RICERCA ITALIANA     

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

Università degli Studi di Camerino

Abstract

The crystals present in an igneous rock and the observed variations in their dimensions and compositions reflect the integrated pressure (P) –temperature (T) – composition (X) – time (t) history of the sample. Because crystals take a finite time to nucleate and grow (rates depending on undercooling, dT, composition, etc), variations in their sizes, size distributions, and compositions can provide insights into magmatic processes and their time-scales. In order to use observed variations in the textures of igneous rocks to gain quantitative insights into the timescales of magmatic processes we require quantitative information on the rates of nucleation (J) and growth (G) as a function of undercooling (dT) for the crystals and melts of interest. Unfortunately, experimental data on nucleation and growth kinetics are not abundant (none for the shoshonitic or trachytic compositions of interest in this proposal) and measured values in other magma types vary by many orders of magnitude, depending on type of crystal, melt composition and degree of undercooling. Furthermore, there are no suitable theoretical models capable of calculating nucleation and growth rates without some experimental constraints. In this proposal we seek funding to allow experimental measurements of crystal nucleation and growth rates in melts of shoshonitic, trachybasaltic and trachytic composition, for application to natural samples from Etna, Stromboli and Campi Flegrei. We will study nucleation and growth induced by both decrease in temperature and by decrease in P(H2O) at constant temperature. The first case simulates magma cooling in a near-surface environment, whereas the second simulates what happens when water-bearing magmas ascend towards the earth’s surface. The second case is of particular interest because even 2-3 wt% dissolved H2O can decrease magma liquidus temperatures by 100-200°C; with decreasing pressure, H2O saturated melts will loose water, resulting in crystallization even if T remains constant. Recent studies have shown that this may be an important process for inducing crystallization in many volcanic systems, including Etna, Stromboli, Campi Flegrei, Mt. St. Helens, Unzen, Pinatubo, Montserrat (see references below). This crystallization can greatly influence magma physical properties (viscosity) and ultimately eruption behavior because the timescales of such processes are similar to eruption timescales (hrs to months, depending on eruption styles), resulting in strong non-linear feedbacks between ascent-rate, degassing, magma crystallization kinetics, and effusion rates. The dimensions and compositions of crystals formed during magma ascent depend strongly on factors such as ascent rate and thus can potentially provide insights into the timescales of magma migration within the crust. However, in order to quantify these timescales we require experimental data on crystal nucleation (J) and growth (G) rates for the melts and crystals of interest and for a range of undercooling values. The main objectives of the work described in this proposal are to acquire the necessary experimental data on J and G as a function of undercooling for shoshonitic, trachybasaltic and trachytic melt compositions, and to apply these data, in combination with measurements of crystal sizes and size distributions (CSDs) and appropriate theoretical models, to evaluation of the dynamics of eruptive processes for recent eruptions at Etna and Stromboli, and for several older eruptions in the Campi Flegrei (e.g., Monte Nuove, Agnano Monte Spina). <<<

Principal Investigator

Michael Robert Carroll Università degli Studi di CAMERINO

Research Objectives

The objective of the research described in this proposal is to obtain experimental, observational and theoretical data necessary for quantitative evaluation of timescales of magmatic processes (storage, ascent, eruption) for alkaline magmas produced at Etna, Stromboli and the Campi Flegrei. In order to used observed textural variations to obtain insights into the dynamics of volcanic systems it is necessary to have quantitative information on the rates of nucleation (J) and growth (G) of crystals in the magma compositions of interest, under P-T-X conditions of interest. Experimental data for the magma compositions of interest (shoshonite, trachybasalt, trachyte) are not available, the few available studies of other magma compositions (mainly basalts and calc-alkaline rhyolites) show that J and G may vary by many orders of magnitude, and there are no suitable theoretical models capable of calculating these parameters. Of particular interest for this study is the kinetics of crystallization of plagioclase and clinopyroxene in shoshonitic and trachybasaltic (hawaiitic) melts, the kinetics of crystallization of alkali feldspar and plagioclase in trachytic melts, and the relations between magma ascent dynamics, crystallization kinetics, and textural evolution of phenocryst and groundmass crystal populations. The major questions to be answered include the following:

1. how does growth rate (G) vary with undercooling [dT=T-T(liquidus)] and melt composition (shoshonitic, trachybasaltic and trachytic)?

