| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 43 | Number 3 | Pages 511-534 | 2002
© Oxford University Press 2002
The Aureole of the Traigh Bhàn na Sgùrra Sill, Isle of Mull: Reaction-Driven Micro-cracking During Pyrometamorphism
1DEPARTMENT OF EARTH SCIENCES, UNIVERSITY OF CAMBRIDGE, DOWNING STREET, CAMBRIDGE CB2 3EQ, UK
2DEPARTMENT OF GEOLOGY AND PETROLEUM GEOLOGY, KING’S COLLEGE, UNIVERSITY OF ABERDEEN, ABERDEEN AB24 3UE, UK
Received March 19, 2001; Revised typescript accepted September 20, 2001
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONThe intrusion of a Tertiary gabbroic sill of 6 m thickness, in which magma flow was locally turbulent, into garnet-grade pelitic gneiss and psammite at 600 bars at Traigh Bhàn na Sgùrra on the Ross of Mull, Scotland, led to the development of a contact aureole of 3 m width. Observed reactions include muscovite breakdown, melting on quartz–feldspar grain boundaries, and isochemical breakdown of biotite and garnet. A simple, two-stage thermal model fitted to the profile of maximum temperature is consistent with a 5 month period of turbulent magma flow. Five generations of micro-cracks occur in the psammite. The oldest pre-dates contact metamorphism and is marked by sub-parallel fluid inclusion arrays. The next resulted from anisotropic thermal expansion caused by magma intrusion. Internally generated stresses related to an increase in volume associated with muscovite breakdown formed two further sets of cracks associated with release of H2O and melting. Melting on quartz–feldspar grain boundaries also resulted in crack formation. The final stage of cracking was the result of anisotropic thermal contraction. Despite the high crack density at the metamorphic peak, little or no melt segregation occurred, demonstrating that micro-cracking alone is not sufficient (at least on this time scale) for melt segregation in static anatectic environments.
KEY WORDS: contact metamorphism; Isle of Mull; melting; micro-cracking
|
INTRODUCTION |
|---|
The geochemical evolution of the Earth’s crust is primarily a consequence of melt migration. Melt migration itself is dependent on the ease with which partial melts can segregate from their source and this process remains an important subject of research. Complete understanding of the segregation process is dependent on constraining both the distribution of melt in the source region and the mechanisms by which it can move at low volume percent (<10%).
Experimental determination of equilibrium melt geometries under hydrostatic conditions in crustal materials demonstrates that melt-filled pores should form an interconnected network (e.g. Jurewicz & Watson, 1985
; Wolf & Wyllie, 1991; Laporte, 1994). Such a network should, in principle, permit the segregation of all but a very small proportion of melt from the source region (McKenzie, 1984). However, the high viscosity of crustal melts is believed to require deformation in a deviatoric stress field to ‘squeeze’ out the melts before they solidify (e.g. Sawyer, 1991; Brown, 1994; Rutter & Neumann, 1995). Recently it has been pointed out that it is probable that pore topologies controlled entirely by textural equilibrium will only very rarely be attained in fluid-producing environments. This is because the rate of reaction and fluid production is generally greater than that of textural adjustment driven by minimization of surface energies (Holness & Siklos, 2000). In such a case the distribution of melts during the early stages of anatexis will be controlled entirely by reaction kinetics. Melt will be concentrated at the sites of formation such as grain boundaries (e.g. Mehnert et al., 1973; Wolf and Wyllie, 1991; Holness, 1999; Holness & Clemens, 1999; Cesare, 2000) or occupy fractures formed by internally generated overpressure (e.g. Connolly et al., 1997; Holness & Siklos, 2000; Watt et al., 2000; Rushmer, 2001). The subsequent evolution of the permeability of the partially melted rock will be a function of the (time-dependent) balance between reaction-induced melt distributions and those controlled by textural equilibrium and compaction.
In this study we present evidence that intrusion of a Tertiary gabbro sill into pelitic and psammitic gneiss resulted in the formation of a pervasive array of melt-filled micro-fractures in the country rocks as a result of a combination of cracking caused by anisotropic thermal expansion and positive volume changes during melt production. The short duration of the metamorphic event resulted in textures that we believe are essentially unmodified by subsequent textural adjustment, permitting an insight into the earliest stages of crustal anatexis.
|
GEOLOGICAL SETTING |
|---|
A swarm of Tertiary inclined sheet-like intrusions is found in the region of Loch Scridain on SW Mull, an island off the west coast of Scotland (Fig. 1). They range in composition from rhyolite to gabbro and were emplaced into Precambrian Moine schists, Mesozoic metasedimentary rocks and Palaeocene lavas (Preston et al., 1998
). The basaltic sheets commonly have glassy chilled margins and a minor but significant proportion of these sheets contains numerous partially fused metasedimentary and ultrabasic or basic xenoliths. Locally, the country rocks to the sills have been partially melted.
