Report on Minerals, Inclusions and Volcanic Processes
A Mineralogical Society of America Short Course
Organized and Edited by:
Keith Putirka, California State University Fresno
Frank Tepley, Oregon State University
Our use of minerals and their inclusions to understand magmatic systems has a long history, extending back to Sorby's (1858) study of inclusions and Darwin's (1844) and King's (1878) field studies of minerals. Although such work has been relatively continuous, notable advances by Barth (1934) and Roedder (1965) sparked interest in thermometry and inclusions respectively, and a series of papers by Eichelberger (1975), Anderson and Wright (1972), Anderson (1976) and Dungan and Rhodes (1978), among others, spurred new interest in using such methods to understand volcanic systems in particular. More recent advances in micro-analytical techniques over the past two decades, perhaps beginning with Davidson et al. (1990), have greatly magnified and accelerated interest in minerals and inclusions, highlighting the potential for scientific advancements and revised views of magma plumbing systems (Marsh, 1995). The short course and the related volume, 'Minerals, Inclusions and Volcanic Processes', was organized with the goal of summarizing where these earlier strands of research have now branched, with the added hope that such a review might initiate new collaborations, based on an alliance of complementary techniques. The prospects for such was perhaps indicated by the range in backgrounds and interests of the 207 people who registered for the short course, which was held on Dec. 13-14 in San Francisco. Here we present a summary of the 16 presentations (one for each chapter in the new volume).
Our first day began with a review by Julia Hammer on crystal kinetics and 'dynamic' experiments (where experimental T or P are not fixed), followed by a review of methods for estimating P-T conditions of crystallization. Hammer's review noted the importance of undercooling, not just to the formation of mineral textures, but also to its apparent requirement for the formation of melt inclusions. Her review also touched on observed decreases in crystal growth rates with time: even at a fixed difference between bulk liquidus T and experimental run T, the thermodynamic undercooling nevertheless decreases with time during an experiment (and in nature), due to crystal growth-driven changes in the composition of a magmatic system, as crystals and liquid approach equilibrium. This decrease in undercooling should decrease crystal growth rates, and so implies that true instantaneous growth rates are probably only captured in the earliest stages of a dynamic experiment. Keith Putirka followed with an analysis of mineral-melt-based thermometers and barometers, calibrated from static experiments (where T and P are fixed, in attempt to achieve equilibrium). He showed that to apply such models, natural systems must approach equilibrium. Tests for equilibrium usually rely on mineral-melt exchange coefficients that are expected to be independent of T and P, such as Fe-Mg exchange between olivine and liquid (Roedder and Emslie, 1970). Putirka showed that dynamic experiments can be used to test our 'tests of equilibrium' and that not all such tests provide a perfect filter. Where possible, multiple tests, (e.g., independent tests of equilibrium, or better, independent P-T estimates) can be crucial for narrowing uncertainty, and even then P estimates tend to be more precise when averaged. Lawford Anderson followed with a wide-ranging talk, showing applications of thermometers and barometers to granitic systems in the Sierra Nevada and the Transverse Ranges of California. In plutons from the latter, the Ti-in-zircon thermometer yields temperatures that are inversely correlated with Hf, with temperatures that range from just above the solidus to near the liquidus-and range to values that are higher than calculated for zircon saturation. This indication of a wide temperature range may be valid, but Anderson warned that the activities of trace components in minerals can be sensitive to modal mineralogy and liquid composition and that new experiments are needed to document such. The next talk, by Thor Hansteen and Andreas Klügel, reviewed methods to estimate pressures of fluid inclusions. Fluid inclusions have the potential for providing very precise P estimates, provided that inclusions are homogeneous, isochoric and remain closed ('Roedder's Rules'). Systems that violate these rules, though, can still be useful; Hansteen and Klügel presented frequency plots of P estimates, indicating how the peaks in such plots can be correlated to multiple depths of magma storage. A key limitation to the approach is the error associated with existing equations of state for fluids, which may not match natural fluid compositions, and are not derived from experiments conducted at high P and T. A promising result, however, is that fluid inclusions and mineral-melt equilibria, in some instances yield very similar pressure ranges.
