Magmatic degassing of multicomponent vapors and assessment of magma depth: application to Vulcano Island (Italy)

Earth and Planetary Science Letters 193 (2001) 467^481 www.elsevier.com/locate/epsl

Magmatic degassing of multicomponent vapors and assessment of magma depth: application to Vulcano Island (Italy) P.M. Nuccio a;b , A. Paonita a; * b

a Istituto Nazionale di Geo¢sica e Vulcanologia, Sezione di Palermo, Via Ugo La Malfa 153, 90146 Palermo, Italy Dipartimento di Chimica e Fisica della Terra ed Applicazione alle georisorse ed ai rischi naturali (CFTA), Universita© di Palermo, Via Archira¢ 36, 90123 Palermo, Italy

Received 10 July 2001; received in revised form 12 September 2001; accepted 13 September 2001

Abstract Degassing of magmatic H2 O, CO2 and rare gases plays a major role in understanding large-scale Earth processes and in the assessment of volcanic activity. Here we describe a quantitative model for magmatic degassing of H2 O^CO2 ^ noble gas^N2 mixtures. Our modeling takes into account non-ideal behaviors by adopting recently developed approaches for the solubility of H2 O^CO2 mixtures in silicate liquids and for noble gas partitioning in H2 O^CO2 bearing magmas. This new approach allows quantitative treatment of inert gas fractionation throughout the degassing of any H2 O^CO2 bearing natural magma in a wide range of thermo-baric conditions. Magma degassing simulations performed by our model have clearly displayed that dissolved H2 O and CO2 in the melt strongly affect inert gas degassing. Due to their modest solubility differences in H2 O-rich magmas, all the inert gases are strongly partitioned into vapor at early degassing extents, after the quick exhaustion of CO2 . In contrast, CO2 -rich melts retain dissolved helium longer because CO2 has firstly to be released from magma, whereas nitrogen and heavy noble gases undergo a similar or higher exsolution than CO2 at early magma degassing extents. We have successfully applied the degassing model to the active volcanic system of Vulcano Island (Italy), where several geochemical parameters have been monitored over the last decade. By quantitatively assessing magma pressure over time, the model has allowed us to reconstruct the rise of magma towards the surface. The results clearly show that the magma has reached low pressure and depth beneath Vulcano, thus interaction with the hydrothermal system, as well catastrophic magma rises may result in hazardous future scenarios. ß 2001 Published by Elsevier Science B.V. Keywords: volcanism; gases; degassing; magma transport; depth; models; Vulcano

1. Introduction

* Corresponding author. Tel.: +39-91-680-9400; Fax: +39-91-680-9449. E-mail address: [email protected] (A. Paonita).

Magma degassing exhibits crucial relevance in the majority of the volcanic processes. The segregation of the main volatiles (H2 O and CO2 ) from magma can induce signi¢cant changes in its physico-chemical properties, such as density and vis-

0012-821X / 01 / $ ^ see front matter ß 2001 Published by Elsevier Science B.V. PII: S 0 0 1 2 - 8 2 1 X ( 0 1 ) 0 0 5 1 2 - X

EPSL 6014 4-12-01

468

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

cosity and liquidus phase relationships [1^7]. The study of magma degassing of minor species (nitrogen, noble gases) is usefully applied for the study of large-scale geological processes, such as mantle evolution, Earth degassing and formation of the atmosphere [8^19]. A fundamental cause of magmatic degassing is magma ascent towards the surface and the related depressurization, vapor supersaturation and consequent release of volatiles [20^25]. In addition, magma ascent may lead to interaction with shallow aquifers, causing phreatic or phreato-magmatic explosions, or dramatic volatile exsolution causing magma instability and the onset of eruptions [26]. Ultimately, volcanic eruption can be considered as the last expression of a magma rise process, hence the detection and quantitative assessment of magma depth assumes particular relevance for the surveillance of active volcanoes. In view of the latter, investigations about the dissolved contents of volatiles in degassing melts and the chemical composition of exsolved vapors appear decisive. In fact, such variables depend on volatile solubilities and abundances in the melt, which, in turn, are more or less directly linked to the magma pressure (P) and depth [15]. Estimations of the depth at which magma stationed prior to its eruption have been possible by interpreting, on the basis of degassing models, the volatile content in igneous rocks [27,28]. More promising results for volcano monitoring come from geochemical approaches to the detection of magma rise by interpreting chemical variations in fumarolic gases on the basis of exsolution processes from the melt [29]. The central idea of such work has been that, by monitoring the chemical variations of various volatiles in magmatic gases and interpreting them in terms of degassing processes, we should be able to quantitatively evaluate magma pressure and its time variation. For this purpose, we have developed a degassing model for the H2 O^CO2 ^noble gas^N2 mixture, dealing with inert gases as minor species in the system. We chose H2 O and CO2 because they almost form the entire magmatic vapor phase, while inert gases are the best tracers of vapor^ melt fractionation, being unmodi¢ed by later reactions. Our model has been applied to the active