|
This study concentrates on Traigh Bhàn na Sgùrra, on the south coast of the Ross of Mull, where several of the Loch Scridain swarm of sheets are emplaced into steeply dipping or vertical Moine metasedimentary rocks. The largest sheet in the area varies from 1·5 to 6 m in thickness and contains abundant xenoliths of country rock and a few of coarse-grained mafic material. The margins of this sheet have already been the subject of a study, which produced convincing field evidence that flow within the sheet was locally turbulent for some significant part of its lifetime (Kille et al., 1986). This result has been corroborated by a study of resetting of K–Ar dates by Ar diffusion in muscovite from the pelites of the wall rocks, which suggested that the localized turbulent flow of basaltic magma continued for a period of 5 months (Wartho et al., 2001).
The large sheet, which we will refer to as the Traigh Bhàn na Sgùrra sill, is exposed in two small bays and also on the headland between them (Fig. 1). The contacts with the almost vertically dipping Moine metasediments are subplanar and dip gently to the NW (Fig. 2). Over parts of the headland, the sill has chilled margins but in the study area no chills are evident. Here the sill reaches its maximum thickness and contains abundant metasedimentary xenoliths. These are mostly concentrated into apparently clast-supported lenticular patches, some 10 m in diameter and up to 2 m thick, at the top surface where it is slightly updomed (Fig. 2). Kille et al. (1986) suggested that these layers may have formed by the xenoliths floating to the top of the sill during terminally waning flow conditions. Gabbroic xenoliths occur near the base, with the difference in location caused presumably by contrasts in rock density (Preston & Bell, 1997). The metasedimentary xenolith population is dominated by psammites, despite the predominant pelitic lithology in the immediate vicinity, suggesting either that significant lateral movement of the xenoliths has occurred or that the pelites are preferentially assimilated by the magma.
|
|
SAMPLING AND METHODS |
|---|
Samples were collected from the metasediments underlying the sill, both from the pelitic gneiss and also from a psammitic horizon of 1 m thickness (Fig. 2). The sampling was undertaken within a narrow traverse located at GR42347 18750, where the thickness of the sill is
6 m and no chilled margins are evident. Here the aureole of the sill reaches its maximum extent (clearly evident in the field from the loss of lustre of the muscovite mica in the pelite; see discussion below). Nearby, the aureole is much narrower and is associated with a chilled margin to the sill—evidence of localization of turbulent flow. For the purposes of the study described in this paper, the sampling of the aureole was confined to an area where the flow in the sill was certainly turbulent (i.e. no chilled margin and a wide aureole).
Previous work (Wartho et al., 2001) showed that the thermal effects in this region were similar on both contacts of the sill, so closely spaced sampling was undertaken on the more easily accessible lower contact. The distance from the basal contact was measured to the nearest few centimetres for the pelitic samples. The psammite bed protrudes some 30 cm into the basal portion of the sill. This is a general feature of the psammitic horizons in the Traigh Bhàn na Sgùrra area and is ascribed to the more refractory nature of the psammite compared with the pelite (Kille et al., 1986). The distance to the contact for the psammite samples was taken to be the distance to the planar contact adjacent to the pelite.
For comparison, additional samples of Moine psammite were collected from nearby localities far from any thermal effects of the Tertiary igneous activity. These were either within 1 km of the Traigh Bhàn na Sgùrra sampling locality, or from Inch Kenneth (in Loch na Keal, to the north of the Ross of Mull).
Polished chips of psammite coated with gold were used for cathodoluminescence (CL) imaging and images were collected using a KE detector fitted to a Phillips XL30 scanning electron microscope at Curtin University of Technology, Perth, Western Australia. The majority of the CL images were collected at a working distance of 17·5 mm from the pole-piece. This equates to a sample–CL detector separation of 0·5 mm. Small working distances and relatively large beam currents (3–5 nA) allowed collection of CL images from even weakly luminescing quartz. Monte Carlo modelling of the interaction zone between a 15 kV primary beam (as used in the majority of images presented here) and silica predicts a 3 µm maximum spatial resolution. Backscattered electron (BSE) images were collected on the same machine, at a shorter working distance (11·5 mm) and with the CL detector removed. The images obtained show a variation in luminescence intensity that is apparent as various shades of grey. Black indicates no (or very weak) luminescence, whereas white indicates very strong luminescence.
The causes of luminescence in quartz are not well understood and result from a complex interplay of factors such as crystal defects (such as lattice defects, trapped electron hole pairs or oxygen vacancies), and minor chemical impurities such as Al, Ti, Na and OH (Marshall, 1988; Waychunas, 1988). The particular mechanisms responsible for the luminescent features described in this study are not known. Quartz precipitated at low T (100–200°C) is typically characterized by low luminescence (D’Lemos et al., 1997; Sullivan et al., 1997). This may be because growth at low temperatures occurs slowly and so the defect density is generally lower than that in quartz grown rapidly at higher temperatures (Watt et al., 2000). High-temperature quartz tends to luminesce strongly (D’Lemos et al., 1997; Watt et al., 2000; Holness & Watt, 2002).
