Journal of Petrology Advance Access originally published online on October 6, 2007
Journal of Petrology 2007 48(11):2125-2148; doi:10.1093/petrology/egm054
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Contact Partial Melting of Granitic Country Rock, Melt Segregation, and Re-injection as Dikes into Ferrar Dolerite Sills, McMurdo Dry Valleys, Antarctica
Morton K. Blaustein Department of Earth and Planetary Sciences, Johns Hopkins University, Baltimore, MD 21218, USA
RECEIVED NOVEMBER 7, 2006; ACCEPTED AUGUST 10, 2007
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONNumerous, interconnected, granitic dikes (<30 cm in width and hundeds of meters in length) cut Ferrar dolerite sills of the McMurdo Dry Valleys, Antarctica. The source of the granitic dikes is partial melting of granitic country rock, which took place in the crust at a depth of about 2–3 km adjacent to contacts with dolerite sills. Sustained flow of doleritic magma through the sill generated a partial melting front that propagated into the granitic country rock. Granitic partial melts segregated and collected at the contact in a melt-rich, nearly crystal-free reservoir adjacent to the initial dolerite chilled margin. This dolerite chilled margin was subsequently fractured open in the fashion of a trapdoor by the granitic melt, evacuating the reservoir to form an extensive complex of granitic dikes within the dolerite sills. At the time of dike injection the dolerite was nearly solidified. Unusually complete exposures allow the full physical and chemical processes of partial melting, segregation, and dike formation to be examined in great detail. The compositions of the granitic dikes and the textures of partially melted granitic wall rock suggest that partial melting was characterized by disequilibrium mineral dissolution of dominantly quartz and alkali feldspar rather than by equilibrium melting. It is also unlikely that melting occurred under water-saturated conditions. The protolith granite contains only
7 vol.% biotite and estimated contact temperatures of 900–950°C suggest that melting was possible in a dry system. Granite partial melting, under closed conditions, extended tens of meters away from the dolerite sill, yet melt segregation occurred only over less than one-half a meter from the dolerite chilled margin where the degree of partial melting was of the order of 50 vol.%. This segregation distance is consistent with calculated length scales expected in a compaction-driven process. We suggest that the driving force for compaction was differential stress generated by a combination of volume expansion as a result of granite partial melting, contraction during dolerite solidification, and relaxation of the overpressure driving dolerite emplacement. On a purely chemical basis, the extent of melt segregation necessary under fractional and batch melting to match the Rb concentrations between melt and parent rock is a maximum of 48 and 83 vol.% melt, respectively. KEY WORDS: Antarctica; dike injection; disequilibrium; granite partial melting; silicic melt segregation
|
INTRODUCTION |
|---|
Crustal melting by basalt injection has been proposed as a fundamental mechanism for crustal evolution (Bergantz, 1989
; Brown, 2006). Numerous field studies of shallow crustal rocks have reported melting of country rock near the contacts of basic intrusions, mainly large dikes and stocks (e.g. Al-Rawi & Carmichael, 1967; Kovach, 1984; Kaczor, 1988; Kitchen, 1989; Philpotts & Asher, 1993; Holness & Watt, 2002; Petcovic & Grunder, 2003; Holness et al., 2005). These occurrences provide important insights into the detailed nature of the melting process, which is most often under transient or disequilibrium conditions wherein the melt composition can be described by a combination of batch and fractional melting. When the new felsic melt re-injects the adjacent basaltic heat source, the critical process of mingling and contamination of the basalt by the granitic melt can also be examined physically and chemically. However, because the most accessible examples of this process are in the uppermost crust, where the country rock, at the time of basalt intrusion, is cold, partial melting is generally limited in spatial extent, with little to no melt extraction or segregation and subsequent transport. At deeper levels in the crust where background temperatures are higher, partial melting is expected to be more extensive and melt segregation and transport much more common. In areas where the crust has undergone rapid thickening (e.g. the Himalayas, the Mojave), high ambient temperatures should favor extensive partial melting, assimilation and melt migration. These conditions may also be approximated in areas of the upper crust where voluminous basaltic magma transport and emplacement has occurred.
A pristine natural laboratory to study this process is the extensive and massive (10 000 km3) Ferrar dolerite sill complex of the McMurdo Dry Valleys, Antarctica, which contains in many areas small granitic dikes cutting adjacent dolerite sills. The dikes are primarily in association with the lowermost sill, the Basement Sill, which was emplaced in pervasive granitic basement rocks at a paleo-depth of about 3–4 km based on the thickness of overlying units capped by comagmatic Kirkpatrick flood basalts (discussed in the next section). In most areas the dikes are actually small dikelets about 3–8 mm in width, yet these dikelets commonly penetrate 50–100 m into the adjoining dolerite sill. In light of the near-instantaneous quenching time of small dikelets (1 min), these occurrences imply that the dolerite was at an insulating temperature at or above the melting range of the granitic dikelets. In Wright Valley, in the Bull Pass and East Dais regions, the granitic dikes commonly reach 20–25 cm in width and extend continuously for hundreds of meters, forming, especially in central Bull Pass, an extensive, approximately orthogonal network. The dikes are extensive and easy to trace; however, the critical exposures of the dikes with the country rock cannot generally be found, owing to poorly exposed or indistinct contact relations. In the upper east wall of central Bull Pass, however, the full field relations are exposed for a complex of these large granitic dikes.
In this paper we present field, petrographic, and geochemical evidence of this unique example of extensive partial melting of water-undersaturated granitic country rock at the contact of a dolerite sill. Prolonged heating by the dolerite melted the adjacent granitic wall rock outward some tens of meters. Granitic melt was extracted from its residue by lateral compaction and porous melt flow and collected into a melt-rich reservoir adjacent to the chilled margin of the dolerite. When the dolerite sill was nearly completely solidified, the chilled margin fractured open like a trapdoor, allowing granitic melt to penetrate as dikes deep into the interior of the dolerite sill. Compaction, melt extraction, and dike propagation were driven by the overpressure of dolerite emplacement, melting, and the contraction associated with emplacement cessation and cooling. The completeness of the field relations and the nature of the magmas and country rock involved make this natural laboratory of fundamental importance to understanding the relationship between basaltic intrusion and the generation of granitic magmas (sensu lato) in the Earth's crust.
|
GENERAL GEOLOGY |
|---|
The McMurdo Dry Valleys (MDV) in Southern Victoria Land are part of the Transantarctic Mountains (TAM), which divide the Antarctic continent and delineate a boundary between Archean–Proterozoic crust of the East Antarctic craton and the Paleozoic–Mesozoic lithospheric blocks of West Antarctica (Borg et al., 1990
). The oldest rocks in the MDV region of southern Victoria Land are late Proterozoic age, multiply deformed, metamorphosed, interbedded marbles, hornsfelses, and schists of the Koettlitz group (Gunn & Warren, 1962; McKelvey & Webb, 1962). These metasediments grade into paragneiss or migmatite immediately adjacent to undeformed granitic plutons [Granite Harbor Intrusives of Gunn & Warren (1962)].
Up to 15 major granitic (sensu lato) plutons in the MDV were emplaced during the late Proterozoic to early Ordovician (Allibone et al., 1993a, 1993b, and references therein). There is much similarity in lithology between plutons as well as varied lithology within plutons from monzodiorite to granodiorite, biotite granodiorite to granite, and monzonite to granite in progressively younger plutons (Allibone et al., 1993a). The youngest group of plutons displays crosscutting relationships with numerous swarms of Vanda mafic and felsic porphyry dikes (Allibone et al., 1993a). Uplift and erosion, following emplacement of the plutons, formed the Kukri Erosion Surface, a peneplain upon which the Devonian to Jurassic age Beacon Supergroup sandstones, conglomerates, siltstones, and minor coal measures were deposited.
Jurassic age (180 Ma) breakup of the Gondwana supercontinent is responsible for extensive intrusion of Ferrar dolerite sills and dikes throughout the TAM and continuing into the Karoo system of South Africa (Elliot & Fleming, 2000; Riley et al., 2006). Comagmatic pyroclastic deposits are further overlain by flood lavas of the Kirkpatrick Basalt (e.g. Elliot & Fleming, 2004). Episodic exhumation and uplift of the TAM began in the early Cretaceous, but most uplift occurred in the early Cenozoic (Fitzgerald & Stump, 1997; Behrendt et al., 1991). The youngest rocks in the MDV are Cenozoic age McMurdo volcanics, which are alkali basalts.
|
FERRAR DOLERITES AND HOSTED GRANITIC DIKES |
|---|
The Ferrar dolerite sills are typically high aspect ratio sheets that vary from 100 to 500 m in thickness and can extend laterally for over 100 km. At least four separate sills have been mapped, with the lowermost sill, the Basement Sill, intruded below the Kukri peneplain in Paleozoic granites, and the uppermost, hypabyssal sill, the Mt. Fleming Sill, leading directly to tephra pipes near the ancient paleo-surface (Gunn & Warren, 1962
; Hamilton, 1965). The sills are often remarkably uniform in position and attitude, and the separation between sills is typically hundreds of meters and constant, although in some places the sills do touch one another. For example, in Bull Pass the Basement Sill climbs upwards and touches the Peneplain Sill. At such contacts, the Basement Sill is always quenched against the Peneplain Sill. The quenched textures are relatively coarse, suggesting that the Basement Sill was intruded after the Peneplain Sill, but while the latter was still significantly warmer than the ambient temperature. In another area, the Friis Hills, the Basement Sill actually pushes into the Peneplain Sill, suggesting that the Peneplain Sill was still mushy during contact. It is inferred that the general sequence of sill emplacement throughout the system progressed from the top downwards (Marsh, 2004).