Jon Blundy then presented work conducted in collaboration with Katherine Cashman, illustrating the utility of merging various petrologic approaches. They showed that at Mount St. Helens, mineral textures, melt inclusions and several different P-T estimation techniques yield an internally consistent picture of the depths and temperatures of partial crystallization and degassing rates at that volcano-a picture that is furthermore consistent with seismic studies that indicate depths of transport. Blundy also showed a P-T diagram illustrating calculated gradients in crystallinity and degree of volatile saturation, emphasizing that minerals within a single magma chamber may sample such variability and that T can be used as a proxy for proximity to a magma chamber wall. Malcolm Rutherford then presented a summary of methods for determining magma ascent rates from experimental investigations, and a survey of natural ascent rates. Rutherford showed that explosive magma ascent rates, as determined from amphibole-breakdown reactions, are similar to rates derived from decompression-induced crystallization (ca. 0.2 m/s for explosive ascent at Mount St. Helens), and both estimates are within an order of magnitude of estimates derived from seismicity (0.6 m/s at Mount St. Helens). Extrusive ascent rates are one to two orders of magnitude lower, and explosive ascent rates appear to be positively correlated with explosivity, indicating an important forensic role for petrology in the assessment of volcanic hazards.
We completed our first day with a review of volatile contents, and models used to estimate such. Nicole Métrich and Paul Wallace emphasized that to calculate the P of melt inclusion entrapment, the inclusion must be vapor saturated and there should be no post-entrapment volatile-loss, otherwise P estimates are systematically low. They also suggested that CO2 loss (to shrinkage bubbles) can be significant. Wallace and Métrich noted, for example, that the highest pressures from olivine inclusions do not exceed 400 MPa and that fluid inclusions tend to yield higher pressures; they conclude that melt inclusions generally record simultaneous degassing and crystallization (as opposed to deep-seated storage). With these caveats, they showed that inclusions from a single eruptive unit can yield a range of saturation pressures; if magma ascent is the cause of degassing, this would seem to imply that inclusion capture is concurrent with magma rise. Their survey also showed that primitive phenocrysts tend to have higher CO2 and H2O contents, and that in some cases CO2 fluxes may trigger H2O loss, which in turn can trigger crystallization. They also showed that melt inclusion S contents are positively correlated with eruptive volume, and can be used to reconstruct volatile budgets for a given eruption, with implications regarding climatic impacts. Gordon Moore concluded our first day with a summary of the characteristics of a good volatile solubility experiment, and tests of models that predict volatile contents or saturation pressures. Among the most important observations are that CO2 and H2O solubilities are highly sensitive to melt composition, which includes an interdependence of the two solubilities. Because of such sensitivities, saturation models should not be extrapolated outside the bounds of their calibration data. At present, only the models of Newman and Lowenstern (2002) (VolatileCalc) and Papale et al. (2006) account for compositional variations in mixed CO2-H2O-bearing melts. Both models appear to work well for rhyolites, while the Papale model appears to work better for more mafic systems, but has difficulties with Ca-rich basalts. For improvements to either model, new solubility experiments using mixed H2O + CO2 systems are critical.
Our second day began with an overview of melt inclusions, and continued with a discussion of isotopic studies. Adam Kent's review of melt inclusions emphasized both their benefits: they capture a wider range of complexity than revealed by whole rocks-and their pitfalls: they may record boundary layer melts, or melts that are otherwise not representative of macro-scale evolutionary processes. He and other presenters emphasized a plot by Faure and Schiano (2005) which shows how Ca/Al ratios can be used to differentiate whether melt inclusions have trapped far-field (i.e., representative) or near-field (non-representative compositions); Kent's survey showed that most inclusions appear to trap far-field compositions. Melt inclusions are also subject to re-equilibration, though their diversity indicates that magma transport rates greatly exceed the rates at which re-equilibration occurs. Frank Ramos and Frank Tepley followed with a summary of isotopic studies of mineral grains. The microsampling procedures they reviewed have been key in identifying potential equilibrium, and certain disequilibrium, between different grains and their host rocks and within individual crystals. Especially interesting are examples where individual crystals yield cores in disequilibrium and rims are in isotopic equilibrium with adjacent glass. Also, intergrain heterogeneity in some localities that has been thought to result from age differences, can in some instances be attributed to mixing between two components. Tepley and Ramos also illustrated