2. how does the mode of inducing undercooling (decrease T at constant P, decrease P(H2O) at constant T) affect G- dT relations?

3. how do the growth mechanisms (interface controlled, diffusion controlled) vary with T and melt composition?

4. how do nucleation rates and densities vary with dT and melt composition?

5. how can we interpret variations in CSDs and groundmass crystallinity for Etna, Stromboli and Campi Flegrei in terms of magma ascent and degassing history – modeling to be done using new and existing textural observations combined with new experimental results on variations in G and J with dT and magma composition).

Obtaining these objectives will require experimental measurement of crystal growth rates and nucleation rates as a function of undercooling in 3 different melt compositions. Undercooling will be produced by both changing temperature at constant pressure (simulating magma cooling at low pressure) and by changing P(H2O) at constant temperature (simulating degassing-induced crystallization during magma ascent). The experimental data, in combination with appropriate theoretical models and observational data on natural samples will be used to evaluate the timescales of magma storage, ascent and eruption processes for selected Stromboli, Etna, and Campi Flegrei eruptions. These volcanic systems present significant differences in eruptive behavior: Stromboli is an open-conduit shoshonitic volcano in ~continuous eruption; Etna is an active volcano with frequent eruptions, with recent magmas showing little difference in major element composition (trachybasalts) but large differences in crystallinity; the Campi Flegrei area has produced many trachytic eruptions but with large variations in eruptive style and intensity. The results of this study will allow the development of quantitative models to explain textural differences observed in volcanic eruption products. To date, petrographic studies have documented significant textural variations among the eruptive products of Etna, Stromboli and Campi Flegrei but without quantitative data on crystal nucleation and growth rates and their variation with undercooling, and on CSDs of natural samples, it is not possible to use the observational data to obtain insights into magma dynamics (e.g., ascent timescales, degassing history, storage history). The work described in this proposal will allow transition from a qualitative to a quantitative treatment of textural variations in alkaline volcanic rocks. <<<

