Journal of Petrology Advance Access published online on December 9, 2008
Journal of Petrology, doi:10.1093/petrology/egn064
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Liquid Immiscibility and Evolution of Basaltic Magma: Reply to S. A. Morse, A. R. McBirney and A. R. Philpotts
1Helmholtz Centre Potsdam–Gfz German Research Centre For Geosciences, Sektion 4.1, Telegrafenberg, 14473 Potsdam, Germany
2Technical University Berlin, Department of Mineralogy And Petrology, Ackerstrasse, 71–76, D-13355 Berlin, Germany
3Earth And Environmental Sciences, University of Munich, Theresienstrasse 41, 80333 Munich, Germany
4Institute of Geology of Ore Deposits, Petrography, Mineralogy and Geochemistry Russian Academy of Sciences, Staromonetny 35, 109017 Moscow, Russia
Received November 6, 2008; Revised typescript accepted November 10, 2008
 |
INTRODUCTION |
|---|
 TOP INTRODUCTION IMMISCIBILITY IN EXPERIMENTSIMMISCIBILITY IN THOLEIITIC...CONCLUDING REMARKSREFERENCES |
|---|
We are pleased that our experimental study (Veksler et al., 2007
) received critical comments from three of the best experts on silicate liquid immiscibility and gabbroic intrusions. Our disagreements are less important than the shared belief that studies of layered intrusions need a new impulse, and existing ideas a thorough revision. This discussion is a good opportunity to emphasize and clarify the key points, better formulate current problems and disagreements, and touch upon some topics that were mentioned only briefly or not at all in our original publication. We continue experimental work on silicate liquid immiscibility (Veksler et al., 2009), and will mention here some new results that are relevant for the discussion.
Like Roedder 30 years ago (Roedder & Weiblen, 1970; Roedder, 1978), we stumbled upon liquid immiscibility by serendipity. Our interest in magma unmixing was sparked by unsought and unexpected melt inclusions in apatite from the Skaergaard intrusion (Jakobsen et al., 2005). These inclusions were impossible to miss. We realized that, surprisingly, some important features of Skaergaard rocks had been overlooked or dismissed by numerous researchers before us. As we continued our work and established close collaboration with experts on the Skaergaard intrusion in Denmark and the UK (see the Acknowledgements) our experimental study became a part of a broader petrographic and geochemical project. This discussion primarily deals with experimental evidence and therefore we will mention petrographic and geochemical observations only briefly. However, those other types of evidence are certainly important and hopefully they will be covered soon by our colleagues in their upcoming publications. We do realize that the idea of early immiscibility in the Skaergaard intrusion is at odds with broadly accepted views, and admit that the concerns expressed by our opponents are fair and well based. We started to consider the immiscibility hypothesis seriously only as the last resort, because everything else (including the compositional convection model advocated by Professor Morse) failed to explain some conflicting facts about the Skaergaard intrusion that were briefly outlined in our original paper and are further discussed below. In this paper, we would also like to point out some possible tests and promising directions for future research.
|
IMMISCIBILITY IN EXPERIMENTS |
|---|
|
TOPINTRODUCTIONIMMISCIBILITY IN EXPERIMENTSIMMISCIBILITY IN THOLEIITIC...CONCLUDING REMARKSREFERENCES |
|---|
Starting compositions
In his criticism of our starting compositions Professor Morse appears to have overlooked the two pages of detailed explanations and rationale that we presented in our original paper. Six out of eight of our starting mixtures were based on compositions of natural immiscible glasses from melt inclusions and mesostasis of volcanic rocks that we compiled from previous publications. Perhaps some of those compositions look unconventional, but we decided to exclude nothing on the basis of preconceptions, and tested them all. Whatever one thinks about appropriate normative compositions of basaltic magma, natural volcanic glasses and their mixtures are arguably much better samples of magmatic melts than the plutonic cumulate rocks that Professor Morse uses as examples in his table 1. The proportions of immiscible liquids in our charges were arbitrary and of course may have differed significantly from those in natural magma. However, if Professor Morse regards experimental reproduction of immiscibility by McBirney & Nakamura (1974
) as ‘the standard of the industry’, he should have no objections to our approach. Our approach was exactly the same as used by McBirney & Nakamura (1974), who, after failing to experimentally reproduce unmixing of their preferred model liquid of the Skaergaard Upper Zone, mixed it with variable and arbitrary amounts of a granophyric liquid, and suddenly had more luck.