Mineral compositions were obtained using a Cameca SX-50 electron microprobe in the Department of Earth Sciences, University of Cambridge. Spectra were recorded with an energy-dispersive (ED) Si(Li) detector system manufactured by Link Analytical Ltd, and processed with the ZAF-4FLS software provided. An accelerating voltage of 20 kV was used, with beam currents of 3 nA. A spot size of 2–3 µm was used for analyses of micas, spinel and garnet, and a defocused beam of 30 µm was used to obtain compositions of feldspars. Detection limits (1
) for a typical silicate analysis were 1 wt % relative for each element.
|
THE PROTOLITH |
|---|
The metasediments at Traigh Bhàn na Sgùrra belong to the Lagan Mor Formation and the overlying Scoor Pelitic Gneiss of the Assapol Group (Holdsworth et al., 1987
). The Scoor Pelitic Gneiss is a coarse-grained garnetiferous pelitic gneiss with numerous discontinuous boudinaged bands of semi-pelite, which locally contain thin calc-silicate horizons. The Lagan Mor Formation is composed of quartzose psammite and pelite, which are interbanded on a metre scale. A strong fabric is defined by a micaceous foliation, which wraps around lenticular quartzo-feldspathic segregations and small garnets in the pelite horizons, and is at a small angle to the fabric defined by compositional layering. The metasediments were originally regionally metamorphosed at 7 kbar and 550 ± 25°C during the Caledonian orogeny (Brearley, 1984). The samples collected as part of this study comprise pelitic material and psammite from a single elongate interbedded boudin.
The pelitic samples contain biotite and muscovite mica, almandine-rich garnets (see Table 1 for representative compositions), plagioclase feldspar (An35–55) and quartz, with abundant accessory apatite. The psammitic horizon contains up to 20 vol. % feldspar, of which the great majority is turbid and untwinned oligoclase (typically Ab78–73, with a minor orthoclase component). Some grains of intermediate alkali feldspar are present. Taking into account the averaging process inherent in using a defocused beam (which averages out fine-scale exsolution), the feldspar population as a whole (Fig. 3) is consistent with recrystallization at high temperatures (Smith & Brown, 1988).
|
|
Up to 10 vol. % paragonitic muscovite occurs in the psammite (activity of muscovite 0·66, activity of paragonite 0·35; see Table 1 for composition), and the grains are aligned parallel to the foliation in the adjacent pelitic horizons. Small quantities (<1 vol. %) of biotite mica and almandine-rich garnet (Table 1) are sometimes present. The quartz grains have a prominent undulose extinction, with abundant sub-grain development, and highly irregular grain boundaries. Accessory phases include rounded (detrital?) grains of apatite, sphene and epidote.
All samples collected from Traigh Bhàn na Sgùrra, including those unaffected by Tertiary intrusions, contain closely spaced arrays of sub-parallel fractures at a high angle to the foliation. The fractures visible in thin section are marked by trails of small fluid inclusions (Fig. 4a). In CL the fractures cutting quartz grains in the psammite are revealed to be filled with weakly luminescing quartz interpreted to have crystallized at low temperature (Fig. 4b). The garnets in the pelite are abundantly cracked, with closely spaced sub-parallel arrays of cracks at a high angle to the foliation (in the same orientation as the cracks in the rest of the rock; Fig. 4c). The cracks are now filled with chlorite, epidote and quartz.
|
–ß transition. Field of view 1·25 mm. (f) ROM99-3 (pelite, 340 cm from the contact). Plane-polarized light. A muscovite grain surrounded by biotite grains. The clouding of the muscovite grains by replacement by biotite + mullite + K-feldspar ± spinel ± corundum should be noted. The reaction starts at the edges of the muscovite grains and moves most rapidly along cleavage planes. Field of view 0·65 mm.In both the psammite and the pelite, some of the muscovite and biotite grains in contact with feldspar have been kinked and bent by the growth of lens-shaped regions of clear albite (Fig. 4d). This effect is currently under investigation (Holness & Parsons, in preparation).
|
TEXTURAL DEVELOPMENT DURING CONTACT METAMORPHISM |
|---|
Psammite
In handspecimen the psammite of the aureole appears unchanged, except close to the contact, where a fine grain-scale network of whitish opaque veins is visible, separating individual quartz grains (e.g. Holness & Clemens, 1999
; Sawyer, 1999). Optical observation of thin sections reveals that the onset of metamorphism is manifest by the development of a network of irregular open cracks cutting across individual quartz and feldspar grains (Fig. 4e). These cracks are easily distinguishable optically from the original crack system, which is marked by sub-parallel fluid-inclusion trails. The distance from the sill at which this occurs is not well constrained because of lack of outcrop (see Fig. 2) but occurs between 8 and 3·5 m from the contact.