The dominant lithology of the sill complex is quartz-normative dolerite. Areally extensive orthopyroxenite, first identified by Gunn & Warren (1962) and Hamilton (1965), is present primarily in the Basement Sill, and in limited sections of the Peneplain sill, but no higher in the stratigraphic sequence. Marsh (2004) interpreted the large, modally abundant orthopyroxene (including 10 vol.% similar Cpx) to be ‘cognate’ xenocrysts or antecrysts entrained from an earlier or related phase of magmatism within the underlying magmatic mush column. These crystals were entrained and transported as a crystal-rich suspension in ascending doleritic magma much as originally hypothesized by Simkin (1967). Both orthopyroxene mode and crystal size within the Basement Sill decrease with increasing distance away from a central location near Bull Pass. The central orthopyroxene zone is hereafter termed the OPX Tongue, following Marsh (2004). The spatial pattern of the OPX Tongue suggests that in this area the Basement Sill was emplaced radially from a central magma feeder in Bull Pass (see Figs 1 and 2). Not coincidentally, the Basement Sill in the area of Bull Pass has the highest areal concentration of cross-cutting granitic dikes. Granitic dikes are also observed on the south wall of Wright Valley, the Dais Intrusion, and the south wall of Victoria Valley. The granitic dikes are numerous, interconnected, up to 30 cm wide, and are occasionally over 100 m in length. Numerous other, much thinner (2–7 mm) granitic dikes are found throughout the region. The granitic dikes are mostly associated with the basal section of the Basement Sill and extend upwards into the sill as far as 200 m.
|
|
|
|
|
In central–east Bull Pass, the pattern of interconnected large granitic dikes at the outcrop scale consists of linear segments with near-orthogonal intersections (see Fig. 3). Where visible, the contact between the granitic dike rock and the surrounding dolerite is sharp both at the outcrop scale and grain scale. This suggests that, at the time of diking, mingling or mixing between granitic melt and doleritic residual melt was unlikely. No granitic dikes of the character found in the Ferrar dolerites have been discovered in any non-Ferrar unit in the MDV region and these granitic dikes should not be confused with the much older granitic dikes and Vanda felsic porphyry dikes of the Granite Harbor Intrusives (see Fig. 2).
|
SOURCE OF GRANITIC DIKES IN FERRAR DOLERITE SILLS |
|---|
Three locations have been found in eastern Bull Pass where granitic dikes intersect the contact between the Basement Sill and granitic country rock (labeled ‘A’, ‘B’, and ‘C’ in Fig. 1). The best exposed example is along a rise in the upper contact of a lobe of the Basement Sill, hereafter referred to as the dolerite feeder, with granitic country rock in east Bull Pass as seen looking from farther west (see Fig. 2c). The contact at this location (77°28·524'S, 161°55·361'E), labeled ‘A’ in Figs 1 and 2, strikes NNE, dips
45° to the north, and coincides with a gentle ridge trending NW–SE and cresting at Mt. Orestes. The paleo-depth at this location is 2–3 km, or equivalent to a pressure of 60–90 MPa, based on the thickness of the overlying Beacon Supergroup and extrusive Ferrar rocks exposed farther west.
At location ‘A’, a granitic dike (30 cm in apparent thickness) in the sill, with a more northerly strike, obliquely intersects the contact between the dolerite chilled margin and the granitic country rock (see Fig. 4). The granitic dike terminates at this contact but is continuous with an 5 cm wide granophyre zone or sheet composed of the same material that parallels the dolerite chilled margin. This granophyre zone decreases monotonically in thickness with increasing distance along the contact in either direction from the termination point of the granitic dike. The granophyre zone is imperceptible in hand sample beyond a distance of 10 m along the contact. Outward from the dolerite chilled margin and beyond the granophyre zone is an 1 cm wide band or sheet of partially melted granite with a distinctly dark gray matrix containing tiny fragments of the dolerite chilled margin (<1 cm in size). This sheet is termed the partially melted granite breccia. Further outward, the partially melted granite changes abruptly to a lighter gray matrix with no fragments of dolerite chilled margin and the gray matrix decreases in mode with increasing distance from the dolerite chilled margin. The local granitic country rock here is the Orestes pluton, which extends outward from the dolerite chilled margin 200 m to a change in lithology that includes hornblende. Vanda felsic porphyry dikes are interspersed in the Orestes granite and because they strike more northerly than the dolerite contact, the dikes contact, but do not cut because they are much older, the dolerite feeder towards the west (see Fig. 2b). At the terminus of the granitic dike and outward from the dolerite chilled margin, the felsic porphyry dikes are located at the ground surface between 9·5 and 29 m and again between 57 and 96 m within the Orestes pluton. It should be noted that in Fig. 2 and Table 1 these distances are corrected to be a projection from the ground surface onto a plane perpendicular to the contact assuming a contact dip of 45°. The older felsic porphyry dikes are part of a large swarm of similar dikes in the area of ‘A’ that range from tens of centimeters to <150 m in thickness. At their greatest abundance, the ratio of dikes to country rock is 4:1 (Allibone et al., 1993a). There is evidence of pervasive partial melting of both Orestes granite and Vanda felsic porphyry dikes along the contact with the dolerite feeder.
The dolerite feeder at location ‘A’ is 100 m thick, contains a wide, coarse OPX Tongue, and has a narrow (0·5 m), well-formed, fine-grained chilled margin. The lower contact of the dolerite feeder is against a wedge of granite of the Bonney pluton that is 50 m at its widest and is in contact with another lobe of the Basement Sill on its far side further down-slope. The granite wedge tapers down in thickness with proximity to central Bull Pass and exhibits evidence of partial melting; however, no granitic dikes have been found originating from the granite wedge. The granitic dike at location ‘A’ cuts into the dolerite feeder for a distance of 70 m, along the surface, before intersecting another, wider granitic dike within the sill that strikes nearly parallel to the dolerite chilled margin.
|
PETROGRAPHY |
|---|
The following sections describe the various rock units along the sampling traverse from the country rock towards the dolerite feeder shown as the black line in Fig. 2b. All distances are measured outwards from the dolerite chilled margin into the country rock. The relative locations of the various rock types at distances less than
10 cm from the dolerite chilled margin are shown in Fig. 4b.
Orestes granite
The Orestes granite is a medium-grained, equigranular, hypidiomorphic rock that contains quartz (26 vol.%), alkali feldspar (40 vol.%), plagioclase (27 vol.%), and a minor amount of biotite (<2·5 mm and 7 vol.%) (Fig. 5). The larger biotite grains frequently host accessory phases of euhedral to subhedral allanite (<0·5 mm), euhedral prismatic apatite (75 µm), and subhedral zircon (10 µm). Many biotite grains are altered to chlorite, unquantified clay minerals, or are reddish brown and have dustings of ilmenite and twinned hematite grains along rims. Quartz and feldspar grains (1·5 mm) tend to form aggregates (1·5 mm), but each grain exhibits a unique angle of extinction. Feldspar grains display twinning, are rarely zoned, and share straight boundaries with quartz, which usually displays undulatory extinction.
|
Partially melted Orestes granite
Pervasive dusting of biotite grains by hematite and lesser ilmenite is common and quartz grains develop cuspate boundaries against feldspar grains but no granophyre is present in Orestes granite at 23 m from the dolerite chilled margin. Fine to coarse-grained (10–250 µm) granophyric zones are present primarily between quartz and feldspar grains in Orestes granite at 5·2 m from the dolerite chilled margin. The granophyre mode increases, minor amounts of unaltered biotite in association with granophyre appear, quartz grain boundaries are more embayed, and feldspars have a more developed sieve texture at their margins (Fig. 5e) with distance towards the dolerite chilled margin. A sliver of partially melted Orestes granite, which varies locally in thickness but is always <1 cm, is sandwiched between the dolerite chilled margin on the right and the granophyre zone on the left, discussed below (see Fig. 4b). This rock contains
50 vol.% (visual estimate) of granophyre with restitic crystals of quartz, feldspar and reacted biotite, and is similar in texture to the partially melted Orestes granite shown on the far left of the leftmost photomicrograph of Fig. 4b.