|
The remaining two mixtures MZ-1 and MZ-2 are the only ones that are directly applicable to the Skaergaard intrusion. They were based not on natural magmatic glasses but on experimental products, and mass-balance calculations. As explained in our original paper, the composition MZ-1 is a replica of a cotectic experimental glass produced by Toplis & Carroll (1995
) at 1105°C and log fO2 = –9·56 [i.e. very close to the fayalite–magnetite–quartz (FMQ) buffer] in a study of a parental Skaergaard composition. Crystal phases coexisting with the melt comprised olivine Fo62, high-Ca clinopyroxene and plagioclase An64. The temperature of the experiment was only 9°C above the onset of ilmenite and magnetite crystallization. Therefore, the melt can be viewed as a close experimental analogue of the Skaergaard liquid at the transition between the LZb and LZc [here and below the names and abbreviations for the Skaergaard stratigraphic units are according to Wager & Brown (1968)]. It is true that the liquidus plagioclase of the MZ-1 composition is somewhat more anorthitic than the natural plagioclase An57–63 from the LZb cumulates (McBirney, 1989). However, the difference can be explained by re-equilibration of natural plagioclase with inter-cumulus liquid at lower temperature. Our starting composition MZ-2 is very similar to MZ-1 but somewhat higher in FeO(t). It is close to the LZc liquid at about 60% crystallization of the Skaergaard magma that was calculated by Nielsen (2004a, 2004b) on the basis of masses and volumes of stratigraphic zones of the intrusion. When we started our experiments 3 years ago these were the most reliable reconstructions of the Skaergaard liquid at the peak of Fe enrichment. A later experimental study of Skaergaard-type dykes by Thy et al. (2006) reproduced very similar liquids at the point of melt saturation in Fe–Ti oxides.
Despite the concern expressed by Professor Morse, the presence of minor normative olivine or quartz in our starting and derivative liquids does not look to us very important. Typical tholeiitic liquids (including the Skaergaard parental melt) are olivine-normative at the start of crystallization, but become quartz-normative at the stage of olivine–pigeonite peritectic reaction (the lower part of the Skaergaard MZ formed exactly at this stage). The Grove projection (Grove, 1993) is a good illustration of the trend (Fig. 1). The topology of the projection, which is based on interpolation of experimental mid-ocean ridge basalt (MORB) and calc-alkaline liquid compositions, implies no barrier between olivine- and quartz-normative liquids evolving along the olivine–augite–plagioclase cotectic. The thermal divide (TD, Fig. 1) is positioned between the olivine and pigeonite fields, deep inside the region of quartz-normative compositions. Notably, all experimental liquids of low-alkali, tholeiitic compositions equilibrated with olivine, augite, low-Ca pyroxene and plagioclase in the study by Longhi & Pan (1988) are quartz-normative, and so is our MZ-1. We also plot in Fig. 1 two alternative experimental Skaergaard trends (McBirney & Naslund, 1990; Toplis & Carroll, 1995). Notably, many of the liquids are quartz-normative. The trends do not pass through the peritectic reaction point, and none of them closely follows the cotectic and peritectic curves [see Veksler (2009) for further discussion]. This probably reflects the difference between Fe-rich tholeiitic liquids and calc-alkaline basaltic melts. The diagram also clearly shows that the alternative Skaergaard experimental trends part at the transition from LZ to MZ.
|
Stable immiscibility vs unmixing during quenching
Professor Philpotts raises a serious issue of distinguishing between super-liquidus immiscibility and unmixing during quenching. The explanation that he offers for the chemical gradients developed in some of our centrifuged glasses is that the gradients were due to settling of FeO powder at the initial stages of melting at slow rotation. In the early stages of our study, we worried about this possible artefact, and therefore tried different loading methods and sintering protocols to achieve perfect homogenization of melts before centrifugation. Finally, while carrying out our most recent series of centrifuge experiments (Veksler et al., 2009
) we learned how to tightly plug inner iron containers and that allowed us to preheat and homogenize charges at ambient gravity for days. We repeated some of the key experiments on compositions used by Veksler et al. (2007) and confirmed that sintering techniques in our previous study had been adequate. Even without that later test, our original study gave enough circumstantial evidence to prove that centrifugation experiments were not fundamentally flawed. If they were, macroscopic chemical gradients would have been present in all the centrifuged glasses. That certainly was not the case. When we examined the centrifuged glasses by scanning and transmission electron microscopy, we did not see any Fe-oxide particles. As discussed in our paper, we saw numerous sub-micrometre, non-crystalline, glassy globules with sharp interfaces, and compositions of typical Fe-rich and silica-rich immiscible liquids. Professor Philpotts does not accept our interpretation of those tiny droplets as products of stable immiscibility, even when they clearly move and coalesce during centrifugation. He is sceptical because of the very small size of those exsolutions, as most of them are indeed two or three orders of magnitude smaller than the micrometre-sized droplets that have been observed in various unmixed basaltic compositions.
Professor Philpotts states that micrometre-sized droplets have never been observed at temperatures above 
1040°C in experiments on natural rock compositions. This statement is not true. Experiments by Krasov & Clocciatti (1979) that we mentioned in our original paper provided an example of high-temperature unmixing, and in fact that study is not unique. Roedder & Weiblen (1970) reported micrometre-sized immiscible droplets in experiments on lunar ferrobasaltic and rhyolitic melts in a temperature interval of 1045–1135°C. The consolute temperature was not determined, but Roedder & Weiblen claimed that it certainly was above 1135 and below 1350°C. In the published photographs, the droplets meet all the morphological criteria of stable exsolutions proposed by Professor Philpotts. Notably, tridymite was reported together with two liquids in products of shorter runs at 1135°C, but was not mentioned in products of slow cooling quenched from 1045°C. This looks like another case of metastable tridymite nucleation, the same as in our experiments.
