RICERCA ITALIANA     

Research program

Crystallization kinetics and time scales of magmatic proceses as recorded in volcanic rock textures

University Co-ordinator

Università degli Studi di CAMERINO - SCIENZE DELLA TERRA - ()

Research Unit Leader

Michael Robert Carroll

Description

Objectives
The proposed experimental studies will focus on measuring plagioclase growth rates (G) in shoshonitic melt and alkali feldspar and plagioclase growth rates in a trachytic melt. Of major interest are:
1. how does growth rate vary with indercooling and melt composition (shoshonitic, trachytic)?
2. how does the mode of inducing undercooling (lower T at constant P, lower P(H2O) at constant T) affect G-dT relations?
3. how do the growth mechanisms (interface controlled, diffusion controlled) vary with dT and melt composition?
4. how do nucleation densities vary with dT and melt composition?

Growth rates are not constant values but instead vary with degree of undercooling (dT) – undercoolings of interest range from tens of degrees to several hundreds of degrees. As discussed previously there are two mechanisms that may act to impose undercooling on a given melt composition. The first is a drop in temperature, as occurs when a magma cools, either within the crust (typically small values of dT), or during magma ascent, eruption and cooling at the earth’s surface (larger values of dT). The second mechanism for imposing undercooling on a melt is to change the composition, and for hydrous magmas, decompression-induced degassing of H2O may undercoolings of hundreds of degrees, even for a system that is nominally isothermal. This method of inducing undercooling is of particular interest for magma ascent processes within the crust and is most useful for studies that seek to link groundmass crystallinity and microphenocryst-groundmass melt compositions to magma ascent history.

Crystal growth due to decompression-induced degassing of rhyolitic melts (in dacitic to andesitic bulk compositions) has been the subject of recent experimental and analytical studies by Blundy and Cashman (2001), Couch et al. (2003a, b), Couch (2003) and Hammer and Rutherford (2002). In addition, Metrich and coworkers (e.g., Clocchiati and Metrich, 1984; Metrich and Clocchiati, 1989, Metrich et al., 1993; Metrich and Rutherford, 1998; Metrich et al., 2001) have presented convincing evidence that shallow-level degassing of H2O drives much of the pre-eruptive crystallization at Etna and Stromboli, without significant cooling of the magmas. The principal objective of the experimental work described in this proposal is to obtain quantitative measurements of feldspar growth rates for a range of undercooling values for use in interpretation and modeling of textural variations (CSD, groundmass crystallinity, microphenocryst compositions) in natural samples. We will not make an explicit attempt to isolate nucleation and growth, as done in previous two-stage studies of nucleation (e.g., James, 1982; Davis and Ihinger, 1999) or studies of growth rates with seed crystals added to the melt (e.g., Muncill and Lasaga, 1987, 1988). Our experimental approach will concentrate on crystal growth rates following undercooling induced by decompression – these results will complement the atmospheric pressure experiments and will allow evaluation crystal growth rates from depth to the earth’s surface: the experimental results will also provide quantitative estimates of nucleation densities (number of crystals per unit volume of melt) for different values of dT. In addition, using measured growth rates we will be able to use CSD data (ln(N) – L correlation) on natural samples to estimate nucleation rates using the fact that N(T) = no*G*t = J*t, where N(T) = total number of crystals, no = nucleation population density at zero crystal size, t = crystallization timescale, G = growth rate, J = nucleation rate (i.e., with knowledge of G, from experiments, we can calculate values for J using CSD of natural samples). This approach will also help circumvent problems concerning heterogeneous versus homogeneous nucleation, which for complex natural compositions with well-defined intervals of crystallization presents problems not present in simple single-component systems.

We also want to investigate differences in crystal growth rate (and growth rate - dT variations) when undercooling is imposed in different ways, either by a temperature drop, or by loss of water during ~isothermal magma ascent (degassing induced undercooling due to raised liquidus temperatures in melts containing less dissolved H2O). Cooling experiments at atmospheric pressure together with our decompression induced crystallization data will allow evaluation of differences/similarities between these two mechanisms for inducing undercooling. The results of Couch (2003) for a granitic composition previously studied in isobaric cooling experiments by Swanson (1977) suggest that maximum growth rates occur at smaller values of dT when growth is induced by decompression. Growth mechanisms (continuous layer growth, screw dislocation, surface nucleation) can be evaluated using variations between the reduced growth rate (Gr: Turnbull and Cohen, 1960) and dT, as shown by Uhlmann (1971) and Kirkpatrick (1975). At higher undercoolings it is likely that diffusion-controlled growth becomes more important and non-planar crystal-melt interfaces develop, resulting in a variety of anhedral crystal morphologies (dendrites, spherulites, hopper shapes). In these cases compositional gradients develop near the melt-crystal interface and in some cases these gradients can be measured by electron microprobe analysis (Muncill and Lasaga, 1987). We will attempt similar measurements on our experiments at large T.

