| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 44 | Number 12 | Pages 2287-2312 | 2003
© Oxford University Press 2003; all rights reserved
Textural and Thermal History of Partial Melting in Tonalitic Wallrock at the Margin of a Basalt Dike, Wallowa Mountains, Oregon
DEPARTMENT OF GEOSCIENCES, OREGON STATE UNIVERSITY, CORVALLIS, OR 97331, USA
* Corresponding author. Telephone: 541-737-1201. Fax: 541-737-1200. E-mail: petcovih{at}geo.orst.edu
RECEIVED SEPTEMBER 15, 2002; ACCEPTED JUNE 18, 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONTHE NATURAL LABORATORYPROGRESSIVE STAGES OF MELTINGCOMPOSITIONAL CHARACTERISTICS OF...DISCUSSIONCONCLUSIONSAPPENDIX: THERMAL MODELREFERENCES |
|---|
Columbia River Basalt Group dikes invade biotite–hornblende tonalite to granodiorite rocks of the Wallowa Mountains. Most dikes are strongly quenched against wallrock, but rare dike segments have preserved zones of partial melt in adjacent wallrock and provide an opportunity to examine shallow crustal melting. At Maxwell Lake, the 4 m thick wallrock partial melt zone contains as much as 47 vol. % melt (glass plus quench crystals) around mineral reaction sites and along quartz–feldspar boundaries. Bulk compositional data indicate that melting took place under closed conditions (excepting volatiles). With progressive melting, hornblende, biotite, and orthoclase were consumed but plagioclase, quartz, and magnetite persisted in the restite. Clinopyroxene, orthopyroxene, plagioclase, and Fe–Ti oxides were produced during dehydration-melting reactions involving hornblende and biotite. Reacting phases became more heterogeneous with progressive melting; crystallizing phases were relatively homogeneous. Progressive melting produced an early clear glass, followed by light (high-K) and dark (high-Ca) brown glass domains overprinted by devitrification. Melts were metaluminous and granitic to granodioritic. Thermal modeling of the partial melt zone suggests that melting took place over a period of about 4 years. Thus, rare dikes with melted margins represent long-lived portions of the Columbia River Basalt dike system and may have sustained large flows.
KEY WORDS: Columbia River Basalt dike; crustal melting; dehydration-melting; tonalite–granodiorite; thermal model
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONTHE NATURAL LABORATORYPROGRESSIVE STAGES OF MELTINGCOMPOSITIONAL CHARACTERISTICS OF...DISCUSSIONCONCLUSIONSAPPENDIX: THERMAL MODELREFERENCES |
|---|
Although there is broad consensus that basalt injection can be fundamental in crustal melting, there are few places where stages of the interaction can be directly sampled. The Wallowa Mountains of northeastern Oregon, however, provide a natural laboratory in which to examine shallow crustal melting. In this area, hundreds of Columbia River Basalt Group (CRBG) feeder dikes cut granitoid rocks of the Wallowa Batholith. The batholith is a biotite- and hornblende-bearing tonalite to granodiorite. Although most dikes were strongly quenched against their wallrock, a few dikes have developed partially melted contact zones, commonly with up to 50 vol. % quenched melt in the wallrock at their margins. This melt is represented by devitrified silicic glass plus plagioclase, pyroxene, and magnetite quench crystals. We have examined the partially melted zone in tonalite at the margin of a CRBG (Grande Ronde) dike where quenched melt is preserved over a distance of 4 m from the dike margin and reaches 47 vol. % near the dike–wallrock contact.
Recent work has shown that dehydration-melting plays a crucial role in the generation of silicic melts in the crust. Dehydration-melting, also called fluid- or vapor-absent melting, is the incongruent reaction of a hydrous mineral assemblage to form melt plus residual minerals. Previous studies of crustal dehydration-melting considered protoliths with either amphibole or mica. Melting of protoliths with both hydrous phases was examined by Rutter & Wyllie (1988)
and Skjerlie & Johnston (1996) at high pressure (10 kbar). This study differs from previous work in that both hornblende and biotite are present in the Wallowa parent rock and melting was shallow.
Related studies of partial melting
Dehydration-melting of mafic to intermediate composition amphibolites at pressures <10 kbar (1000 MPa) have produced trondhjemitic, to tonalitic, to granodioritic melt coexisting with a restite of clinopyroxene + orthopyroxene + plagioclase ± quartz ± Fe–Ti oxides (Beard & Lofgren, 1991; Rapp et al., 1991; Rushmer, 1991; Wolf & Wyllie, 1994; Patiño Douce & Beard, 1995). Biotite dehydration-melting reactions at <10 kbar have produced granitic to granodioritic melt and a restite of orthopyroxene + plagioclase + Fe–Ti oxides (generally ilmenite and/or magnetite) ± alkali feldspar ± quartz from protoliths as diverse as biotite tonalites (Patiño Douce & Beard, 1995; Singh & Johannes, 1996a, 1996b), a biotite-bearing metagreywacke (Vielzeuf & Montel, 1994), pelites (Le Breton & Thompson, 1988; Vielzeuf & Holloway, 1988), and a high-F tonalitic gneiss (also containing 
2 wt % hornblende; Skjerlie & Johnston, 1992, 1993). At pressures around 10 kbar and greater, garnet was a crucial phase in the restite for nearly all experimental protoliths.
Partial melting experiments have been performed on protoliths containing both biotite and amphibole at 10 kbar. A garnet–biotite–hornblende tonalite yielded up to 40 vol. % melt (melt composition not given) with a restite of orthopyroxene + clinopyroxene + rutile + garnet (Rutter & Wyllie, 1988). A biotite–hornblende–epidote gneiss produced about 35 vol. % peraluminous, granodioritic to granitic melt and a restite of orthopyroxene + clinopyroxene + plagioclase + garnet + ferro-pargasitic amphibole (Skjerlie & Johnston, 1996).
Whereas work on piston-cylinder experiments has rarely reported coexisting, compositionally distinct melts, the occurrence of multiple melts is common in natural examples and rock core experiments. Shallow (<1 kbar) partial melting of Sierra Nevada biotite granite at the contact with a trachyandesite plug produced both brown and clear glass coexisting with relict plagioclase + sanidine + quartz, with magnetite, rutile, and Mg-cordierite replacing biotite (Al-Rawi & Carmichael, 1967; Kaczor et al., 1988; Tommasini & Davies, 1997). Brown and clear glass coexisting with a restite of plagioclase + quartz +orthoclase + Fe–Ti oxides + minor orthopyroxene were also noted by Green (1994) in partially melted biotite granodiorite xenoliths hosted in andesitic to dacitic dikes. Philpotts & Asher (1993) noted abundant disequilibrium melting textures, such as sieve-textured feldspar, in partially melted biotite gneiss at the contact of a basalt dike at paleodepth of about 10 km. Knesel & Davidson (1996) melted 2 cm cubes of biotite alkali granite under atmospheric conditions, yielding coexisting clear and brown melts plus a restite of quartz + plagioclase + Fe–Ti oxides ± alkali feldspar. Multiple composition melts attributed to muscovite and biotite (+ quartz + plagioclase) dehydration-melting reactions in pelite at 7 kbar were also noted by Rushmer (2001).
|
THE NATURAL LABORATORY |
|---|
|
TOPABSTRACTINTRODUCTIONTHE NATURAL LABORATORYPROGRESSIVE STAGES OF MELTINGCOMPOSITIONAL CHARACTERISTICS OF...DISCUSSIONCONCLUSIONSAPPENDIX: THERMAL MODELREFERENCES |
|---|
The Wallowa Batholith and CRBG dikes
The Wallowa Mountains are largely composed of the Wallowa Batholith, a series of Late Jurassic plutons (140–160 Ma; Armstrong et al., 1977
) intruded into island arc terranes accreted to the western margin of North America (Fig. 1). Dikes exposed in the batholith are part of the Chief Joseph dike swarm, which fed the Columbia River flood basalts (Taubeneck, 1970). Imnaha Basalt (17·3–17·0 Ma; Baksi, 1989) is preserved as erosional remnants on some peaks of the Wallowa Mountains and as dikes. Most dikes in the batholith are of Grande Ronde Basalt (16·9–15·6 Ma; Baksi, 1989). Paleodepth at the time of Grande Ronde dike emplacement was as great as 2·5 km, as estimated from a thickness of about 1 km of Imnaha flows unconformably overlying about 1·5 km of relief in the Wallowa Mountains.
|
Individual basalt dikes within the Wallowa Mountains can extend several kilometers along strike, are a few centimeters to 50 m thick (average 7–10 m), and are steeply dipping (average 70°). Overall, dikes strike N10°W, although strikes may vary from N55°W to N30°E within sub-swarms where dikes occur in zones of concentration of 7–12 dikes per km2 (Taubeneck & Duncan, 1997
; Petcovic et al., 2001). Dikes have one or more of the following morphologies: dikes with quenched margins and no interaction with the wallrock, dikes with partially melted wallrock at their margins, dikes that have eroded their wallrock, and dikes containing whole to disaggregated crustal xenoliths that constitute locally as much as 30% of the dike (Grunder & Taubeneck, 1997). The majority of Wallowa dikes have an aphanitic quench zone (a few centimeters to 20 cm thick) at their margins. Rarely, these dikes may have localized zones of partial melting that extend into wallrock for up to about 50 cm from the dike–wallrock contact. Typically, localized melting zones occur where dikes have eroded their own quenched margins so that coarse-grained basalt is directly in contact with the wallrock. Partial melting is most common in narrow wallrock screens or in wallrock trapped between two cross-cutting dikes.