The examples that Professor Philpotts presents in his comments give an impression that stable immiscible droplets nucleate and grow easily. Unfortunately, this is not always the case. As mentioned above, McBirney & Nakamura (1974) could not, according to their own account, reproduce immiscibility in pre-homogenized Fe-rich Skaergaard melts, and had to mix them with silicic granophyre to achieve unmixing. Longhi (1990) had to use different loading methods and noted that sharp liquid–liquid interfaces were absent in some of his experimental products. It seems that liquid immiscibility is better reproduced during slow cooling (e.g. Dixon & Rutherford, 1979) than in isothermal experiments. In any case, one would expect the nucleation density and size distribution of the droplets to vary broadly from one experiment to another in response to variable over-saturation, diffusion rates, viscosity, interfacial energy, and other physical properties.
Our disagreements with Professor Philpotts are indeed very similar to those between Visser & Koster van Groos (1976, 1977), Roedder (1977) and Freestone & Hamilton (1977), who argued about the correct interpretation of opalescent glasses and sub-micron exsolutions in the classical system K2O–FeO–Al2O3–SiO2. We worked only briefly on that system (Veksler et al., 2009) and will not comment on that case. However, we were aware of conflicting views on the origin of opalescent glasses, and that was exactly the reason why we used centrifugation. Settling of colloidal emulsions by in situ centrifugation is a straightforward and reliable way of distinguishing them from quench exsolution. However, in some cases one can arrive at a correct interpretation simply by careful optical examination of opalescent glasses recovered from conventional static runs. Let us look, for example, at a fragment of a polished bead of glass (Fig. 2) produced in our recent static experiment (Veksler et al., 2009) on a composition that was taken from a region of stable, low-temperature immiscibility in the system K2O–CaO–FeO–Al2O3–SiO2 (Hoover & Irvine, 1978). After 24 h at 1090°C the charge quenched to turbid, opalescent glass composed of sub-micron dispersed Fe-rich glassy droplets in a silica-rich matrix glass. We interpret the emulsion as formed by stable unmixing rather than exsolution during quenching. Our conclusion is based on the pattern of flow textures around slowly rising air bubbles trapped inside the charge. One can see that the bubbles pushed plumes and clouds of dispersed Fe-rich phase upwards and aside leaving behind a tail of almost pure and clean silica-rich liquid. Such flow textures could not form instantly in a few seconds of quenching time and therefore imply protracted existence of the emulsion at high temperature. Furthermore, centrifugation experiments on this composition at the same temperature confirmed stable unmixing beyond reasonable doubt. Centrifugation greatly enhanced coalescence of the droplets, and after 4 h of forced separation nearly half of the emulsion settled down to a condensed bottom layer with a razor-sharp interface [see Veksler et al. (2009) for further details].
|
Professor Philpotts brings up examples of metastable sub-micron exsolution in technological glasses, but does not mention that these are not always due to quenching. In fact, some examples of colloidal exsolution have been shown to form slowly in under-cooled liquids above the glass transition. For instance, Toplis & Reynard (2000
) observed time-dependent structural changes in a P-bearing Na aluminosilicate melt using in situ Raman spectroscopy, and attributed them to sub-liquidus immiscibility above the glass transition. Unmixing developed within 3 h at 700–775°C, slower at higher temperatures, and faster at greater under-cooling. Notably, glasses that were quenched after dwelling times of a few hours were optically transparent and showed no heterogeneity under scanning electron microscope at a resolution of 1 µm. Another instructive example comes from the classical glass-forming system Na2O–CaO–SiO2 that is used for manufacturing common window glass. Burnett & Douglas (1970) performed a series of annealing experiments on under-cooled melts in the region of sub-liquidus immiscibility of the system. At 640°C they observed nanoscale emulsions and documented three distinct stages of sub-micron coarsening that lasted for at least 27 h in total.
Of course, the above are examples of metastable (sub-liquidus) immiscibility, whereas we were investigating immiscibility above the liquidus. Nevertheless, with regard to kinetics, we see no fundamental difference between sub- and super-liquidus unmixing. As long as crystals do not nucleate, silicate liquids do not know whether they are above or below the liquidus. What really matters for the kinetics and resulting emulsion morphology are the relationships between super-saturation, viscosity, interfacial energy, and diffusion rates that define the rates of nucleation and growth, and may provide favourable conditions for long-term stability of sub-micron droplets. Investigations of liquid unmixing in materials science are more advanced, and they have already revealed high-temperature immiscible sub-micron silicate emulsions that remained stable for hours and even days. In situ spectroscopic methods developed in glass technology, such as small-angle X-ray scattering (SAXS; Mazurin & Porai-Koshits, 1984), may be used for observation of initial, latent stages of unmixing of natural magmatic melts. Future progress, in our view, would require better experimental constraints on the kinetics of immiscibility, the stability factors of immiscible emulsions at high temperatures, direct measurements of interfacial energy, and advanced models of liquid–liquid and crystal–liquid phase relationships for geologically relevant melt compositions.