some practical aspects of determining disequilibrium within and between crystals using both micromilling and laser ablation sample removal techniques, and low-blank chemical processing. Ilya Bindeman then followed with a survey of O isotope ratios from single crystals. Like their radiogenic counterparts, individual crystals yield a much wider range of O isotope ratios than their whole rock hosts, and much more heterogeneity than contained within bulk mineral separates. Among the more important results is that O isotopes provide a powerful tool for tracing the recycling of hydrothermally altered crust into magmatic systems. Oxygen also has an advantage of being slowly diffused compared to cations, and so O isotope heterogeneity can be preserved over longer times scales than other diffusing species. Our discussion of isotopes concluded with a review of crystal ages obtained from U-series disequilibria by Kari Cooper and Mary Reid. Cooper and Reid began their presentations with a discussion on how U-Th-Ra disequilibria are measured, plotted and interpreted in volcanic rocks. This led to a more in-depth discussion on each system (e.g., U-Th and Ra-Th disequilibria), including practical applications to volcanic systems. Their review of methods indicated that collection of such data is a highly time-intensive process, but with substantial rewards. For example, they showed that at Lacher See volcano in Germany, some flows yielded minerals with ages identical to (whole-rock derived) eruption ages, but that stratigraphically lower, more evolved whole rocks (presumably from the top of the magma chamber) hosted minerals that were 17 ka older than recorded by the whole rock system-perhaps representing material inherited from an earlier magmatic episode, and/or indicating the minimum subterranean life span for the nominally 12 ka Lacher See eruptive system. Cooper and Reid also showed U-series disequilibria at Mount St. Helens, which are suggestive of magma residence times on the order of decades to centuries.
The second day concluded with a series of talks related to diffusive time scales, mineral textures and physical models of magma extraction. Fidel Costa provided a detailed overview of the techniques used to obtain time-scales from diffusion profiles. Costa emphasized that diffusion profiles can yield times scales that are much shorter than can be accessed by radioisotope methods, and that the very young ages (compared to radioisotope methods) determined from diffusion profiles (mostly < 100y), reflect entrainment of older crystals and periods of crystal overgrowth-effectively re-equilibration, as opposed to the original time of crystal growth, as measured, for example, by U-series methods. Costa used the Bishop Tuff as an example, where diffusive time scales for Ti in quartz are very short (ca. 100 y) and likely reflect late stage re-heating and later crystal overgrowth, which may be especially important in large magmatic systems. As further evidence of such complexity, Martin Streck followed with a review and update of the textural observations that spurred the renewed interest in volcanic minerals more than 30 years ago. Like Bindeman, Streck emphasized that when individual crystals are viewed in detail, genetic terms as xenocryst and antecryst (older crystals from a related, but earlier magmatic episode) effectively lose their meaning. Individual crystals often record several episodes of growth and re-equilibration, and thus many phenocrysts (in the original sense of Iddings, simply a crystal that is conspicuous from groundmass or surrounding crystals) may have xenocryst cores, antecryst zones, and equilibrium rims, as the crystals are passed from one rock/magma to another. Streck showed how traditional methods of optical mineralogy, and other imaging methods, reveal different types of zoning, and that at Arenal, such textural studies of basaltic andesite lava flows allowed a precise enumeration of magmatic events (five) that contributed or affected the crystal population. The following discussion, however, indicated the discouraging trend that fewer students than ever are being trained to use a petrographic microscope.
Pietro Armienti followed Streck with a review of crystal size distributions (CSD) to estimate crystallization and magma transport rates. Armienti first showed that it is possible to derive the same CSD for a given sample regardless of the scale of sampling (which in the example from Mt. Etna range from a 7cm2 thin section, to a photo covering >800 cm2), provided one is careful with small- and large-end truncation effects. Armienti's review then showed that within a frequency plot of numbers of crystals vs. crystal length, the slope of the curve directly reflects the ratio of nucleation rate to growth rate, which should be constant when undercooling is constant. A key aspect of CSD then is that their slope directly reflects undercooling, and any changes in slope (at least in a closed system) can be interpreted as such. An interesting finding at Mt Etna is that CSD for rocks sampled near the vent are identical to those sampled downstream, indicating that crystallization occurred prior to eruption, not during downhill flow. Armienti also showed some potential new applications of CSD at Stromboli, where peaks in CSD may indicate degassing.