Some 300 cm from the base of the sill muscovite has broken down, resulting in a clouding of the muscovite grains associated with the development of a weak brown/clear pleochroism, and the growth of a thin (1–2 µm) clear feldspar rim (Fig. 4f). This reaction initiates along cleavage planes at the ends of the muscovite laths and is first developed where the muscovite grain is in direct contact with alkali feldspar. Detailed transmission electron microscopy work by Brearley (1986) on a xenolith of Moine pelitic gneiss enclosed within a nearby sill showed that this results from a complex replacement reaction involving the growth of sub-microscopic plates of biotite associated with potassium feldspar, spinel, ± corundum ± mullite. The (001) planes of the biotite reaction product are parallel to those of the precursor muscovite, whereas the mullite needles form radiating sprays with no preferred orientation within the muscovite basal plane.
An important and novel result of this study is the observation that the muscovite-breakdown reaction results in the formation of two sets of micro-cracks. The first set comprises abundant narrow cracks visible in both BSE and CL images as bright zones. From the lack of contrast between these cracks and the feldspar rims on the reacted muscovites, we believe the cracks to be filled with a feldspar-rich material. Many of these cracks appear to reuse the healed fracture array inherited from the protolith (Fig. 5a). The mobility of the material in the cracks is also clear from its presence on quartz–feldspar interfaces adjacent to reacting muscovites (Fig. 5b), although it is plausible that some of this material formed in situ by melting at the quartz–feldspar interface. We believe the material in these cracks to be a solidified melt.
|
Rarely, a second set of microfractures is visible in CL (but not in BSE) images, radiating from the sites of reacted muscovite grains. These cracks are thinner, are clearly associated with fluid inclusion arrays (Fig. 5c and d) and represent the partially healed pathways for an aqueous fluid phase. We believe these to have formed as a consequence of release of H2O formed by breakdown of the muscovite. Their relative scarcity is perhaps due to the ease with which H2O would dissolve in a nearby H2O-undersaturated melt. It is noteworthy that the H2O-filled fractures have healed, at least in part, whereas the melt-filled fractures remain as parallel-sided features with no sign of even incipient healing to form arrays of melt inclusions (see Cesare et al., 1997).
Between 200 and 175 cm from the contact with the sill the feldspar grains develop clear cuspate margins where they are adjacent to quartz and highly elongate clear apophyses of feldspar extend from individual grains along quartz–quartz grain boundaries (Fig. 6a and b). The quartz grain boundaries are highly irregular and the feldspar extensions along the grain boundaries follow these irregularities. Cuspate extensions also protrude into quartz grains, and where this occurs they commonly end in a fine crack, sometimes marked by a fluid inclusion trail (Fig. 6a and b). The cuspate extensions and the outer few microns of the feldspar grains generally have a slightly different birefringence from their host grain but the same extinction position under crossed polars. These features are noticeably more marked in the vicinity of clusters of reacted muscovite grains, but even isolated feldspar grains enclosed within single grains of quartz may develop extensions related to micro-cracks.
|
At 137 cm from the base of the sill the reacted muscovite grains are replaced by an extremely fine-grained, turbid, non-pleochroic aggregate and, where they are adjacent to quartz, are surrounded by a thick rim (10 µm) of polycrystalline alkali feldspar (with a range of compositions between rims from Ab37Or61An2 to Ab56Or38An6). These rims have a marked cuspate outer margin where they are in contact with quartz (Fig. 6c). The cuspate appearance is due to the development of euhedral terminations to small projections of quartz, optically continuous with the bulk of the adjacent quartz grain, and this feature is seen from here all the way up to the contact. Closer than 137 cm to the sill the grain size of the mullite breakdown product becomes large enough to discern its high birefringence and pale lilac pleochroism. The mullite needles, which are commonly bent, attain their largest size in the feldspar rim to the reacted muscovite, sometimes protruding into the adjoining quartz (but no further than the maximum dimension of the euhedral terminations).
At 80 cm from the base of the sill the sites of the reacted muscovite grains are marked by smoothly rounded lozenges containing prominent clusters of (partially or wholly pinitized) euhedral cordierite grains apparently nucleated on the central part of the pseudomorph (Fig. 6d). The lozenges also contain fine-scale radiating granophyric intergrowths, which are rarely associated with small grains of biotite. Cracks similar to those observed to radiate from reacted muscovite grains in lower-grade samples are now very prominent, with wide bases in which granophyric intergrowths are clearly evident.