Comparisons with other experimental studies of Skaergaard liquids
There would have been no need for us to perform centrifugation experiments on Skaergaard liquids if the problem of magmatic evolution at Skaergaard had been successfully resolved by other, conventional experimental methods. Unfortunately, it was not. We do not mind when our opponents call our experiments questionable and we share their concerns with regard to chemical equilibrium. Experiments on ferrobasaltic and rhyolitic liquids at temperatures below 1100°C are anything but easy and the results should certainly be treated with caution. Reaching chemical equilibrium in silicate systems at low temperatures has been recognized as a great challenge for decades. For example, Kennedy (1948) showed that 240 h in air was not long enough to reach ferrous–ferric equilibrium in partly crystallized basalt at 1100°C. Crystal–liquid equilibria are also notoriously difficult at lower temperatures. For instance, neither Toplis & Carroll (1995) nor Thy et al. (2006) were able to reproduce the reappearance of fayalitic olivine in the final stages of crystallization of Skaergaard model liquids. The important lesson from our study is that low-temperature silicate liquid immiscibility is not immune to nucleation and kinetic problems either.
Relationships between extreme Fe-enrichment and liquid immiscibility appear to be complex. One of us (Veksler) has recently searched through experimental databases for cases of extreme Fe-enrichment in basaltic liquids, specifically, melts with total FeO contents above 22 wt %. A detailed review of those cases will be published elsewhere (Veksler, 2009). In a nutshell, the conclusions are simple. Extreme Fe-enrichment up to 30 wt% FeO(t) and even higher has been documented in melts equilibrated with fayalitic olivine, high-Ca pyroxene, plagioclase, and Fe–Ti oxides at reducing conditions, atmospheric pressure, and temperatures slightly above 1000°C (e.g. Longhi & Pan, 1988). The enrichment may result from crystallization combined with liquid immiscibility, or fractional crystallization alone. However, the latter is possible only in bulk compositions with very low total alkalis. Alkali contents in the parental Skaergaard magma [e.g. broadly accepted starting compositions compiled by Nielsen (2004a)] seem to allow Fe-enrichment only to about 22 wt % FeO(t) regardless of redox conditions. This is exactly the maximum FeO(t) that Toplis & Carroll (1995) and Thy et al. (2006) reported in their experiments on parental Skaergaard liquids. Any further Fe-enrichment would require redistribution (removal) of alkalis by some process other than fractional crystallization, and silicate liquid immiscibility appears to be the only sensible option.
Phase equilibria limitations are compounded by mass-balance constraints. The sequence of Skaergaard liquids produced by partial melting of cumulates (McBirney & Naslund, 1990) shows no pronounced enrichment in total alkalis (see the AFM diagram in Fig. 3), and this is a clear violation of mass balance. Hunter & Sparks (1990) completely dismissed those experiments and the trend as fundamentally flawed, but we are not so sure [see Veksler (2009) for further discussion]. We believe that the experiments on cumulates may carry an important message about the real Skaergaard inter-cumulus liquids. However, some features of the McBirney & Naslund (1990) trend are puzzling. One has to explain, for example, why the sixfold increase in P2O5 from 0·2 wt% in LZa to 1·2 wt % UZa is not accompanied by a similar rise of the K2O concentrations (the latter increased only by a factor of 1·7). By all accounts, K is approximately as incompatible as P, the element that McBirney & Naslund (1990) used to control the degree of melting in their experiments. Although fractional crystallization in the absence of apatite and/or alkali feldspars cannot fractionate K from P, these elements are easily separated by silicate liquid immiscibility. Therefore, apparent decoupling of K and P in the partial melts of Skaergaard cumulates (McBirney & Nalsund, 1990), may provide circumstantial evidence for early unmixing and liquid–liquid fractionation starting from LZb–LZc, in agreement with our centrifuge experiments on the composition MZ-1.
|
|
IMMISCIBILITY IN THOLEIITIC MAGMA |
|---|
|
TOPINTRODUCTIONIMMISCIBILITY IN EXPERIMENTSIMMISCIBILITY IN THOLEIITIC...CONCLUDING REMARKSREFERENCES |
|---|
Our original paper dealt with experimental evidence and was not meant to discuss implications for the Skaergaard or any other gabbroic intrusion in detail. Our critics have a superb knowledge of mafic intrusions and other products of basaltic magma. However, at times they criticize things that we never actually said, and we feel that we have to respond and clarify our position.
Liquid immiscibility vs compositional convection
Professor Morse believes that compositional convection in large basaltic intrusions always ‘trumps’ liquid immiscibility. With all due respect, we do not see how such trick can be pulled off. We view immiscibility and compositional convection as processes that are, in certain ways, opposite to each other. In the model favoured by Professor Morse (his Discussion), compositional convection is driven by chemical gradients at crystallization fronts and it works towards equilibration and complete homogenization of the liquid. In contrast, immiscibility splits a homogeneous melt into two liquid phases, and therefore works towards greater heterogeneity. Immiscibility is a self-sufficient, equilibrium process of differentiation, totally independent of crystallization. Liquidus crystals and immiscible droplets can be kept in equilibrium indefinitely, and with time phase separation by gravity should only improve. Furthermore, because both conjugate liquids are required to be in thermodynamic equilibrium with the same liquidus crystal assemblage, percolation and separation of immiscible liquids in a crystal mush appears to be a perfect way of making modal layering. We do not see how any other process can rival liquid immiscibility in this respect.