Only a handful of dikes in the Wallowas exhibit extensive partial melting (i.e. zones of partial melting that are meters thick and continuous for tens to hundreds of meters). These dikes consistently lack an aphanitic quench zone at the dike margin. In these dikes, melted margins are typically one-quarter to one-third of the width of the dike, and in cases where dikes are not vertical, the hanging wall has a thicker melted margin (Grunder & Taubeneck, 1997).
The Maxwell Lake Dike
This study focuses on a single, well-exposed Grande Ronde basalt dike with well-developed partially melted wallrock margins (Fig. 2). The dike strikes N20°E, dips steeply to the west (averaging about 75°), and is from 2·6 to 7·8 m thick. It extends as en echelon segments for at least 1 km along strike. Paleodepth at the time of dike emplacement was at most 2 km, as reconstructed from regional geology.
|
The partial melt margin along the hanging wall (western margin) of the dike is generally 2–2·5 m thick but reaches nearly 5 m thick at the southern end of the outcrop (Fig. 2). The footwall partial melt margin is about 1·5 m thick. The thickness of the melt zone in the hanging wall is typically 1·7–1·5 times as thick as that of the footwall and is 0·3–0·5 times as thick as the dike.
The dike margins are divided into four mappable zones based on outcrop-scale textural characteristics (Fig. 2). These are: the unmelted country rock, the mafics-out zone, the mottled zone, and the mush zone. These textural zones are 10 cm to 2 m wide with gradational transitions from one zone to the next. Zones parallel the dike and, like the partial melt zone overall, are proportionally wider in the hanging wall.
The unmelted wallrock is a hypidiomorphic granular hornblende biotite granodiorite by IUGS classification. Based on the compositional classification of Barker (1979) it is a tonalite; we refer to it as a tonalite owing to the paucity of orthoclase relative to plagioclase feldspar (Table 1; Fig. 3). The mafics-out zone is characterized by the absence of biotite and hornblende. Instead, fine-grained pyroxene and Fe– Ti oxides occupy former biotite and hornblende mineral sites. Glass may be present as thin seams surrounding quartz and feldspar grains. The mottled zone is characterized by a blue–gray mottled texture made up of residual quartz and feldspar grains that lack distinct margins, mafic mineral reaction domains, and brown glass seams that surround grains. The mush zone is a discontinuous, 10–50 cm wide zone paralleling the dike margin. This zone contains sparse amorphous grains of quartz and feldspar in a fine-grained, blue–gray groundmass. The presence of the mush zone appears to correlate with thicker parts of the dike. A dense network of blue–gray veins and a cataclastic texture occur in the partially melted tonalite at the southern end of the outcrop.
|
|
|
PROGRESSIVE STAGES OF MELTING |
|---|
|
TOPABSTRACTINTRODUCTIONTHE NATURAL LABORATORYPROGRESSIVE STAGES OF MELTINGCOMPOSITIONAL CHARACTERISTICS OF...DISCUSSIONCONCLUSIONSAPPENDIX: THERMAL MODELREFERENCES |
|---|
Quenched partial melt zones in the margins of the Maxwell Lake dike have captured a continuum of textural reactions, most of which can only be viewed in thin section. We have grouped samples collected from the hanging wall partial melt zones into five progressive stages based on interpretations of the melt reaction textures. Samples representing Stage 1 were collected from the unmelted wallrock zone (Fig. 2) at a distance of >4 m from the dike–wallrock contact. Samples for Stages 2 and 3 were collected from the mafics-out zone (
2–4 m from the contact), and samples representing Stages 4 and 5 were collected from the mottled zone (0·5–2 m from the contact) (Fig. 2). These stages of reaction range from unmelted wallrock (Stage 1) to nearly 47 vol. % quenched melt (Stage 5).
Stage 1
The unmelted tonalite is medium to coarse grained with an inhomogeneous distribution of biotite and hornblende (Fig. 4). Biotite rims are commonly altered to chlorite whereas hornblende grains may contain small quartz inclusions and/or clinopyroxene cores. Magnetite and trace phases, including (in decreasing order of abundance) apatite, zircon, and titanite, are associated with biotite and hornblende and also may be enclosed within grains.
|
Stage 2
Stage 2 contains trace amounts (<1 vol. %) of glass, thus representing the onset of melting (Table 1, Fig. 3). Hornblende contains sub-microscopic dusty reaction products whereas biotite contains dusty opaque reaction minerals along cleavage planes and up to 0·5 mm inward from crystal edges. No glass is present on quartz–feldspar boundaries, but thin (<10 µm) seams of clear glass are present along fractures within quartz crystals. These seams locally grade into rare small pools of yellow–brown glass at the margins of biotite and hornblende crystals. The sample of Stage 2 has a cataclastic overprint as evidenced by abundant veins, minutely fractured quartz and feldspar crystals, and offset twinning in plagioclase.
Stage 3
Stage 3 is most conspicuously characterized by the appearance of continuous glass seams and the absence of both hornblende and biotite (Table 1, Fig. 3). Dusty brown reaction products rimmed by optically aligned pyroxene grains occur in hornblende reaction sites (Fig. 5a). Sites of biotite consumption are occupied by a fine-grained intergrowth of glass, dusty magnetite and lesser ilmenite, orthopyroxene, and plagioclase feldspar with opaque oxides aligned in parallel bands (Fig. 6a). When in contact with glass, plagioclase generally has a spongy texture and poorly developed fritted margins. Individual cells in spongy plagioclase are rounded and filled with brown glass. We estimate 4 vol. % glass is trapped within spongy plagioclase.
|
|
About 12 vol. % devitrified brown glass is localized around sites where biotite and hornblende have been consumed and along quartz–plagioclase grain boundaries (Fig. 7a). Seams are as thin as a few tens of microns and as thick as 1 mm. The brown glass occurs as two textural domains that grade into one another (Fig. 7b). The dominant light brown glass domain is characterized by radiating sheaves of microfibrous crystals that we interpret as fine spherulitic devitrification. Opaque oxides occur as sparse needles. The dark brown domain remains largely glassy and is characterized by dusty appearance under the microscope. Light brown domains surround dark brown domains, and both domains are irregularly distributed around biotite and hornblende reaction sites and along seams between quartz and feldspar (Fig. 7a). Microlites of acicular to hopper plagioclase, acicular pyroxene, and equant magnetite are associated with the brown glass domains (Fig. 7b). These quench crystals make up typically 2 vol. % of the bulk mode, but up to 20 vol. % of the quenched groundmass. Minor clear, granular domains have sharp boundaries with the brown glass domains (Fig. 7b). These clear granular domains are somewhat crystalline (i.e. not completely isotropic) and almost always associated with embayed quartz crystals.
|
Stage 4
Stage 4 is characterized by the absence of orthoclase; the modal proportion of quartz is <3 vol. %, and the sample contains about 18 vol. % glass and about 1 vol. % quench crystals (Table 1, Fig. 3). Small (<100 µm long), optically aligned clinopyroxene and orthopyroxene crystals occupy sites where hornblende has been consumed. Magnetite and lesser ilmenite are concentrated towards the center of the hornblende reaction sites. A fine-grained intergrowth of glass, aligned magnetite and lesser ilmenite, orthopyroxene, and plagioclase occupies sites of biotite consumption. The three devitrified glass domains described for Stage 3 are also present in Stage 4. The sample of Stage 4 has a cataclastic overprint as evidenced by brecciated crystal fragments, microscopic fractures in crystals, and offset twinning in plagioclase.