volcanic system of Vulcano Island (Aeolian Islands, Italy), where the chemical composition of fumarolic gases has been well monitored in the last decade [30^36]. 2. Previous studies involving magma degassing models Vapor saturation of magma can occur as a consequence of magma ascent and depressurization, whereas a similar e¡ect may derive from the isobaric crystallization of anhydrous minerals [20^ 25,37,38]. As suggested by both theoretical and experimental investigations [21^25], growing bubbles in the melt quickly reach and maintain chemical equilibrium with the surrounding liquid, even in silicic magmas rising with ascent rates three orders of magnitude higher than their common pre-eruptive values (W1034 MPa/s or 0.01 m/s). Indeed, non-equilibrium exsolution might just be due to dramatic depressurization occurring during explosive volcanic activity [24], however, here we will not be dealing with such a process. Based on vapor^melt equilibrium, several workers have developed models focused on pre-eruptive degassing of H2 O^CO2 mixtures [27,28,37,39^ 41]. Other models have focused on magmatic degassing of minor volatiles from silicate melts [8,11,15^19,42^44], particularly noble gases and nitrogen from mid-ocean ridge basalt (MORB), although the in£uence of main volatiles on the degassing of minor species was neglected and ideal behavior of minor gases in the vapor was supposed. Bottinga and Javoy [21] overcame the latter assumption in their degassing model for the H2 O^CO2 ^He^Ar mixture from basaltic melts, nevertheless the H2 O e¡ect on the solubilities of the other species was not still evaluated. Recently, Paonita et al. [45] have shown that the CO2 /He ratio in a degassing basalt would evolve along a sharply di¡erent path if anhydrous or H2 O bearing (5 wt%) melt degases, although the e¡ect of the gradual degassing of water on He solubility, as well as the non-ideal helium behavior in vapor, was neglected. Referring to the modalities of bubble release from natural magmas, closed system and open

EPSL 6014 4-12-01

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

system degassing have generally been the processes adopted [8,11,15^19,21,27,28,39^45]. The former predicts that the exsolved vapor stays in contact with the melt during the entire degassing process, whereas the latter assumes that in¢nitesimal aliquots of volatiles are exsolved in equilibrium from the liquid phase and immediately removed. On the other hand, an intermediate process between the open and closed system has been described by Nuccio and Valenza [29], and it will be hereafter referred to as discrete system degassing. According to the authors, magma normally rises by discontinuous separate events, so the degassing process would also occur by a series of distinct degassing events, each one exsolving its own gas phase. The vesiculation causes poorly soluble species (i.e. CO2 , He or N2 ) to be preferentially partitioned into the gas bubbles, so the vapor will be enriched in these species, resulting in an increased output of low solubility volatile components when bubbles reach the surface of the Earth [29]. 3. Methods of magma degassing calculation As previously a¤rmed, our modeling of magma degassing includes H2 O, CO2 , N2 and noble gases as volatile components. As the model is based on the attainment of chemical equilibrium between vapor and melt, the solubilities of all the involved volatiles and their reciprocal e¡ects assume crucial relevance. In this work, we have not developed new descriptions for the solubility of the gas mixtures but we have adopted existing models in literature, given that the most recent achievements in this ¢eld meet our requirements. 3.1. Solubility models selected from literature Noble gases always remain trace components in both melt and vapor of natural magmatic systems, while N2 may become at most a minor component only in the vapor phase, reaching concentrations around 0.1 mol% [12,15,17,46]. Therefore, we may assume that the thermodynamic characteristics of the H2 O^CO2 ^silicate