In all samples nearer the sill, the feldspar grains, including those entirely enclosed by single quartz grains, are surrounded by a granophyric rim of 40 µm thickness (Fig. 6e). Small grains of feldspar are completely replaced by granophyre. Granophyre-filled cracks emanate from many of the feldspar rims in some samples (Fig. 6f), although in several samples the cracks appear to be more closely associated with the sites of reacted muscovite grains (Fig. 7a).
|
The next sample towards the sill, collected from a distance of 74 cm, contains polycrystalline quartz paramorphs after tridymite in the granophyric intergrowths (Fig. 7b). The sample collected from the rounded protuberance of the psammite bed into the sill itself contains many granophyre-filled cracks, both inter- and intra-granular, many of which have thick rims of euhedral quartz containing abundant inclusions, possibly of glass (Fig. 7c and d).
The biotite grains in the contact metamorphosed psammite are generally replaced by an Fe-stained, fine-grained chlorite in association with ilmenite and K-feldspar. Ilmenite forms clusters of oriented plates at the margins of the biotite grains, first visible 263 cm from the contact. These plates have three orientations, resulting in the formation of a lattice. This reaction is not stable at any pressure with the observed compositions of coexisting chlorite and biotite (using THERMOCALC, Powell & Holland, 1988; Holland & Powell, 1990) and is probably a retrogressive reaction associated with the movement of aqueous fluids through the cooling rock mass. An association of chloritization with an increase in turbidity of the feldspars supports this. Retrogressive, brown-stained chlorite is very common throughout the rocks of the aureole less than 3 m from the contact, surrounding the rare garnet grains, and occurring in cracks that cut across the granophyric areas.
Pelite
The onset of metamorphism in the pelitic gneiss is clearly evident in the field as the appearance of a dull dusty pink colour to the edges of the otherwise clear and reflective muscovite grains. This dull coloration, which is first visible in hand specimen 320 cm from the basal contact in the sampling traverse (and at 340 cm in thin section), gradually extends into the body of the grains, completely replacing them by 300 cm from the base of the sill. This is manifest in thin section by the appearance of a pale brown turbid alteration, which is first seen at the edges of muscovite grains and moves along the cleavage planes with increasing grade, gradually replacing them by a fine-grained turbid aggregate (e.g. Fig. 4f). The associated thin radiating cracks filled with solidified melt so evident in the psammites are also present in the pelites but are not so noticeable in thin section because of the common presence of nearby biotite. This means that the muscovite grains are typically not in direct contact with quartz or feldspar, which are the phases that crack. However, the pelites are abundantly cracked, with a network of open, unhealed, cracks within individual quartz and feldspar grains, which clearly post-dates the metamorphic peak.
Biotite grains in the pelitic gneiss are generally altered, with the replacement of biotite by chlorite generally increasing in intensity towards the sill, although it is rare for the replacement to be complete.
Between 120 and 100 cm from the basal contact thick (50 µm), often heavily chloritized, granophyric rims form on the grain boundaries between quartz and feldspar (Fig. 8a). The quartz grains bounding these rims are invariably smooth, whereas the feldspar grains have euhedral overgrowths in optical continuity, suggesting that the melt composition was in the feldspar primary phase volume at the peak temperature. Such blocky margins to former melt pools are a common feature of solidified melt (e.g. Jurewicz & Watson, 1985; Wolf & Wyllie, 1991; Sawyer, 2001).
|
The biotite reacts to form a new strongly pleochroic fox-red biotite as small, lozenge-shaped grains aligned parallel to the lattice of the old biotite grains (Fig. 8b). The new biotite grains are set in a poikilitic matrix of K-feldspar with numerous very small grains of hercynitic spinel [Fe/(Fe + Mg) = 0·63–0·68] and magnetite. Brearley (1987), working on a xenolith from a nearby sill, used TEM to establish that the reaction is associated with the breakdown of the Fe–Al biotite to a more Mg-rich biotite.
In the sample located 90 cm from the basal contact, the sites of the reacted muscovite grains have developed a thick granophyric rim where they are in contact with quartz.
The garnet porphyroblasts in the pelite outside the aureole are slightly chloritized on their margins and along internal cracks. As the contact is approached the amount of chloritization increases slightly. Between 60 and 70 cm from the basal contact the garnets are observed to have reacted to form an aggregate of euhedral hercynitic spinel [Fe/(Fe + Mg) = 0·78] and pyroxene (identifiable only from their shape, as they are completely replaced by chlorite) poikilitically enclosed in large plagioclase (Ab46An50Or4) and quartz grains. Garnet is completely replaced at 35 cm from the contact. The presence of plagioclase within the pseudomorphs suggests that this reaction occurred in the presence of melt, as plagioclase is not one of the reaction products. Instead, the feldspar crystallized from a ubiquitous melt phase, enclosing the solid products of the garnet reaction.
