Journal of Petrology Advance Access published online on October 9, 2007
Journal of Petrology, doi:10.1093/petrology/egm057
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Contrasting Cooling Rates in the Lower Oceanic Crust at Fast- and Slow-spreading Ridges Revealed by Geospeedometry
1School of Earth and Ocean Sciences, Petch Building, University of Victoria, PO Box 3055 STN CSC, V8W 3p6, Victoria, B.C., Canada
2Department of Geology, University of Leicester, University Road, Leicester LE1 7RH, UK
Received October 23, 2006; Revised typescript accepted September 6, 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONANALYTICAL TECHNIQUESGEOSPEEDOMETRY METHODSVARIATIONS IN COOLING RATE...DISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Two approaches to determining the high-temperature (
1000°C to 600°C) cooling rate of the lower oceanic crust and upper mantle are presented and critically evaluated. The first is based on the down-temperature diffusive exchange of Ca between olivine and clinopyroxene. The second, less well-constrained, approach is based on the down-temperature diffusive exchange of Mg and Fe between olivine and spinel. Cooling rates based on olivine–spinel geospeedometry are approximately an order of magnitude faster than those from Ca-in-olivine geospeedometry. In contrast, cooling rates derived from thermochronology and remanent magnetism are approximately an order of magnitude slower than those derived by Ca-in-olivine geospeedometry; this is probably because they record cooling at lower temperatures. Using the Ca-in-olivine geospeedometer, the cooling rate of samples from the lower oceanic crust and upper oceanic mantle formed in the Oman ophiolite and in the three main ocean basins has been determined. Samples from the lower oceanic crust formed at fast-spreading ridges show a large decrease in cooling rate between the top and base of the gabbroic section, with most of the variation occurring within the upper kilometre. This is consistent with vertical heat loss (within the crustal frame of reference) dominating the thermal evolution at fast-spreading ridges. Samples from Ocean Drilling Program Hole 735B, which formed at the slow-spreading Southwest Indian Ridge, show no variation in cooling rate over 1500 m depth range and cooled substantially faster than rocks from the deeper portion of the gabbros in the Oman ophiolite, where the change in cooling rate with depth is limited. These observations are consistent with heat loss from small plutons emplaced in cool lithosphere at the slow-spreading ridge. Alternatively, they could be explained by cooling through the Ca-in-olivine closure interval during uplift towards the surface. KEY WORDS: geospeedometry; lower oceanic crust; Hess Deep; Pito Deep; ODP Hole 735B; ODP Leg 153
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOSPEEDOMETRY METHODSVARIATIONS IN COOLING RATE...DISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
The rate of cooling of the lower oceanic crust depends on the interplay between heat supply through both magma intrusion and conduction of heat from the mantle and heat extraction via conduction and hydrothermal circulation. Hydration reactions may provide a further heat source but this is probably of relatively minor importance (e.g. Allen & Seyfried, 2004
). The conductive portion of this heat balance is reasonably well constrained from a theoretical standpoint (e.g. Sleep, 1975). However, the advective components (magmatic and hydrothermal) are poorly understood. We do not know the size of magma batches added to the lithosphere nor where they are emplaced. Furthermore, we do not know the size and distribution of fluid flow pathways nor the partitioning of the total heat flux between conductive and hydrothermal cooling.
One approach to understanding the cooling of the oceanic lithosphere is to use geospeedometry techniques to determine the cooling rate of samples of the oceanic lower crust and upper mantle. These techniques are based on modelling the extent of down-temperature diffusion of an element from one mineral into another as described below. For geospeedometry to constrain cooling rates accurately both the temperature dependence of the equilibrium partitioning of an element between the two phases and the diffusion coefficient of the element in each phase must be known (e.g. Lasaga, 1983). Using this approach Coogan et al. (2002) modelled the down-temperature diffusion of Ca from olivine into clinopyroxene to determine the relative cooling rates of a suite of gabbros from a section through the Oman ophiolite. However, absolute cooling rates could not be determined, because of uncertainty in the diffusion coefficient for Ca in olivine. Coogan et al. (2005) have re-determined this diffusion coefficient to allow more accurate determination of cooling rates using this approach. Here we provide further discussion of the applicability of this approach. We also discuss the exchange of Mg and Fe between olivine and spinel as an alternative geospeedometer that has previously been applied to the oceanic lithosphere (Ozawa, 1983, 1984). This approach can be applied in rocks in which the Ca-in-olivine may not be applicable (e.g. dunites) and comparison of cooling rates determined by independent methods provides checks on the methodologies.
We have applied these geospeedometry techniques to determine the cooling rate of samples of the oceanic lithosphere from the following five areas.
(1) Ocean Drilling Program (ODP) Hole 735B, which recovered 1500 m of gabbroic rocks formed at the Southwest Indian Ridge (SWIR; full spreading rate 14 mm/year; Dick et al., 2000; Hosford et al., 2003) and exposed at the seafloor at the Atlantis Bank area.
(2) ODP Holes 920D and 923A, which recovered short cores (
200 m) of mantle and plutonic rock, respectively, from the Mid-Atlantic Ridge south of the Kane fracture zone (the MARK area; full spreading rate 25 mm/year; Cannat et al., 1995). These lithologies have been tectonically emplaced on the seafloor in both locations. These sites are approximately 22 km apart within the same spreading segment.
(3) Hess Deep in the equatorial Pacific, where crust that formed at the East Pacific Rise (EPR) is rifted apart exposing the entire upper crust (lavas and dikes) and the upper part of the plutonic complex (full spreading rate 130 mm/year; Lonsdale, 1988; Karson et al., 2002). Additionally, drilling on a rift horst at Hess Deep by ODP Leg 147 recovered shallow-level gabbros, but because the base of the sheeted dike complex is not exposed here the depth in the lower crust could not be determined (Gillis et al., 1993).
(4) Pito Deep, in the southern Pacific, where crust that formed at the EPR is rifted apart exposing the entire upper crust (lavas and dikes) and the upper part of the plutonic complex (full spreading rate 140 mm/year; Francheteau et al., 1988; Perk et al., 2007).
(5) The Oman ophiolite, where a complete section through the gabbroic portion of the crust and a thick upper mantle section is exposed. The samples discussed here come from the Wadi Abyad and Wadi Bani Kharus sections in the Nakhl-Rustaq block of the ophiolite. Regional mapping and the dip of layering in the deepest gabbros provide constraints on the dip of the ophiolite section studied and thus the thickness of the units. This ophiolite is thought to have formed at an intermediate- to fast-spreading ridge.
These data provide insight into the relative cooling rates in different locations—both vertically within a single lithospheric section and between areas formed at different spreading rates. In turn, this provides important constraints on magmatic, tectonic and hydrothermal processes during lower crustal accretion at fast- and slow-spreading ridges.
|
ANALYTICAL TECHNIQUES |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOSPEEDOMETRY METHODSVARIATIONS IN COOLING RATE...DISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
All data were collected using a JEOL 8600 Superprobe at the University of Leicester. Analyses of Ca in olivine used the operating conditions described by Coogan et al. (2002
) and all other analyses were performed using standard analytical conditions. Because the olivine–spinel thermometry used here depends on determining the Fe3+ content of spinel from an electron probe analysis the accuracy of this approach was checked using spinel crystals previous analysed by Mössbauer spectroscopy (Canil et al., 1990). The Fe3+ content of spinel determined by electron probe analysis, and calculated following the method of Droop (1987), agrees well with the Fe3+ content determined by Mössbauer spectroscopy (Fig. 1).
|
|
GEOSPEEDOMETRY METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOSPEEDOMETRY METHODSVARIATIONS IN COOLING RATE...DISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
In a system in which the partitioning of elements between two phases is temperature dependent, the subsolidus distribution of the element in a slowly cooled rock can be used to calculate a ‘closure temperature’ for the effective cessation of diffusive exchange between the two phases if a suitable thermometer is available (Dodson, 1973
). Assuming that the phases maintain equilibrium during initial cooling the closure temperature is controlled by the down-temperature transition from kinetic processes (diffusion) being rapid enough to maintain equilibrium, to kinetics being too slow to significantly alter the distribution of the element (Fig. 2). The temperature interval between the lowest temperature at which crystals maintain complete equilibrium and the highest temperature at which no measurable diffusive composition change occurs is called the closure interval. For a given crystal size, the closure temperature decreases with decreasing cooling rate because slower cooling allows a greater time in any given temperature interval for equilibrium to be established. However, at any given cooling rate the closure temperature will decrease with decreasing grain size, as equilibrium can be achieved more rapidly over shorter distances. Likewise, the rim of a crystal will record a lower closure temperature than its core because of the shorter diffusion distance to the exchanging phase. Interpretation of the closure temperature of a system with more than two phases is more complex if the element of interest is exchanging between all of the phases (e.g. Jenkin et al., 1994). The following sections discuss the use of this approach to determining the cooling rate of samples from the oceanic lithosphere using the down-temperature exchange of Ca from olivine to clinopyroxene and Mg–Fe exchange between olivine and spinel.
|
The Ca-in-olivine geospeedometer
Following Coogan et al. (2002
) we determine the cooling rate of samples of the lower oceanic crust by measuring the partitioning of Ca between olivine and clinopyroxene and modelling this as the result of down-temperature diffusion of Ca from olivine into clinopyroxene. This approach provides insight into the cooling rate at temperatures between 600°C and 1000°C, with the exact temperature range dependent on the cooling rate. The general approach is briefly reviewed here but the reader is referred to Coogan et al. (2002) for details. The basic approach follows Dodson (1986), who provided a method to determine the zoning of an element in a phase after cooling for cases where the element's distribution between two phases is temperature dependent. We use the distribution of Ca between olivine and clinopyroxene, which is highly temperature sensitive, with higher concentrations of Ca in olivine at high temperature than at lower temperatures. The temperature dependence of the partition coefficient (kd) of Ca between these phases has been calibrated (Kohler & Brey, 1990) and we have recently undertaken experiments to determine the diffusion coefficient (D) of Ca in olivine (Coogan et al., 2005). Armed with these data, and Dodson's model, the zoning of Ca in olivine can be used to determine the cooling rate of the sample.