Stage 5
Stage 5 contains about 31 vol. % glass and 9 vol. % quench crystals (Table 1, Fig. 3), thus representing the maximum degree of partial melting at the Maxwell Lake dike. Optically aligned orthopyroxene with minor magnetite occurs in hornblende consumption sites (Fig. 5b). Aligned magnetite and lesser ilmenite intergrown with orthopyroxene and plagioclase occurs in biotite reaction sites (Fig. 6b). Although still very fine, crystals are coarser than in Stage 3. Relict quartz crystals are rounded and embayed, and may be surrounded by haloes of acicular clinopyroxene. Plagioclase crystals in contact with glass have well-developed spongy texture, fritted margins, and a thin (<25 µm wide) optically distinct rim. Spongy plagioclase may contain up to 7 vol. % glass; 30% of the plagioclase has a spongy texture (Fig. 3).
As in Stages 3 and 4, glass is localized around reacted mafic sites and as seams up to 2 mm thick on quartz– plagioclase grain boundaries. Stage 5 glass occurs as dominant, microfibrous light brown domains, glassy dark brown domains, and minor clear granular domains. Acicular to hopper quench crystals of plagioclase, pyroxene, and magnetite are localized in the brown glass domains, typically making up 25 vol. %, but up to 50 vol. %, of the groundmass.
|
COMPOSITIONAL CHARACTERISTICS OF PROGRESSIVE MELTING |
|---|
|
TOPABSTRACTINTRODUCTIONTHE NATURAL LABORATORYPROGRESSIVE STAGES OF MELTINGCOMPOSITIONAL CHARACTERISTICS OF...DISCUSSIONCONCLUSIONSAPPENDIX: THERMAL MODELREFERENCES |
|---|
Analytical methods
Whole-rock bulk analyses were performed using the Rigaku 3370 X-ray fluorescence (XRF) spectrometer at Washington State University's GeoAnalytical Laboratory. Analyses of crystals and glass were performed using the CAMECA SX-50 Electron Microprobe at Oregon State University. Analyses of hornblende, pyroxene, feldspar, and biotite used a beam current of 30 nA, an accelerating voltage of 15 kV, and a 3–5 µm diameter beam. Glass was analyzed using the same conditions and a broad (20 µm) beam. Sodium was counted first in glass and crystals because of its susceptibility to migration. Additional details have been provided by Petcovic (2000)
.
Bulk composition
Most analyses of major and trace elements for the five stages are identical within analytical error (Table 2). The samples from Stage 2 and, to a lesser degree Stage 4, are slightly altered, which is reflected in low Na2O values. Nevertheless, minor loss of Na2O during melting is possibly indicated. Some variations greater than analytical error, such as higher La, Ce, and Th in Stages 3, 4, and 5 compared with Stage 1, probably reflect differences in the proportion of trace phases between samples. Overall, however, the process of partial melting of the Wallowa tonalite was a closed system, excepting volatile components, based on the near identity of bulk-rock analyses of unmelted and partially melted rock.
|
Phase compositions
Hornblende and biotite
Hornblende (Table 3, Fig. 8a) is present only in Stages 1 and 2 (Fig. 3). Relative to unmelted rock, Stage 2 hornblende is compositionally more heterogeneous, and has lost virtually all Cl, half of the F, and much of the K (Fig. 8). Whereas Stage 1 hornblende has a normal trend of increasing FeO with decreasing MgO, Stage 2 hornblende has a scattered increase in FeO with increasing MgO (Fig. 8b).
|
|
Like hornblende, biotite is present only in Stages 1 and 2 (Fig. 3, Table 3). Stage 1 biotite is tightly clustered in composition, relative to Stage 2 biotite, which has generally lower FeO and higher TiO2 and highly variable MgO concentrations (Fig. 9). Stage 2 biotite has as great as 10-fold enrichment in Na2O with relatively modest loss in K2O (Fig. 9b). Fluorine concentrations are also enriched in most Stage 2 biotite and correspond to high TiO2 compositions (Fig. 9c). Chlorine concentration is broadly the same in Stage 1 and in Stage 2 (Cl of 0·5–0·15 wt %) but Stage 2 biotite with low F (<0·3 wt %) also has low Cl (<0·05 wt %; Fig. 9c).
|
Feldspar
Plagioclase and orthoclase together make up nearly 50 vol. % of the bulk unmelted wallrock (Fig. 3). Orthoclase is absent from the restite assemblage by Stage 4 (Table 4, Fig. 3), indicating that it is consumed during the early stages of melting. Stage 2 orthoclase is less potassic than Stage 1 orthoclase (Fig. 10a) and contains up to three times more Ba (0·54–1·69 wt % in Stage 2 vs 0·08–0·55 wt % in Stage 1).
|
|
Plagioclase makes up the bulk of the mode in all stages (Fig. 3). In the partially melted wallrock, it occurs as a residual phase (Stages 2–5), as interstitial material in biotite reaction sites (Stages 3–5), and as quench crystals associated with glass seams (Stages 3–5) (Table 4). Changes in residual plagioclase composition are not systematic with increased melting (Fig. 10). Stage 5 plagioclase is less different from Stage 1 than are the intervening stages. On the whole, residual plagioclase becomes richer in K2O, CaO, FeO, and MgO, and poorer in Na2O with continued melting (Fig. 10). Optically distinct plagioclase rims in Stage 5 are also compositionally distinct, having higher concentrations of CaO, FeO, and MgO than Stage 5 relict plagioclase (Fig. 10b).
Plagioclase from reacted biotite sites in Stages 3–5 (Fig. 6) is predominantly labradorite (Table 4). Plagioclase in biotite reaction sites is generally more calcic than residual plagioclase and contains <1 wt % K2O, but is compositionally variable. MgO and FeO concentrations are slightly higher than in residual plagioclase (up to 0·34 wt % MgO and 1·54 wt % FeO in Stage 5), and concentrations of these oxides increase with continued melting.
Andesine to labradorite quench crystals were analyzed in Stages 3–5 (Table 4). The compositional range of quench plagioclase is similar to that of relict plagioclase, but quench crystals may be slightly less potassic. FeO and MgO concentrations in quench plagioclase are slightly higher in more advanced stages of melting. Stage 5 quench crystals contain up to 0·13 wt % MgO and 1·31 wt % FeO.
Pyroxene
Excepting rare cores in hornblende, pyroxene is not present in the wallrock assemblage until Stage 3, indicating that it is produced during partial melting reactions (Fig. 3). Texturally, pyroxene occurs in Stages 3–5 as microlites associated with hornblende reaction sites, as interstitial material in biotite reaction sites, and as acicular quench crystals associated with brown glass seams. Both orthopyroxene and clinopyroxene occur in hornblende reaction sites as optically aligned microlites (Fig. 5). In Stage 3, augite, pigeonite, and lesser enstatitic orthopyroxene occupy decomposed hornblende sites (Table 5). However, by Stage 5, nearly all of this pyroxene is enstatitic orthopyroxene (Table 5). Early clinopyroxene and orthopyroxene crystals have heterogeneous compositions, but pyroxene compositions become more homogeneous with continued reaction (Fig. 11). Concentrations of Na2O in augite decrease from 1·3 wt % in Stage 3 to <0·4 wt % in Stage 4. Stage 3 augite contains up to 8 wt % Al2O3 whereas Stage 4 augite contains <5 wt %.
|
|
Orthopyroxene intergrown with plagioclase and opaque oxides occurs in biotite reaction sites in Stages 3–5 (Fig. 6). Enstatitic orthopyroxene in biotite reaction sites is slightly more Mg- and Al-rich and Ca- and Na-poor than enstatitic orthopyroxene in hornblende reaction sites (Table 5).
Enstatitic orthopyroxene quench crystals are associated with brown glass seams, and sparse augite quench crystals also occur in Stage 5 (Table 5). Quench enstatitic orthopyroxene is similar in composition to enstatitic orthopyroxene in decomposed hornblende sites, but is Al-poor in comparison with enstatitic orthopyroxene in decomposed biotite sites (typically 0·8–1·8 wt % Al2O3 vs >3 wt % for biotite). Quench augite is also Al-poor in comparison with augite in reacted hornblende sites from Stages 3 and 4.