Partially melted Orestes granite breccia
There is a sharp transition to a 1 cm wide zone of partially melted Orestes granite, hereafter referred to as partially melted granite breccia. Restite grains of plagioclase have a relatively less pronounced sieve texture compared with the large feldspar grain aggregates that characterize the partially melted Orestes granite farther away from the dolerite chilled margin. This band is also distinguished by finer-grained (10 µm) interstitial granophyre, which contains small plagioclase grains and clusters, up to several millimeters in size, of an equal mix of orthopyroxene (50 µm) and anorthite-rich plagioclase grains representing pieces of reacted dolerite chilled margin. These clusters are not necessarily coherent and orthopyroxene grains are observed in the interstitial granophyre far from the clusters. Another population of larger orthopyroxene grains (200 µm) is also dispersed throughout the interstitial granophyre.
Vanda felsic porphyry
Vanda felsic porphyry dikes contain 20 vol.% phenocrysts of alkali feldspar, plagioclase, quartz, biotite, and hornblende. Feldspar grains (<3·5 mm) are subhedral, commonly twinned, and range from showing no to moderate zoning. Quartz grains (<2 mm) are anhedral, have slight undulatory extinction, and commonly form aggregates of a few grains. Biotite and hornblende grains (<2 mm) often host apatite and zircon grains, are greenish brown, altered to chlorite or clay minerals, and dusted with opaque minerals. Groundmass is fine-grained granite composed of near-equal amounts of anhedral feldspar and quartz grains that are difficult to distinguish from granophyre associated with partial melting. Therefore, no estimate of the mode of granophyre in the Vanda felsic porphyry dikes rock was attempted. Perhaps, in future work, cathodoluminescence could be used to differentiate quartz generations.
Granophyre zone
The granophyre zone extends continuously into the granitic dike at location ‘A’. Along the sampling traverse, the granophyre zone is 5 cm wide and is in contact with partially melted Orestes granite breccia on the left and in contact with the thin sliver (<1 cm wide) of partially melted Orestes granite attached to the dolerite chilled margin on the right as shown in Fig. 4. The granophyre zone is a holocrystalline, porphyritic granophyre which hosts <10 vol.% large quartz, plagioclase, and alkali feldspar grains and coherent reacted biotite in addition to much larger polycrystalline aggregates of these same phases. These large grains are texturally similar to the restite grains of the partially melted Orestes granite that border both sides of the granophyre zone.
Granitic dikes in Ferrar dolerite sills
All of the granitic dike rocks observed and sampled are porphyritic and host from <1 to 10 vol.% large grains and polycrystalline aggregates in a groundmass mix of fine-grained granite and granophyre. As in the granophyre zone, the large grains are texturally similar to the restite phases in the partially melted granitic country rocks. The relative abundance of fine-grained granite (0·15–0·5 mm) vs granophyre (0·25 mm) in the groundmass of each section varies and one can be found without the other. The most common large grains are primarily feldspar and quartz, usually <3 mm in size but with rare grains >7 mm. Some thin sections contain minor amounts of reacted biotite (<5 vol.%), reacted hornblende (<1 vol.%), and opaque minerals (<1 vol.%). Alteration phases, <10 vol.% in some sections, occur mostly as breakdown products of hydrous phases and include chlorite and prehnite.
Dolerite chilled margin
The dolerite chilled margin contains a texturally and mineralogically distinctive 2·5 mm wide planar zone immediately adjacent and parallel to the contact with the sliver of partially melted Orestes granite. This zone, hereafter referred to as the dolerite reaction zone, differs from the rest of the dolerite chilled margin by a marked increase in grain size, the presence of biotite and orthopyroxene, and lack of clinopyroxene.
|
WHOLE-ROCK AND MINERAL CHEMISTRY |
|---|
Analytical methods
All whole-rock chemical analyses were performed by using X-ray fluorescence (XRF; SGS Canada Inc. Mineral Services). Detection limits for major oxides are 0·01% and for trace elements are 2 µg/g, except for Ba, which had a limit of detection of 20 µg/g. Mineral compositions were quantified by using wavelength-dispersive spectrometry (WDS) and X-ray maps by using energy-dispersive spectrometry (EDS) on the JEOL Superprobe 8600 electron probe microanalyzer (EPMA) at Johns Hopkins University. Natural and synthetic mineral standards and a ZAF correction scheme were used (Armstrong, 1988
). A beam current of 20 nA, accelerating voltage of 15 kV, and a beam diameter of 0–10 µm were employed.
Bulk composition
Whole-rock major and trace element chemical data from samples of unmelted Orestes granite, partially melted Orestes granite, partially melted Vanda felsic porphyry, partially melted Orestes granite breccia, granophyre zone, and granitic dike rocks are presented in Table 1. All samples were collected near location ‘A’ (Fig. 1). The distance column refers to the distance of the sample location into the dolerite sill from the chilled margin in a transect orthogonal to the strike of the dolerite chilled margin contact (i.e. the black line in Fig. 2b). It should be noted that, as before, these distances are corrected to be a projection from the ground surface onto a plane perpendicular to the contact assuming a contact dip of 45°. All samples of partially melted granite are from the Orestes pluton, except samples TH5 and TH6, which are from a Vanda felsic porphyry dike.
Whole-rock normative compositions plotted on an anorthite–albite–orthoclase ternary and a quartz–albite–anorthite ternary are shown in Figs 6 and 7, respectively, for unmelted Orestes granite, partially melted Orestes granite, partially melted Vanda felsic porphyry, partially melted Orestes granite breccia, granophyre zone, and granitic dike samples. A tight cluster of similar compositions of unmelted Orestes granite and partially melted granitic (Orestes and Vanda) samples from >50 cm from the dolerite chilled margin is distinct from the compositional trend of decreasing orthoclase component of partially melted Orestes granite samples beginning at least at a distance of 12·3 cm and closer to the dolerite chilled margin (see Fig. 6). The compositions of the granophyre zone define another tight cluster aligned with the previous trend but are more enriched in orthoclase component. A less obvious but similar trend is shown by Fig. 7b; however, the quartz component is relatively constant. The granophyre zone samples plot to the right of the water-saturated granite minimum at P(H2O) = 200 MPa (Tuttle & Bowen, 1958). With decreasing An/(An + Ab) content, the eutectic composition crosses to the upper right of, but does not intersect, the granophyre zone samples. Harker diagrams of major and trace elements (Fig. 8) also reflect the compositional trends observed in the ternary diagrams. For reference, also shown is a least-squares line fitted between samples of partially melted Orestes granite at <12·3 cm from the contact and the granophyre zone. For most elements, the partially melted granitic (Orestes and Vanda) samples >50 cm from the contact plot near this line. Granitic dike samples also plot near this line, particularly for oxides, Al2O3, CaO, and MgO.
|
|
|
) and granophyre (
). All of the plotted data are from Dolerite reaction zone
Five, 512 x 512 pixel X-ray maps at 100x magnification with a cumulative 1 s count time per pixel were collected perpendicular to the contact between the sliver of partially melted Orestes granite and the dolerite chilled margin. These maps include the sliver of partially melted Orestes granite, dolerite reaction zone, and dolerite, covering a distance of about 5 mm. Line averages parallel to the contact, representing averages of 51 columns, for CaO and SiO2 are shown in Fig. 9. The purpose of this windowing average is to smooth out the concentrations of a single column average, which is an attempt to mask the uncertainty associated with averaging over an area of insufficient size. The line averages demonstrate relative uniformity of oxide content (wt%) within zones but reveal large jumps between rock types. Line averages for other oxides are relatively constant when compared with CaO and SiO2 and are not shown.
|
Mineral compositions
Feldspar
Plagioclase is present in all rock units whereas alkali feldspar is restricted to unmelted Orestes granite, partially melted Orestes granite and breccia, partially melted Vanda felsic porphyry, granophyre zone, and granitic dike rocks. Plagioclase occurs as a primary phase in unmelted Orestes granite, as a primary phase in restite in partially melted Orestes granite, and as a reaction product formed from the breakdown of biotite in the presence of other phases. Plagioclase is also in the granophyre zone and granitic dikes as restite grains entrained from partially melted Orestes granite and as a primary phase in the dolerite reaction zone and unreacted dolerite chilled margin. Representative feldspar compositions are shown in Table 2 in oxide weight per cent and in cations per eight oxygens. With proximity to the dolerite chilled margin restitic plagioclase is on average more anorthitic in the partially melted Orestes granite, but is less anorthitic than dolerite reaction zone and dolerite chilled margin plagioclase as seen on an anorthite–albite–orthoclase ternary ternary (Fig. 7). Restitic alkali feldspar in partially melted Orestes granite is on average more K-rich with distance towards the dolerite chilled margin. The sieve texture of the feldspars most probably reflects melting along cleavage planes (Philpotts & Asher, 1993
, and references therein). There is no evidence for strong major element compositional variation from core to rim within feldspar grains.
|
Alkali feldspar is the only feldspar in the granophyre, with the exception of two analyzed plagioclase grains thought to be pieces broken free from sieve-textured restitic plagioclase grains. Alkali feldspar from the granophyre zone is on average more K-rich with distance towards the dolerite chilled margin (Fig. 7). Two-feldspar thermometry of restite grains yields temperatures of 635–700°C using SOLVCALC (Wen & Nekvasil, 1994
) and clearly records equilibration at sub-solidus temperatures.