Modelling the zoning of Ca in olivine to determine the cooling rate of a sample requires making the assumption that the outer surface of the olivine maintains equilibrium with clinopyroxene and that Ca is transferred from the interior of the olivine to the surface by volume diffusion. This process results in Ca zoning in the olivine after cooling with lower Ca contents (i.e. lower closure temperatures) at the crystal rim as a result of the shorter diffusion distances to the grain boundary. Further, it must also be assumed that near the closure temperature the system can be treated as a binary exchange of Ca; that is, no exchange of Ca with other phases can occur and no net-transfer reactions can occur. These restrictions are likely to be met, even in the presence of plagioclase, because of the extremely slow diffusion of Ca in plagioclase (Grove et al., 1984). This kinetic barrier means that Ca in plagioclase cannot exchange diffusively with the adjacent olivine and clinopyroxene. An ExcelTM spreadsheet and macro to perform these calculations is available from L.A.C. or G.R.T.J. upon request. In the following sections we discuss some of the assumptions made in this modelling; further discussion has been provided by Coogan et al. (2002).
Temperature dependence of Ca partitioning
We use the calibration of Kohler & Brey (1990) for the temperature dependence of Ca partitioning between olivine and clinopyroxene. This thermobarometer is calibrated from 900 to 1400°C and from 2 to 60 kbar. The calibration is divided into high- and low-temperature regions with the cross-over at 1002°C at low pressures. The low-temperature thermometer reproduces the nine experiments used to calibrate this region to better than 22°C with a standard deviation of 13°C. Because we are interested in the subsolidus partitioning of Ca between olivine and clinopyroxene during cooling we use the lower temperature calibration throughout. This thermometer accounts for pressure explicitly but does not account for the major element composition of the phases in controlling the kd. Both olivine–melt Ca partition coefficients (Libourel, 1999) and phase relations in the forsterite–fayalite–monticellite–kischsteinite system (Davidson & Mukhopadhyay, 1984) suggest that Ca is more soluble in more fayalitic olivines. However, more fayalitic olivines coexist with more Fe-rich (hedenbergitic) clinopyroxene and there has been no experimental calibration of the change in Ca partitioning between olivine and clinopyroxene with changing Mg–Fe of the bulk system to our knowledge. Coogan et al. (2002, fig. 7) showed that there is no systematic difference in the cooling rate calculated from the Ca-in-olivine geospeedometer for closely spaced samples that have different olivine forsterite contents. This provides empirical evidence that, over the range of olivine compositions studied (Fo70–90), there is little or no change in the Ca partition coefficient with changing olivine forsterite content. Thus, in the absence of an experimental investigation of the role of the major element composition of the bulk system in controlling the partition coefficient, we use the calibration of Kohler & Brey (1990). We caution against the use of this geospeedometer in systems with highly variable Mg/Fe or in Fe-rich systems.
The diffusion coefficients
There have been a number of studies of the diffusion coefficient of Ca in olivine. Morioka (1981) measured the diffusion coefficient in pure (synthetic) forsterite in air. Because the formation of defects in olivine is dependent on oxidation of Fe (Nakamura & Schmalzried, 1983; Chakraborty, 1997; Dohmen & Chakraborty, 2007) these experiments did not produce a diffusion coefficient relevant to diffusion in olivine in the lower oceanic crust. Jurewicz & Watson (1988) measured the diffusion coefficient of Ca in natural olivine crystals by suspending them in a melt. More recently, Coogan et al. (2005) measured the diffusion coefficient over a wider range of temperatures than either of the previous studies (900–1500°C). The latter two studies found an activation energy for diffusion of 200 kJ/mol but Coogan et al. (2005) found a smaller pre-exponential factor than was determined by Jurewicz & Watson (1988). We favour the value determined by Coogan et al. (2005) because of the more controlled experimental design and much larger temperature range of the calibration. However, use of a different value for Do simply shifts all of the cooling rates and preserves the relative variations we observe.
Coogan et al. (2005) found no major dependence of the Ca diffusion coefficient on the olivine forsterite content, over a limited range (Fo83–92), or on the crystallographic orientation. Their diffusion coefficient for Ca along the c-axis of olivine is used throughout this study. This is
|
|
log fO2 = log[fO2*] – log[10–12], with fO2* being the fO2 conditions of interest in bars. It is clear from equation (1) that the diffusion coefficient for Ca in olivine is dependent on the oxygen fugacity. This is shown in Fig. 3, where the temperature dependence of the diffusion coefficient is compared at fixed fO2 and along the quartz–magnetite–fayalite (QM–F) oxygen fugacity buffer. Thus, before using Ca-in-olivine geospeedometry to determine the cooling rate we must consider the role of changing oxygen fugacity during cooling.
|
We can constrain the oxygen fugacity during crystallization using the partitioning of Fe between plagioclase and olivine following the method of Sugawara (2001
). Because Fe diffusion in plagioclase is likely to be slow (because it is dominantly Fe3+ and trivalent cations generally diffuse more slowly than divalent ones of similar size) this should record the Fe content of plagioclase at near-solidus temperatures and thus the fO2 conditions at these temperatures. This approach suggests crystallization close to the QM–F buffer (Fig. 4). This is also broadly consistent with the Fe3+/Fe2+ ratio in mid-ocean ridge basalt (Christie et al., 1986; Bezos & Humler, 2005). The redox path followed by an olivine crystal in an oceanic olivine gabbro during cooling is less readily constrained. In samples containing ilmenite and magnetite the ilmenite–magnetite coupled thermometer and oxygen barometer of Ghiorso & Sack (1991) gives redox conditions close to QM–F also (Fig. 4). However, this is a closure temperature and ‘closure fO2’, which are difficult to interpret confidently in terms of a T–fO2 path. Even if this faithfully records cooling of oxide-bearing gabbros along the QM–F buffer, it is unclear if this is the appropriate fO2 condition to assume in calculating the Ca diffusion coefficient in olivine, for at least three reasons. First, the equilibrium fO2 during cooling may be different in oxide-bearing and oxide-free samples. Second, the system may not remain in equilibrium during cooling. Third, the point defects controlling Ca diffusion may change with temperature from ones that are controlled by redox conditions to ones that are not (e.g. Chakraborty, 1997; Dohmen & Chakraborty, 2007). These issues are discussed in the following three paragraphs.
|
The equilibrium T–fO2 path during cooling of oxide-free gabbros (which most of the samples analysed are) is complicated by the absence of any phases in which Fe3+ is a major component. This prevents any simple buffering reaction involving exchange of Fe between a Fe3+-bearing phase and a Fe2+-bearing phase (e.g. the QM–F buffer). Instead, Fe3+-bearing components in the nominally Fe3+-free phases may control the fO2 or it may be externally buffered by gases (e.g. CO–CO2 equilibria) or internally buffered by sulphides (FeS–FeO equilibria). The inset in Fig. 4 shows a model for the evolution of the fO2 during cooling in an olivine gabbro with fixed bulk Fe3+ content based on the pMELTS model (Ghiorso et al., 2002
). This is shown only as an illustration because pMELTS does not account for Fe in plagioclase and because it is plausible that the real system may be open to oxygen. However, this model illustrates the difference in fO2 evolution in a (Fe3+-rich) spinel-bearing and spinel-free system. If spinel is present during cooling the fO2 follows a pathway approximately parallel to the QM–F buffer. If spinel is artificially suppressed in the same bulk composition, removing this sink for Fe3+, the fO2 path followed during cooling is substantially more oxidizing. Without more sophisticated solution models for Fe3+-bearing components in plagioclase and clinopyroxene we cannot accurately predict the equilibrium fO2 during cooling.
Because Fe3+ in clinopyroxene and plagioclase substitutes for Al and/or Si, Fe3+ reduction during cooling may require diffusion of one of these, slowly diffusing, species for charge balance. Thus, the kinetics of redox equilibria may be more sluggish in an oxide-free than oxide-bearing system. This raises doubts about whether the equilibrium fO2 pathway will be followed by the bulk system. Even in olivine, reduction of Fe3+ requires net-transfer reactions (Dohmen et al., 2003) that may inhibit olivine maintaining equilibrium with the bulk system at low temperatures.
The fO2 dependence of Ca diffusion in olivine determined by Coogan et al. (2005) is calibrated only for high temperatures and oxidizing conditions. Thus, we must consider whether this fO2 dependence can be extrapolated to lower temperatures and more reducing conditions. The data of Coogan et al. (2005, fig. 8) show an apparent decrease in the fO2 dependence of the diffusion coefficient with decreasing temperature. This would be expected if the point defects controlling Ca diffusion change with decreasing temperature from being ones created by redox reactions to ones created by non-redox related impurities within the olivine [i.e. a shift from transition metal extrinsic diffusion to pure extrinsic diffusion in the terminology of Chakraborty (1997)]. This transition has been observed to occur for Mg–Fe diffusion at 900°C by Dohmen & Chakraborty (2007) and is expected to occur for Ca diffusion too. Thus, irrespective of the T–fO2 path followed by the system, it is unclear whether it is appropriate to model diffusion of Ca at fixed fO2 or at decreasing fO2 with cooling.
In summary, because of the uncertainties discussed above we opt to model Ca diffusion at a fixed fO2 of 10–12 bars. This is between the fO2 at the solidus (around 10–8 to 10–10 bars based on olivine–plagioclase equilibria; Fig. 4) and the upper fO2 defined by ilmenite–magnetite equilibria at temperatures below the closure temperature of Ca exchange (10–15 to 10–16 bars).
The diffusion coefficient for Ca in clinopyroxene is expected to play little role in the exchange of Ca from olivine to clinopyroxene. This is because of both the much higher concentration of Ca in clinopyroxene and the larger amounts of clinopyroxene than olivine in these rocks, which allows the clinopyroxene to act as an almost infinite reservoir of Ca (e.g. Dodson, 1973, 1986; Jenkin et al., 1994). However, to test this assumption we have undertaken a series of numerical experiments (see ‘modelling approach’) that account explicitly for Ca diffusion in clinopyroxene as well as olivine and compare these with calculations using the Dodson model.