Magnetite and ilmenite
Magnetite is ubiquitous in all stages, occurring as a primary phase in the unmelted wallrock, as a residual phase in Stages 2–5, texturally associated with hornblende and biotite reaction sites in Stages 2–5 (Figs 5 and 6), and as quench crystals associated with brown seams of glass. Residual magnetite, as well as magnetite associated with decomposed hornblende and biotite sites, may contain up to 15 wt % TiO2. Sparse ilmenite is present as scattered grains in Stages 3 and 5, in Stage 4 decomposed hornblende sites, and in reacted biotite sites from Stages 3–5. The small size of Fe–Ti oxides made microprobe analysis difficult and often yielded low analytical totals (Table 6, and see Petcovic, 2000). Only a few magnetite–ilmenite pairs in Stages 3 and 4 yielded analyses suitable for geothermometry (see thermal modeling section, below).
|
Glass
Glass analyzed from Stages 3–5 yields a different composition for each textural domain (Table 7, Fig. 12). The light brown, microfibrous glass domains are metaluminous to mildly peraluminous (generally <1% normative corundum) and granitic with up to 8 wt % K2O in Stage 3 and 6·5 wt % K2O in Stage 5 (Table 7). The dark brown glass domains are metaluminous and tonalitic in composition. The two domains are best distinguished by the amount of K2O relative to CaO (Fig. 12a). With respect to Na2O, TiO2, MgO, FeO, and Al2O3, the brown glass domains largely overlap at each stage. With increased degree of melting (from Stage 3 to Stage 5), K2O and SiO2 concentrations decrease and FeO, MgO, and TiO2 concentrations increase slightly in both brown glass domains (Fig. 12).
|
|
The clear granular domains of Stages 3–5 are essentially composed of SiO2 (Table 7, Fig. 12b). The clear to yellow–brown glass domains in Stage 2 were not successfully analyzed. These seams proved to generally be too thin and/or altered for successful analysis.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONTHE NATURAL LABORATORYPROGRESSIVE STAGES OF MELTINGCOMPOSITIONAL CHARACTERISTICS OF...DISCUSSIONCONCLUSIONSAPPENDIX: THERMAL MODELREFERENCES |
|---|
Partial melting at the margin of the Maxwell Lake CRBG dike in the Wallowa Mountains provides a macrocosm of textural and compositional information about progressive melting of continental crust at a scale between experimental charges and granitoid intrusions. We first discuss the nature of the melting reactions, including the nature of melts produced. We then compare this example of natural melting with other natural examples and experimental work on similar protoliths. Finally, the conditions of melting are evaluated, followed by development of a thermal model that predicts the timescale over which melting took place.
Nature of the melting reactions
Closed-system melting
Within the range of samples we have examined, partial melting appears to have taken place under closed conditions. Nearly all major elements (as well as most trace elements) were conserved from Stage 1 to Stage 3 to Stage 5 (Table 2), indicating that these samples represent a chemically closed system. Stages 2 and 4 are weighted less in this discussion because of the cataclastic overprint and slight bulk compositional differences attributed to alteration. We do not consider water and other volatile components here. Compositional near-identity during progressive melting also suggests that melts did not separate from their restite. Indeed, we observed no textural evidence in thin section or in outcrop that suggests large-scale melt flow.
The melting reactions
During the progress of partial melting observed over 4 m of wallrock adjacent to the Maxwell Lake dike, hornblende, biotite, and orthoclase were entirely consumed. The composition of these phases became highly variable as they broke down. Plagioclase, quartz, and magnetite were partially consumed yet persisted in the restite with as much as 31 vol. % glass. With progressive melting, relict plagioclase developed a spongy residuum by the reaction of andesine to produce labradorite plus an albitic melt partially trapped within the plagioclase.
During progressive melting, orthopyroxene, clinopyroxene, secondary plagioclase, magnetite, sparse ilmenite, and melt (now represented by devitrified glass + quench crystals) were produced. The compositions of new minerals were initially variable, yet became more homogeneous with continued melting. Melt developed within and around decomposed biotite and hornblende sites, and also along seams up to 2 mm thick between relict quartz and plagioclase crystals. In Stage 3, aligned clinopyroxene and orthopyroxene microlites, minor magnetite and rare ilmenite, and glass occupy decomposed hornblende sites, suggesting that dehydration-melting reactions involving hornblende produced these phases. By Stage 5, all of the clinopyroxene in hornblende decomposition sites has reacted to produce enstatitic orthopyroxene. In Stages 3–5, biotite dehydration-melting reactions produced aligned magnetite and lesser ilmenite in an intergrown matrix of orthopyroxene, plagioclase, and glass. Because partial melting took place in a closed system, components released from phases that break down must have been accommodated either in the restite or in the melt.
Consideration of the difference between modal proportion of phases between Stage 1 and Stage 3 (Fig. 3), along with consideration of the composition of these phases, allows us to determine a general, initial melt-producing reaction:
 | (1) |
18 vol. % melt (12 vol. % glass + 2 vol. % quench crystals + 4 vol. % glass trapped in spongy plagioclase; Table 1). Between Stage 3 and Stage 5 (Fig. 3), wallrock closer to the dike margin experienced higher temperature conditions, suggesting that additional melt was produced from the following reaction:
 | (2) |
Although we have established general melt-producing reactions, other work on partial melting in crystalline rocks (e.g. Wolf & Wyllie, 1991; Philpotts & Asher, 1993; Hammouda et al., 1996; Knesel & Davidson, 1996) has suggested that dehydration-melting reactions are locally controlled by stoichiometry of melt reactions and kinetics rather than by the overall bulk assemblage. For example, melt in reaction (1) could have been produced by local reactions such as hornblende + quartz, biotite + plagioclase, biotite + hornblende + orthoclase, or quartz + plagioclase. This process results in a disequilibrium assemblage of heterogeneous local melts, residual minerals and secondary minerals in different stages of reaction. Abundant disequilibrium textures (e.g. spongy plagioclase) and initially variable phase compositions suggest that partial melting reactions in the Wallowa tonalite were also controlled by local assemblages.
Composition of the melts produced
We analyzed three distinct glass domains in the partially melted wallrock: a dominant, light brown, high-K glass exhibiting feathery devitrification; a less abundant, dark brown, high-Ca glass; and a sparse, clear, high-Si glass. The distribution of both brown glasses is highly irregular, occurring in seams between quartz and plagioclase, adjacent to hornblende and biotite reaction sites, and within these reaction sites. The compositions of each glass domain largely overlap for each stage, except for high K2O in the light brown glass (4–8 wt %) and high CaO in the dark brown glass (3–5 wt %). It is possible that the high-K glass was produced by dehydration-melting reactions dominated by biotite, whereas the high-Ca glass was produced by reactions dominated by hornblende. Compositionally, high-K glass data in Fig. 12 largely overlap with the biotite dehydration-melting field in Fig. 13, and high-Ca glass data overlap with the compositional field for amphibole-derived melts. Additionally, the high-K glass is more abundant than the high-Ca glass.
|
Texturally, we find that an overprint of devitrification has obscured whether the light and dark brown glasses were initially separate melts originating from biotite and hornblende breakdown reactions, respectively. Light brown glass, which exhibits spherulitic devitrification, surrounds cores of dark brown, largely isotropic glass. This textural relationship occurs in melt seams and adjacent to biotite and hornblende breakdown sites. We are currently unable to determine whether the dark and light brown glass domains are relicts of coexisting melts, or whether they are the product of Ca–K migration during devitrification. However, it is likely that locally controlled melting reactions initially produced distinct melts of variable composition, that these melts homogenized to some degree during continued melting, and that they experienced devitrification after quenching.
The clear granular domains observed in Stages 3–5 are nearly always associated with embayed to resorbed quartz grains. Texturally, clear domains are somewhat crystalline with undulose to patchy extinction. They are composed essentially of SiO2, suggesting that they represent fully (re)crystallized polymorphs of quartz. It is unclear whether these domains were ever melted. Clear glass localized on quartz fractures in Stage 2 is probably a product of initial melting reactions involving quartz and feldspar.