We never proposed that immiscible liquids in Skaergaard or any other intrusion separated like oil and water into two major, condensed, continuous layers. We experimented on crystal-free compositions but never claimed that unmixing of natural magma was super-liquidus. Crystal phases were certainly present in the Skaergaard magma already during its emplacement. In natural magmatic systems, we view immiscibility as a continuous process that goes in parallel to crystallization and cooling. At any given moment, we expect to have emulsion clouds mingling and interacting with crystals. For simplicity, immiscible droplets can be viewed as another crystal phase, more buoyant than plagioclase, and with a chemical composition between that of alkali feldspar and pure silica.
We agree with Professor Morse that there are clear signs of significant vertical redistribution of FeO, alkalis and silica in the Skaergaard magma chamber, and do not deny that compositional convection may have played a role before immiscibility started. Rejected silicic components may go up either in plumes (Professor Morse proposes that), or as clouds of immiscible silicic droplets. However, as soon as the liquid hits the liquid–liquid binodal (solvus), one cannot have it both ways. When Professor Morse asks: ‘Will the flux of light solute physically [and chemically] interfere with the evolving liquid pairs?’, the question strikes us as very strange. We do not see how light plumes of rejected solute could form in the presence of two immiscible liquids. As soon as the Fe-rich liquid crystallizes a few mafic crystals, it would also nucleate a few silica-rich immiscible droplets. As crystallization and immiscibility proceed, the evolving Fe-rich liquid is likely to become even more Fe-rich and dense because of the broadening of the miscibility gap with falling temperature (Fig. 4). We believe that emulsion clouds are more effective carriers than convection plumes because the former are stable and do not dissipate with time. Plumes in a convecting liquid blend in and eventually vanish, whereas stable emulsion droplets would grow even bigger as they move.
|
The thermodynamic model that Professor Morse presents in his fig. 7 (see his Discussion) is dimensionless and too schematic for a serious discussion. We do not find the analogy with atmospheric circulation very useful because air, rain clouds and silicate melts differ dramatically in viscosity, density, heat conductivity, and a number of other important physical properties. We doubt that the P–T cycle of coupled over- and under-saturated adiabats can be sustained at the scaling parameters of the Skaergaard magma chamber. For instance, the establishment of adiabats implies, by definition, that melt and crystals in descending plumes sink faster than they exchange heat with the ambient magma. Such a scenario does not seem to us physically plausible. In effect, Professor Morse proposes in his Fig. 7 that melt and crystals in small, fast-cooling intrusions are in thermal and chemical equilibrium throughout the magma volume, whereas in large, slowly cooling intrusions equilibrium is reached only at the top and bottom crystallization fronts. This notion seems to us paradoxical and counter-intuitive, if not plain wrong.
Plagioclase issues
Professor Morse discusses plagioclase buoyancy, adcumulus crystal growth and compositional zoning. These are no doubt interesting and important subjects, but they have little to do with the main topics of our experimental study. We do not see why the variations in plagioclase zoning that Toplis et al. (2008) documented in the Skaergaard Layered Series contradict the idea of liquid immiscibility and liquid–liquid fractionation starting at the top of the LZ. Furthermore, the extensive reverse zoning that Stripp et al. (2007) documented in plagioclase crystals from reactive symplectites is believed to arise from liquid immiscibility and liquid–liquid fractionation in the crystal mush. However, a serious discussion of plagioclase zoning in Skaergaard gabbros should be better left for another paper. Here we would only express our disagreement with the statement that ‘the two-liquid hypothesis cannot account for the presence of cumulus plagioclase in the rocks of a dense, Fe-rich liquid layer’ (Morse, Discussion). Our model Fe-rich Skaergaard liquid MZF yields 38 wt % of normative An40, and it crystallized early liquidus plagioclase An55. Therefore, plagioclase is produced by the Fe-rich immiscible liquid early and in abundance. Density calculations imply that plagioclase crystals would float in the Fe-rich Skaergaard liquid, immiscible or not, but plagioclase intergrowths with mafic minerals and Fe–Ti oxides would sink (see also Philpotts & Dickson, 2000). In general, we do not see how immiscibility should be inconsistent with cumulus plagioclase.
Immiscibility in the Skaergaard intrusion
As mentioned above, McBirney & Nakamura (1974) demonstrated liquid immiscibility in experiments on mixtures of silicic granophyre and Fe-rich partial melts of the Skaergaard UZ. They proposed that certain types of melanogranophyre segregations in the UZc formed by immiscibility. However, the trend of extreme Fe-enrichment throughout the MZ after the start of Fe–Ti oxide crystallization remained, as discussed above, contentious. The case for immiscibility at Skaergaard was further weakened when it was found that the distribution of excluded trace elements between melanogranophyre and host ferrogabbro (or ferrodiorite) was inconsistent with experimental liquid–liquid partition coefficients (McBirney, 2002). The discovery of contrasting silicic and Fe-rich melt inclusions in apatite (Jakobsen et al., 2005) was so important because it finally confirmed liquid immiscibility in the Skaergaard UZ rocks beyond reasonable doubt. The inclusions proved immiscibility but, again, they explained neither the continuous Fe enrichment during the formation of MZ and UZa, nor how (and when) the Skaergaard magma had managed to reach the miscibility gap. That is why we decided to revisit experimental evidence on immiscibility and Fe-enrichment.
