :j Partington, G.A., McNaughton, N.J. and Williams, 1.S., 1995, A review of the geology, mineralization, and geochronology of the . .. :1 Greenbushes pegmatite, western Australia: Economic Geology, v. 90, p. 616-635. Pearce, J.A. and Cann, J.R., 1973, Tectonic setting of basic volcanic rocks determined using trace element analyses: Earth and Planetary •II! Science Letters, v. 19, p. 290-300. i'l Pearce, J. A., Harris, N.B.W. and Tindle, A.G., 1984, Trace element discrimination diagrams for the tectonic interpretation of granitic II I' " rocks: Journal of Petrology, v. 25, p. 956-983. .. Pichavant, M. and Hammouda, T., 1995, Control of redox state and Sr isotopic composition of magmas: Source or processes?: U.S. Geological Survey Circular 1129, p. 116-117. Pichavant, M., Kontak, D.l, Briqueu, L., Valencia Herrera, land aark, A.H., 1988a, The :Miocene-Pliocene Macusani volcanics, SE •.. Peru: II. Geochemistry and origin of a felsic peraluminous magma: Contributions to Mineralogy and Petrology, v. 100, p. 325-338. .­ Pichavant, M., Kontak, DJ., Valencia Herrera, J. and aark, A.H., 1988b, The :Miocene-Pliocene Macusani volcanics, SE Peru: 1. :Mineralogy and magmatic evolution of a two-mica aluminosilicate-bearing ignimbrite suite: Contributions to :Mineralogy and .. Petrology, v. 100, p. 325-338. Pichavant, M., Montel, J.-M. and Richard, L.R., 1992, Apatite solubility in peraluminous liquids: Experimental data and an extension of .. the Harrison-Watsonmodel: Geochimica et Cosmochimica Acta, v. 56, p. 3855-3861. .. Pitcher, W.S., 1982, Granite type and tectonic environment, in Hsu, K. J., Motmtain building processes: Academic Press, London, p. .- 19-40. Pollard, P.J., Nakapadungrat, S. and Taylor, R.G., 1995, The Phuket supersuite, southwest Thailand: Fractionated I-type granites associated with tin-tantalum mineralization: Economic Geology, v. 90, p. 586-602. • Ramo, O.T. and Haapala, 1.,1995, One hundred years ofrapakivi granite: Mineralogy and Petrology, v. 52, p. 129-185. ..= =~ 151

Ray, G.E., Webster, LC.L. and Ettlinger; A.D., 1995, The distribution of skarns in British Columbia and the chemistry and ages of their related plutonic rocks: Economic Geology, v. 90, p. 920-937. Ryan, J.G. and Langmuir, C.H., 1993, The systematics of boron abundances in young volcanic rocks: Geochimica et Cosmochimica Acta, v. 57, p. 1489-1498. Ryerson, P.J. and Watson, E.B., 1987, Rutile saturation in magmas: implications for Ti-Nb-Ta depletion in island-arc basalts: Earth and Planetary Science Letters, v. 86, p. 225-239. Rytuba, J.J. and McKee, E.H., 1984, Peralkaline ash flow tuffs and calderas of the McDermitt volcanic field, southeast Oregon and north central Nevada: Journal of Geophysical Research, v. 89, p. 8616-8628. Schilling, J.G., Bergeron, M.B. and Evans, R., 1980, Halogens in the mantle beneath the North Atlantic: Royal Society of London, Philosophical Transactions, v. A297, p. 147-178. ' Shaw, S.E., and Flood, R.H., 1981, The New England Batholith, eastern Australia: Geochemical variations in time and space: Journal of Geophysical Research, v. 86, p. 10530-10544. Stoesser, D.B. and Camp, V.E., 1985, Pan-African microplate accretion of the Arabian Shield: Geological Society of America Bulletin, v. 96, p. 817-826. Tacker, R.C., and Candela, P.A., 1987, Partitioning of molybdenum between magnetite and melt: A preliminary experimental study of partitioning of ore metals between silicic magmas and crystalline phases: Economic Geology, v. 82, p. 1827-1838. Takahashi, M., Aramake, S. and