To reconstitute the bulk melt composition from the glass domains and quench crystals, we employed three methods (Table 8). (1) A composite melt analysis was achieved by averaging electron microprobe point analyses from a grid of evenly spaced points in a region encompassing glass domains and quench crystals. (2) Using a melt modal reconstitution, the melt composition was calculated by summing the compositions of the devitrified glass domains and plagioclase, pyroxene, and magnetite quench crystals, each weighted by their modal abundance. (3) Using a bulk modal reconstitution, the difference between modally weighted composition of restite phases and whole-rock bulk composition yielded the composition of the melt.
|
On the whole, we think the grid analysis average is the best approximation of the real bulk composition, because it is not subject to errors in modal estimates. There is poor agreement among the three compositions reconstituted for Stage 3 and good agreement for Stage 5 (Table 8). The bulk modal reconstitution was difficult to complete because of the variability in mineral phase compositions, and due to the difficulty in evaluating the mode of extremely fine mineral reaction products in decomposed biotite and hornblende sites. For example, the differences in Stage 3 are probably due to difficulties in modal estimates, in particular overestimation of the abundance of Fe–Ti oxides leading to high values of FeO in the reconstructed melt.
The reconstructed melt in Stage 3 is metaluminous to mildly peraluminous (0·3 wt % normative corundum) and granitic (Fig. 13a). Stage 5 reconstructed melt is metaluminous and granodioritic to granitic (Fig. 13a). With increased melting, concentrations of SiO2 and K2O decrease whereas concentrations of most other components (notably Al2O3, CaO and FeO) increase (Fig. 13).
The natural vs the experimental laboratory
The Maxwell dike locality bears out the conclusion that progressive partial melting in crystalline rocks produces a heterogeneous assemblage of compositionally variable melts, based on other examples of melting (e.g. Büsch et al., 1974; Kaczor et al., 1988; Wolf & Wyllie, 1991; Philpotts & Asher, 1993; Hammouda et al., 1996; Knesel & Davidson, 1996; Rushmer, 2001). In contrast, dehydration-melting experiments of powdered starting materials do not produce such melt heterogeneity. Despite the inherent disequilibrium nature of melting in the natural examples, we observe that the overall modal and compositional record of progressive melting is similar to that of equilibrium experiments. In the following comparisons, we bear in mind that the Wallowa tonalite differs from most other melting studies of crustal rocks in that it contains subequal proportions of amphibole and mica, and that melting took place at <1 kbar. However, melting of granitoid protoliths appears to be insensitive to pressures as great as 8 kbar (Singh & Johannes, 1996a), which suggests that the results of melting in the Wallowa example are applicable to much of the upper crust.
Comparison of mineral compositions
Plagioclase formed part of the parent rock in all granitoid natural and experimental protoliths, and residual plagioclase was a component of restite assemblages under nearly all pressure–temperature conditions up to 10 kbar. In these studies, as in the Wallowa rocks, plagioclase was consumed only at large extents of melting. Dissolution features, such as fritted margins and spongy textures observed in reacted Wallowa plagioclase, were also observed by Büsch et al. (1974), Kaczor et al. (1988), Green (1994), and Philpotts and Asher (1993), who attributed fritted margins to melting along cleavage planes. By Stage 5, Or and Ab components of residual Wallowa plagioclase had increased, and relict rims were high in An, FeO, and MgO. High-An rims in relict plagioclase were also observed by Büsch et al. (1974). Vielzeuf & Montel (1994) reported an increase in the Or component of residual plagioclase, and many studies (Büsch et al., 1974; Kaczor et al., 1988; Beard & Lofgren, 1991; Philpotts & Asher, 1993; Singh & Johannes, 1996b) observed an increase in An in residual plagioclase relative to starting plagioclase.
Amphibole dehydration-melting reactions produced clinopyroxene + lesser orthopyroxene in amphibole reaction sites under nearly all pressure–temperature conditions up to 10 kbar (e.g. Beard & Lofgren, 1991; Rapp et al., 1991; Rushmer, 1991; Patiño Douce & Beard, 1995). In the Wallowa rocks, dehydration-melting reactions involving hornblende initially produced aligned augite + pigeonite + lesser enstatitic orthopyroxene + sparse magnetite. Beard & Lofgren (1991) observed enstatitic orthopyroxene, augite, and sparse pigeonite as amphibole reaction products, but their augite contained <5 wt % Al2O3. In contrast, Wallowa augite contained up to 8 wt % Al2O3.
Studies involving biotite dehydration-melting reactions document orthopyroxene + Fe–Ti oxides ± alkali feldspar (rarely + amphibole) in biotite reaction sites (e.g. Büsch et al., 1974; Kaczor et al., 1988; Vielzeuf & Montel, 1994; Patiño-Douce & Beard, 1995; Singh & Johannes, 1996a; Rushmer, 2001). Reactions involving biotite breakdown in the Wallowa wallrock produced orthopyroxene + plagioclase + high-Ti magnetite + sparse ilmenite. Similar to what we observed from Stage 3 to Stage 5, Patiño Douce & Beard (1995) and Singh & Johannes (1996b) observed an increase in En component in orthopyroxene with rising temperature. In contrast to other studies involving biotite breakdown, alkali feldspar was not observed. It is possible that the high water content of initial dehydration melts and low total pressure destabilized alkali feldspar as a reaction product [compare experimental work by Naney (1983) and Johnson & Rutherford (1989)]. On the other hand, hornblende dehydration-melting reactions may have contributed Ca to form plagioclase in biotite reaction sites.
Comparison of melts
In crystalline rocks such as the Wallowa tonalite, in situ melt of multiple compositions (preserved as glass or granophyre) has been reported as seams around crystals, along fractures within crystals, and localized around decomposed mafic sites (e.g. Büsch et al., 1974; Kaczor et al., 1988; Philpotts & Asher, 1993; Green, 1994; Knesel & Davidson, 1996; Tommasini & Davies, 1997; Rushmer, 2001). In the partially melted granite studied by Kaczor et al. (1988), clear glass (high SiO2, alkalis, and Rb) localized around quartz was a product of biotite breakdown and early quartz–feldspar melting. Brown glass (high CaO, Al2O3, MgO, FeO, and TiO2), localized around oxides and spongy feldspar, became more abundant at higher extents of melting. Philpotts & Asher (1993) reported two distinct melt compositions: a high-K glass located between quartz and orthoclase, and a low-K glass between quartz and andesine.
In their granite cube experiment, Knesel & Davidson (1996) observed clear (trachytic) glass locally grading into brown (mugearitic) glass that surrounded Fe–Ti oxides replacing biotite. Rushmer (2001) observed voluminous, granitic glass derived from muscovite breakdown along grain boundaries and cracks in quartz grains, with minor, biotite-derived glass (higher FeO and TiO2) localized around spinel in biotite reaction sites. Locally, melt compositions mixed along grain boundaries, producing an intermediate-composition glass. Previous workers have pointed out that progressive disequilibrium melting of this type produces melts that were initially enriched in many incompatible elements, particularly Rb as well as 87Sr/86Sr derived from the breakdown of biotite– hornblende, whereas successive melts were enriched in the restite component (particularly Sr) released with plagioclase (e.g. Kaczor et al., 1988; Hammouda et al., 1996; Knesel & Davidson, 1996; Tommasini & Davies, 1997; Davies & Tommasini, 2000).
Similar to other studies, melt in the Wallowa tonalite occurs on quartz–plagioclase boundaries, trapped within spongy plagioclase, and around decomposed hornblende and biotite sites. The heterogeneity in quenched melt from the Wallowa samples may, in part, be attributable to devitrification textures; however, comparison with other studies suggests that local reactions also played a role. Dehydration-melting reactions dominated by either biotite or hornblende could account for the high-K and high-Ca brown glasses, respectively. Sparse clear (Stage 2) glass, which we have not analyzed, probably represents an early product of melting on quartz–feldspar boundaries, similar to the clear glass observed by Kaczor et al. (1988).
The Stage 3 bulk (reconstructed) melt was produced by dehydration-melting reactions involving biotite, hornblende, plagioclase, quartz, and orthoclase. Although biotite and hornblende are modally subequal in the unmelted Wallowa tonalite, the Stage 3 bulk melt lies within the field for melts produced from biotite-bearing protoliths (Fig. 13). Biotite dehydration-melting reactions, however, may produce 2–3 times more melt than amphibole dehydration-melting reactions (Patiño Douce & Beard, 1995). The composition of Stage 5 bulk melt (Fig. 13) reflects continued melting and the consumption of clinopyroxene and plagioclase.