Timescale

24 months

National and international background

Studies of the equilibrium properties of magmatic systems have a long history and today the basic phase relations are known for a wide variety of magma types under a broad range of P-T-X conditions. There also exist quite advanced thermodynamic models for calculating phase relations in complex natural systems (e.g., Ghiorso et al., 1983, and numerous subsequent revisions) The same cannot be said for kinetic aspects of magmas, such as the rates of crystal nucleation and growth, yet it is these parameters that ultimately control the textures that are observed. The crystals present in an igneous rock and the observed variations in their dimensions and compositions reflect the integrated pressure (P) –temperature (T) – composition (X) – time (t) history of the sample. Because crystals take a finite time to nucleate and grow, variations in their sizes, size distributions, and compositional variations can provide insights into magmatic processes and their time-scales. Such information is complementary to time scale information provided by studies of short-lived radiogenic isotopes (e.g., U-series isotopes), which typically provide information on timescales on the order of 10^3 to 10^5 years (e.g., Hawkesworth et al., 2000; Turner et al., 2000). Recent improvements in analytical methods for U-series isotopes should soon allow more reliable extension of these isotopic methods to overlap the timescales (hours to days to centuries) sampled by studies of textural variations and Crystal Size Distributions (CSDs; e.g. Marsh, 1988; Cashman and Marsh, 1988; Cashman, 1990).
Traditionally, studies concerning magma crystallization processes have considered heat loss and cooling as the principal driving force for crystallization. Crystallization induced by cooling has been investigated by studies of natural samples (e.g., Kirkpatrick, 1977; Cashman and Marsh, 1988; Armienti et al., 1994; Higgins, 1996a, b), by laboratory experiments (e.g., Fenn, 1977; Swanson, 1977; Lofgren, 1980; Muncill and Lasaga, 1987, 1988; Davis et al., 1997; McCoy and Lofgren, 1999), and in theoretical models (e.g., Hort, 1997; Hort and Spohn, 1991; Granasy and James, 1998). The results of such studies of cooling-induced crystallization have found numerous applications, from estimating magma residence times between eruptions (Magnan, 1990; Resmini and Marsh, 1995; Higgins, 1996b) to evaluating timescales of magma-mixing events and magma solidification (Ohnennestetter and Brown, 1992; Venezky and Rutherford, 1997; Crisp et al., 1994; Cashman et al., 1999), to developing models of magma chamber dynamics (Solomatov and Stevenson, 1993; Simakin et al., 1994).
However, another process that may strongly influence magma crystallization involves the fact that the presence of dissolved H2O in magma can strongly reduce liquidus temperatures (by hundreds of degrees for a few wt% H2O). This implies that degassing of H2O-bearing magmas during ascent will lead to increased liquidus temperatures and thus significant crystallization can occur even if magma ascent is under near-isothermal conditions (Wilson et al., 1980). The importance of degassing induced crystallization during magma ascent has been documented for a number of recent eruptions, including Mt. St Helens (Cashman and Blundy, 2000; Blundy and Cashman, 2001), Mt Pinatubo (Hammer and Rutherford, 2002), Montserrat (Couch et al., 2003a,b), Unzen (Nakada and Motomura, 1999), Stromboli (Metrich et al, 2001), and Etna (Metrich and Clocchiatti, 1989; Simakin et al., 1999). Furthermore, extensive microlite crystallization on the timescales of magma ascent can be expected to have significant effects on eruptive dynamics because of changes in rheological properties of magmas, due to increased crystal content, crystallization-induced changes in melt composition, and degassing-induced changes in melt H2O content (Papale and Dobran, 1994; Sparks, 1997; Melnik and Sparks, 1999; Nakada and Motomura, 1999; Hammer and Rutherford, 2002; Couch et al., 2003a).
For improving our understanding of the importance of degassing-induced crystallization during magma ascent it is clear that the timescales of ascent must be related to the timescales of crystal nucleation and growth. From available data on crystal growth rates (not actually very abundant), magma ascent time-scales of minutes/hours (explosive eruptions) to days/months (effusive eruptions) will have major effects on crystals with dimensions from tens to hundreds of microns. For this reason, studies of microlite sizes and groundmass textures offer the best possibilities to relate textural observations to eruptive time scales and volcanic eruption dynamics. Likewise, studies of CSDs of phenocryst populations provide the opportunity to estimate timescales of magma chamber crystallization processes (pre-eruptive).
The use of textural observations to gain insights into magmatic processes and their timescales requires quantitative data on nucleation and growth rates of crystals to complement observations and measurements on natural samples. For example, the use of measurements of crystal size distributions (CSDs – discussed below) to estimate magma residence times require knowledge of G, the crystal growth rate (e.g., Marsh, 1988; Cashman and Marsh, 1988). Likewise, use of observations concerning groundmass crystallinity to constrain magma ascent dynamics (e.g., Couch et al., 2003a,b; Hammer and Rutherford, 2002) requires knowledge of crystal growth rates and their variations with undercooling (the difference between temperature experienced and actual liquidus temperature for the crystal of interest – without undercooling there is no driving force for crystallization). However, crystal growth rates can vary by many orders of magnitude, depending mainly on type of crystal, melt composition (including water content) and degree of undercooling (Kirkpatrick, 1981). Interest in crystal nucleation and growth has a long history and much published work is found outside of the usual earth science literature. Quantitative data on crystal nucleation and growth relevant to compositions of geological interest are not very abundant, with many studies limited to simple synthetic systems which are quite different from complex natural systems (e.g., as reviewed in Kirkpatrick, 1981). As an introduction to the state of the art for quantitative textural studies of igneous rocks we first briefly describe some general ideas concerning nucleation and growth of crystals and the theory of CSD studies.