The role of deformation
Plutonic rocks from slow-spreading ridges commonly display a wide variety of crystal plastic deformation microstructures. The strain and recrystallization associated with this could lead to enhanced mobility of Ca within olivine, and thus equilibration of the olivine to lower temperatures, than would occur in an unstrained system. For example, recrystallization can reduce grain sizes decreasing diffusion distance, and produce more grain boundaries that can act as fast diffusion pathways. Subgrain walls may also act as fast diffusion pathways. Alternatively, lattice strain without recrystallization may change the diffusion rate through creating defects.
To examine the role of deformation we analysed detailed traverses across olivine crystals showing deformation bands and recrystallized textures (Fig. 5). There are systematic decreases in olivine Ca content toward both subgrain boundaries and deformation bands, demonstrating that these act as fast diffusion pathways allowing Ca to diffuse out of olivine into these regions. However, different boundaries behaved in different ways. For example, the drop in calcium content of the olivine at the subgrain boundaries at A, B and C in Fig. 5 is similar to that at the edge of the crystals and the zoning is also approximately symmetrical. These observations suggest that (1) diffusion along the subgrain boundaries at A, B and C was almost as rapid as diffusion along the olivine–plagioclase grain boundaries, and (2) these subgrain boundary existed at temperatures at, or above, the start of the Ca closure interval (
900°C). In contrast, the kink boundaries crossed at points D–F (Fig. 5b) have more limited drops in Ca adjacent to them. This indicates that either these were not as effective diffusion pathways as grain boundaries (perhaps because of smaller lattice mismatch) and/or they formed at lower temperatures within the Ca closure interval. The likelihood of complex three-dimensional cut effects makes testing between these models beyond the scope of this study. However, these data stress the importance of using only undeformed crystals for geospeedometric determinations of cooling rates using the Ca-in-olivine geospeedometer; this is the approach used here. These observations also suggest that the distribution of Ca in plastically deformed olivines could be used to investigate the temperatures at which different deformation mechanisms operated in natural olivine samples.
|
Modelling approach
We have used Dodson's (1973
, 1986) analytical model to determine the cooling rates of olivine–clinopyroxene-bearing rocks following Coogan et al. (2002). Using this model involves assuming that clinopyroxene acts as an infinite reservoir of Ca, and that the initial distribution of Ca prior to cooling is irrelevant to the final distribution (i.e. that complete equilibration is achieved during the initial stages of cooling). Coogan et al. (2002) evaluated these assumptions empirically, and concluded that they are reasonable approximations for the system being studied. Here we test these assumptions further by comparing the Dodson model with numerical models that explicitly account for: (1) diffusion of Ca in clinopyroxene; (2) the relative proportions of olivine and clinopyroxene; (3) the initial conditions.
For the numerical experiments we used an explicit finite-difference model (e.g. Crank, 1975) with a spherical olivine crystal with a diameter of 1 mm exchanging with clinopyroxene. The finite-difference models use the Ca diffusion coefficient in clinopyroxene from Dimanov & Jaoul (1998; E = 284 kJ/mol; Do = 5 x 10–11 m2/s) for temperatures <1230°C, which is similar to earlier measurements of this parameter. We used the partition coefficient from Kohler & Brey (1990), which we note does not strictly adhere to Dodson's (1973, 1986) assumption of a linear change in rim Ca content of olivine with time. Despite this, numerical models that approximate the assumption that clinopyroxene acts as an infinite reservoir of Ca are always within <2°C of the Dodson model for the core Tc and show identical curvature.
Figure 6a compares the closure profiles predicted by the numerical experiments with those predicted by the Dodson model for four cooling rates assuming a clinopyroxene-to-olivine ratio of five. This is typical for oceanic olivine gabbros (Coogan, 2007). The Dodson model underpredicts the core closure temperature by 15°C for all cooling rates. Thus, if the Dodson model is used to extract cooling rates from the distribution of Ca in olivine for samples that have a clinopyroxene-to-olivine ratio of five, the cooling rate will be overestimated by 0·1–0·2 log units; that is, a faster cooling rate will be necessary in the Dodson model to freeze in as much Ca as is observed. This difference, although non-trivial, is of the same order as expected from uncertainties in the shape of the olivine crystal and the cooling history and does not warrant using a different approach to calculating cooling rates. Numerical experiments show that even for equal proportions of olivine and clinopyroxene the bulk closure temperature is overpredicted by only 20°C (Fig. 6b).
|
The finite-difference models used to calculate the closure profiles shown in Fig. 6a were all calculated assuming equilibrium at an initial temperature of 1200°C. For the fastest cooling rates (100 000°C/year) this is within the closure interval over which Ca is frozen into olivine. Numerical experiments show that even at this rapid cooling rate there is <10°C difference in the core closure temperature across the starting temperature range 1150–1600°C. This maximum temperature is effectively infinite for this system. For starting temperatures <1150°C the core closure temperature is progressively lower by increasing amounts relative to that predicted by the Dodson model. This is due to the system not maintaining equilibrium even during the highest-temperature stage of cooling (e.g. Jenkin et al., 1994
). Instead, the low starting temperature is partially frozen into the olivine. If the Dodson model is used to extract cooling rates from the zoning of Ca in olivine for samples that have cooled at these fastest cooling rates, the cooling rates will be underestimated; that is, a slower cooling rate than the true cooling rate will be required in the Dodson model to allow the Ca content of the olivine to decrease to the measured values. For slower cooling rates the initial temperature is basically irrelevant provided that the olivine crystallized from a basaltic magma (e.g. <5°C decrease in core closure temperature when decreasing the starting temperature from 1200°C to 1050°C for a cooling rate of 10 000°C/year). In summary, although the assumptions in the Dodson model are not completely valid both empirical and numerical tests indicate that the uncertainties introduced by using this model are relatively small (Fig. 6). The various other unconstrained uncertainties inherent in extracting cooling rates from closure temperatures (for example, the three-dimensional shape of the olivine is never known and the details of the temperature–time path are unknown) mean that using a more sophisticated (and laborious) approach is unwarranted.
Error estimation
There are various ways in which the errors associated a geospeedometer could be evaluated. Quantifiable systematic inaccuracies in the Dodson (1973) model come principally from uncertainties in the partition coefficient (kd) for Ca between olivine and clinopyroxene, and in the diffusion coefficient (Do and E) of Ca in olivine. We have evaluated the potential inaccuracies in cooling rates arising from uncertainties in these parameters by performing Monte Carlo simulations as follows. Cooling rates for a given measured partition coefficient were calculated using the Dodson formulation 2000 times allowing kd, E and Do to vary within given bounds. The closure temperature calculated from a given kd is allowed to vary about the true value with a standard deviation of 20°C. This value was chosen to match the largest difference between the calculated and experimental temperature in the calibration experiments (Kohler & Brey, 1990) and is greater than the standard deviation of this misfit (13°C). The activation energy was allowed to vary around the measured value of 207 kJ/mol with a standard deviation of 8 kJ/mol [the uncertainty in the experimental calibration of Coogan et al. (2005)]. Because Do and E cannot vary independently and still fit the experimental data on Ca diffusion, Do was calculated as a function of E to best fit the experimental data of Coogan et al. (2005). The predicted uncertainty (1
) in the cooling rate based on these simulations is a function of cooling rate, being ±0·3 log units at a cooling rate of 30°C/Myr to ±0·2 log units at a cooling rate of 10 000°C/Myr. The decrease in uncertainty at higher closure temperatures is because these lie closer to the central point of the diffusion coefficient calibration where the uncertainties are smallest. It should be noted that these systematic errors are unimportant when comparing relative cooling rates.
Non-systematic inaccuracies come from uncertainties in the shape and size of the olivine crystals. These can be estimated based on the variation in cooling rate calculated from different crystals in the same sample and from perpendicular analytical traverses in the same crystal. As discussed below, these amount to approximately ±0·3 log units uncertainty in any single cooling rate determination.
The spinel–olivine Mg–Fe geospeedometer
During cooling of a rock containing olivine and spinel the partitioning of Mg and Fe between these phases changes significantly with decreasing temperature, providing a driving force for Mg to diffuse out of spinel into olivine and for Fe to do the opposite (e.g. Mori, 1977; Fabriès, 1979; Roeder et al., 1979; O’Neill & Wall, 1987; Ballhaus et al., 1991; Sack & Ghiorso, 1991). The closure temperature for this exchange in the oceanic crust and upper mantle is between 600°C and 900°C, with the exact temperature being dependent on the cooling rate. It should be noted that the lowest temperatures are still above the highest temperature at which serpentinization is expected to occur, which is consistent with the lack of any change in olivine composition towards serpentine veins. Ozawa (1983, 1984) presented a detailed analysis of the potential of Mg–Fe exchange between olivine and spinel as a geospeedometer. Here we follow his approach except that we have modified the model to include updated diffusion coefficients in both olivine and spinel and a more recent calibration of the temperature dependence of Mg–Fe exchange between these phases as described below. The following sections describe the input required for the olivine–spinel geospeedometer and discuss the assumptions that are made and their validity.
Temperature dependence of Mg–Fe partitioning between olivine and spinel
The olivine–spinel Mg–Fe exchange thermometer has been through numerous iterations (e.g. Mori, 1977; Fabriès, 1979; Roeder et al., 1979; O’Neill & Wall, 1987; Ballhaus et al., 1991; Sack & Ghiorso, 1991). There is considerable discrepancy between the temperatures calculated by these different calibrations (e.g. Kessel et al., 2007). We investigated the Ballhaus et al. (1991) and Sack & Ghiorso (1991) calibrations and found temperature differences of up to 150°C for some pairs of analyses. Based on the analysis of Kessel et al. (2007) we have used the thermometer of Sack & Ghiorso (1991) to determine the closure temperature of spinel–olivine Mg–Fe exchange here. However, the large discrepancy between different calibrations, and the inability of any calibration to reproduce all experimental temperatures accurately (e.g. Kessel et al., 2007), suggests that caution must be exercised in the use of olivine–spinel Mg–Fe thermometry.