Overall, the bulk Wallowa melts are metaluminous to barely peraluminous, in contrast to peraluminous (1–5% normative corundum) melts produced from biotite-bearing protoliths. Melts from amphibole-bearing protoliths are trondhjemitic to granodioritic, metaluminous to peraluminous (0–7% normative corundum), silicic, and low in mafic oxides; Wallowa melts differ in that they are more K-rich and Al-poor (Fig. 13). Our bulk melt composition reflects the involvement of both biotite and hornblende in dehydration-melting reactions. With increased degree of melting (Stage 3 to Stage 5), Wallowa melts become slightly more mafic, more aluminous, less potassic, and less silicic, as is consistent with other partial melting studies. Overall, Stage 3 Wallowa bulk melt is similar to average A-type granite (as compiled by Skjerlie & Johnston, 1992) and some high-silica rhyolites (see Streck, 2002). Stage 5 Wallowa bulk melt is similar to metaluminous rhyodacites thought to be mainly of a crustal-melt origin (see Feeley & Grunder, 1991; Johnson & Grunder, 2002).
Thermal history and implications of diking
Thermal conditions of melting
Temperatures recorded by minerals in the melt zone represent an integrated thermal effect from heating during diking followed by cooling when basalt flow ceased. The maximum thermal gradient at the edge of the Maxwell Lake dike was about 1090°C, assuming a geothermal gradient of 25°C/km, a depth of 2 km to the dike, and a basalt magma temperature of 1140°C [liquidus temperature from Murase & McBirney (1973)]. The wallrock could have been hotter than estimated if it had been subjected to heating by previous dike intrusion. Mineral thermometry and comparison with experimental phase equilibria have been used to estimate the temperature conditions represented by each progressive melting stage (Fig. 14a).
|
The small grain size and heterogeneity of Fe–Ti oxides compromise their use in geothermometry. Nevertheless, potential equilibrium pairs of magnetite and ilmenite were identified in Stages 3 and 4 on the basis of the Mg–Mn partitioning criteria of Bacon & Hirschmann (1988)
(Table 6). Stage 3 magnetite– ilmenite pairs give a range of values with an average temperature of about 830°C (Fig. 14a). Stage 4 pairs suggest an average temperature of 1060°C (Fig. 14a). Stage 5 contained no equilibrium pairs. Pyroxene pairs did not produce equilibrium tie lines (after Lindsley, 1983), and were therefore not successful for thermometry.
Comparison with experimental phase equilibria from intermediate composition protoliths can help constrain the temperature at which melt-forming reactions took place. From a series of low-pressure experiments designed to locate the solidus of a synthetic biotite tonalite, Singh & Johannes (1996a, 1996b) concluded that the onset of melting in biotite-bearing rocks may be as low as 700°C. However, 1–3 kbar melting experiments on a meta-greywacke indicated a biotite breakdown range of 800–860°C (Vielzeuf & Montel, 1994), and 3 kbar experiments on a synthetic biotite tonalite indicated a range of 850–930°C (Patiño Douce & Beard, 1995). We have chosen an initial biotite-breakdown temperature of 800°C, which is therefore a minimum estimate for Stage 2 where the dehydration-melting of biotite produced thin glass seams (Fig. 14a). Under pressure conditions of 
5 kbar, hornblende in most intermediate composition amphibolites reacted out between about 850 and 925°C (Beard & Lofgren, 1991; Patiño Douce & Beard, 1995). Because Stage 3 lacks both biotite and hornblende, temperature at this location probably reached at least 925°C (Fig. 14a).
Thermal model of wallrock melting
We have constructed a one-dimensional thermal model to estimate how long basalt flowed in the dike to provide sufficient thermal flux to the wallrock to propagate partial melt zones and related temperatures as far as 4 m from the dike–wallrock contact (additional details are given in the Appendix). In Part A of the model (Fig. 14a), the dike was held at a constant temperature of 1140°C to simulate magma flow while the dike was active. The final conditions of this simulation provided the initial conditions for Part B (Fig. 14b and c), in which magma flow has ceased and the dike and wallrock were allowed to cool. Both parts of the model assume all heat flow was via conduction. Additionally, we assume that the latent heat of crystallization of the basalt was equal to the latent heat of melting of the wallrock, and that both were released linearly over the entire temperature interval. Latent heat of crystallization of the wallrock was neglected.
Results for Part A of the model (for an intermediate thermal diffusivity value of 4·11 x 10-7 m2/s; Fig. 14a) suggest that tonalite located within the first few centimeters of the dike–wallrock contact began to undergo partial melting within a few hours of dike intrusion. This timescale is similar to that of experimental studies, such as the water-saturated melting experiments of Büsch et al. (1974) and the granite cube experiments by Knesel & Davidson (1996), where small volumes of rock underwent melting over timescales of hours to days. Within about 1 year of dike injection, all Stage 5 wallrock (0–1 m from the dike–wallrock contact) had reached temperatures in excess of 925°C. Stage 4 wallrock (1–2 m from the contact) had begun to melt after about 1 year, whereas Stage 3 wallrock (2–3 m from the contact) underwent melting after nearly 3 years. Results suggest that about 4 years were required for rocks at 4 m from the dike–wallrock contact to reach 800°C. This, therefore, is the maximum time that the dike was active; wallrock beyond this distance shows no evidence of biotite breakdown or onset of partial melting.
Results of modeling the cooling history of both the dike and wallrock are shown in Fig. 14b. Because the heat pulse continued to propagate during cooling, wallrock further than about 3 m from the contact experienced additional heating. However, at 4 m from the contact (the critical transition between unmelted and partially melted wallrock), there was only about 10°C of additional heating (Fig. 14c) between 0·05 and 1 year following cessation of dike flow. Isotherms were steep enough so that wallrock at distances >4 m never experienced temperatures in excess of 800°C and therefore experienced no partial melting.
We found that both parts of the model were sensitive to thermal properties of the dike and wallrock [expressed as the thermal diffusivity value (
)]. For example, a minimum thermal diffusivity value (1·22 x 10-6 m2/s) yielded model results indicating that wallrock at 4 m from the contact began to break down after 1·2 years. A maximum thermal diffusivity value (1·38 x 10-7m2/s) suggested that breakdown began after about 10·6 years. The model was rather insensitive to dike thickness. Part A was independent of dike thickness, whereas both dike and wallrock cooled more slowly with a thicker dike in Part B, yet the shape of the isotherms remained largely unaffected. For example, a dike of 16 m thickness was 200°C warmer after 20 years of cooling than the results shown in Fig. 14b.
The 4 year heating interval implied by this model is a maximum time if the wallrock were preheated by prior dike activity. Preheating the wallrock to 200°C at the time of dike injection allowed temperatures to reach 800°C at 4 m from the contact in 2·6 years, one year faster than with wallrock at 55°C. On the other hand, it is more likely that the 4 year timescale is a minimum if intermittent flow in the dike or cooling by means other than conduction occurred. Long pauses in flow of the dike could potentially cause complex overprints of melting and crystallization reactions. Although we cannot preclude some fluctuations, the regular textural progression observed in the melt zone is consistent with continuous or pulsating flow. Cooling of dike and wallrock by means other than conduction, such as cooling by circulating groundwater [as suggested by Delaney (1987)], may account for the rapid cooling history implied by the presence of glass and quench crystals in the partial melt zones. If such cooling took place, the calculated heating times based on conductive cooling are minima.
Implications of the thermal model for CRBG volcanism
Dikes with substantial wallrock melting are rare in the Wallowas; in mapping of four sub-swarms we have found only two dikes with well-developed wallrock partial melt zones. In both cases where there is substantial partial melt in wallrock adjacent to a dike, the dike is not prominently quenched. Instead, coarse-grained basalt extends to the dike–wallrock contact. The thermal model of the Maxwell Lake dike suggests that magma flowed through this dike for at least several years. We believe that, in general, the higher thermal flux experienced by dikes with partial melt at their margins indicates that they had a prolonged history of activity and therefore were more likely to have fed major CRBG flows. Dikes with quenched margins and no interaction with the wallrock, on the other hand, probably represent conduits used for short periods of time before solidifying. We take this to be analogous to Hawaiian-style eruptions where early dike-fed fissure eruptions become localized to yield long-lived central vent eruptions that feed flows.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONTHE NATURAL LABORATORYPROGRESSIVE STAGES OF MELTINGCOMPOSITIONAL CHARACTERISTICS OF...DISCUSSIONCONCLUSIONSAPPENDIX: THERMAL MODELREFERENCES |
|---|
In rare instances, wallrock adjacent to Columbia River Basalt Group dikes in the Wallowa Mountains has undergone partial melting, providing a unique opportunity to examine crustal melting in a natural setting. The unmelted wallrock is a hornblende–biotite granodiorite to tonalite, a lithology that is rarely examined in the experimental partial melting literature but is probably common in natural settings. Samples collected from the margin of a Grande Ronde dike at Maxwell Lake represent progressive stages of closed-system partial melting over a distance of about 4 m from unmelted tonalite (Stage 1) to about 47 vol. % quenched melt (Stage 5). Partial melt reactions took place at a paleodepth of about 2–2·5 km.