Pyroxene
Pyroxene is present in the partially melted Orestes granite, partially melted Vanda felsic porphyry, within pieces of dolerite chilled margin contained in partially melted Orestes granite breccia, and in the dolerite reaction zone and dolerite chilled margin. Orthopyroxene occurs as either a reaction product from biotite breakdown, a product from reactive dissolution of clinopyroxene in the dolerite chilled margin at the contact, or as xenocrysts in the dolerite sills and, relatively less commonly, in the dolerite chilled margin. Representative pyroxene compositions are shown in Table 3 in oxide weight per cent and in cations per six oxygens. Orthopyroxene is more Mg-rich in the dolerite reaction zone than in partially melted Orestes granite (Fig. 10). It also increases systematically in Mg within the dolerite reaction zone, reaching a maximum at the boundary with the unreacted dolerite chilled margin (Fig. 11).
|
|
|
Clinopyroxene occurs solely in the dolerite sill, dolerite chilled margin, and within the unreacted centers of some pieces of dolerite chilled margin within the partially melted Orestes granite breccia. Clinopyroxene and orthopyroxene near the sharp contact between the dolerite reaction zone and unreacted dolerite chilled margin plot on an experimentally determined tie-line (Lindsley, 1983
; Andersen et al., 1993) indicating that the two pyroxenes were in equilibrium at temperatures of 900–950°C at the contact (Table 3).
Biotite
Biotite is present in the unmelted Orestes granite and partially melted granitic (Orestes and Vanda) rocks, as a crystallization product, and in the dolerite chilled margin. Most biotite grains are partially altered to clays and give poor EMPA totals. However, a representative biotite composition is given in Table 4 in oxide weight per cent and in cations per 11 oxygens. Magmatic magnetite and ilmenite occur primarily within reacted biotite; however, sub-solidus hydrothermal alteration is the likely cause for abundant hematite within reacted biotite. Representative Fe–Ti oxide compositions are given in Table 5 in oxide weight per cent and in cations per three or four oxygens. The oxidation state of Fe for each analysis was calculated by simultaneously solving linear equations for the total number of cations and charge balance. Rare coexisting magnetite and ilmenite and small grain size precluded their use for thermometry.
|
|
|
NATURE OF GRANITE PARTIAL MELTING AND REACTION RELATIONSHIPS |
|---|
A significant amount of field and laboratory work has been carried out to understand how granitic rocks undergo partial melting (Brown & Rushmer, 2006
, and references therein). Observations of natural partial melting have included either partial melting and disaggregation of granite or granodiorite xenoliths (Bacon, 1992; Green, 1994) or partial melting of granite, tonalite, gneiss, pegmatite, or metasediments at the walls of mafic dikes, plugs, sills, and feeder conduits at depths between 2 and 10 km (Al-Rawi & Carmichael, 1967; Kaczor, 1988; Kitchen, 1989; Philpotts & Asher, 1993; Holness & Watt, 2002; Petcovic & Grunder, 2003; Holness et al., 2005). The most salient observation in all these studies, despite varied protolith compositions, is the occurrence of multiple, compositionally distinct melts (as glass or granophyre) representing the products of melting reactions controlled by local mineral assemblages and not by general equilibrium minimum melting. Some of these studies have observed the breakdown of hydrous phases by dehydration melting (Patiño Douce, 1999). They stress the importance of dehydration reactions for the initiation of melting, further melt-fluxing caused by the release of water, and permeability enhancement by crack generation owing to the positive volume change during partial melting, albeit with little or no melt mobilization [e.g. Philpotts & Asher (1993) and Holness & Watt (2002), respectively].
During partial melting of the Orestes granite, all major phases including plagioclase, alkali feldspar, and quartz were consumed to varying degrees. Biotite was consumed completely. Stages of biotite breakdown as a result of dehydration melting are observed in the partially melted Orestes granite. In order of increasing degree of breakdown and, therefore, possibly melting, these stages include a change of color from greenish brown to reddish brown, dusting of Fe–Ti oxides at biotite rims, and the complete breakdown of biotite to an intergrowth of Fe–Ti oxides and plagioclase with orthopyroxene bordering the original biotite grain boundary. This sequence is similar to that reported in other studies (e.g. Al-Rawi & Carmichael, 1967; Brearley, 1987). The absence of alkali feldspar as a reaction product of biotite breakdown is common in other field (e.g. Petcovic & Grunder, 2003) and experimental studies (Naney, 1983; Johnson & Rutherford, 1989; Patiño Douce & Beard, 1995). Alkali feldspar may not be a stable reaction product, possibly because of a high water content of the initial dehydration melts and also the low pressure of melting (Naney, 1983; Johnson & Rutherford, 1989). In this study, the sequence of biotite breakdown is generally observed in partially melted Orestes granite as one approaches the dolerite chilled margin. However, evidence of more than a single stage can be observed within single samples, particularly with distance away from the dolerite chilled margin. Alteration of biotite to chlorite is probably a hydrothermal phenomenon and occurs after the partial melting event; chlorite is unstable at high temperature (Graphchikov et al., 1999). The presence of unreacted biotite (i.e. no breakdown products) in partially melted Orestes granite with higher amounts of granophyre is inferred to be a secondary crystallization product during solidification of the granitic melt (see Fig. 5g).
The relative contributions of plagioclase, compared with alkali feldspar, to the melt composition, as indicated by the whole-rock composition of the granophyre zone, is small as observed by the anorthite–albite–orthoclase ternary and quartz–albite–anorthite ternary (Fig. 7). That is, the albite component added to the melt has evidently come from the alkali feldspar and not the plagioclase. The whole-rock compositions of the granophyre zone are depleted in both albite and anorthite component relative to the composition of the unmelted Orestes granite and partially melted Orestes granite samples collected at >50 cm from the dolerite chilled margin (Fig. 6). However, the whole-rock granophyre zone compositions do not intersect a line representing simple mixing of alkali feldspar domains and quartz (Fig. 7b). This suggests either that the whole-rock composition of the granophyre zone does not represent the true composition of the granophyre zone because of the inclusion of grains of restitic plagioclase or that many of the plagioclase domains have been subsequently hydrothermally altered. The two plagioclase grains plotted (Fig. 7) are, as mentioned above, probably unmelted fragments of restitic plagioclase. Given that the An/(An + Ab) value for the host Orestes granite is 0·27, then the position of the whole-rock granophyre zone composition in Fig. 7b is shifted to the albite apex away from the expected melt composition between the two eutectic compositions (circles labeled ‘e’) at An/(An + Ab) values of 0·21 and 0·36. This suggests that partial melting is characterized by disequilibrium between melt and solid.
The 2·5 mm wide dolerite reaction zone at the outer edge of the dolerite chilled margin was initially normal chilled dolerite. As a result of the large chemical potential gradient at near-solidus temperatures across this contact, however, reactive dissolution of clinopyroxene with concomitant precipitation of orthopyroxene occurred in response to diffusion in both directions. A similar reaction relationship in the chilled margin of a diabase feeder dike was reported by Philpotts & Asher (1993) and is discussed below. A scaling of the diffusion equation can be used to estimate the duration for the growth of this reaction zone, provided the melt produced in the dolerite as a result of reactive dissolution is beyond the (interconnected) percolation threshold. If chemical transport is governed purely by a realistic chemical diffusivity in the melt of D 10–10 m2/s, operating over a length scale (L) of 2 mm, then the diffusion time scale (L2/D) is of the order of days. Alternatively, if the growth of this reaction zone is dominated by diffusion in the solid, with a realistic D of the order of 10–14 m2/s, the time scale is instead of the order of 40 years. It is likely that the true duration is between these two estimates; experiments under similar conditions would clarify the appropriate diffusivity for this process. Ultimately, the growth time of this reaction zone places an important constraint on the duration of partial melting of granitic wall rock by the adjacent dolerite sill.
|
MODELS OF GRANITE PARTIAL MELTING AND MELT SEGREGATION |
|---|
The upper lobe of the Basement Sill, or dolerite feeder, was emplaced into cool granitic country rock, producing a thin, well-formed chilled margin of dolerite along the contact. The presence of a strongly chilled margin suggests that possible pre-heating of country rock as a result of the previous emplacement of stratigraphically higher sills at this location was negligible. Given a geothermal gradient of 25°C/km, the ambient country rock temperature was about 100°C. That the dolerite feeder acted as a conduit for sustained flux of magma and was not a single magma injection is supported by a calculated contact temperature of
715°C (at 100 MPa). This model temperature was calculated by solving the full heat conduction equation for cooling and solidification of an instantaneously injected sill (e.g. Turcotte & Schubert, 1982, p. 172) assuming a temperature of 1200°C for the doleritic magma based on two-pyroxene thermometry (Simon & Marsh, 2005). This temperature (i.e. 715°C) is approximately at or just above the water-saturated granite solidus for pressures between 60 and 90 MPa (Tuttle & Bowen, 1958) and is much lower than the temperature range of 900–950°C necessary to produce the observed extent of melting. More probably, the dolerite feeder acted as a conduit for a sustained flux of magma for some period of time, pinning the contact temperature above the granite minimum, and causing a partial melting front to propagate outward into the granitic country rock. This is consistent with the preferred model of Petcovic & Dufek (2005) wherein they reconciled wall rock melting by invoking advective magma transport within a Columbia River flood basalt feeder dike.