The diffusion coefficients
We need to know the inter-diffusion coefficients for Mg–Fe in both olivine and spinel to model the cooling rate from the composition of these phases. For olivine we have used the (intrinsic) diffusion coefficient of Dohmen & Chakraborty (2007) that is consistent with, but more widely calibrated than, the diffusion coefficient of Chakraborty (1997). This is fO2 independent at the closure temperature for olivine–spinel Mg–Fe exchange in the samples that we studied. We have not included a change in olivine diffusion coefficient with changing composition during diffusion because the increased computational costs are not justified by the small changes in diffusion coefficient (e.g. at 800°C the diffusion coefficient for Fo90 = 1·1 x 10–20 m2/s and for Fo93 = 9·3 x 10–21 m2/s; this difference is equivalent to a 9°C change in temperature). Mg–Fe diffusion along the c-axis in olivine is approximately six times faster than along the a- and b-axes (Chakraborty, 1997; Dohmen & Chakraborty, 2007). Because we do not know the lattice orientations of the olivine crystals surrounding the spinel grains we modelled diffusion using a simple average diffusion coefficient.
The Mg–Fe inter-diffusion coefficient is less well known for spinel than for olivine. Recent experiments by Liermann & Ganguly (2002) probably provide the best constraints but are for almost Cr-free spinel. Those workers gave Mg–Fe diffusion rates 0·7–1·1 orders of magnitude faster in spinel than in olivine between 1000°C and 600°C. Both experimental (Freer & O’Reilly, 1980) and empirical (Ozawa, 1983) evidence suggests that Mg–Fe inter-diffusion is significantly slower in Cr-rich spinel than in aluminous spinel but there are no rigorous experimental studies in the Cr-bearing system. Liermann & Ganguly (2002) performed their diffusion experiments using a graphite–O2 buffer and did not investigate the effect of varying fO2 on the diffusion coefficient. To avoid introducing arbitrary compositional and redox corrections we use the diffusion coefficient of Liermann & Ganguly (2002) throughout. Although Liermann & Ganguly (2002) calculated separate diffusion coefficients for Fe and Mg these are within error of each other; based on this we use an average rather than calculating an inter-diffusion coefficient that changes with spinel Mg-number.
Both the olivine and spinel Mg–Fe diffusion coefficients are slightly larger than the olivine Ca diffusion coefficient. Based on this, the closure temperature for Mg–Fe exchange between olivine and spinel is expected to be slightly lower than for Ca exchange between olivine and clinopyroxene for a given cooling rate and grain size.
Net-transfer reactions
If olivine and spinel are the only phases in the rock, Mg–Fe exchange between them would appear to be a relatively simple system to model by assuming no significant transfer of Cr, Al, Fe3+ or Si. In contrast, in systems containing other phases (e.g. pyroxenes, plagioclase) multiple exchange and net-transfer reactions may occur simultaneously. Only if these reactions all cease prior to the upper temperature of the olivine–spinel Mg–Fe closure interval will a simple model of binary exchange be applicable.
Evidence that multiple reactions occur during cooling in pyroxene-bearing systems can be seen in harzburgite samples from the Oman ophiolite. Thermodynamic modelling of subsolidus phase relations using the MELTS algorithm (Ghiorso & Sack, 1995) predicts that during subsolidus cooling spinel should grow and that its Cr-number should decrease. This is consistent with the zoning in Cr and Al observed in orthopyroxene and spinel in harzburgites from the Oman ophiolite (Fig. 7). The Cr and Al contents of orthopyroxene decrease towards its boundary with a spinel crystal, consistent with loss of these elements to the growing spinel. The spinel crystal has a rim with lower Cr-number than the core, consistent with growth from Cr and Al derived from the orthopyroxene (which has a lower Cr/Al than the spinel) and with the predictions from MELTS for the Cr-number of spinel to decrease with decreasing temperature. It is not our aim here to fully unravel the multicomponent reactions involved in generating the zoning shown in Fig. 7 but simply to note that in pyroxene-bearing systems these kinds of reactions are both predicted and observed.
|
Spinel growth at the expense of Cr and Al from pyroxene may have important implications for the interpretations of spinel–olivine Mg–Fe exchange thermometry in harzburgites. Spinel growth will lead to an increase in the Mg-number of adjacent silicate phases as a result of the lower Mg-number of spinel compared with all silicate phases. In turn, this will induce zoning in the Mg-number of the silicate phases around the spinel that is unrelated to the Mg–Fe zoning induced by exchange reactions driven by cooling. Additionally, changing the Cr-number of the spinel, and probably the Fe3+ content too, will change the partition coefficient for Mg–Fe exchange between olivine and spinel. These effects will severely complicate the interpretation of Mg–Fe-based geospeedometry in these rocks. In an attempt to avoid complexities associated with multiple reactions we choose to avoid, as far as possible, complications related to other phases by analysing olivine–spinel pairs only where either the sample is a pyroxene-free dunite or the spinel is totally enclosed in olivine (at least in the plane of the thin section and where the relative grain sizes suggest that this is true in three dimensions).
Despite avoiding samples in which elements other than Mg and Fe might be expected to be exchanging between phases we still found evidence for net-transfer reactions complicating Mg–Fe exchange between olivine and spinel. Spinel crystals in these samples are weakly zoned in Cr-number (decreasing towards the rim) and larger spinel crystals generally have higher core Cr-numbers than smaller ones. We interpret these data as indicating that Cr and Al dissolved in olivine at high temperatures are extracted to form overgrowths on spinel during cooling. This is consistent with the much lower abundances of these elements in olivine in dunites than in olivine phenocrysts in mid-ocean ridge basalt (MORB). Despite the concentrations of both of these elements typically being between 100 and 1000 times higher in spinel than in olivine at solidus temperatures, the much larger amount of olivine in the system (dunites are typically 99% olivine and 1% spinel) means that this retrograde reaction is not insignificant for either the mode, or composition, of spinel. Spinel growth during cooling will lead to: (1) a moving boundary condition that cannot be incorporated into numerical models without knowledge of the timing of spinel growth; (2) increased Mg/Fe in the adjacent olivine as a result of the lower Mg/Fe of the growing spinel; (3) a decrease in spinel Cr-number, which will change the partition coefficient for Mg and Fe between olivine and spinel; (4) dilution of the Fe3+ content in the spinel, and hence a relative reduction in the system, if it is closed to O; in turn, this will change the partition coefficient for Mg and Fe between olivine and spinel. All of these will introduce inaccuracy into cooling rates calculated from the closure temperature of Mg–Fe exchange between olivine and spinel.
Further evidence for net-transfer reactions involving minor components comes from the zoning of Ca in olivine adjacent to an accessory clinopyroxene in a dunite (Fig. 8). The olivine surrounding this clinopyroxene shows a marked decrease in Ca content towards the clinopyroxene crystal as expected because of down temperature re-equilibration (Kohler & Brey, 1990; Coogan et al., 2002; see above). Uncertainty in the three-dimensional shape of this Ca-depleted zone prevents a complete evaluation of this. However, making the assumption that this Ca depletion is approximately spherical in three dimensions, and that the background CaO concentration has not changed (0·15 wt% CaO), the volume of clinopyroxene that could be formed from this Ca can be estimated; this is similar to the volume of clinopyroxene that is observed (Fig. 8). If the Ca content of the olivine was originally higher than the background value, as it would be if the dunite was originally in equilibrium with MORB as is widely believed (e.g. Kelemen et al., 1995), then substantially larger amounts of clinopyroxene would be formed by retrograde loss of Ca from olivine. For a system open to oxygen, this kind of ‘exsolution’ of Ca from olivine to form clinopyroxene (and spinel) can be described in terms of redox reactions such as
|
| (2) |
) and can occur in a system closed to oxygen; for example, |
|
|
Diffusion along grain boundaries
Before modelling Mg–Fe exchange between olivine and spinel we need to understand the diffusion pathway between the rims of olivine and spinel crystals. An approach to understanding this is to study the Mg–Fe zoning around a spinel–olivine grain boundary and across adjacent olivine–olivine grain boundaries (Fig. 9). If grain boundary diffusion is an efficient transport mechanism then similar lengths of diffusion profile should be observed in each case (Fig. 9d). Alternatively, if grain boundary diffusion is an inefficient transport mechanism then little or no Mg–Fe zoning will be observed at olivine–olivine grain boundaries (Fig. 9c). Unfortunately, none of the dunite samples that we have studied to determine cooling rates of the oceanic lithosphere are fresh enough to allow us to test for Mg–Fe zoning at olivine–olivine grain boundaries. Instead, we have studied a very fresh dunite from Dun Mountain (New Zealand). There is a smooth increase in olivine forsterite content approaching the olivine–spinel grain boundary (Fig. 9f) but little or no increase at adjacent olivine–olivine grain boundaries (Fig. 9g–i). The smooth increase in olivine forsterite content approaching the spinel crystal suggests diffusive exchange between these phases where they are in contact. In contrast, the lack of an increase in olivine forsterite content at the olivine–olivine grain boundaries implies that these are not efficient pathways for Mg–Fe exchange between olivine and spinel. As grain boundaries are likely to be more efficient transport pathways than subgrain boundaries it seems unlikely that these act as fast-diffusion pathways either.
|
Modelling approach
Three uncertainties in using olivine–spinel Mg–Fe exchange as a geospeedometer have been discussed above: (1) the calibration of the olivine–spinel Mg–Fe exchange thermometer; (2) the Mg–Fe diffusion coefficient in Cr-rich spinel; (3) the role of net-transfer reactions. Because the significance of these problems to calculating cooling rates using this geospeedometer is uncertain, and because relative cooling rates should be less affected than absolute cooling rates, we have applied the geospeedometer to a series of dunites from the Oman ophiolite as an empirical test of this approach. Based on the lack of evidence for large Mg–Fe fluxes along grain boundaries, and the sub-equant shape of spinel crystals, we follow Ozawa (1984
) in modelling olivine as a spherical shell around a spherical spinel crystal. Cation exchange is permitted only along their mutual grain boundary. We analysed spinel crystals that were as far from any other spinel crystals as possible to avoid complications related to overlapping diffusion haloes and thus we assume an infinite volume of olivine around the spinel. We use an explicit finite-difference model (e.g. Crank, 1975) in which the cooling rate is constant and that utilizes the diffusion coefficients and thermometer discussed above.