With the onset of melting, a trace amount of a clear melt was produced, now preserved along fractures in quartz. Dehydration-melting reactions involving both biotite and hornblende, plus plagioclase, orthoclase and quartz, produced melt (preserved as variably devitrified glass and quench crystals) localized around decomposed mafic sites, on quartz–feldspar boundaries and in spongy plagioclase. Comparison with other natural examples and experimental work indicates that a dominant high-K (light brown) glass resulted from biotite dehydration-melting, leaving aligned magnetite and ilmenite intergrown with plagioclase and orthopyroxene in the former biotite sites. A less abundant high-Ca (dark brown) glass was produced during dehydration-melting of hornblende leaving a dusty intergrowth of clinopyroxene, lesser orthopyroxene, and sparse magnetite in former hornblende sites. Approximately 18 vol. % melt were produced in these early stages of melting. Up to 29 vol. % additional melt were produced by the reaction of orthoclase, clinopyroxene, quartz, plagioclase and magnetite. This reaction was terminal for both orthoclase and clinopyroxene, leaving optically aligned orthopyroxenes in former hornblende sites. With progressive melting, phases being consumed became strikingly more heterogeneous in composition, whereas reaction products were relatively homogeneous. The progress of disequilibrium melting reactions as well as the composition of reaction products are broadly similar to those observed in other natural and experimental case studies.
The bulk composition of the reconstructed early melts was granitic and metaluminous to barely peraluminous and closely approximates the composition of many A-type granites. With progressive reaction, the melt became more granodioritic and metaluminous and closely mimics the composition of rhyodacitic volcanic rocks thought to be the products of crustal melting. In general, the Wallowa bulk melt, produced by simultaneous dehydration-melting of biotite and hornblende, was intermediate in composition between granitic peraluminous liquids produced from biotite dehydration melting and the tonalitic liquids produced from amphibole dehydration melting.
Thermal modeling suggests that for a dike of 8 m thickness carrying magma at 1140°C, about 4 years were required to initiate breakdown reactions in wallrock at a distance of 4 m from the dike–wallrock contact. Depending on the choices of thermal properties of the dike and host rock, 1–10 years were required to initiate the melting reactions at 4 m. If flow of magma in the dike was intermittent, or cooling was enhanced by groundwater circulation, then the dike could have been active longer. We think that dikes with substantial partially melted margins represent long-lived portions of the Columbia River Basalt feeder system and may have sustained large flows, possibly analogous to Hawaiian-style central vent eruptions. In contrast, the majority of dikes, which are quenched against the wallrock, represent dike propagation and fissure eruption events that were short-lived.
|
APPENDIX: THERMAL MODEL |
|---|
|
TOPABSTRACTINTRODUCTIONTHE NATURAL LABORATORYPROGRESSIVE STAGES OF MELTINGCOMPOSITIONAL CHARACTERISTICS OF...DISCUSSIONCONCLUSIONSAPPENDIX: THERMAL MODELREFERENCES |
|---|
Symbols used in model
|
* Values used to calculate for model results shown in Fig. 14 are in bold.
Intermediate value used for both dike and wallrock.
Part A
To simulate heating of wallrock while magma is flowing through the dike, we model the system as heat flow via conduction in a semi-infinite solid (e.g. Carslaw & Jaeger, 1959). The general heat equation, given an initial condition of T(x > a, 0) = Tw and boundary conditions of dike margin constant at Td [T(x = a, t) = Td], and T must be Tw at infinity [T(x 
, t) = 0] is
 | (A1) |
 | (A2) |
 | (A3) |
(i.e. that |dfs/d| = 1), the solution is
 | (A4) |
 | (A5) |
Part B
To simulate cooling of the dike and wallrock, we model the system as an infinite solid. At t = tp, dike injection is stopped, and dike and wallrock are allowed to cool. The general solution to equation (A4) at t = tp is (after Carslaw & Jaeger, 1959)
 | (A6) |
|
ACKNOWLEDGEMENTS |
|---|
We would especially like to thank Bill Taubeneck for his assistance with the petrographic analysis and his guidance on the field component of this research. We would also like to thank Roger Nielsen of the Oregon State University Electron Microprobe Laboratory for his advice on and assistance with data collection. George Bergantz and Roy Haggerty provided valuable ideas and input for the thermal modeling. Discussions with Peter Reiners, John Dilles, and Joe Dufek helped to clarify this work. Reviews by Tracy Rushmer and Mike Williams substantially improved the manuscript. Thanks go to Mike Winkler, Brandon Browne, Jesse Dickinson, Lang Farmer, and George Bergantz for participation in field work. This research was supported in part by the Geological Society of America grant number 6514-99 awarded to H.L.P.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONTHE NATURAL LABORATORYPROGRESSIVE STAGES OF MELTINGCOMPOSITIONAL CHARACTERISTICS OF...DISCUSSIONCONCLUSIONSAPPENDIX: THERMAL MODELREFERENCES |
|---|
Al-Rawi, Y. & Carmichael, I. S. E. (1967). A note on the natural fusion of granite. American Mineralogist 52, 1806–1814.[Web of Science]
Armstrong, R. L., Taubeneck, W. H. & Hales, P. O. (1977). Rb–Sr and K–Ar geochronometry of Mesozoic granitic rocks and their Sr isotopic composition, Oregon, Washington, and Idaho. Geological Society of America Bulletin 88, 397–411.
Bacon, C. R. & Hirschmann, M. M. (1988). Mg/Mn partitioning as a test for equilibrium between coexisting Fe–Ti oxides. American Mineralogist 73, 57–61.[Abstract]
Baksi, A. K. (1989). Reevaluation of the timing and duration of extrusion of the Imnaha, Picture Gorge, and Grande Ronde Basalts, Columbia River Basalt Group. In: Reidel, S. P. & Hooper, P. R. (eds) Volcanism and Tectonism in the Columbia River Flood-Basalt Province. Geological Society of America, Special Papers 239, 105–111.
Barker, F. (1979). Trondhjemite: definition, environment and hypothesis of origin. In: Barker, F. (ed.) Trondhjemites, Dacites, and Related Rocks (Developments in Petrology, 6). Amsterdam: Elsevier, pp. 1–12.
Beard, J. S. & Lofgren, G. E. (1991). Dehydration melting and water-saturated melting of basaltic and andesitic greenstones and amphibolites at 1, 3, and 6·9 kb. Journal of Petrology 32, 365–401.
Büsch, W., Schneider, G. & Mehnert, K. R. (1974). Initial melting at grain boundaries, part II: melting in rocks of granodioritic, quartzdioritic and tonalitic composition. Neues Jahrbuch für Mineralogie, Monatshefte 8, 345–370.
Carslaw, H. S. & Jaeger, J. C. (1959). Conduction of Heat in Solids, 2nd edn. Oxford: Oxford University Press.
Davies, G. R. & Tommasini, S. (2000). Isotopic disequilibrium during rapid crustal anatexis: implications for petrogenetic studies of magmatic processes. Chemical Geology 162, 169–191.[CrossRef][Web of Science]
Deer, W. A., Howie, R. A. & Zussman, J. (1992). An Introduction to the Rock-Forming Minerals, 2nd edn. Hong Kong: Addison Wesley Longman.
Delaney, P. T. (1987). Heat transfer during emplacement and cooling of mafic dykes. In: Halls, H. C. & Fahrig, W. L. F. (eds) Mafic Dyke Swarms. Geological Association of Canada, Special Papers 34, 31–46.
Feeley, T. C. & Grunder, A. L. (1991) Mantle contribution to the evolution of middle Tertiary silicic magmas during early stages of extension: the Egan Range Volcanic Complex, east–central Nevada. Contributions to Mineralogy and Petrology 106, 154–169.[CrossRef][Web of Science]
Ghiorso, M. S. & Sack, R. O. (1991). Fe–Ti oxide geothermometry: thermodynamic formulation and the estimation of intensive variables in silicic magmas. Contributions to Mineralogy and Petrology 108, 485–510.[CrossRef][Web of Science]
Green, N. L. (1994). Mechanism for middle to upper crustal contamination: evidence from continental-margin magmas. Geology 22, 231–234.
Grunder, A. L. & Taubeneck, W. H. (1997). Partial melting of tonalite at the margins of Columbia River Basalt Group dikes, Wallowa Mountains, Oregon. Geological Society of America, Abstracts with Programs 29, 18.