The operative granite solidus here is expected to be hotter than the water-saturated granite minimum. A minimum estimate of the water content of the unmelted Orestes granite is 0·33 wt%, based on assuming complete biotite breakdown at an initial modal content of 7·8 vol.%, obtained by detailed modal analysis (i.e. manual tracing of all biotite grains on a digitized image of a single thin section). For comparison, partial melting experiments at 50 MPa involving a natural granite with an assemblage of primarily quartz, plagioclase, alkali feldspar, and muscovite with a total water content of 0·58 wt% yielded 5 vol.% melt at 850°C and with the addition of 1 wt% H2O yielded 40 vol.% melt at 800°C (Attrill & Gibb, 2003). Manual tracing of granophyre regions in thin sections between restite phases in partially melted Orestes granite provides an estimate of the melt fraction at solidification and, hence, not the degree of melting. The degree of melting was most certainly greater than this estimate for partially melted Orestes granite at <50 cm from the dolerite chilled margin. At 12·3 cm, 1·2 m, and 5·2 m there is, respectively, 55 vol.%, 31 vol.%, and 23 vol.% granophyre. Furthermore, these melt fractions can be considered to be minimum estimates because some melt may have solidified as overgrowths on restite grains. Given the uncertainty in the biotite composition and mode, and the potential presence of at most 0·1–0·3 wt% intergranular H2O (Whitney, 1988) and water stored in grain boundaries [0·1 g of H2O m–2 estimated by Holness et al. (2005) in a pelite], it is possible that melt fractions of the magnitude observed in the experiments of Attrill & Gibb (2003) could be generated at similar temperatures.
It is clear from Figs 6 and 7 that the silicic melt forming the granophyre zone was segregated from partially melted Orestes granite over the interval >12·3 cm but <50 cm from the dolerite chilled margin. The Harker diagrams of Fig. 8 are consistent with this interpretation and are analogous with the restite-unmixing model of Chappell et al. (1987). The chemical relations of Fig. 6 also suggest segregation of a monotonically decreasing volume of melt from partially melted Orestes granite with increasing distance from the dolerite chilled margin. A simple mass-balance calculation based on normative alkali feldspar from whole-rock compositions indicates that 0·45, 0·55, and 0·65 mass fraction removals of granitic melt of the composition of the granitic dikes from unmelted Orestes granite can restore the compositions of partially melted Orestes granite at distances of 12·3, 6·7, and 5·6 cm. These estimates represent minimum mass fractions of melt removed because the partially melted Orestes granite still contains interstitial granophyre between restite phases that is assumed to have been granitic melt.
Another estimate of the mass fraction of granitic melt removed from the partially melted Orestes granite can be made through application of batch and fractional melting models using trace element partitioning between melt and restite. End-member fractional melting is not likely based on the observation of granophyre within partially melted Orestes granite, representing residual granitic melt left after segregation. A fractional melting model is, nevertheless, useful for providing a lower limit on the degree of granitic melt removed provided that end-member batch melting is also not likely. Based on whole-rock trace element compositions (Table 1), Rb is the least compatible element in this assemblage. Partition coefficients (crystal/melt) for Rb in alkali feldspar, plagioclase, quartz, and biotite are, respectively, 0·5, 0·04, 0·04, and 4·2 (Ragland, 1989). A bulk partition coefficient was calculated by using the results of a CIPW norm calculation on a whole-rock unmelted Orestes granite composition normalized after adding biotite. The two melting models, using three different initial mass fractions of biotite, are plotted as a function of melt mass fraction using the calculated bulk partition coefficient for Rb in Fig. 12. Also shown in this figure are Rb concentrations for samples A-395-4, A-395-1, and A-395-3, which are partially melted Orestes granite collected at, respectively, 12·3, 6·7, and 5·6 cm from the dolerite chilled margin. Batch melting predicts 0·20, 0·31, and 0·83 mass fractions of melt removed, whereas a fractional melting model predicts 0·18, 0·24, and 0·48 mass fractions of melt removed from partially melted Orestes granite at, respectively, 12·3, 6·7, and 5·6 cm from the dolerite chilled margin. Except for the batch melting model at 5·6 cm, both melting models predict a smaller mass fraction of melt removed at each location than the simple mass-balance calculation based on normative alkali feldspar from whole-rock compositions.
|
|
MECHANICS OF GRANITIC MELT SEGREGATION |
|---|
Interstitial pore melt will segregate from its matrix in response to a gradient in pore pressure. A gradient in pore pressure can be generated by body forces, such as melt buoyancy relative to the matrix, by changes in hydrostatic or deviatoric stress, and positive volume changes associated with partial melting. In this instance, it is unlikely, for two reasons, that melt buoyancy is responsible for melt segregation. First, the contact between the dolerite chilled margin and granitic country rock dips at
45°, with the granitic country rock being the hanging wall and the dolerite chilled margin being the foot wall (see Fig. 2). This orientation is opposite to that required for granitic melt to collect buoyantly at the contact with the dolerite chilled margin. Second, this gravitational instability is unlikely to develop because the time scale for melt to buoyantly rise through partially molten granitic rock is much larger than the time scale for the dolerite sill to cool. This also prevents segregated melt that collects at the dolerite chilled margin from buoyantly rising as a single volume along the contact. For example, a Stokes-flow (i.e. creeping flow) scaling for the velocity of buoyant melt rise (not porous melt flow) in a host of partially molten granitic rock gives a time scale of ts = µ/
gW 1014 s, where µ is the effective viscosity of the solid matrix, which is 1017 Pa s for Westerly granite containing a melt fraction of 10% (Rutter & Neumann, 1995), is the density contrast between melt and rock (102 kg/m3), g is gravitational acceleration, and W is the thickness of the melt layer (1 m). In contrast, a time scale for conductive cooling of a solidifying dolerite sill of thickness L = 100 m is tc = L2/
1010 s, where is thermal diffusivity (10–6 m2/s). That is, the ratio of the characteristic time for significant buoyant flow relative to that for cooling is ts/tc 104. Melt segregation caused by externally imposed changes in hydrostatic stress is also not likely owing to lack of a plausible mechanism for causing large changes in hydrostatic stress, such as introduction or release of overburden. Even if these mechanisms did occur, the length scale over which melt segregation has operated here is much less than the tectonic length scales over which changes in hydrostatic stress would exist. That is, the expected gradients in hydrostatic stress are much too broad to work effectively over this highly restricted area.
Partial melting commonly involves a local positive volume change and, therefore, creates an elevated pore pressure directed over a selected spatial region. Evidence for reaction-driven deformation has been observed in the contact aureoles of plutons, for example, as mostly intra-crystalline micro-cracks thought to contain melt (Holness & Watt, 2002). No evidence of cracking has been observed in the partially melted granitic (Orestes and Vanda) rocks described here, possibly because of the lack of reactions with large positive volume changes, such as muscovite breakdown (Rushmer, 2001). Even with local pore melt pressure gradients at the sites of reaction there is no general preference to drive melt flow in the direction of the dolerite chilled margin, a condition necessary for melt segregation to occur. This conclusion is supported by the absence of evidence for melt segregation in the contact aureole studied by Holness & Watt (2002). Nevertheless, elevated pore melt pressure caused by partial melting of the granitic rocks is probably, at least in part, responsible for generating the overpressure necessary to fracture the dolerite chilled margin and create space for the formation of the granitic dikes.