From the point at which garnet broke down, the sites of the reacted muscovite grains, which previously were turbid fine-grained aggregates comprising radiating sheaves of mullite needles set in K-feldspar, begin to be replaced. The radiating sheaves of mullite needles are pseudomorphed by fine-grained spinel aggregates (Fig. 8c and d). This replacement occurs at the margins of the muscovite pseudomorphs and is essentially complete 35 cm from the contact. The reaction by which this occurs is not clear because of the extremely fine grain size.
A sudden increase in the amount of melt (discernible as fine-grained aggregates of euhedral feldspar crystals set in a granophyric matrix, Fig. 8e) occurs at 57 cm from the contact. The feldspars are either plagioclase (An35–40 with an orthoclase component of up to 5 mol %) or an intermediate alkali feldspar (Or60–65 with an anorthite component of up to 5 mol %). The original plagioclase grains in the pelite have a sieve-like appearance as a result of melting within the grains, with euhedral overgrowths of an intermediate alkali feldspar (Or60–65), which most probably grew during solidification of the melt phase. The increase in the amount of melt is related to the final replacement of biotite by elongate aggregates of fine spinel grains, aligned parallel to the (001) cleavage plane of the reacted mica, enclosed by grains of pinitized cordierite (recognizable from the hexagonal prism outline), plagioclase (Ab56) and K-feldspar (Fig. 8f). The spinel is closely associated with fine ilmenite plates and grains of magnetite.
At 10 cm from the contact, the trails of spinel lose their coherence, presumably as a result of minor movement of the melt phase, which now occupies 70 vol. %. A few rounded polycrystalline restitic aggregates of quartz still remain. Quartz paramorphs after tridymite are very rare in the partially melted pelites.
|
INTERPRETATION OF REACTION TEXTURES |
|---|
Microprobe analysis of individual grains of muscovite and plagioclase feldspar (Wartho et al., 2001
) and biotite and garnet in the pelites (this study) revealed no detectable intra-grain compositional zoning. These observations are consistent with very limited diffusion distances and lack of chemical communication during the short-lived heating event.
Muscovite breakdown
The onset of metamorphism is marked by the breakdown of muscovite to a fine-grained biotite- and mullite-bearing aggregate, similar to the reaction described by Brearley (1986). A previously undocumented characteristic of this reaction is the development of both an alkali-feldspar rim to the site of the reacted muscovite grain and cracks radiating from the sites of reaction, the majority of which are filled at the muscovite end by what is possibly feldspar.
The cracking associated with muscovite breakdown closely resembles that produced by fluid-absent melting of muscovite within whole-rock samples of a muscovite-bearing quartzite at 8 kbar by Connolly et al. (1997) (Fig. 9). We believe that this textural similarity and the mobility of the material in the cracks (Fig. 5) demonstrate that the material in the cracks is also solidified melt. At higher grades the cracks are filled with granophyric intergrowths, which crystallized from a melt phase (e.g. Figs 6f and 7a).
|
At pressures lower than 5 kbar, muscovite breaks down in the presence of quartz to form K-feldspar + Al2SiO5 + H2O ± biotite (Fig. 10). In contrast, melt production associated with muscovite breakdown in the Traigh Bhàn na Sgùrra rocks suggests that the reaction described here is a metastable one, directly analogous to that achieved during experimental melting of muscovite–quartz rocks at low pressures (Brearley & Rubie, 1990). A possible form of the reaction is
|
 |
found that experimental overstepping of the (sub-solidus) muscovite breakdown reaction (by 180°C in their experiments) at 1 kbar (Fig. 10) resulted in an outer rim of melt forming around the muscovite grains, which were replaced by aggregates of mullite, biotite, K-feldspar, spinel and corundum. These textures were achieved in run times ranging from 2 to 5 months—a similar duration to that inferred for the metamorphism of the walls of the Traigh Bhàn na Sgùrra sill (Wartho et al., 2001). The feldspar rims we describe resemble the experimentally produced melt rims (Brearley & Rubie, 1990). We suggest that they formed by solidification of a melt phase, in agreement with Brearley (1986), who, on the basis of textural evidence from pelitic xenoliths from a nearby sill, suggested the former presence of melt in reacted muscovite grains.