Because the changes in olivine forsterite content around a spinel are typically small, and may be modified by net-transfer reactions, modelling zoning profiles is not an efficient way to extract cooling rates from olivine–spinel Mg–Fe exchange (see also Ozawa, 1984). Instead, following Ozawa (1984), we have modelled the closure temperature of the core of spinel crystals of different sizes, for different cooling rates, and compare these with measured core closure temperatures for spinel crystals of different sizes.
Figure 10 shows apparent size vs measured core closure temperature for spinel crystals from three dunite samples from the Oman ophiolite. For each sample the change in Mg–Fe partitioning between olivine and spinel with changing temperature was parameterized based on the average spinel and olivine compositions using the Sack & Ghiorso (1991) thermometer assuming only Mg–Fe exchange with the coexisting olivine during cooling. Because the change in Mg–Fe partition coefficient with cooling is very dependent on the Cr-number of the spinel, different samples show somewhat different closure temperatures for a given cooling rate and spinel size (Fig. 10). The size of the crystals was determined using backscattered electron imaging of the samples under the electron microprobe and is thus a minimum crystal diameter. Because of this we consider the maximum crystal size measured for any given core closure temperature to be the closest to the true size at that closure temperature. The data for spinel in dunites from the Oman ophiolite define curvilinear correlations between size and closure temperature as expected for diffusive exchange during cooling because smaller crystals can continue to equilibrate to lower temperatures. The cooling rates determined using this approach are fastest near the sole (2001OL44), where early obduction (Hacker, 1994) probably led to rapid cooling, slowest in the middle of the mantle section (2002K4), and intermediate at the top of the mantle section (2001OL33).
|
Figure 10 shows that the closure temperature decreases more slowly with decreasing size than predicted by the models. There are several possible explanations for this including: (1) an increase in cooling rate with decreasing temperature; (2) inaccuracies in the diffusion and/or partition coefficients; (3) deformation of the spinel meaning that apparent single grains are composed of multiple sub-grains. The current data do not allow us to unambiguously distinguish between these alternatives.
In addition to the aforementioned dunites we studied a troctolite from the lower crust at Hess Deep. In this sample small spinel crystals occur enclosed in both olivine and plagioclase. A single spinel enclosed in an olivine crystal is much more Fe-rich (Mg-number 50) than spinel of a similar size, and of the same Cr-number, enclosed in plagioclase (Mg-number 65). This is consistent with down-temperature diffusive exchange of Mg–Fe between the spinel and olivine. Based on the closure temperature of the crystal enclosed in olivine (765°C) a cooling rate of 7000°C/Myr was determined.
A comparison of cooling rates calculated using different approaches
Cooling rates determined from Ca-in-olivine geospeedometry are tabulated in Electronic Appendix 1 (available for downloading from http://www.petrology.oxfordjournals.org) and are compared with rates determined using other approaches in Fig. 11. A critical factor in this kind of comparison is the closure interval for the systems being considered. If the cooling rate varies significantly with temperature then comparing the cooling rate recorded in different closure intervals is not a meaningful way to evaluate different approaches.
|
Closure temperatures calculated from Ca-in-olivine geothermometry and olivine–spinel geothermometry are generally similar; thus, we expect these systems to provide similar cooling rates. Figure 11 shows that the cooling rates calculated from the olivine–spinel geospeedometer are consistently faster than those from the Ca-in-olivine geospeedometer for samples expected to have cooled at a similar rate (i.e. collected near one another). Based on the discussion above the most likely problems are the uncertainties in the following: (1) the calibration of the olivine–spinel Mg–Fe exchange thermometer; (2) the Mg–Fe inter-diffusion coefficient in Cr-rich spinel; (3) the impact of net-transfer reactions on the olivine–spinel Mg–Fe geospeedometer. Irrespective of these or other uncertainties, relative cooling rates determined by either approach should be reliable unless the spinel shows a very broad range of composition that will affect both the thermometer and diffusion coefficient.
Cooling rates determined from thermochronology for ODP Hole 735B (Southwest Indian Ridge; John et al., 2004) are based on the difference in the closure age of zircon (Tc 850°C) and biotite (Tc 350°C). This gives an average cooling rate over a temperature interval 200°C lower than the closure interval for Ca-in-olivine geospeedometry in samples from ODP Hole 735B. Cooling rates calculated by thermochronology are, on average, 0·5 log units slower than those derived from Ca-in-olivine geospeedometry (Fig. 11). Both the lower temperature interval for which the cooling rates are determined by thermochronology and uncertainties in the appropriate closure temperatures to use in thermochronology could explain the difference between these estimates.
Cooling rates for ODP Hole 923A (Mid-Atlantic Ridge) have been determined based on remanent magnetism. These cooling rates are based on the magnetic signals of the Jaramillo (1· 07–0·99 Ma), Matuyama (0·99–0·78 Ma), and Brunhes (0·78-present) chrons in some rocks, which suggest temperatures of 500°C at 0·99 Ma and 390–450°C at 0·78 Ma (Gee & Meurer, 2002). These temperatures are substantially lower than the closure temperature for Ca-in-olivine in samples from this drill hole (800–900°C). Thus remanent magnetism studies are expected to give slower cooling rates than Ca-in-olivine geospeedometry assuming a simple decrease in cooling rate with decreasing temperature. This is what is observed; cooling rate estimates based on remanent magnetism are slower than those based on Ca-in-olivine by 1 log unit (Fig. 11; Gee & Meurer, 2002).
In summary, cooling rates determined from modelling the Ca content of olivine in equilibrium with clinopyroxene are consistent with independent estimates based on olivine–spinel Mg–Fe geospeedometry, thermochronology and remanent magnetism to within approximately one order of magnitude. Much of the observed differences may be due to the lower temperature at which cooling rates are recorded by thermochronology and remanent magnetism. Faster cooling rates calculated from olivine–spinel Mg–Fe geospeedometry may reflect the uncertainties in this thermometer and the diffusion coefficient in Cr-rich spinel.
|
VARIATIONS IN COOLING RATE WITHIN THE LOWER OCEANIC CRUST |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOSPEEDOMETRY METHODSVARIATIONS IN COOLING RATE...DISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
We have studied an extensive suite of samples from the oceanic lithosphere to determine the variation in cooling history both with position in the lithosphere and with spreading rate (Fig. 12). By far our largest dataset is for cooling rates determined based on Ca-in-olivine geospeedometry and thus we focus on this approach in the following discussion. As discussed above, using one geospeedometer for all rate determinations ensures that relative variations in cooling rate are robust. As shown in the previous section, the accuracy of this approach is probably better than ±1 order of magnitude. The precision of any single cooling rate estimate can be estimated from analyses of multiple olivine crystals in the same sample, analysing perpendicular profiles through the same crystal, and analysis of olivine in closely spaced samples assumed to have cooled at the same rate. Based on these three approaches (Fig. 12) the precision of cooling rates can be conservatively estimated to be ±0·3 log units.
|
Fast-spreading ridges
Figure 12a compares the cooling rates derived by Ca-in-olivine geospeedometry for samples from the Oman ophiolite and samples formed at the East Pacific Rise (i.e. samples believed to have formed at fast-spreading ridges). Samples from the Wadi Abyad section of the Oman ophiolite show a progressive decrease in cooling rate with depth in the crust as previously reported by Coogan et al. (2002
). Gabbroic samples collected within the mantle section in the Oman ophiolite cooled at approximately the same rate as those from the base of the crust.
Samples from the plutonic section of East Pacific Rise crust exposed at Hess Deep and Pito Deep are compared with the more extensive sample suite from the Oman ophiolite in Fig. 12a. The two samples from the upper 600 m of the plutonic section exposed at Pito Deep have cooling rates identical to those of samples from the same depth in the Oman ophiolite. The samples from Hess Deep appear to have cooled a little more rapidly than those from either the Oman ophiolite or Pito Deep. It should be noted that the sample shown at 2000 m depth below the sheeted dyke complex (NZ 9-6; Hekinian et al., 1993) was collected from talus on the slope of the intra-rift ridge—thus its precise depth of formation is unknown. Because of this, and the fact that a small decrease in the sampling depth of sample 2218-1221 (at 400 m depth below sheeted dykes) would remove the differences between these areas, we do not stress this difference in cooling rate between Hess Deep and the other areas. A single sample from ODP Hole 894G on the intra-rift ridge at Hess Deep (Gillis et al., 1993) cooled as fast as the uppermost gabbros in the Oman ophiolite, suggesting that the gabbros recovered from this location are very shallow-level gabbros.
In summary, the data suggest that there is a systematic decrease in cooling rate, over the Ca-in-olivine closure interval, of two to three orders of magnitude between the top and base of the lower crust at fast-spreading ridges. Any models of crustal accretion and hydrothermal circulation should reproduce this variation.
Slow-spreading ridges
Cooling rates have been determined for core from drill holes from two areas at slow-spreading ridges (Fig. 12b and c; note the different vertical scales). ODP Hole 735B recovered 1500 m of gabbro formed at the slow-spreading Southwest Indian Ridge (14 mm/year full rate). Remanent magnetism in this drill core suggests 19° of tectonic tilting (Dick et al., 2000); thus this core represents 1420 m of paleo-vertical crust. This small difference is not accounted for in our figures to allow direct comparison with published data, which are all in terms of depth in the hole. ODP Holes 920D and 923A were drilled 30 km apart just south of the Kane fracture zone in the Atlantic (25 mm/year full rate; 23°N; MARK area) and recovered mantle peridotite and gabbro, respectively. The gabbroic samples from the crustal sections in both areas cooled at a restricted range of rates (1000–10 000°C/Myr), possibly suggesting that this cooling rate is characteristic of gabbroic intrusions formed at slow-spreading ridges. Surprisingly, there is no variation in cooling rate with depth in the deep (1500 m) drill core from ODP Hole 735B. This contrasts with samples from faster-spreading ridges. which, at depths where the cooling rate is as high as in the slower-spreading setting (i.e. in the upper 1000 m of the plutonic section), show a significant decrease in cooling rate with depth. Another notable feature is that two lithologies that are interleaved in ODP Hole 923A, compositionally primitive troctolitic gabbros and more evolved gabbros, apparently cooled at the same rate.