Changes in deviatoric stress related to both melting in the wall rock and thermal contraction in response to dolerite solidification, following flow cessation, are perhaps most likely to have generated the pore melt pressure gradient necessary to drive melt segregation and collection into a melt-rich reservoir. Let us consider the schematic two-dimensional model shown in Fig. 13. Here it is assumed that melt segregation takes place after partial melting and not during melting. In the first frame, the pore melt pressure is at the level of the hydrostatic stress (
1 = 2 = 3) and the pore melt is static. A decrease in 2 as a result of, for example, a tensional strain, at the right interface causes a planar crack to be initiated in the partially melted granite immediately adjacent and parallel to the dolerite chilled margin. The tear occurs exactly where the partially melted granite has the least strength (i.e. highest melt fraction). Tiny slices of partially melted granite on the dolerite chilled margin side of the crack remain welded to the dolerite chilled contact. The melt pore pressure in the crack is less than in the partially melted granite. This pressure gradient drives melt into the crack by porous melt flow with commensurate pore space reduction, or lateral compaction, of the partially melted granite on both sides of the crack. With continued strain, the melt-filled crack widens to form a staging melt reservoir for granitic dike emplacement.
|
If matrix deformation is assumed to occur by viscous deformation of the crystals, then a length scale over which compaction occurs adjacent to an impermeable boundary is given by the compaction length,
[(
+ 4/3
)
/µ]1/2 101/2 (McKenzie, 1984). Here, and are the effective bulk and shear viscosities of the matrix [1015 Pa s following Rabinowicz & Vigneresse (2004)]. Dynamic viscosity of the granitic melt, µ, is 106 Pa s and permeability, , is 10–8 m2 using the Rumpf–Gupte relation from Hersum et al. (2005) with a mean crystal size of 0·1 mm and a melt fraction of 0·5. With these properties, the compaction length scale is 3 m. This compaction length represents an upper bound because the model assumes a partially melted host that contains a constant melt fraction when in fact melt may be progressively lost and never achieve this state. However, for this situation, permeability will decrease sharply with distance away from the dolerite chilled margin as the melt fraction correspondingly decreases. The compaction length is approximately one order of magnitude larger than the distance for which partially melted Orestes granite has a composition that differs from the unmelted Orestes granite, which suggests that melt can be removed by this process. The magnitude of the deviatoric stress necessary to initiate and then fill a crack by segregating melt is probably driven by pressure reduction within the doleritic magma upon cessation of flow in concert with contraction owing to solidification of the magma feeder itself. Each process has a distinct characteristic strain rate that would dictate both the mechanism of matrix deformation and also whether the rate of filling the granitic melt reservoir is controlled by the rate of porous melt flow or matrix deformation. A complete continuum-scale model of simultaneous heat and mass transport is under development to test these scenarios.
|
EMPLACEMENT OF GRANITIC DIKES |
|---|
There are few documented natural examples of partial melt from country rock intruding the adjacent magmatic rock that initially generated the melting. Kaczor (1988
) reported that partially melted granite, up to 56% melt, around a trachyandesite plug in the Sierra Nevada, California intruded back into the plug after the plug had cooled enough to develop jointing. There is no evidence, however, that melt segregated from the restite minerals or that any mixing between the granitic melt and the trachyandesite occurred (Kaczor, 1988). A field example similar to the present study area was presented by Philpotts & Asher (1993), who described multiple events of partial melt, derived from gneiss and pegmatite country rock screens, intruding the Higganum diabase dike in Connecticut. The first set of melts entering the diabase dike are preserved as felsic wisps (<15 µm wide) that penetrate at most about 0·5 cm into partially solidified diabase chilled margin. These wisps were subsequently stretched and thinned by further flow of the diabasic magma. Upon complete solidification of the diabase, a network of interconnected granophyric veins, each up to 4 cm wide, entered the diabase. These veins are oriented perpendicular to the screens, traversing up to several meters between intervening screens. Near the chilled margins of the diabase, the veins abruptly thin and merge with networks of thinner (<100 µm) veinlets that connect with grain boundary granophyre in the partially melted country rock screens (Philpotts & Asher, 1993). Although this last example more closely resembles the Ferrar granitic dikes, the volume of intruding melt, the volume of partially melted wall rock, and the extent of intrusion are relatively minor relative to the Ferrar granitic dikes. The granophyric magma zone sandwiched between partially melted Orestes granite and the dolerite chilled margin, which is connected directly to granitic dike rock at location ‘A’ (Fig. 1), unequivocally identifies this zone as the staging reservoir for the melt forming the plexus of dikes within the dolerite. Based on the volume of the granitic dikes, it is apparent that the granophyric magma zone was probably significantly wider prior to the evacuation of melt that formed the granitic dikes. Moreover, the wide distribution of these large dikes in the Bull Pass area suggests that the dikes originated from a number of locations by the same process. Another notable aspect of this occurrence is the high rigidity of the dolerite chilled margin, which acted as a mechanical barrier to prevent mixing of granitic melt with doleritic melt, allowing formation of a reservoir of granitic melt from which to generate the granitic dikes.
Granitic dike emplacement clearly occurred late in the cycle of dolerite emplacement and solidification. To make fracture-like boundaries, the dolerite had to have acquired enough strength to allow fracturing rather than viscous flow. In partially molten rock, fracturing will occur only at relatively low degrees of melt fraction (<50%; e.g. Marsh, 2002). Neither the dolerite nor the granitic dike show chilling against one another, which suggests that the whole system was still hot, perhaps above the solidus of the dike magma. This condition allowed the granitic magma to move without confronting the challenge of transporting sluggish viscous silicic melt through a narrow dike in cool country rock (Rubin, 1995). Also, unlike at the dolerite chilled margin adjacent to the granophyre melt at location ‘A’, there is no corresponding reaction zone in the dolerite at the contact between the dolerite and the granitic dike. The lack of a corresponding reaction zone is found only at the contact between the dolerite and the granitic dike within 1 m of the dolerite chilled margin and it is unknown if a reaction zone exists deeper into the interior of the sill as outcrops of the contact are altered. Nevertheless, in the proximity of the dolerite chilled margin, the lack of a reaction zone suggests that the emplacement of the granitic dikes occurred after granophyric melt began collecting at the dolerite chilled margin (and at a lower temperature).
Another interesting feature of the field area is the complete absence of granitic dikes directed away from the dolerite feeder into the partially melted granitic (Orestes and Vanda) rocks. This suggests that the dolerite feeder was nearly solidified at the time of granitic dike formation and, therefore, still in the process of cooling. This may provide an explanation for the driving force and thus the direction of diking. The dolerite chilled margin formed a rigid wall separating a contraction on the dolerite side from an overpressure on the wall rock side. The contraction was due to 10% volume reduction associated with dolerite solidification (based on the density contrast between melt and solid rock) and the overpressure owing to excess pore melt pressure associated with melting. This approximately dipole pressure field apparently directed all dike propagation into the dolerite.
|
IMPLICATION FOR CONTAMINATION OF BASALTIC MAGMA IN THE LOWER CRUST |
|---|
The motivation of this research is, in part, to use the observations for melt generation and segregation from a shallow crustal example to infer how similar processes operate in the lower crust. However, there are several caveats including, for example, that the length scale of melt segregation (centimeter to meter in this study as opposed to kilometer to tens of kilometers in the deep crust) and the rheological behavior (i.e. brittle vs ductile) are likely to be very different. Also, major element melt compositions in the deep crust are less likely to be out of chemical equilibrium than in more shallow and rapid melting environments such as contact aureoles. Another limiting factor is the geometry of the contact. In the present study, the contact between dolerite and granite was
45° from horizontal. Perhaps the key to large-scale crustal anatexis (i.e. felsic melt production) is underplating of dolerite such that heat is advected in a purely vertical direction from the dolerite into the overlying granite. Such a model has been advocated in other field and numerical studies (see Bergantz, 1989).
The field relations, essentially crustal anatexis arrested during development, allow us to speculate on the causes of contamination of magmas in traversing continental crust. Here, contamination is defined as modification of incompatible and not major element composition of the intruding magma by the host country rock. This has been shown countless times in isotopic studies (e.g. Carter et al., 1979; Taylor, 1980; Stewart & DePaolo, 1990; Zeng et al., 2005), but the actual physical processes by which this happens remain unconstrained. Although stoping is an obvious viable mechanism (it is seen to a very limited extent in these sills), contamination is evidently caused by much more than settling and partial digestion of stoped blocks. Many lavas and intrusive rocks showing contamination show no evidence whatsoever of partially digested or fragmented xenoliths, suggesting that there must be other, much more subtle and difficult to recognize, common contamination processes. One such process is suggested by an extension of the process described here. This involves the reactivation of a magma feeder containing the granitic dikes generated as a consequence of a previous episode of magma flow. Plutonic magmatic systems must have a vertically extensive underlying plumbing style similar to that observed for typical volcanic systems. That is, a major eruptive episode is generally marked by a series of eruption events separated by repose times. At depth this pulsative sequence manifests itself in magmatic feeders as times of local inflation and magma flow interspersed with periods of deflation and partial solidification.