Our final talk by George Bergantz illustrated some of the latest work in attempting to understand physical mechanisms of magma storage and transport, especially as related to high-Si eruptions. Bergantz showed that many arc-related systems reveal a gap in SiO2 among erupted products, an observation now named for Reginald Daly (Daly Gap), while others are strikingly homogeneous (monotonous intermediates of Hildreth, 1981). Bergantz showed how numerical studies indicate that convection, in different manifestations, can explain both suites. Sluggish convection can create or redistribute heterogeneities, as plumes move material from one part of a system to another, so producing thermal and chemical gradients, especially if the Reynolds number is low (Re<1, i.e., laminar flow). Even at high Re (>104, i.e., turbulent flow), heterogeneities can also be produced if convection is limited to a single overturn. Bergantz observes that monotonous intermediates preserve much microscopic-scale heterogeneity, and so reflect multiple overturn events, despite the fact that such flows tend to be rich in SiO2 and crystal-rich, and so are viscous and resistant to forces that drive convection.
Although recent advancements initiated the organization of the short course and volume, the presentations and discussions that followed showed there is much work to be done. Undoubtedly, there remains a clear need to refine various tools through additional experiments, so as to better determine volatile saturation conditions and equations of state for complex fluids, as well as crystal kinetic parameters, and the development of mineral textures. The short course also highlighted that many current lines of investigation are highly complementary, and can be used to great effect in concert. For example, U-series ages appear to indicate the earliest stages of magma generation while diffusion profile ages inform us of the latest phases of transport. Similarly, mineral-melt barometers appear to inform us about the deeper parts of volcanic systems, and volatile-saturated equilibria inform us of the shallower-while fluid inclusions appear to record both, perhaps with higher precision. An alliance of methods can also provide key tests of our assumptions and interpretations. For example, if kinks in CSD are an indicator of degassing, then melt inclusions from related rocks with non-kinked CSD should yield higher volatile contents; if melt inclusions are being trapped largely upon upward transport, the times scales of such residence and transport, as might be inferred from diffusion profiles, should be consistent with transport times that might be inferred from a combination of ascent rate estimates from hornblende reaction-rims, and estimates of the depths from which magmas are delivered, as determined from geobarometers. To the extent that such tests yield a coherent picture of a volcanic system, the advances outlined at the short course and within the volume illustrate the promise of petrology and mineralogy for affording fundamental tests of the evolution of magma storage, transport and eruption.
Anderson AT (1976) Magma mixing: petrological process and volcanological tool. J Volc Geoth Res 1:3-33.
Anderson AT, Wright TL (1972) Phenocrysts and glass inclusions and their bearing on oxidation and mixing of basaltic magmas, Kilauea volcano, Hawaii. Am Min 57:188-216.
Barth TW (1934) Temperatures in lavas and magmas and a new geologic thermometer. Naturen 6:187-192.
Darwin C (1844) Geological observations on the volcanic islands. London, Smith-Elder
Davidson, JP, DeSilva, SL, Holden, P, Halliday, AN (1990) Small-scale disequilibrium in a magmatic inclusion and its more silicic host. J Geophys Res 95:17661-17675
Dungan MA, Rhodes JM (1978) Residual glasses and melt inclusions in basalts from DSDP legs 45 and 46: evidence for magma mixing. Contr Min Petrol 67:417-431.
Eichelberger JC (1975) Origin of andesite and dacite: evidence of mixing at Glass
Mountain in California and at other circum-Pacific volcanoes. Geol Soc Am Bull 86:1381-1391
Faure F, and Schiano P (2005) Experimental investigation of equilibration conditions during forsterite growth and melt inclusion formation. Earth Planet Sci Lett 236: 882-898
Hildreth W (1981) Gradients in silicic magma chambers: implications for lithospheric magmatism. J Geophys Res 86:10153-10192
King C (1878) Systematic Geology. U.S. Govt Print Office, Washington D.C.
Marsh BD (1995) Solidification fronts and magmatic evolution. Mineral Mag 60:5-40
Newman S, Lowenstern JB (2002) VolatileCalc: a silicate melt-H2O-CO2 solution model written in Visual Basic for Excel. Comp Geosci 28:597-604
Papale P, Moretti R, Barbato D (2006) The compositional dependence of the saturation surface of H2O+CO2 fluids in silicate melts. Chem Geol 229:78-95
Roedder, E (1965) Liquid CO2 inclusions in olivine-bearing nodules and phenocrysts from basalts. Am Min 50:1746-1782
Sorby HC (1858) On the microscopic structures of crystals, indicating the origin of minerals and rocks. Geol Soc London Q J 14:453-500