Hammouda, T., Pichavant, M. & Chaussidon, M. (1996). Isotopic equilibration during partial melting: an experimental test of the behaviour of Sr. Earth and Planetary Science Letters 144, 109–121.[CrossRef][Web of Science]
Holtz, F., Pichavant, M., Barbey, P. & Johannes, W. (1992). Effects of H2O on liquidus phase relations in the haplogranite system at 2 and 5 kbar. American Mineralogist 77, 1223–1241.[Abstract]
Johnson, J. A. & Grunder, A. L. (2000). The making of intermediate composition magma in a bimodal suite: Duck Butte Eruptive Center, Oregon, USA. Journal of Volcanology and Geothermal Research 95, 175–195.[CrossRef][Web of Science]
Johnson, M. C. & Rutherford, M. J. (1989). Experimentally determined conditions in the Fish Canyon Tuff, Colorado, magma chamber. Journal of Petrology 30, 711–737.
Kaczor, S. M., Hanson, G. N. & Peterman, Z. E. (1988). Disequilibrium melting of granite at the contact with a basic plug: a geochemical and petrographic study. Journal of Geology 96, 61–78.[Web of Science]
Kitchen, D. E. (1989). The disequilibrium partial melting and assimilation of Caledonian granite by Tertiary basalt at Barnesmore, Co. Donegal. Geological Magazine 126, 397–405.[Abstract]
Knesel, K. M. & Davidson, J. P. (1996). Isotopic disequilibrium during melting of granite and implications for crustal contamination of magmas. Geology 24, 243–246.
Le Breton, N. & Thompson, A. B. (1988). Fluid-absent (dehydration) melting of biotite in metapelites in the early stages of crustal anatexis. Contributions to Mineralogy and Petrology 99, 226–237.[CrossRef][Web of Science]
Lindsley, D. H. (1983). Pyroxene thermometry. American Mineralogist 68, 477–493.[Abstract]
Murase, T. & McBirney, A. R. (1973). Properties of some common igneous rocks and their melts at high temperatures. Geological Society of America Bulletin 84, 3563–3592.
Naney, M.T. (1983) Phase equilibria of rock-forming ferromagnesian silicates in granitic systems. American Journal of Science 283, 993–1033.
Patiño Douce, A. E. & Beard, J. S. (1995). Dehydration-melting of biotite gneiss and quartz amphibolite from 3 to 16 kbar. Journal of Petrology 36, 707–738.
Petcovic, H. L. (2000). Partial melting of tonalite at the margins of a Columbia River Basalt Group dike, Wallowa Mountains, northeastern Oregon. M.S. thesis, Oregon State University, Corvallis.
Petcovic, H. L., Grunder, A. L. & Taubeneck, W. H. (2001). Distribution and geometry of feeder dikes from Columbia River basalt subswarms, NE Oregon and W Idaho. Geological Society of America, Abstracts with Programs 33, 148.
Philpotts, A. R. & Asher, P. M. (1993). Wallrock melting and reaction effects along the Higganum diabase dike in Connecticut: contamination of a continental flood basalt feeder. Journal of Petrology 34, 1029–1058.
Rapp, R. P., Watson, E. B. & Miller, C. F. (1991). Partial melting of amphibolite/eclogite and the origin of Archean trondhjemites and tonalites. Precambrian Research 51, 1–25.[CrossRef][Web of Science]
Rushmer, T. (1991). Partial melting of two amphibolites: contrasting experimental results under fluid-absent conditions. Contributions to Mineralogy and Petrology 107, 41–59.[CrossRef][Web of Science]
Rushmer, T. (2001). Volume change during partial melting reactions: implications for melt extraction, melt geochemistry, and crustal rheology. Tectonophysics 342, 389–405.[CrossRef][Web of Science]
Rutter, M. J. & Wyllie, P. J. (1988). Melting of vapour-absent tonalite at 10 kbar to simulate dehydration-melting in the deep crust. Nature 331, 159–160.[CrossRef]
Singh, J. & Johannes, W. (1996a). Dehydration melting of tonalites. Part I: beginning of melting. Contributions to Mineralogy and Petrology 125, 16–25.[CrossRef][Web of Science]
Singh, J. & Johannes, W. (1996b). Dehydration melting of tonalites. Part II: composition of melts and solids. Contributions to Mineralogy and Petrology 125, 26–44.[CrossRef][Web of Science]
Skjerlie, K. P. & Johnston, A. D. (1992). Vapor-absent melting at 10 kbar of a biotite- and amphibole-bearing tonalitic gneiss: implications for the generation of A-type granites. Geology 20, 263–266.
Skjerlie, K. P. & Johnston, A. D. (1993). Fluid-absent melting behavior of an F-rich tonalitic gneiss at mid-crustal pressures: implications for the generation of anorogenic granites. Journal of Petrology 34, 781–815.
Skjerlie, K. P. & Johnston, A. D. (1996). Vapour-absent melting from 10 to 20 kbar of crustal rocks that contain multiple hydrous phases: implications of anatexis in the deep to very deep continental crust and active continental margins. Journal of Petrology 37, 661–691.
Stormer, J. C., Jr (1983). The effects of recalculation on estimates of temperature and oxygen fugacity from analyses of multicomponent iron–titanium oxides. American Mineralogist 68, 586–594.[Abstract]
Streck, M. J. (2002). Partial melting to produce high-silica rhyolites of a bimodal suite: compositional constraints among rhyolites, basalts, and metamorphic xenoliths from the Harney Basin, Oregon. International Journal of Earth Sciences DOI 10.1007/s00531-001-0246-7.
Taubeneck, W. H. (1970). Dikes of Columbia River Basalt in northeastern Oregon, western Idaho, and southeastern Washington. In: Gilmour, E. H. & Stradling, D. (eds) Proceedings of the Second Columbia River Basalt Symposium. Cheney, WA: Eastern Washington State College, pp. 73–96.
Taubeneck, W. H. (1995). A closer look at the Bald Mountain Batholith, Elkhorn Mountains, and some comparisons to the Wallowa Batholith, Wallowa Mountains, northeastern Oregon. In: Vallier, T. L. & Brooks, H. C. (eds) Geology of the Blue Mountains Region of Oregon, Idaho and Washington. US Geological Survey, Professional Papers 1438, 45–123.
Taubeneck, W. H. & Duncan, R. A. (1997). Preferential occurrence of eruptive axes of dikes of the Columbia River Basalt Group in unmetamorphosed Mesozoic granitic intrusives, with 40Ar/39Ar dates for four dikes and two of the earliest flows of basalt, Northeast Oregon and Western Idaho. Geological Society of America, Abstracts with Programs 29, 48.
Tommasini, S. & Davies, G. R. (1997). Isotope disequilibrium during anatexis: a case study of contact melting, Sierra Nevada, California. Earth and Planetary Science Letters 148, 273–285.[CrossRef][Web of Science]
Vielzeuf, D. & Holloway, J. R. (1988). Experimental determination of the fluid-absent melting relations in the pelitic system. Contributions to Mineralogy and Petrology 98, 257–276.[CrossRef][Web of Science]
Vielzeuf, D. & Montel, J. M. (1994). Partial melting of metagreywackes. Part I: Fluid-absent experiments and phase relationships. Contributions to Mineralogy and Petrology 117, 375–393.[CrossRef][Web of Science]
Wolf, M. B. & Wyllie, P. J. (1991). Dehydration-melting of solid amphibolite at 10 kbar: textural development, liquid interconnectivity and applications to the segregation of magmas. Mineralogy and Petrology 44, 151–179.[CrossRef][Web of Science]
Wolf, M. B. & Wyllie, P. J. (1994). Dehydration-melting of amphibolite at 10 kbar: the effects of temperature and time. Contributions to Mineralogy and Petrology 115, 369–383.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. G. Hersum, B. D. Marsh, and A. C. Simon Contact Partial Melting of Granitic Country Rock, Melt Segregation, and Re-injection as Dikes into Ferrar Dolerite Sills, McMurdo Dry Valleys, Antarctica J. Petrology, November 1, 2007; 48(11): 2125 - 2148. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. KURITANI, H. KITAGAWA, and E. NAKAMURA Assimilation and Fractional Crystallization Controlled by Transport Process of Crustal Melt: Implications from an Alkali Basalt-Dacite Suite from Rishiri Volcano, Japan J. Petrology, July 1, 2005; 46(7): 1421 - 1442. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||