Two ultramafic samples recovered from ODP Hole 920D, which was drilled into a mantle peridotite massif south of Hole 923A, were also analysed to determine their cooling rate. One sample, 920D 15R3 18–21, is a pyroxenite vein that contains olivine. The other sample, 920D 22R7 22–29, is a lherzolite. Cooling rates from these samples are similar to each other, suggesting that they record the cooling rate for this peridotite massif. The mantle samples record a slightly slower cooling rate than any of the crustal samples; the implications of this are discussed below.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOSPEEDOMETRY METHODSVARIATIONS IN COOLING RATE...DISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Using geospeedometry to understand the cooling of the oceanic lithosphere
Perhaps the most important conclusion of the above sections is that, although there are complexities, useful insights into the cooling history of the oceanic lithosphere (and other mafic bodies) can be achieved using geospeedometry. This is very clear from the differences in cooling rates for different bodies shown in Fig. 12. Future studies that use this approach, alongside thermochronology (e.g. John et al., 2004
), should rapidly enhance our understanding of the interplay between magmatism, tectonism and hydrothermal circulation at mid-ocean ridges. Future development of geospeedometers that are independent of, or only weakly dependent on, the redox conditions [e.g. diffusion of Mg out of plagioclase into the adjacent mafic phases during cooling; see Coogan (2007, fig. 10f)] will provide even better constraints on the cooling history of the oceanic lithosphere. Existing methods provide important constraints on the relative cooling rates of samples both from a given body and between different bodies and these are emphasized in the following discussion. We stress that relative cooling rates derived from even a poorly calibrated geospeedometer can provide useful information about geological processes provided that the exchange process is well understood. We also note that all studies that use the compositions of minerals in slowly cooled rocks from the oceanic lower crust and mantle must consider the role of subsolidus exchange and not simply assume that these record primary compositions.
Cooling rates at fast-spreading ridges
The simplest possible model that could potentially explain the smooth decrease in cooling rate with depth in the lower crust for samples from fast-spreading ridges (Fig. 12a) is the conductive cooling of a half-space. In this model the upper crust would be cooled by hydrothermal circulation and the lower crust would cool conductively as it moved off-axis. In the reference frame of a single column of crust, this can be modelled as one-dimensional vertical conductive cooling of a semi-infinite body with a fixed surface temperature and an initial constant temperature. Here we assume that the top of the lower crust is held at 400°C and that initially the entire crust and upper mantle is at 1300°C. We model the cooling of this half-space using equation 4.124 of Turcotte & Schubert (2002) for a cooling interval of 100°C centred on the approximate closure temperature for a given depth.
Despite the physical unlikelihood of such a simple model, the conductive cooling of a half-space model (Fig. 13a) in fact yields overall variations in cooling rate with depth very similar to those observed in the data. A detailed thermal model that assumes a gabbro glacier mode of lower crustal accretion, and conductive cooling of the lower crust, also fits the data reasonably well (Maclennan et al., 2005). In contrast, thermal models that include hydrothermal circulation in the lower crust (Maclennan et al., 2005), which are consistent with either the gabbro glacier style of crustal accretion (Quick & Denlinger, 1993) or a hybrid of crystal subsidence and crystallization in sills (Boudier et al., 1996), do not show as much variation in cooling rate with depth and are offset to faster cooling rates (Fig. 13a). The slightly slower cooling rates calculated from the Ca-in-olivine geospeedometer than for cooling a half-space may either reflect inaccuracy in the geospeedometer or indicate a different thermal history from that modelled. For example, convection within the mantle may reduce the cooling rate at depth; if convection maintains a temperature of 1300°C at 10 km depth to some distance off-axis then a linear geotherm would hold the base of the crust close to the Ca-in-olivine closure temperature.
|
How could the lower oceanic crust cool at a similar rate to that predicted by such a simple model? One explanation would be if the lower oceanic crust does cool conductively with hydrothermal circulation being largely restricted to the upper crust. This would require the latent heat of crystallization to be largely removed from the axial magma chamber at the base of the sheeted dyke complex (top of the lower crust) because most of the lower crust crystallizes at the axis (e.g. Dunn et al., 2000
; Crawford & Webb, 2002). As noted above, a detailed thermal model of this process from Maclennan et al. (2005) fits the data reasonably well (Fig. 13a). Alternatively, some heat may be extracted from the lower crust by hydrothermal circulation but then the lower crust may be reheated after metasomatic minerals clog the permeability (e.g. Cochran & Buck, 2001; Fisher, 2003). Meaningful modelling of such ‘thermal rebound’ is not straightforward because neither the thickness of the hydrothermally cooled layer nor its temperature structure at the start of thermal rebound is known a priori. Models do show that the lower part of the crust could be reheated to temperatures greater than the Ca-in-olivine closure temperature and thus record slow cooling as the conductive layer gradually thickens (e.g. Cochran & Buck, 2001).
The very rapid cooling rates in the uppermost gabbros in all three areas studied have important implications for where latent heat is likely to be extracted from the lower crust. These rapid cooling rates are consistent with models of efficient hydrothermal cooling of the top of the plutonic section (Cann et al., 1985; Henstock et al., 1993; Phipps Morgan & Chen, 1993; Quick & Denlinger, 1993). Seismic experiments consistently observe a reflector, interpreted to be the top of a magma sill, at this level in the crust (e.g. Detrick et al., 1987). This is underlain by a crystal mush zone; that is, a region from which the latent heat of crystallization has been largely removed. Maintaining a magma sill at this level, despite the rapid cooling rates, requires regular replenishment of this body and that the crystals formed are removed, presumably downwards, to form at least a portion of the underlying mush zone (Fig. 13a; Henstock et al., 1993; Phipps Morgan & Chen, 1993; Quick & Denlinger, 1993; Coogan et al., 2002; Maclennan et al., 2005). This does not, however, discount a portion of the lower crust crystallizing in situ at depth at fast-spreading ridges, as has been suggested (Reuber, 1990; Bedard, 1991; Boudier et al., 1996; Kelemen et al., 1997).
Cooling rates at slow-spreading ridges
Three simple thermal models are compared with the cooling rates obtained from ODP Hole 735B in Fig. 13b: (1) cooling as a half-space; (2) cooling of plutons emplaced into thick, constant temperature, lithosphere; (3) cooling of the lithosphere as it is uplifted beneath the ridge axis.
Cooling as a half-space predicts systematic decreases in cooling rate with depth (Fig. 13b) that are not observed. Instead, the cooling rates at ODP Hole 735B are high and constant. The half-space model is characterized by smaller decreases in cooling rate with depth at deeper levels. However, if the rocks from Hole 735B had originally formed deep enough for the decrease in cooling rate with depth not to be resolvable (equivalent to deeper than 1500 m below the base of the sheeted dyke complex in Oman; Fig. 12a) then they would have cooled more slowly than the rocks from the base of the crust in the Oman ophiolite, which they did not; this model is thus rejected. If the lower crust was cooled rapidly then reheated to temperatures near the Ca-in-olivine closure temperature, as discussed above for fast-spreading ridges, this would also lead to decreasing cooling rates with depth.
Plutons emplaced into a constant temperature lithosphere may show little variation in cooling rate with depth if they cool conductively. An example of this scenario is shown in Fig. 13b, where a 1 km high sill is assumed to have been emplaced into wall-rock of constant temperature (500°C) and cooled by one-dimensional conduction. A large suite of possible body sizes, shapes and wall-rock temperatures would also fit the observed cooling rates and no meaning should be read into the specific conditions of this model. The model simply shows that plutons emplaced into approximately constant temperature lithosphere could produce the observed cooling rates. If hydrothermal circulation extracted heat from the roof of the magma body more efficiently than from its base cooling rates would decrease with depth. Likewise, if there was a significant thermal gradient in the wall-rocks prior to emplacement of the magma then cooling rates would be expected to decrease with depth. Irrespective of these complications, cooling of sills emplaced into thick, cool, lithosphere provides a plausible explanation for the observed cooling rates.
A final model that we consider is that the rocks recovered from ODP Hole 735B may have crystallized within the lithosphere at depths above the closure temperature for Ca in olivine and then cooled as the gabbros were transported up towards their present location at the seafloor. In this scenario the cooling rate would be expected to be approximately constant at all depths, as it is simply controlled by the uplift rate. Cooling rates for the temperature interval between 900°C and 800°C are shown for different uplift rates in Fig. 13b assuming steady-state one-dimensional advection and conduction [equation 7 of Carslaw & Jaeger (1959), ignoring heat production]. Although this model is simplified, the modelled cooling rates are similar to those observed, providing another plausible explanation for the data.
Further insight into lower crustal accretion at slow-spreading ridges is provided by the short drill core from ODP Hole 923A from the Mid-Atlantic Ridge (Fig. 12c). This core is made up of two interlayered lithologies (primitive troctolitic gabbros and more evolved, olivine-poor gabbros). These have been interpreted in two ways. One model suggests that they result from the co-crystallization of two distinct magmas, possibly with the primitive lithologies crystallizing from sills injected into a crystal mush that crystallized from the more evolved one (e.g. Ross & Elthon, 1997; Coogan et al., 2000). Alternatively, it has been suggested that the drill core is made up of sills emplaced over an extended time interval, based on the observation that different lithologies record different magnetic signatures (Gee & Meurer, 2002; Meurer & Gee, 2002; see the section ‘A comparison of cooling rates calculated using different approaches’).