Experiments
Shoshonite experiments:
Experiments on plagioclase growth kinetics in a shoshonitic melt subjected to isothermal decompression will concentrate on a sample corresponding to the crystal-poor gold colored pumices periodically produced during larger than normal explosive events (paroxisms of Barberi et al., 1993). The major difference between scoriae from typical Stromboli activity and gold pumice is that the gold pumices are crystal-poor (~5% crystals), despite having major element compositions similar to the normal scoriae products of Stromboli (~50% crystals, mainly plagioclase). It is generally proposed that the volatile-rich and crystal-poor magma responsible for the gold pumice is the primary magma arriving in the Stromboli system, and the black scoriae represent the material after “processing” (cooling, degassing and crystallization) in the shallower Stromboli plumbing system (Francalanci et al. 1989, 1993; Metrich et al, 2001). The Stromboli shoshonites and Etna hawaiites (e.g., sample studied by Metrich and Rutherford, 1998) are actually very similar in major element composition: Stromboli has 0.67 wt% more SiO2, 0.34 wt% more TiO2, 1.23 wt% more Al2O3, 1.2 wt% less FeO, 0.8 wt% less MgO, less Na2O and more K2O but almost equal total Na2O+K2O (4.29 vs. 4.18 wt%), and similar pre-eruptive melt H2O contents (2.5-3 wt%). Given the similarity in composition (and thus liquidus temperatures and melt viscosities) we believe that the crystal growth data for the shoshonitic composition should be applicable also to the Etna composition; eventual differences will probably be within the uncertainties on the growth rate and nucleation density measurements. This assumption will be verified using the more easily obtained (in terms of time and experimental costs) 1 atm experimental data . If necessary a small number of high pressure experiments may be made on the Etna composition, concentrating on plagioclase growth kinetics and using the published phase equilibrium data of Metrich and Rutherford (1998) to constrain equilibrium liquidus temperatures.

Crystallization produced by water loss accompanying magma ascent will be evaluated in experiments done at constant temperature but with variable P(H2O). Evaluation of crystallization kinetics also requires knowledge of equilibrium crystallization temperatures as a function of P(H2O) in order to evaluate the effective dT at different pressures. For this reason the Plag liquidus temperature will be determined for P(H2O) between 70 MPa and atmospheric pressure for the shoshinitic composition of interest. For the crystallization kinetics experiments, the samples will be first held at 70 MPa, resulting in melt H2O contents of ~2.5 wt%, in agreement with pre-eruptive H2O contents estimated by Metrich et al (2001) using analyses of olivine-hosted melt inclusions. After initial equilibration for several hours at conditions slightly above the liquidus at 70 MPa (T to be determined in separate experiments), pressure will be dropped by a certain amount and the sample will be left at the final pressure for different lengths of time before quenching. This procedure will be repeated for final pressures ranging from 60 to ~10 MPa, allowing access to a range of undercooling values, with undercooling increasing with decreasing final pressure of the experiment (the lowest final pressure to be investigated will be limited by capsule size and amount of water exsolved; if too much water exsolves the pressure inside the capsule may exceed the external confining pressure and the capsule will rupture – the minimum pressure accessible remains to be verified but previous experience suggests that it will be in the range 10-15 MPa). Undercooling at the different pressures investigated will be evaluated using the equilibrium experiments designed to define the plagioclase liquidus temperature between 70 MPa and atmospheric pressure.