It is certainly true that immiscibility in the Skaergaard magma, as everything else, started from nothing and evolved into something (Morse, Discussion). However, it would be helpful to be more specific. The Skaergaard magma approached the miscibility gap from the mafic, Fe-rich side, and unmixing must have started by nucleation of a few droplets of silicic liquid in a large volume of ferrobasaltic melt. Mass balance between the bulk composition MZ-1 and the conjugate liquid compositions Lfe and Lsi from our centrifuge experiments (Veksler et al., 2007) implies 15–20 wt % of the silicic liquid. This gives a rough idea about the proportions of immiscible liquids at the start of immiscibility when crystallization of the Layered Series proceeded from LZb to LZc and 40–45 wt % of the Skaergaard intrusion remained molten (Nielsen, 2004a). However, by the time of UZb formation the proportions must have been different. As temperature further decreased by 50°C (McBirney & Naslund, 1990), the Skaergaard magma crystallized large amounts of mafic cumulates, and the degree of crystallization rose to 90 wt % (Nielsen, 2004a). Everybody seems to agree that at this late stage the remaining liquid portion of magma was a mixture of two immiscible melts. Professor Morse suggests that the Fe-rich melt remained predominant in the UZ but our mass-balance calculations do not support such view. For the mass-balance calculation in the UZ, one needs to know the compositions of coexisting liquids and the bulk composition of the emulsion. For the former, one can take the compositions of the experimental immiscible glasses reported by McBirney & Nakamura (1974), or the most contrasting compositions of apatite-hosted melt inclusions found in UZb and UZc (Jakobsen et al., 2005). For the latter, one can use the partial melt of the UZb cumulates (McBirney & Naslund, 1990), or the bulk UZb liquid calculated by mass balance (Nielsen, 2004a, 2004b). All these data are presented in Table 1. As it turns out, the experimental immiscible liquids (McBirney & Nakamura, 1974) and the UZb partial melt (McBirney & Naslund, 1990) do not balance at all. The UZb partial melt clearly cannot represent the bulk of the emulsion because it is lower in silica than the experimental Fe-rich immiscible liquid Lfe, and therefore mass balance for silica would result in negative fractions of Lfe. On the other hand, the compositions of apatite-hosted melt inclusions, and the bulk UZb liquid calculated by Nielsen (2004a, 2004b) fit the mass-balance equations reasonably well. Silica, alumina, FeO(t) and Na2O give consistent Lfe fractions of 25–33 wt%. Therefore, it follows that the Fe-rich immiscible liquid was not predominant during UZb formation. These mass-balance relationships are illustrated in Fig. 4 in a form of pseudo-binary T–x diagram. The diagram shows that, starting from LZc, massive crystallization of Fe–Ti oxides and other mafic minerals drove the bulk composition of the emulsion towards silica enrichment, while the Fe-rich liquid continued to evolve towards further Fe-enrichment and silica depletion as a result of the widening of the miscibility gap with falling temperature. This is how, in our view, liquid immiscibility resolves the issue that many people historically called the Fenner–Bowen Skaergaard controversy.
After a few hours of centrifugation, only a small fraction of the silicic liquid floated to the top of the charge in our experiments. In a similar way, we believe that only a small fraction of the silicic conjugate melt reached the Upper Border Series (UBS) in the Skaergaard intrusion. However, the evidence from the UBS is very important and probably deserves a revision. Some encouraging hints towards the effects of immiscibility and large-scale liquid–liquid fractionation can be found in the paper by Naslund (1984), who noted that the UBS rocks ‘are systematically richer in K2O and SiO2 ... than their Layered Series counterparts’ and the differences in K2O and SiO2 ‘do not appear to be explicable by any mixture of cumulus minerals or cumulus minerals and trapped Layered Series liquid’. The rest of the silica-rich immiscible liquid should have contributed to the cumulate zones of the Layered Series, and formed the melanogranophyre segregations that Professor McBirney describes in his comments. Like the UBS rocks, Skaergaard melanogranophyres may deserve a more detailed study. Judging by the bulk-rock analyses that have been published so far, we do not see any significant, systematic difference between the granophyric zones in pegmatites from the LZ and melanogranophyres from UZc with regard to their P2O5 and alkali contents. Pegmatitic granophyres with P2O5 concentrations varying from 0·02 to 0·9 wt % (Larsen & Brooks, 1994) do not look more P-rich than segregation and pod-like melanogranophyres that contain 0·07–0·81 wt % P2O5 (McBirney, 1989). Furthermore, the statement that granophyric zones in pegmatites are ‘normal products of crystal fractionation’ (McBirney, Comments) certainly contradicts the opinion of Larsen & Brooks (1994), who wrote that contact relations between the gabbroic and granophyric zones of pegmatites were ‘consistent with the production of the granophyric component by liquid immiscibility followed by gravity separation from the conjugate mafic melt of the pegmatitic system’. Larsen & Brooks (1994) viewed pegmatitic pockets in the lower zones of the Layered Series as ‘micro-magma chambers’ that followed the same path of crystallization and immiscibility as the whole Skaergaard intrusion.