469

melt system will not be perturbed by the very slight amounts of such minor volatiles. Thus, the already existing models developed for the H2 O and CO2 solubility in silicate melts can work under these new conditions [47]. Solubility of H2 O^CO2 mixtures in silicate liquids has often been treated assuming Henrian behavior of the two volatiles in the melt and their ideal mixing in the vapor ([39,40,48] and references therein), the developed models working with speci¢c compositions of silicate melts or narrow compositional ranges. Recently, Papale [41] has calibrated a thermodynamic model for the solubility of H2 O^ CO2 mixtures in melts, which avoids the above assumptions and is able to calculate the solubilities of any mixed H2 O^CO2 vapor in any silicate melt composition. We have thus selected it to perform our calculations. On the other hand, the behavior of noble gases in both the silicate melt and coexisting vapor strongly depends upon the presence of the main volatiles [45,47]. Nuccio and Paonita [47] have recently developed a complete description in their Extended Ionic Porosity (EIP) model, which is currently the only tool capable of quantitatively evaluating Henry's constants of noble gases in any natural H2 O^CO2 bearing melts. The EIP model includes an Equation Of State (EOS) for the calculation of the noble gas fugacities in H2 O^ CO2 -dominated vapors. Both the EIP model and the EOS have been chosen for our purposes. As they had been only developed for noble gases, here we have extended them in order to describe the N2 solubility in magmas (see Appendix A). 3.2. Degassing calculations The adopted solubility models can be incorporated in an iterative procedure describing the fractionation of the involved volatiles throughout magma degassing. We initially ¢x both the magma pressure and the initial H2 O, CO2 and inert gas contents in melt. The latter may be selected so as to make the melt saturated at the initial pressure. Afterwards, the pressure is incremented downwards of a given step, leading to a supersaturated melt, and the dissolved volatiles have to be transferred to vapor phase in accordance

EPSL 6014 4-12-01

470

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

with their solubilities and concentrations. This means that the chemical equilibria deriving from the adopted solubility models, as well as mass balances between vapor and melt for each volatile (see Eqs. 1^5 of Table 1), have to be simultaneously satis¢ed at the new pressure, and all the involved variables are computed (the concentrations of all the volatiles in both vapor and melt). Again, the pressure is lowered by the given step and the above calculation is repeated, so as to compute the new concentrations of volatiles in the liquid and gas phase at the new pressure.

Both open and closed system degassing can be modeled by the above approach. In closed system degassing, the initial amounts of all the volatile in the magma remain the same (Eq. 6 in Table 1), whereas they decrease in the open system because the vapor is entirely removed from the magma at the end of each step of depressurization (Eq. 7 in Table 1). 3.3. Discrete system degassing Based on the de¢nition of discrete system de-

Table 1 Equation set used to model the degassing process No.

Equation

Reference

1

PH2 O W(13yCO2 )WP = QH2 O WxH2 O W(fH2 O ³)

[41]

2

PCO2 WyCO2 WP = QCO2 WxCO2 W(fCO2 ³)

[41]

xTH2 O 3xH2 O

xTCO2 3xCO2

[41]

5

ˆ …13yCO2 †3xH2 O yCO2 3xCO2 K hHe P He yHe ˆ P xHe xTHe …13xCO2 3xH2 O † ‡ xHe …xTH2 O ‡ xTCO2 31† yHe ˆ xTH2 O 3xH2 O ‡ xTCO2 3xCO2

6

Constant xTH2 O , xTCO2 and xTHe for closed system

This work

7

(xTH2 O )i = (xH2 O )i31 , (xTCO2 s )i = (xCO2 )i31 , (xTHe )i = (xHe )i31 for open and discrete system