Ishihara, S., 1980, Magnetite-serieslilmenite-series vs. I-type/S-type granitoids: Mining Geology Japan Special Issue, v. 8, p. 13-28. Taylor, S.R. and McLennan, S.M., 1985, The continental crust: its composition and evolution: Blackwell Scientific Publications, Oxford, 312p. Tegtmeyer, K~J. and Farmer, G.L., 1990, Nd isotopic gradients in upper crustal magma chambers: Evidence for in situ magma-waIl-rock interaction: Geology, v. 18, p. 5-9. Thompson, R.N., Morrison, MA., Hendry, G.L. and Parry, S.1., 1984, An assessment of the relative roles of crust and mantle in magma genesis: an elemental approach. Royal Society of London, Philosophical Transactions,v. A310, p. 549-590. Twist, D. And Harmer, R.E.J" 1987, Geochemistry of contrasting siliceous magmatic suites in the Bushveld complex: Genetic aspects and implications for tectonic discrimination diagrams: Journal of Volcanology and Geothe=al Research, v. 32, p. 83-98. van Middelaar, W.R. and Keith, J.D.; 1990, Mica chemistry as an indicatory of oxygen and halogen fugacities in the CanTung and other W-related granitoids in the North American Cordillera: in Stein, H.J. and Hannah, J.L., eds, Ore-bearing granite systems; petrogenesis and mineralizing processes. Geological Society of America Special Paper 246, 205-220. Watson, E.B. and Harrison, T.M., 1983, Zircon saturation revisited: Temperature and composition effects in a variety of crustal magma types: Earth and Planetary Science Letters, v. 64, p. 295-304. Webster, J.D. and Duffield, W.A., 1991, Volatiles and lithophile elements in Taylor Creek Rhyolite: Constraints from-glass inclusion analysis: American Mineralogist, v. 76, p. 1628-1645. Webster, J.D., Taylor, R.P. and Bean, C., 1993, Pre-eruptive melt composition and constraints on degassing of a water-rich pantellerite magma, Fantale Volcano, Ethiopia: Contributions to Mineralogy and Petrology, v,-114, p. 53-62. Whalen, J.B., Currie, K.L. and Chappell, B.W., 1987, A-type granites: geochemical characteristics, discrimination and petrogenesis: Contributions to MiDeralogy and Petrology, v. 95, p. 407-419. \ White, A.J.R. and Chappell, B.W., 1983, Granitoid types and their distribution in the Lachlan Fold Belt, southeastern Australia: Geological Society of America Memoir 159, p. 21-33. White, W.H., Bookstrom, A.A., Kamilli, R.J., Ganster, M.W., Smith, R.P., Ranta, D.E. and Steininger, R.C., 1981, Character and origin "of climax-type molybdenum deposits: Economic Geology 75th Anniversary Volume, p. 270-316. Wolff, lA. and Storey, M., 1984, Zoning in highly alkaline magma bodies: Geological Magazine, v. 121, p. 563-575. Yogodzinki, G.M., Kay, R.W., Volynets, O.N., Koloskov, A.V. and Kay, S.M., 1995, Magnesian andesite in the western Aleutian Komandorsky region: Implications for slab melting and processes in the mantle wedge.: Geological Society of America Bulletin, v. 107, p. 505-519. Zeck, 1970, An erupted migmatite from Cerro del Hoyazo, SE Spain: Contributions to Mineralogy and Petrology, v. 37, p. 351-358. Zen, E., 1988, Phase relations of peraluminous granitic rocks and their petrogenetic implications: Annual Reviews of Earth and Planetary Sciences, v. 16, p. 21-5l. Zoback, M.L., McKee, E.H., Blakely, R.J. and Thompson, G.A., 1994, The Northern Nevada Rift: Geological Society of America, v. 106, p. 371-382. Zhu Zhengshu, .1990, The geological characteristics and mineralization types of Yanbei tin deposit, Huichang country, Jiangxi province: Mineral Deposits (in Chinese), v. 9, p. 325-330.