In Figure 14a–d, we schematically show the sequence of events involving a basaltic magma containing granitic dikes of the general nature found in the Ferrar system, which leads to crustal contamination of the basaltic feeder. The active feeder at near-solidus temperatures and containing silicic dikelets, generated as described herein, is reactivated by the introduction and flow of near-liquidus temperature magma from deeper in the system. The ensuing flow causes a melting front to propagate progressively outwards towards the margins. The original basaltic rock of the feeder and the silicic dikelets are each reheated and remobilized and, with subsequent magma flow and melt back, are systematically caught up in the flow. The melted dikelets are stretched, as a result of the developing parabolic magma velocity profile, and progressively thinned until chemical diffusion becomes effective in dispersing the silicic material within the basaltic magma. That is, the characteristic time for chemical diffusion (tD) depends on the square of the dikelet half-thickness (L2) divided by the governing chemical diffusivity (D), tD L2/D. Because D is typically very small (e.g. 10–10 m2/s), diffusion times for any appreciable value of L are commonly much longer than local solidification times. However, if the distance of magma flow is large, thinning becomes extreme, L is greatly reduced, and chemical diffusion becomes highly effective. All physical traces of the dikelets are removed, leaving a systematic and diagnostic distribution of incompatible element and unusual isotopic abundances. For example, in this process more silicic dikelets are likely to exist and be assimilated near the margins than in the center of the basaltic feeder, and this would produce an elevated, concave profile of 87Sr/86Sr within the feeder (see Fig. 14e). This signal would travel with the feeder magma and upon emplacement as a lava or sill would show up as a diagnostic feature. Spatial signals of this nature can be revealed through systematic analysis of sample profiles through dikes, sills, and plutons. A concave variation of this very nature has been found in 87Sr/86Sr by Hergt et al. (1989) in a profile through the 150 m thick Ferrar dolerite sill at Portal Peak in the Transantarctic Mountains. The variation in 87Sr/86Sr is large (0·7090–0·7110; see Fig. 14f) and the process outlined above provides a realistic explanation of the variation in strontium isotopes. Another factor is that initial disequilibrium melts may be richer in 87Sr/86Sr than the bulk-rock composition, as demonstrated in experiments by Knesel & Davidson (1999).
|
With increasing depth in the crust and increasingly hotter wall rock, this contamination process becomes increasingly effective in all its critical aspects. Melting is more extensive, solidification is slower, melt segregation is more effective, the dipolar pressure field is larger, more and larger granitic dikes are likely to form, and reactivation of the feeder zone is more likely given a wider window of cooling time. Conversely, diking is obviously limited to crustal depths at which strain associated with thermal contraction is accommodated by fracture and not ductile flow. Also, the process described above should not be confused or applied to understanding the larger-scale generation of hybrid granitic magmas through intrusion of basaltic magmas into the deep crust.
|
CONCLUSIONS |
|---|
A natural example of silicic melt generation, segregation, and emplacement of granitic dikes into Ferrar dolerites of the McMurdo Dry Valleys, Antarctica provides an opportunity to examine fully these sequential processes on a convenient spatial scale. The granitic dikes are numerous, long (hundreds of meters), fairly thick (
10–30 cm), interconnected, and fine-grained. The root source of one dike is completely exposed at the upper contact of the Basement Sill and granitic country rock. The dike emanates from a thin (5 cm) melt sheet separating chilled dolerite from partially melted granite. Residual interstitial granophyric melt decreases away from the contact from 55% to zero within a distance of <20 m. Higher than expected dolerite contact temperatures of 900–950°C calculated using two-pyroxene thermometry suggest that the dolerite feeder acted as an open conduit for a sustained flux of magma. As a consequence of this flow, the contact temperature was pinned above the ‘dry’ granite minimum, the most restrictive condition necessary to generate granitic melt. Closed-system partial melting of granite occurred beyond 50 cm from the dolerite chilled margin. However, compositional moment balances on a anorthite–albite–orthoclase ternary between the alkali feldspar-enriched melt sheet and granitic dike whole-rock compositions are reconciled by melts segregated from increasingly alkali feldspar-depleted partially melted granite at 12·3 cm and closer to the dolerite chilled margin. Melting models and mass-balance calculations predict a range of between 48 and 83% maximum volumes of segregated granitic melt, but these are only estimates as the samples are not exclusively residuum. If granitic melt segregation occurs by viscous compaction of the restitic crystal matrix, then, employing commonly used values of the critical parameters, the compaction length scale is 3 m. This is an upper bound, as the compaction model assumes a constant melt fraction. Nevertheless, the result is only an order of magnitude larger than the distance over which the partially melted granite has a composition that differs from unmelted granite. Contraction attending cessation of doleritic magma flow in addition to solidification-induced contraction probably generated deviatoric stresses within the partially melted zone that initiated crack formation at and parallel to the contact, allowing interstitial melts to flow outward into the dolerite in response to a pore pressure gradient. Excess pore pressure within this granitic melt reservoir along the contact subsequently tore open the brittle dolerite chilled margin, in the fashion of a trapdoor, and emplaced, essentially by evacuation, granitic dikes into the solidifying dolerite. Granite partial melting, segregation, and dike emplacement probably occurred within a maximum period of several tens of years, as suggested by the time estimated to produce, by interdiffusion between the granitic melt and dolerite, a thin (2·5 mm) distinctive planar dolerite reaction zone within the dolerite chilled margin. Reactivation of similarly injected basaltic feeders deeper in the crust, with dikelet stretching and absorption by simultaneous diffusion, presents a possible means of extensive and subtle crustal contamination of basaltic magma. |
ACKNOWLEDGEMENTS |
|---|
We thank Mike Weiss, Amanda Charrier, and participants of the 2005 NSF-funded ‘Magmatic Field Laboratory Workshop’ for field assistance. Special thanks go to the PHI pilots and the McMurdo support staff for their dedicated, high-quality help. We thank Ken Livi for assistance with microprobe analysis, and John Ferry for comments on an initial draft of the manuscript. George Bergantz, Alberto Patiño Douce, and an anonymous reviewer are thanked for constructive reviews. Editorial assistance and a review from Marjorie Wilson are gratefully appreciated. This work is supported by NSF Grants OPP 0229306 and OPP 0440718 to Johns Hopkins University.
|
FOOTNOTES |
|---|
Present address: Department of Geoscience, University of Nevada at Las Vegas, 4505 Maryland Pkwy, Las Vegas, NV 89154, USA. 
*Corresponding author. Present address: Lamont–Doherty Earth Observatory, 61 Route 9W, Palisades, NY 10964, USA. Telephone: (845) 365-8662. Fax: (845) 365-8155. E-mail: hersum{at}ldeo.columbia.edu
|
REFERENCES |
|---|
Allibone AH, Cox SC, Graham IJ, Smillie RW, Johnstone RD, Ellery SG, Palmer K. Granitoids of the Dry Valleys area, southern Victoria Land: plutons, field relationships, and isotopic dating. New Zealand Journal of Geophysical Research (1993a) 36:281–297.
Allibone AH, Cox SC, Smillie RW. Granitoids of the Dry Valleys area, southern Victoria Land: geochemistry and evolution along the early Paleozoic Antarctic Craton margin. New Zealand Journal of Geophysical Research (1993b) 36:299–316.
Al-Rawi Y, Carmichael ISE. Natural fusion of granite. American Mineralogist (1967) 52:1806–1814.[Web of Science]
Andersen DJ, Lindsley DH, Davidson PM. QUILF: a PASCAL program to assess equilibria among Fe–Mg–Ti oxides, pyroxenes, olivine, and quartz. Computers and Geosciences (1993) 19:1333–1350.[CrossRef]
Armstrong JT. Quantitative analysis of silicate and oxide minerals: Comparison of Monte Carlo, ZAF and phi–rho–z procedures. In: Microbeam Analysis—1988—Newbury DE, ed. (1988) San Francisco, CA: San Francisco Press. 239–246.
Attrill PG, Gibb FGF. Partial melting and recrystallization of granite and their application to deep disposal of radioactive waste. Lithos (2003) 67:103–117.[CrossRef][Web of Science]
Bacon CR. Partially melted granodiorite and related rocks ejected from Crater Lake caldera, Oregon. Transactions of the Royal Society of Edinburgh (1992) 83:27–47.
Behrendt JC, LeMasurier WE, Cooper AK, Tessensohn F, Trehu A, Damaske D. Geophysical studies of the West Antarctic Rift System. Tectonics (1991) 10:1257–1273.[Web of Science]
Bergantz GW. Underplating and partial melting—implications for melt generation and extraction. Science (1989) 245:1093–1095.
Borg SG, DePaolo DJ, Smith BM. Isotopic structure and tectonics of the central Transantarctic Mountains. Journal of Geophysical Research (1990) 95:6647–6667.
Brearley AJ. A natural example of the disequilibrium breakdown of biotite at high temperature: TEM observations and comparison with experimental kinetic data. Mineralogical Magazine (1987) 51:93–106.[CrossRef][Web of Science]
Brown M. Melt extraction from lower continental crust of orogens: the field evidence. In: Evolution and Differentiation of the Continental Crust—Brown M, Rushmer T, eds. (2006) New York: Cambridge University Press. 331–383.
Brown M, Rushmer T. Introduction. In: Evolution and Differentiation of the Continental Crust—Brown M, Rushmer T, eds. (2006) New York: Cambridge University Press. 1–20.