Optical and BSE imaging (on the Camerino SEM) will be used to measure plagioclase crystal sizes as a function of time spent at the subliquidus temperature (pressure) in order to evaluate crystal growth rates for a range of undercoolings (10s of degrees to ~200°C). Nucleation densities will be estimated following the methods of Hammer et al. (1999) and appropriate stereologic corrections (e.g. Cheng and Lemlich, 1983; Armienti et al., 1994). Growth rate will be estimated for 10 largest microlites in each sample using G=(LW)^0.5 / 2t where L and W are measured lengths and widths and t is time; this is the method used in a number of previous studies (Fenn, 1977; Swanson, 1977; Hammer and Rutherford, 2002; Couch et al, 2003) and results yield maximum growth rates that are more easily compared with previous studies. Results from this method are also consistent with result from with more detailed and time-consuming CSD analysis (Hammer et al., 1999). Mineral and melt compositions in selected experiments will be determined using the electron microprobe in the CNR laboratories in Rome (300 euro per day – estimated 6 days). These experiments will be done in Ar-pressurized TZM pressure vessels using a Ni-NiO buffer to fix oxygen fugacity. We have requested funds in this proposal for setting up a rapid-quench TZM vessel in Camerino. This involves a SiC-heated vertical furnace, TZM (Ti-Zr-Mo alloy) pressure vessels and rapid-quench extensions (quench times of a few seconds; see Carroll and Blank, 1997 for details), a pump to produce Ar at high pressures and a pressure gauge. Additional equipment (valves, high P tubing, closed-circuit cooling system, electronic components, etc.) is already available in the Camerino experimental laboratories.

Trachyte experiments
Experiments on alkali feldspar and plagioclase growth kinetics in a trachytic melt will concentrate on decompression-induced crystallization using a trachytic composition for which we are currently studying the water-saturated phase relations at pressures of 20-250 Mpa (using hydrothermal pressure vessels already set up in Camerino – these vessels, made of a Ni-rich alloy, cannot be used for the shoshonite experiments because they cannot be used above maximum T of ~900°C). The trachytic starting composition comes from the compositionally zoned Breccia Museo eruption (Campi Flegrei; Melluso et al., 1995) and is representative in terms of major element composition of many of the intermediate trachytic products of the Campi Flegrei area. At 200 MPa, NNO, it crystallizes small amounts of biotite (Bt) and magnetite (Mt) near the liquidus (~850°C) followed by significant amounts of sanidine near 820°C and albitic plagioclase near 800°C. We propose to investigate the growth kinetics of alkali feldspar and plagioclase in this trachyte using undercooling imposed by decompression-induced water loss from the melt in order to evaluate feldspar growth rates in trachytic melts subjected to near-surface decompressions. The growth rates obtained from this work will allow us to better interpret variations in groundmass crystallinities in trachytic magmas in terms of dynamics of magma ascent, as has been done recently for a number of eruptions involving calc-alkaline dacites to andesits with rhyolitic groundmass compositions (e.g., Mt. St. Helens; Blundy and Cashman, 2001: Montserrat; Couch et al., 2003a,b: Pinatubo; Hammer and Rutherford, 2002). The proposed experiments for a trachytic melt will be similar to those proposed for the shoshonitic composition. The experimental starting material is a powder made from almost crystal-free obsidian blocks found in the Breccia Museo deposit (Melluso et al., 1995). In the experiments, samples will be first held slightly above the liquidus at 200 MPa (~850°C) and then they will be ~instantaneously (less than 1 minute) decompressed at constant T to different final pressures (different dT values) and left to allow crystal growth for different time durations. Increase in crystal size with time at a given final pressure will be used to estimate growth rate and series of experiments at different final pressures will be used to evaluate how growth rate varies with undercooling (using methods previously described for shoshonitic decompression experiments). Samples will be analyzed optically and by BSE-SEM for measuring crystal sizes, and by electron microprobe to determine phase compositions (estimated 10 days microprobe time). Since we are already studying the phase relations of this composition it will not be necessary to conduct additional phase equilibrium experiments in order to constrain undercooling values at the various crystal growth pressures investigated. The results of these experiments will be used in collaboration with researchers from the Osservatorio Vesuviano who have recently begun measurements of crystal size distributions for a variety of Campi Flegrei eruption products (Monte Nuove, Agnano-Monte Spina)

The timetable for work of the RU Camerino is as follows:
Year 1
Finish setting up TZM apparatus (to start immediately using funds available – FD); decompression and phase equilibrium (Plag liquidus) experiments on shoshonitic composition to measure growth rates and nucleation densities as function of undercooling; analysis of experiments and integration with 1atm results of Pisa group for application to eruptive dynamics of Stromboli and Etna
Year 2
Continued integration with 1atm results of Pisa group for application to eruptive dynamics and CSD measurements for Stromboli and Etna; preparation of publication(s) (together with Pisa researchers) concerning Stromboli and Etna dynamics; experiments and analytical work on plagioclase and alkali feldspar growth in trachytic melt subjected to various degrees of decompression to induce undercooling; evaluation of feldspar growth in trachtic melts and applications to selected Campi Flegrei eruptions, in collaboration with Osservatorio Vesuviano researchers; preparation of publication concerning feldspar growth rates in trachytic melts.