Immiscibility in the Holyoke basalt
Thanks to the extensive research efforts of Professor Philpotts, large tholeiitic lava flows in Connecticut (USA) have become the best studied examples of differentiated volcanic bodies that underwent crystallization, crystal settling, and immiscibility. Basaltic lava in Connecticut is compositionally close to the Skaergaard magma but the former cooled and crystallized much faster. As a result, unquestionable traces of liquid immiscibility have been preserved in the basaltic groundmass as micron-sized Fe-rich globules dispersed in rhyolitic matrix glass (Philpotts, 1979). We value the keen observations and careful experiments by Philpotts (1979) and Philpotts & Doyle (1983) but we are not entirely sure that those studies correctly bracketed the onset of immiscibility. According to Philpotts (1979), microphenocrysts in the groundmass show strong compositional zoning. This implies that the groundmass crystals and melts may have been out of equilibrium and became compromised during the original fast cooling of the lava flow. Reheating of such disequilibrium groundmass crystal–glass assemblages would not necessarily reproduce the equilibrium course of crystallization and immiscibility. To better clarify the sequence of events in the cooling lava flow, one should probably investigate in detail the extent of crystal zoning in the groundmass, and carry out reheating experiments also on plagioclase-hosted melt inclusions that apparently show immiscibility at higher temperature. In our view, with regard to the timing of immiscibility in the Holyoke basalt the jury is still out.
|
CONCLUDING REMARKS |
|---|
|
TOPINTRODUCTIONIMMISCIBILITY IN EXPERIMENTSIMMISCIBILITY IN THOLEIITIC...CONCLUDING REMARKSREFERENCES |
|---|
Although our involvement with the Skaergaard intrusion started only recently and our knowledge of subtle geological details may be incomplete, objections from our more experienced opponents has not convinced us so far that early immiscibility in the Skaergaard magma is a bad idea. We do not deny that some of the results of our original study are unexpected or controversial, and we have tried to address those issues in new experiments since the original paper was published. However, we suspect that experimental methods in application to ferrobasalts at low temperatures may be already pushed to the limits and the ultimate, decisive evidence in favour of or against early unmixing in the Skaergaard and similar intrusions is more likely to come from natural rocks rather than experimental studies.
|
ACKNOWLEDGEMENTS |
|---|
We have been fortunate in recent years to closely collaborate with Dr Marian Holness, Miss Gemma Stripp, Dr Charles Kent Brooks, Dr Jakob K. Jakobsen, Dr Troels F. D. Nielsen, and Dr Christian Tegner. This collaboration started at the Kent Brooks symposium in 2004, and we would like to emphasize the crucial role that Kent played in launching this joined Skaergaard enterprise. Without his encouragement, and also indispensable contributions from other colleagues in the group, we would never dare to challenge the paradigms of the classical Skaergaard intrusion.
*Corresponding author. E-mail: veksler{at}gfz-potsdam.de
|
REFERENCES |
|---|
|
TOPINTRODUCTIONIMMISCIBILITY IN EXPERIMENTSIMMISCIBILITY IN THOLEIITIC...CONCLUDING REMARKSREFERENCES |
|---|
Burnett DG, Douglas RW. Liquid–liquid phase separation in the soda–lime–silica system. Physics and Chemistry of Glasses (1970) 11:125–135.[Web of Science]
Carmichael ISE. The petrology of Thingmuli, a Tertiary volcano in eastern Iceland. Journal of Petrology (1964) 5:435–460.
Dixon S, Rutherford MJ. Plagiogranites as late-stage immiscible liquids in ophiolite and mid-ocean ridge suites: an experimental study. Earth and Planetary Science Letters (1979) 45:45–60.[CrossRef][Web of Science]
Freestone IC, Hamilton DL. Liquid immiscibility in K2O–FeO–Al2O3–SiO2: Discussion. Nature (1977) 267:559.
Grove TL. Corrections to expressions for calculating mineral components in ‘Origin of Calc-Alkaline Series Lavas at Medicine Lake Volcano by Fractionation, Assimilation and Mixing’ and ‘Experimental Petrology of normal MORB near the Kane Fracture Zone: 22°–25°N, Mid-Atlantic Ridge’. Contributions to Mineralogy and Petrology (1993) 114:422–424.[CrossRef]
Hoover JD, Irvine TN. Liquidus relations and Mg–Fe partitioning on part of the system Mg2SiO4–Fe2SiO4–CaMgSi2O6–CaFeSi2O6–KAlSi3O8–SiO2. Carnegie Institution of Washington Yearbook (1978) 77:774–784.
Hunter RH, Sparks RSJ. The differentiation of the Skaergaard Intrusion. A reply. Contributions to Mineralogy and Petrology (1990) 104:248–254.[CrossRef][Web of Science]
Jakobsen JK, Veksler IV, Tegner C, Brooks CK. Immiscible iron- and silica-rich melts in basalt petrogenesis documented in the Skaergaard intrusion. Geology (2005) 33:885–888.
Kennedy GC. Equilibrium between volatiles and iron oxides in igneous rocks. American Journal of Science (1948) 246:529–549.
Krasov NF, Clocchiatti R. Immiscibility in silicate melts and its possible petrogenetic importance, as shown by study of melt inclusions. Transactions (Doklady) of the USSR Academy of Science (1979) 248:92–95.
Larsen RB, Brooks CK. Origin and evolution of gabbroic pegmatites in the Skaergaard intrusion, East Greenland. Journal of Petrology (1994) 35:1651–1679.