Carter SR, Evensen NM, Hamilton PJ, O’Nions RK. Basaltic magma sources during the opening of the North Atlantic. Nature (1979) 281:28–30.[CrossRef][Web of Science]
Chappell BW, White AJR, Wyborn D. The importance of residual source material (restite) in granitic petrogenesis. Journal of Petrology (1987) 28:1111–1138.
Elliot DH, Fleming TH. Weddell triple junction: The principal focus of Ferrar and Karoo magmatism during initial breakup of Gondwana. Geology (2000) 28(6):539–542.
Elliot DH, Fleming TH. Occurrence and dispersal of magmas in the Jurassic Ferrar Large Igneous Province, Antarctica. Gondwana Research (2004) 7:223–237.[CrossRef][Web of Science]
Fitzgerald PG, Stump E. Cretaceous and Cenozoic episodic denudation of the Transantarctic Mountains, Antarctica; new constraints from apatite fission track thermochronology in the Scott Glacier region. Journal of Geophysical Research (1997) 102:7747–7765.[CrossRef]
Graphchikov AA, Konilov AN, Clemens JD. Biotite dehydration, partial melting, and fluid composition: Experiments in the system KAlO2–FeO–MgO–SiO2–H2O–CO2. American Mineralogist (1999) 84:15–26.[Abstract]
Green NL. Mechanism for middle to upper crustal contamination: Evidence from continental-margin magmas. Geology (1994) 22:231–234.
Gunn BM, Warren G. Geology of Victoria Land between the Mawson and Mulock Glaciers, Antarctica. New Zealand Geological Survey Bulletin (1962) 71:157.
Hamilton W. Diabase Sheets of the Taylor Glacier Region Victoria Land, Antarctica. US Geological Survey, Professional Papers (1965) 456-B.
Hergt JM, Chappell BW, Faure G, Mensing TM. The geochemistry of Jurassic dolerites from Portal Peak, Antarctica. Contributions to Mineralogy and Petrology (1989) 102:298–305.[CrossRef][Web of Science]
Hersum TG, Hilpert M, Marsh BD. Permeability and melt flow in simulated and natural partially molten basaltic magmas. Earth and Planetary Science Letters (2005) 237:798–814.[CrossRef][Web of Science]
Holness MB, Watt GR. The aureole of the Traigh Bhan na Sgurra Sill, Isle of Mull: reaction-driven microcracking during pyrometamorphism. Journal of Petrology (2002) 43:511–534.
Holness MB, Dane K, Sides R, Richardson C, Caddick M. Melting and melt segregation in the aureole if the Glenmore Plug, Ardnamurchan. Journal of Metamorphic Petrology (2005) 23:29–43.[CrossRef]
Johnson MC, Rutherford MJ. Experimentally determined conditions in the Fish Canyon Tuff, Colorado, magma chamber. Journal of Petrology (1989) 30:711–737.
Kaczor SM. Disequilibrium melting of granite at the contact with a basic plug: a geochemical and petrographic study. Journal of Geology (1988) 96:61–78.[Web of Science]
Kitchen DE. The disequilibrium partial melting and assimilation of Caledonian granite by Tertiary basalt at Barnesmore, Co. Donegal. Geological Magazine (1989) 126:397–405.[Abstract]
Knesel KM, Davidson JP. Sr isotope systematics during melt segregation by intrusion of basalt into continental crust. Contributions to Mineralogy and Petrology (1999) 136:285–295.[CrossRef][Web of Science]
Kovach LA. Dynamics of magma ascent through the Sierra Nevada, California. (1984) Baltimore, MD: Johns Hopkins University. 287. PhD dissertation.
Lindsley DH. Pyroxene thermometry. American Mineralogist (1983) 68:477–493.[Abstract]
Marsh BD. On bimodal differentiation by solidification front instability in basaltic magmas, I: Basic mechanics. Geochimica et Cosmochimica Acta (2002) 66:2211–2229.[CrossRef][Web of Science]
Marsh BD. A magmatic mush column Rosetta Stone: The McMurdo Dry Valleys of Antarctica. EOS Transactions, American Geophysical Union (2004) 47:497–502.
McKelvey BC, Webb PN. Geological investigations in southern Victoria Land, Antarctica. New Zealand Journal of Geophysical Research (1962) 5:143–162.
McKenzie D. The generation and compaction of partially molten rocks. Journal of Petrology (1984) 25:713–765.
Naney MT. Phase equilibria of rock-forming ferromagnesian silicates in granitic systems. American Journal of Science (1983) 283:993–1033.
Patiño Douce AE. What do experiments tell us about the relative contributions of crust and mantle to the origin of granitic magmas? In: Understanding Granites. Integrating New and Classical Techniques. Geological Society, London, Special Publications—Castro A, Fernandez C, Vigneresse JL, eds. (1999) 158:55–75.
Patiño Douce AE, Beard JS. Dehydration-melting of biotite gneiss and quartz amphibolite from 3 to 15 kbar. Journal of Petrology (1995) 36:707–738.
Petcovic HL, Dufek JD. Modeling magma flow and cooling in dikes: Implications for emplacement of Columbia River flood basalts. Journal of Geophysical Research (2005) 110. B10201, doi:10.1029/2004JB003432.
Petcovic HL, Grunder AL. Textural and thermal history of partial melting in tonalitic wallrock at the margin of a basalt dike, Wallowa Mountains, Oregon. Journal of Petrology (2003) 44:2287–2312.
Philpotts AR, Asher PM. Wallrock melting and reaction effects along the Higganum diabase dike in Connecticut: contamination of a continental flood basalt feeder. Journal of Petrology (1993) 34:1029–1058.
Rabinowicz M, Vigneresse JL. Melt segregation under compaction and shear channeling: Application to granitic melt segregation in a continental crust. Journal of Geophysical Research (2004) 109. B04407, doi:10.1029/2002JB002372.
Ragland PC. Basic Analytical Petrology (1989) New York: Oxford University Press.
Riley TR, Curtis ML, Leat PT, Watkeys MK, Duncan RA, Millar IL, Owens WH. Overlap of Karoo and Ferrar magma types in KwaZulu Natal, South Africa. Journal of Petrology (2006) 45:541–566.
Rubin AM. Getting granite dikes out of the source region. Journal of Geophysical Research (1995) 100:5911–5390.[CrossRef]
Rushmer T. Volume change during partial melting reactions: implications for melt extraction, melt geochemistry and crustal rheology. Tectonophysics (2001) 342:389–405.[CrossRef][Web of Science]
Rutter EH, Neumann DHK. Experimental deformation of partially molten Westerly granite under fluid-absent conditions, with implications for the extraction of granitic magmas. Journal of Geophysical Research (1995) 100:15697–15715.[CrossRef]
Simkin T. Flow differentiation in the picritic sills of North Skye. In: Ultramafic and Related Rocks—Wyllie PJ, ed. (1967) New York: John Wiley. 64–69.
Simon AC, Marsh B. Pyroxene thermometry in the OPX Tongue of the Basement Sill, McMurdo Dry Valleys, Antarctica. EOS Transactions, American Geophysical Union (2005) 86(52). Fall Meeting Supplement, Abstract V13H-08.
Stewart BW, DePaolo DJ. Isotopic studies of processes in mafic magma chambers; The Skaergaard Intrusion, East Greenland. Contributions to Mineralogy and Petrology (1990) 104:125–141.[CrossRef][Web of Science]
Taylor HP. The effects of assimilation of country rocks by magmas on 18O/16O and 87Sr/86Sr systematics in igneous rocks. Earth and Planetary Science Letters (1980) 47:2483–2499.
Turcotte DL, Schubert G. Geodynamics Applications of Continuum Physics to Geological Problems (1982) New York: John Wiley.
Turnbull IM, Allibone AH, Forsyth PJ, Heron DW. Geology of the Bull Pass–St. Johns Range Area, Southern Victoria Land, Antarctica. (1994) Lower Hutt, New Zealand: Institute of Geological & Nuclear Sciences Limited. Institute of Geological & Nuclear Sciences Geological Map 14.
Tuttle OF, Bowen NL. Origin of granite in the light of experimental studies in the system NaAlSi3O8–KAlSi3O8–SiO2–H2O. Geological Society of America Bulletin (1958) 104:1–153.
von Platen H. Experimental anatexis and genesis of migmatites. In: Controls of Metamorphism; a Symposium—Pitcher WS, Flinn GW, eds. (1965) Edinburgh: Oliver & Boyd. 203–216.
Wen S, Nekvasil H. SOLVCALC: an interactive graphics program for calculating the ternary feldspar solvus and for two-feldspar geothermometry. Computers and Geosciences (1994) 20:1025–1040.[CrossRef]
Whitney JA. The origin of granite: The role and source of water in the evolution of granitic magmas. Geological Society of America Bulletin (1988) 100:1886–1897.
Zeng L, Asimow PD, Saleeby JB. Coupling anatectic reactions and dissolution of accessory phases and the Sr and Nd isotope systematics of anatectic melts from a metasedimentary source. Geochimica et Cosmochimica Acta (2005) 69:3671–3682.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||