Crystal nucleation and growth
Nucleation involves the addition of atoms or molecules to a nucleus having the structure of the solid. Because of the energetic cost of maintaining a crystal-melt interface, nuclei below a certain critical size will not spontaneously grow. The critical size varies with the amount of undercooling (dT= T(liquidus)-T(actual)) and varies from an infinite size above the liquidus to smaller sizes at higher values of undercooling. For silicate liquids the nucleation rate typically varies from zero to a maximum value at some dT,
and then decreases for greater dT because diffusion rates become too slow (Fig. 1).



Nucleation is said to be homogenous if it occurs within the liquid, or heterogeneous if it occurs on the surface of another phase (which is often easier from an energetic point of view). Formal theories for nucleation are limited to single component systems and even for these there are significant differences between theory and experiment (Kirkpatrick, 1981; Cashman, 1990). There are no useful theoretical models for predicting nucleation rates in complex natural systems.
Crystal growth involves attachment of atoms or molecules to a stable nucleus and may be controlled by either the rate of attachment at the melt-crystal interface (interface controlled) or by the rate of diffusion of atoms through the melt to the growing crystal face (diffusion controlled). Growth may also be controlled by diffusion of latent heat away from the crystal-melt interface but for relatively viscous silicates (as opposed to liquid metals), heat diffusion is orders of magnitude faster that melt component diffusivities and thus should not represent a limiting factor for crystal growth. Because atom mobility decreases with decreasing temperature, the growth rate varies with dT, from zero at the liquidus (dT=0), rising to a maximum, and then decreasing for larger dT. On a diagram of rate versus dT, the maximum in growth rate frequently occurs at smaller values of dT compared with the maximum in nucleation rate. Models of growth rates in simple systems are more successful than models for nucleation rates in reproducing experimental observations but extension of such models to multicomponent natural systems require further work (e.g., Lasaga, 1982; Muncill and Lasaga, 1987, 1988).
The theory of crystal size distributions, introduced for geological problems by Marsh (1988), was originally developed for chemical engineering applications (Randolph and Larson, 1972). Cashman (1990) provides a review of CSD studies, theories and possible complications or problems of interpretation. A CSD analysis is based on measuring, for a given mineral, the cumulative number (N) of crystals as a function of crystal size (L); because crystal-size data are typically obtained on planar surfaces it is necessary to introduce stereologic corrections to obtain valid volumetric (3D) information (e.g., DeHoff and Rhines, 1972; Cashman and Marsh, 1988; Pareschi et al, 1990). The cumulative population data (N) are recast as a population density (n) by finding the slope of the cumulative size curve for different crystal sizes (i.e., n(L) = dN/dL). The data are presented as a plot of ln(N) versus L and it has been observed that many (but not all) volcanic rocks show regular, approximately linear relations between L and ln(n) (Fig 2).



For an open system at steady state, the slope of the ln(n) – L trend is equal to –1/Gt where G is crystal growth rate (assumed constant) and t is time, which for this case can be considered as residence time within the system. Another end-member case is a closed system lacking input or output of material (e.g., intrusion into a magma chamber, crystallization for a certain period of time, then rapid eruption). In this case, for a constant growth rate, L=Gt and the age of initial nucleation (magma intrusion) is give by the size L of the largest crystal. Obviously there are many possible complications to these two end-member cases but a major strength of the CSD approach is that it allows a quantitative representation of petrographic data for subsequent evaluation, modeling and interpretation. In particular, CSD can be accurately described on the basis of equations that relate the numbers of crystals and their sizes to kinetic factors of nucleation and growth as well as system boundary conditions (mixing, crystal settling, etc.). It should be clear however that any meaningful interpretation of CSD data requires quantitative knowledge of crystal growth and nucleation rates for the minerals and potential undercoolings of interest. One final goal of the proposed work is to arrive at the point where we can integrate thermodynamic equilibrium models for crystallization with models that account for kinetic factors, thus allowing the possibility to quantitatively model crystallization processes in terms of mass and energy balances and also time-scale. <<<