The primitive troctolitic gabbros studied here occur as <5 m thick bodies that if emplaced into cool wall-rocks would have cooled very rapidly. For example, if the wall-rocks were at 500°C (they could not be much hotter and still preserve a remanent magnetic signature), and these represent sills, then purely conductive cooling would lead to cooling rates over the closure interval for Ca-in-olivine >4 orders of magnitude greater than those observed. The lack of any systematic difference in cooling rate between these lithologies, and the relatively slow cooling rates (Fig. 12c), suggests that irrespective of how these rocks were accreted they were all emplaced above the closure temperature for Ca-in-olivine (800–900°C in the rocks). This is much higher than the Curie temperature, suggesting that the differences in remanent magnetism are not due to sill emplacement over an extended period of time. Instead, the remanent magnetism is probably controlled by the formation of secondary magnetite during slow cooling (Gee et al., 1997). This is supported by the observation that the magnetic signal is carried by very pure magnetite (Gee & Meurer, 2002) in contrast to the Ti–Al-bearing magmatic magnetite in these rocks. In summary, the rocks from Hole 923A show no evidence of protracted accretion over 0·2 Myr; instead, these rocks appear to have solidified and cooled together.
Finally, two clinopyroxene-rich samples from the mantle section drilled in the MARK area at ODP Hole 920D were studied to compare the cooling rates in the mantle with those in the adjacent crust. These mantle samples both record slightly (3–4 times) slower cooling rates than the gabbroic samples from the same ridge segment. There is no significant difference in the cooling rate recorded by the pyroxenite vein and the lherzolite, and both contain olivine Fo90, suggesting that the cooling rate is representative of the mantle as a whole, not simply of late intrusions into it. The measured cooling rates of 500–1000°C/Myr are consistent with cooling rates expected from uplift of the mantle to the seafloor at slow-spreading rates (Fig. 13b).
Assuming that the process of upwelling and crustal accretion is generally similar along the segment that contains both Holes 923A and 920D, the difference in cooling rate between the mantle and crustal rocks is noteworthy. This suggests that the crustal rocks cannot have crystallized within the mantle at temperatures above the closure temperature for Ca-in-olivine. Although the geographical separation of these sampling localities prevents firm conclusions being drawn, if this difference in cooling rate is confirmed in areas where gabbro bodies occur encased in mantle rocks it will provide important constraints on the depth of crystallization at slow-spreading ridges.
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOSPEEDOMETRY METHODSVARIATIONS IN COOLING RATE...DISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
ACKNOWLEDGEMENTS |
|---|
Journal reviews by John Maclennan and two anonymous reviewers are gratefully acknowledged, as is input from editor Colin Devey. Discussions with, and comments from, a number of other people have aided this study at various stages. These include Graham Baines, Dante Canil, Kathy Gillis, Peter Kelemen and Kaz Ozawa. John Maclennan and William Wilcock are thanked for making us think harder about reheating in the off-axis. Janis Lloyd provided valuable assistance with the study of the cooling rates of samples from Pito Deep. Dante Canil, Roger Hekinian, Jeff Karson, Emily Klein and Yaoling Niu loaned us either samples or standards, for which they are thanked. Khalil Al-Rashdi is thanked for collecting one of the dunite samples from the Oman ophiolite. Sumit Chakraborty's advice at numerous stages in this project proved invaluable and is greatly appreciated. Financial support came from an NERC Fellowship (NER/I/S/2001/00774) and an NSERC Discovery Grant, both to L.A.C.
*Corresponding author. Telephone: (1) 250 472 4018. E-mail: lacoogan{at}uvic.ca
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOSPEEDOMETRY METHODSVARIATIONS IN COOLING RATE...DISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Allen DE, Seyfried W. E. Jr. Serpentinization and heat generation: Constraints from Lost City and Rainbow hydrothermal systems. Geochimica et Cosmochimica Acta (2004) 68:1347–1354.[CrossRef][Web of Science]
Ashworth JR, Chambers AD. Symplectic reaction in olivine and the controls of intergrowth spacing in symplectite. Journal of Petrology (2000) 41(2):285–304.
Ballhaus C, Berry RF, Green DH. High pressure experimental callibration of the olivine–orthopyroxene–spinel oxygen barometer: implications for the oxidation state of the upper mantle. Contributions to Mineralogy and Petrology (1991) 107:27–40.[CrossRef][Web of Science]
Bedard JH. Cumulate recycling and crustal evolution in the Bay of Islands ophiolite. Journal of Geology (1991) 99:225–249.[Web of Science]
Bezos A, Humler E. The Fe3+/Fe ratios of MORB glasses and their implications for mantle melting. Geochimica et Cosmochimica Acta (2005) 69(3):711–725.[CrossRef][Web of Science]
Boudier F, Nicolas A, Ildefonse B. Magma chambers in the Oman ophiolite: fed from the top and the bottom. Earth and Planetary Science Letters (1996) 144:239–250.[CrossRef][Web of Science]
Canil D, Virgo D, Scarfe CM. Oxidation state of mantle xenoliths from British Columbia, Canada. Contributions to Mineralogy and Petrology (1990) 104:453–462.[CrossRef][Web of Science]
Cann JR, Strens MR, Rice A. A simple magma-driven thermal balance model for the formation of volcanogenic massive sulphides. Earth and Planetary Science Letters (1985) 76:123–134.[CrossRef][Web of Science]
Cannat M, Karson JA, Miller DJ, eds. Proceedings of the Ocean Drilling Program, Initial Reports, 153 (1995) College Station, TX: Ocean Drilling Program.
Carslaw HS, Jaeger JC. Conduction of Heat in Solids (1959) Oxford: Oxford Science Publications.
Chakraborty S. Rates and mechanisms of Fe–Mg interdiffusion in olivine at 980–1300°C. Journal of Geophysical Research (1997) 102(B6):12317–12331.[CrossRef]
Christie DM, Carmichael ISE, Langmuir CH. Oxidation states of mid-ocean ridge basalt glasses. Earth and Planetary Science Letters (1986) 79:397–411.[CrossRef][Web of Science]
Cochran JR, Buck WR. Near-axis subsidence rates, hydrothermal circulation, and thermal structure of mid-ocean ridge crests. Journal of Geophysical Research (2001) 106:19233–19258.[CrossRef]
Coogan LA. The lower oceanic crust. In: Treatise on Geochemistry—Turekian K, Holland HD, eds. (2007) Amsterdam: Elsevier.
Coogan LA, Saunders AD, Kempton PD, Norry MJ. Evidence from oceanic gabbros for porous melt migration within a crystal mush beneath the Mid-Atlantic Ridge. Geophysics, Geochemistry, Geosystems (2000) 1. Paper No. 2000GC000072.
Coogan LA, Jenkin GRT, Wilson RN. Constraining the cooling rate of the lower oceanic crust: a new approach applied to the Oman ophiolite. Earth and Planetary Science Letters (2002) 199:127–146.[CrossRef][Web of Science]
Coogan LA, Hain A, Stahl S, Chakraborty S. Experimental determination of the diffusion coefficient for calcium in olivine between 900°C and 1500°C. Geochimica et Cosmochimica Acta (2005) 69(14):3683–3694.[CrossRef][Web of Science]
Coogan LA, Howard KA, Gillis KM, Bickle MJ, Chapman HJ, Boyce AJ, Jenkin GRT, Wilson RN. Chemical and thermal constraints on focused fluid flow in the lower oceanic crust. American Journal of Science (2006) 306:389–427.
Crank J. The Mathematics of Diffusion (1975) Oxford: Oxford Scientific Publications.
Crawford WC, Webb SC. Variation in the distribution of magma in the lower crust and at the Moho beneath the East Pacific Rise at 9°–10°N. Earth and Planetary Science Letters (2002) 203:117–130.[CrossRef][Web of Science]
Davidson PM, Mukhopadhyay DK. Ca–Fe–Mg olivines: phase relations and a solution model. Contributions to Mineralogy and Petrology (1984) 86:256–263.[CrossRef][Web of Science]
Detrick RS, Buhl P, Vera E, Mutter J, Orcutt J, Madsen J, Brocher T. Multi-channel seismic imaging of a crustal magma chamber along the East Pacific Rise. Nature (1987) 326(6108):35–41.[CrossRef]
Dick HJB, Natland JH, Alt JC, et al. A long in situ section of the lower oceanic crust: results of ODP Leg 176 drilling at the Southwest Indian Ridge. Earth and Planetary Science Letters (2000) 179:31–51.[CrossRef][Web of Science]
Dimanov A, Jaoul O. Calcium self-diffusion in diopside at high temperatures: implications for transport properties. Physics and Chemistry of Minerals (1998) 26:116–127.[CrossRef][Web of Science]
Dodson MH. Closure temperature in cooling geochronological and petrological systems. Contributions to Mineralogy and Petrology (1973) 40:259–274.[CrossRef][Web of Science]
Dodson MH. Closure profiles in cooling systems. Material Science Forum (1986) 7:145–154.
Dohmen R, Chakraborty S. Fe–Mg diffusion in olivine II: Point defect chemistry, change of diffusion mechanisms and a model for calculation of diffusion coefficients in natural olivine. Physics and Chemistry of Minerals (2007) doi:10.1007/s00269-007-0158-6.
Dohmen R, Becker HW, Chakraborty S. Point defect equilibration and diffusion in olivine at low temperatures (T < 1000°C). DMG, Supplement. European Journal of Mineralogy (2003) 15:42.
Droop GTR. A general equation for estimating Fe3+ concentrations in ferromagnesian silicates and oxides from microprobe analyses, using stoichiometric criteria. Mineralogical Magazine (1987) 51:431–435.[Web of Science]
Dunn RA, Toomey DR, Solomon S. Three-dimensional seismic structure and physical properties of the crust and shallow mantle beneath the East Pacific Rise at 9°30'N. Journal of Geophysical Research (2000) 105:23537–23555. (B10).[CrossRef]
Fabriès J. Spinel–olivine geothermometry in peridotites from ultramafic complexes. Contributions to Mineralogy and Petrology (1979) 69:329–336.[CrossRef][Web of Science]
Fisher AT. Geophysical constraints on hydrothermal circulation: observations and models. In: Energy and Mass Transfer in Marine Hydrothermal Systems, Volume 89.—Halbach PE, Tunnicliffe V, Hein JR, eds. (2003) Dahlem University Press. 29–52.