Evidence that melt pools did surround the reacted muscovite grains, at least in the higher-grade psammites, is provided by the irregular outer margins to their feldspar rims (Fig. 6c). The irregularities are a consequence of the development of many small protrusions with euhedral terminations on the enclosing quartz grain. Large mullite crystals occur both within the feldspar rims surrounding the reacted muscovite grains and, importantly, within the euhedral protrusions growing in optical continuity on the surrounding quartz. This can only have occurred if the marginal quartz crystallized after the growth of the mullite crystals. The most likely explanation is that the marginal quartz crystallized from a melt phase, consistent with the growth of euhedral terminations, and also indicating that quartz must have participated in the melting reaction. In common with other examples of similar textures formed during solidification (e.g. Platten, 1981; Jurewicz & Watson, 1985; Wolf & Wyllie, 1991; Holness & Clemens, 1999; Holness & Watt, 2001; Sawyer, 2001) the early crystallizing quartz grew as overgrowths on pre-existing grains. That quartz participated in the melting reaction in the pelites is also suggested by the presence of wide feldspar rims to reacted muscovite grains only where they are in contact with quartz. A suggested reaction is
 |
) and it is the consequent overpressuring that was responsible for the associated grain-scale fracturing (Connolly et al., 1997; Rushmer, 2002). The balanced reaction given by Brearley (1986) for isochemical muscovite breakdown in a xenolith from a nearby sill yields a volume increase of 5%, similar to that of 2·1% calculated by Connolly et al. (1997). In higher-grade psammite, the muscovite-out reaction was no longer isochemical and involved the adjacent quartz. Rushmer (2002) calculated a volume increase of 6·8% for a reaction similar to that inferred for the aureole:
 |
Melting on quartz–feldspar interphase boundaries
Two separate reactions occurring at grain boundaries between quartz and feldspar grains can be distinguished on textural grounds. The lower-temperature reaction resulted in the formation of a thin, clear, optically continuous, rim to the feldspar grain and the formation of elongate apophyses on the feldspar grain along quartz grain boundaries and within cracks in quartz grains. The higher-temperature reaction resulted in the formation of thick granophyric layers on all quartz–feldspar grain boundaries.
The elongate apophyses of feldspar related to the first reaction are clearly evident only in psammite samples, although some examples are seen in the pelites. This difference is due perhaps to the greater number of quartz–feldspar grain boundaries in the psammite. This texture has already been ascribed to melting where it is observed in other contact aureoles (e.g. Platten, 1981; Holness & Clemens, 1999; Clemens & Holness, 2000; Rosenberg & Riller, 2000). The observation that the apophyses in individual quartz grains commonly are the base for a micro-crack with associated fluid inclusions suggests that they are related to fluid-present melting. They could have formed either as the H2O required for melting was transported to the melt site, or as the H2O-bearing melt reached saturation during solidification. The source of the fluid is unclear but a plausible source is the reacting muscovite.
The higher-temperature reaction results in a texture highly reminiscent both of those produced during experimental melting of arkosic rocks (Mehnert et al., 1973) and of those observed in partially melted rocks in small contact aureoles (e.g. Holness, 1999). The thickness of the rims is approximately constant at 50 µm in any thin section (taking into account the expected variation in a random 2-D section) and varies little with distance to the contact. An important feature of this melting reaction is that it appears abruptly 100 cm from the contact, and occurs on all quartz–feldspar boundaries, including those isolated from the grain boundary network (i.e. associated with apparently isolated inclusions of one phase in single grains of the other). It is also associated with an increase in the number of melt-filled micro-cracks, suggesting a positive volume change on reaction.
The presence of two clearly distinguishable melting reactions in the psammite, both of which involve only quartz and feldspar as solid reactants, is notable. We suggest that the lower-temperature one occurs in the presence of H2O, and that the higher-temperature reaction is melting under H2O-absent conditions. The lower-temperature reaction produced only minor amounts of melt (<1 vol. % inferred from the development of the feldspar apophysis + fluid inclusion texture) concentrated in the vicinity of muscovite grains, which are inferred to have been the source of the limited amounts of H2O required for reaction. In contrast, the higher-temperature reaction was ubiquitous and produced a much greater melt volume (20 vol. %).
Other reactions inferred to have occurred in the pelites of the aureole appear to involve isochemical (or nearly so) breakdown of biotite and garnet. The biotite breakdown reaction has been studied previously by Brearley (1987), who found small pockets of melt within reacted biotite grains. It is probable that biotite breakdown in the Traigh Bhàn na Sgùrra rocks was also accompanied by melting.
The garnet breakdown reaction cannot be constrained because of the complete replacement of one of the reaction products (which we suspect to be pyroxene).
|
CONTROL OF MELTING BY H2O AVAILABILITY AND REACTION KINETICS |
|---|
If the psammite in the aureole melted under conditions close to chemical equilibrium (i.e. if the melting reaction were controlled by the rate of heat supply) then the existence of restitic cores to feldspar grains would constrain maximum temperatures to those of the cotectic in the Qtz–Ab–An–Or ± H2O system. The survival of feldspar in the psammite almost up to the contact with the sill points to a significant factor limiting completion of melting, given the small temperature range (a few tens of degrees) of the cotectic in systems with a granitic composition. This suggests that either reaction kinetics or the availability of reactants (i.e. H2O) was important in controlling the extent of reaction.
The availability of H2O was clearly an important control on the progress of the lower-temperature (H2O-present) melting reaction. The small amounts of melt present in the psammite inferred from textural criteria would then signify small amounts of H2O, rather than low maximum temperatures.