Longhi J. Silicate liquid immiscibility in isothermal crystallization experiments. Proceedings of the 20th Lunar and Planetary Science Conference (1990) 13–24.
Longhi J, Pan V. A reconnaisance study of phase boundaries in low-alkali basaltic liquids. Journal of Petrology (1988) 29:115–147.
Mazurin OV, Porai-Koshits EA. Phase Separation in Glass (1984) Amsterdam: North-Holland. Pergamon, New York.
McBirney AR. The Skaergaard layered series: I. Structure and average compositions. Journal of Petrology (1989) 30:363–397.
McBirney AR. The Skaergaard Layered Series. Part VI. Excluded trace elements. Journal of Petrology (2002) 48:535–556.
McBirney AR, Nakamura Y. Immiscibility in late-stage magmas of the Skaergaard intrusion. Carnegie Institution of Washington Yearbook (1974) 73:348–352.
McBirney AR, Naslund HR. The differentiation of the Skaergaard intrusion. A discussion of Hunter and Sparks (Contrib. Mineral. Petrol. 95: 451–461). Contributions to Mineralogy and Petrology (1990) 104:235–240.[CrossRef][Web of Science]
Naslund HR. Petrology of the Upper Border series of the Skaergaard intrusion. Journal of Petrology (1984) 25:185–212.
Nielsen TFD. The shape and volume of the Skaergaard intrusion, Greenland: implications for mass balance and bulk composition. Journal of Petrology (2004a) 45:507–530.
Nielsen TFD. A mass balance approach to the differentiation of the Skaergaard intrusion, East Greenland. (2004b) 68. 14th Annual Goldschmidt Conference, June 5-11, 2004: Copenhagen, Denmark. A630. The Geochemical Society and European Association of Geochemistry, Geochimica et Cosmochimica Acta.
Philpotts AR. Silicate liquid immiscibility in tholeiitic basalts. Journal of Petrology (1979) 20:99–118.
Philpotts AR, Dickson LD. The formation of plagioclase chains during convective transfer in basaltic magma. Nature (2000) 406(6791):59–61.
Philpotts AR, Doyle CD. Effects of magma oxidation state on the extent of silicate liquid immiscibility in a tholeiitic basalt. American Journal of Science (1983) 283:967–985.
Roedder E. Liquid immiscibility in K2O–FeO–Al2O3–SiO2: Discussion. Nature (1977) 267:558–559.
Roedder E. Silicate liquid immiscibility in magmas and in the system K2O–FeO–Al2O3–SiO2: an example of serendipity. Geochimica et Cosmochimica Acta (1978) 43:1597–1617.
Roedder E, Weiblen PW. Lunar petrology of silicate melt inclusions, Apollo 11 rocks. In:. Proceedings of the Apollo 11 Lunar Science Conference. Geochimica et Cosmochimica Acta Supplement 1, (1970) 1:507–528.
Stripp G, Holness M, Nielsen T, Veksler I, Tegner C. Reactive symplectite formation as tracers of late-stage liquids: the Skaergaard Intrusion. In: The Mineralogical Society of Ireland and Great Britain. Frontiers in Mineral Sciences 2007, Programme and Abstracts—Harte B, Carpenter M, eds. (2007) Cambridge. 227.
Thy P, Lesher CE, Nielsen TFD, Brooks CK. Experimental constraints on the Skaergaard liquid line of descent. Lithos (2006) 92:154–180.[CrossRef][Web of Science]
Toplis MJ, Carroll MR. An experimental study of the influence of oxygen fugacity on Fe–Ti oxide stability, phase relations, and mineral–melt equilibria in ferro-basaltic systems. Journal of Petrology (1995) 36:1137–1171.
Toplis MJ, Reynard B. Temperature and time-dependent changes of structure in phosphorus containing aluminosilicate liquids and glasses: in situ Raman spectroscopy at high temperature. Journal of Non-Crystal Solids (2000) 263–264:123–131.
Toplis MJ, Brown WL, Pupier E. Plagioclase in the Skaergaard intrusion: Part 1: Core and rim compositions in the layered Series. Contributions to Mineralogy and Petrology (2008) 155:329–340.[CrossRef][Web of Science]
Veksler IV. Extreme iron enrichment and liquid immiscibility in mafic intrusions: experimental evidence revisited. In: Lithos (2009) (in press), doi:10.1016/j.lithos.2008.10.003.
Veksler IV, Dorfman AM, Borisov AA, Wirth R, Dingwell DB. Liquid immiscibility and evolution of basaltic magma. Journal of Petrology (2007) 48:2187–2210.
Veksler IV, Dorfman AM, Rhede D, Wirth R, Borisov AA, Dingwell DB. Liquid unmixing kinetics and the extent of immiscibility in the system K2O–CaO–FeO–Al2O3–SiO2. Chemical Geology (2008) 256:119–130.
Visser W, Koster van Groos AF. Liquid immiscibility in K2O–FeO–Al2O3–SiO2. Nature (1976) 264:426–427.
Visser W, Koster van Groos AF. Liquid immiscibility in K2O–FeO–Al2O3–SiO2: Reply to Roedder, Freestone and Hamilton. Nature (1977) 267:560.
Wager LR, Brown GM. Layered Igneous Rocks (1968) Edinburgh: Oliver & Boyd.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||