Francheteau J, Patriat P, Segoufin J, Armijo R, Doucoure M, Yelleschaouche A, Zukin J, Calmant S, Naar DF, Searle RC. Pito and Orongo Fracture Zones—the northern and southern boundaries of the Easter Microplate (Southeast Pacific). Earth and Planetary Science Letters (1988) 89:363–374.[CrossRef][Web of Science]
Freer R, O’Reilly W. The diffusion of Fe2+ ions in spinels with relevance to the process of maghemitization. Mineralogical Magazine (1980) 43:889–899.[CrossRef][Web of Science]
Gee J, Meurer WP. Slow cooling of middle and lower crust inferred from multicomponent magnetisations of gabbroic rocks from the Mid-Atlantic Ridge south of the Kane fracture zone (MARK) area. Journal of Geophysical Research (2002) 107(B7). ), doi:10.1029/2000JB000062.
Gee JS, Lawrence RM, Hurst SD. Remanence characteristics of gabbros from the MARK area: implications for crustal magnetisation. In: Proceedings of the Ocean Drilling Program, Scientific Results, 153—Karson JA, Cannat M, Miller DJ, Elthon D, eds. (1997) College Station, TX: Ocean Drilling Program. 429–436.
Ghiorso MS, Sack RO. Fe–Ti oxide geothermometry: thermodynamic formulation and the estimation of intensive variables in silicic magmas. Contributions to Mineralogy and Petrology (1991) 108:485–510.[CrossRef][Web of Science]
Ghiorso MS, Sack RO. Chemical mass transfer in magmatic processes IV. A revised and internally consistent thermodynamic model for the interpolations of liquid–solid equilibria in magmatic systems at elevated temperatures and pressures. Contributions to Mineralogy and Petrology (1995) 119:197–212.[Web of Science]
Ghiorso MS, Hirschmann MM, Reiners PW, Kress VCI. The pMELTs: a revision of MELTS for improved calculation of phase relations and major element partitioning related to partial melting of the mantle to 3 GPa. Geochemistry, Geophysics, Geosystems (2002) 3. doi:10.1029/2001GC000217.
Gillis KM, Mével C, Allan J, eds. Proceedings of the Ocean Drilling Program, Initial Reports, 147 (1993) College Station, TX: Ocean Drilling Program.
Grove TL, Baker MB, Kinzler RJ. Coupled CaAl–NaSi diffusion in plagioclase feldspar: Experiments and applications to cooling rate speedometry. Geochimica et Cosmochimica Acta (1984) 48:2113–2121.[CrossRef][Web of Science]
Hacker BR. Rapid emplacement of young oceanic lithosphere: argon geochronology of the Oman ophiolite. Science (1994) 265:1563–1565.
Hekinian R, Bideau D, Francheteau J, Cheminée JL, Armijo R, Lonsdale P, Blum N. Petrology of the East Pacific Rise crust and upper mantle exposed in Hess Deep (East Equatorial Pacific). Journal of Geophysical Research (1993) 98(B5):8069–8094.
Henstock TJ, Woods AW, White RS. The accretion of oceanic crust by episodic sill intrusion. Journal of Geophysical Research (1993) 98(B3):4143–4161.
Hosford A, Tivey M, Matsumoto T, Dick H, Schouten HHK. Crustal magnetization and accretion at the Southwest Indian Ridge near the Atlantis II fracture zone, 0–25 Ma. Journal of Geophysical Research (2003) 108. doi:10.1029/2001JB000604.
Jenkin GRT, Farrow CM, Fallick AE, Higgins D. Oxygen isotope exchange and closure temperature in cooling rocks. Journal of Metamorphic Geology (1994) 12:221–235.[CrossRef][Web of Science]
John BE, Foster DA, Murphy JM, Cheadle MJ, Baines AG, Fanning CM, Copeland P. Determining the cooling history of in situ lower oceanic crust—Atlantis Bank, SW Indian Ridge. Earth and Planetary Science Letters (2004) 222:145–160.[CrossRef][Web of Science]
Jurewicz AJG, Watson EB. Cations in olivine, Part 2: Diffusion in olivine xenocrysts, with applications to petrology and mineral physics. Contributions to Mineralogy and Petrology (1988) 99:186–201.[CrossRef][Web of Science]
Karson JA, Klein EM, Hurst SD, Lee C, Rivizzigno P, Curewitz D, Morris AR, Party HDS. Structure of uppermost fast-spread oceanic crust exposed at the Hess Deep Rift: Implications for subaxial processes at the East Pacific Rise. Geochemistry, Geophysics, Geosystems (2002) 3. doi:10.1029/2001GC000155.
Kelemen PB, Shimizu N, Salters VJM. Extraction of mid-ocean-ridge basalt from the upwelling mantle by focused flow of melt in dunite channels. Nature (1995) 375:747–753.[CrossRef]
Kelemen PB, Koga K, Shimizu N. Geochemistry of gabbro sills in the crust–mantle transition zone of the Oman ophiolite: implications for the origin of the oceanic lower crust. Earth and Planetary Science Letters (1997) 146:475–488.[CrossRef][Web of Science]
Kessel R, Beckett JR, Stolper EM. The thermal history of equilibrated ordinary chondrites and the relationship between textural maturity and temperature. Geochimica et Cosmochimica Acta (2007) 71:1855–1881.[CrossRef][Web of Science]
Kohler TP, Brey GP. Calcium exchange between olivine and clinopyroxene calibrated as a geothermometer for natural peridotites from 2 to 60 kb with applications. Geochimica et Cosmochimica Acta (1990) 54:2375–2388.[CrossRef][Web of Science]
Lasaga AC. Geospeedometry: an extension of geothermometry. In: Kinetics and Equilibrium in Mineral Reactions, Volume 3—Saxena SK, ed. (1983) Berlin: Springer. 82–114.
Libourel G. Systematics of calcium partitioning between olivine and silicate melt: implications for melt structure and calcium content of magmatic olivines. Contributions to Mineralogy and Petrology (1999) 136:63–80.[CrossRef][Web of Science]
Liermann H.-P, Ganguly J. Diffusion kinetics of Fe2+ and Mg2+ in aluminous spinel: experimental determination and applications. Geochimica et Cosmochimica Acta (2002) 66(16):2903–2913.[CrossRef][Web of Science]
Lonsdale P. Structural pattern of the Galapagos microplate and evolution of the Galapagos Triple Junctions. Journal of Geophysical Research (1988) 93(B11):13551–13574.[CrossRef]
Maclennan J, Hulme T, Singh SC. Cooling of the lower oceanic crust. Geology (2005) 33(5):357–360.
Meurer BP, Gee J. Evidence for protracted construction of slow-spread oceanic crust by small magmatic injections. Earth and Planetary Science Letters (2002) 201:45–55.[CrossRef][Web of Science]
Mori T. Geothermometry of spinel lherzolites. Contributions to Mineralogy and Petrology (1977) 59:261–279.[CrossRef][Web of Science]
Morioka M. Cation diffusion in olivine—II. Ni–Mg, Mn–Mg, Mg and Ca. Geochimica et Cosmochimica Acta (1981) 45:1573–1580.[CrossRef][Web of Science]
Nakamura A, Schmalzried H. On the nonstoichiometry and point defects of olivine. Physics and Chemistry of Minerals (1983) 10:27–37.[CrossRef][Web of Science]
O’Neill HSC, Wall JV. The olivine–orthopyroxene–spinel oxygen geobarometer, the nickel precipitation curve, and the oxygen fugacity of the Earth's upper mantle. Journal of Petrology (1987) 28(6):1169–1191.
Ozawa K. Evaluation of olivine–spinel geothermometry as an indicator of thermal history for peridotites. Contributions to Mineral Petrology (1983) 82(1):52–65.[CrossRef]
Ozawa K. Olivine–spinel geospeedometry: analysis of diffusion-controlled Mg–Fe2+ exchange. Geochimica et Cosmochimica Acta (1984) 48:2597–2611.[CrossRef][Web of Science]
Perk N, Coogan LA, Karson JA, Klein EM, Hanna H. Primitive cumulates from the upper crust formed at the East Pacific Rise. Contributions to Mineral Petrology (2007) 154:575–590.[CrossRef]
Phipps Morgan J, Chen YJ. The genesis of oceanic crust—magma injection, hydrothermal cooling, and crustal flow. Journal of Geophysical Research (1993) 98(B4):6283–6297.
Quick JE, Denlinger RP. Ductile deformation and the origin of layered gabbro in ophiolites. Journal of Geophysical Research (1993) 98(B8):14015–14027.[CrossRef]
Reuber I. Diapiric magma intrusions in the plutonic sequence of the Oman ophiolite traced by the geometry and flow patterns of the cumulates. In: Symposium on Diapirism with Special Reference to Iran (1990) Hormozgan, Teheran: Teheran University Government. 315–338.
Roeder PL, Campbell IH, Jamieson HE. A re-evaluation of the olivine–spinel geothermometer. Contributions to Mineralogy and Petrology (1979) 68:325–334.[CrossRef][Web of Science]
Ross K, Elthon D. Cumulus and postcumulus crystallisation in the oceanic crust: Major and trace-element geochemistry of Leg 153 gabbroic rocks. In: Proceedings of the Ocean Drilling Program, Scientific Results, 153—Karson JA, Cannat M, Miller DJ, eds. (1997) College Station, TX: Ocean Drilling Program. 333–353.
Sack RO, Ghiorso MS. Chromian spinels as petrogenetic indicators: thermodynamics and petrological applications. American Mineralogist (1991) 76:827–847.[Abstract]
Sleep NH. Formation of oceanic crust: some thermal constraints. Journal of Geophysical Research (1975) 80:4037–4042.
Sugawara T. Ferric iron partitioning between plagioclase and silicate liquid: thermodynamics and petrological applications. Contributions to Mineralogy and Petrology (2001) 141:659–686.[Web of Science]
Turcotte DL, Schubert G. Geodynamics: applications of continuum physics to geological problems (2002) Cambridge: Cambridge University Press. 456.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  F. Costa, R. Dohmen, and S. Chakraborty Time Scales of Magmatic Processes from Modeling the Zoning Patterns of Crystals Reviews in Mineralogy and Geochemistry, January 1, 2008; 69(1): 545 - 594. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||