The muscovite-out reaction generally occurred at a lower temperature than the onset of melting on quartz–feldspar grain boundaries. Were the muscovite the source of the H2O, then either the muscovite persisted metastably to temperatures close to those of melting (e.g. Brearley, 1986), or the H2O released by muscovite breakdown remained in the rock until ambient temperatures reached those of the Qtz–Ab–An–Or–H2O melting reaction. Were the former case to have held, one would expect the amount of melting in the rock to reflect the amount of muscovite. The amounts of muscovite in the psammite protolith determined by point-counting are consistent with a bulk H2O content of 0·02–0·15 wt %. Assuming saturation of a granitic melt with 1 wt % H2O (Johannes & Holtz, 1996), these amounts of H2O can potentially produce far more melt than is observed. This suggests that most of the H2O released by muscovite breakdown either escaped along fractures or dissolved in an early melt phase before the melting point in the Qtz–Or–Ab–An–H2O system was reached.
The thickness of the granophyric rims inferred to have formed during H2O-absent melting is most probably a function of reaction kinetics. Melting of dry granite occurs at a rate several orders of magnitude slower than the rate of melting of the same rock in the presence of H2O (Arzi, 1978a). Formation of the 50 µm melt rims surrounding restitic feldspar grains would take a few hours in the presence of H2O, but up to a year if dry (Arzi, 1978a). The maximum time during which the rocks within 80 cm of the contact were above the dry solidus was a few months (Wartho et al., 2001; this study, see below) consistent with insufficient time for complete reaction. The amount of restitic feldspar is thus related to the average grain size of the feldspars.
|
PRESSURE AND TEMPERATURE OF CONTACT METAMORPHISM |
|---|
The presence of tridymite paramorphs in the crystallized melt pools can be an indicator of very low pressures of metamorphism, as the requisite high temperatures to crystallize tridymite in its primary phase field are attained only by melts solidifying at low pressure. It is important to stress that tridymite is well known to crystallize far from equilibrium—it grows at surface temperatures in lavas and tuffs well after eruption. The assumption in the following sections that the tridymite crystallized in its primary phase field is supported by the similarity of the thermal profiles thus obtained with those obtained using Ar-diffusion data (Wartho et al., 2001
). An estimate of the pressure of metamorphism can be obtained using the locations in the psammite of the wet melt-in, dry melt-in and tridymite-in reactions in the Qtz–Ab–An–Or ± H2O system (assuming none of these three reactions was significantly overstepped, which is perhaps plausible given the high temperatures involved).
From the simplified phase diagram (Fig. 11), three pressure regions can be defined using the order of appearance of the three reactions. At pressures lower than 100 bars, the quartz–tridymite inversion occurs at temperatures below the solidus and so high quartz would not be seen on the liquidus of either of the melting reactions. At pressures greater than 500 bars, the first reaction to occur in the Qtz–Ab–An–Or ± H2O system would be wet melting, followed by dry melting. Solidification of all partially melted rocks would result in crystallization of quartz, except for rocks that had melted to a great extent. These would crystallize both tridymite and high quartz on cooling. At intermediate pressures the H2O-present melting reaction is followed by the inversion of high quartz to tridymite. Melts formed in the absence of H2O would then have tridymite on the liquidus rather than high quartz. The order of appearance of the three reactions in the psammites in the aureole of the Traigh Bhàn na Sgùrra sill is consistent with pressures greater than 500 bars.
|
The maximum possible pressure of contact metamorphism can be constrained from a consideration of the contact temperature determined using the composition of the gabbro in the sill to be 1130°C (Kille et al., 1986). This maximum possible temperature for the aureole constrains the pressure to no greater than 1·3 kbar in order to crystallize tridymite (within its primary phase field) from a silica-rich melt (Fig. 11).
The dry melting reaction in the system Qtz–Ab–An–Or is relatively insensitive to pressure variations in the region 0·5–1·3 kbar. When taken in conjunction with the pressure-sensitive quartz–tridymite inversion, the likely contact temperature and their location within the aureole, the dry melting reaction can be used to constrain further the pressure of metamorphism (Fig. 12). It is important to emphasize that we are assuming neither the quartz–tridymite inversion nor the dry quartz–feldspar melting reaction to be significantly overstepped. Plausible profiles of the maximum temperature in the proximal part of the aureole are obtained only for pressures of 600 bars.
|
At low pressures, melting in the Qtz–Ab–An–Or–H2O system is highly sensitive to changes in pressure. The range of melting temperatures for all feldspar compositions observed in the psammite at 600 bars is shown in Fig. 12. Melting occurs on a very localized scale in the aureole and thus the temperature at which it occurs is controlled by the composition of each individual feldspar grain. This variability is reflected in the relatively wide spatial distribution of the onset of melting—at the lower-temperature end of the error box in Fig. 12 (shaded), only a few feldspars show evidence of melting. At the higher-temperature end (white part of the box), the majority of grains show signs of melting.
Other reacti