This work

3 4

[47] This work

The equation set Eqs. 1^5 describes the solubility of a gas mixture H2 O^CO2 ^He in silicate melts, on the condition that helium is a minor component in the melt and its solubility follows Henry's law. Two equations like Eqs. 4 and 5 can be added to the set to include any other inert gas in the system, providing it may be treated as an Henrian component. Eqs. 1^3 compute vapor^melt equilibrium of H2 O^CO2 mixtures and their mass balance among dissolved, exsolved and total concentrations of volatiles. By using the interaction parameters and reference state for H2 O and CO2 (the hypothetical pure solute at P, T) de¢ned in [41], these three relations are thermodynamically self-consistent in a system having H2 O and CO2 as volatile species. Eqs. 1^3 with the parameters in [41] can still be used in a system that includes minor volatiles (i.e. rare gases in magmas), because it may easily be assumed that minor species do not a¡ect appreciably the activities of the main components in both vapor and melt. This has also been proven, within the experimental uncertainties, by runs involving the system He^H2 O^ CO2 ^silicate melt (unpublished data), although it is a very common condition in Earth Sciences investigations. Eqs. 4 and 5 compute vapor^melt equilibrium of the inert gas in H2 O^CO2 bearing silicate melts and its mass balance, assuming that the inert gas is a minor component in the melt. The value of Henry's constant is computed by the EIP model [47] as a function of P, T, dissolved H2 O and CO2 and melt composition. The reference state adopted for inert gas by [47] is the hypothetical pure solute at P, T and depends on the composition of the H2 O^CO2 bearing melt. The fugacity coe¤cient of the inert gas in vapor is computed by the EOS in [47] and depends on P, T and H2 O and CO2 proportions in the vapor. The EOS also predicts a Henrian behavior of the inert gases in vapor up to several mol% of concentration. Finally, Eq. 6 constrains the degassing process to follow the closed system modality, whereas Eq. 7 describes open and discrete system degassing. Symbols: x and y are the molar fraction of the volatile in the melt and vapor, respectively; P is the fugacity coe¤cient of the volatile in the vapor and Q is its activity coe¤cient in the melt; f³ is the fugacity of the volatile in its reference state; Kh is the Henry's constant of the inert gas (in pressure units; see [47]); superscript `T' indicates the total molar fraction of the volatile in the system. Subscript `i' is a counter of the depressurization steps and assumes integer values between 0 and e, where e is the total number of steps.

EPSL 6014 4-12-01

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

gassing (see Section 2 and [29]), this can easily be treated as intermediate condition with respect the previous ones. We simply leave the magma to degas by closed system until reaching a ¢nal pressure and, likewise, a ¢nal CO2 concentration in the exsolved vapor. At this point, the system becomes open (Eq. 7 in Table 1) and the existing vapor is entirely removed, leaving the melt saturated at the ¢nal pressure. The whole process represents a discrete event of magma degassing. We may again increment the pressure downwards and repeat the entire procedure until reaching a new ¢nal pressure value (or CO2 molar fraction in the vapor), so as to de¢ne a further event of discrete magma degassing. The above calculation procedures have been implemented in a computer routine, by which the degassing curves shown in this work were obtained. 4. Computed behavior of inert gases during magma degassing Our degassing model predicts that inert gas be-

Fig. 1. Residual fractions of dissolved inert gases and CO2 throughout closed system degassing of a H2 O-rich rhyolite at 900³C (thick curves) and 1200³C (dashed curves). The initial pressure and CO2 molar fraction in the vapor were 500 MPa and 0.2, respectively; the initial concentrations of inert gases derived from the average ratio between inert gases and CO2 in silicic magmas [43]. It is noteworthy that the curves for inert gases remain unmodi¢ed by varying the initial concentrations of the inert gases. Rhyolite composition from [58]. Curves for N2 and Ar are practically coincident.

471

Fig. 2. Residual fractions of helium throughout closed and open system degassing of a H2 O-rich MORB at 1200³C. The initial pressure, as well as CO2 and He molar fractions in the vapor were as in Fig. 1. Curves named `simpler models' were computed by using the equations in [43], which do not take into account the H2 O and CO2 e¡ect on the inert gas solubility. Liquid composition from [60].

havior throughout protracted magma degassing strongly depends on several boundary parameters, such as H2 O and CO2 proportions in the melt, silicate melt composition, pressure and temperature. In fact, these parameters in£uence the inert gas solubility throughout the entire degassing process, outlining a very complex scenario. In this light, numerical simulations of open or closed system degassing are useful to investigate the di¡erent variables. Helium solubility in H2 O-rich basalts or rhyolites [47] can be 15 times higher than that of CO2 in the same melt [41]. Hence, during gas fractionation from H2 O-rich rhyolitic and basaltic melts, helium will be extensively degassed when the less soluble CO2 is highly depleted in the melt (Figs. 1 and 2). This occurs at relatively early degassing extents because the CO2 amounts are low and the very soluble H2 O cannot prevent large He partition into the vapor phase. Helium concentration in vapor remains approximately constant until the CO2 exhaustion in gas phase (Fig. 3A,B). As soon as helium can be e¤ciently degassed, its depletion in the melt causes a decreasing concentration in vapor too. In H2 O-rich rhyolitic and basaltic magmas, heavier noble gases and nitrogen display similar solubility to that of He [47]; hence, their fractionation with respect to CO2 shows large

EPSL 6014 4-12-01

472

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

Fig. 3. N2 and helium molar fractions as a function of CO2 molar fraction in the vapor throughout open (thick lines) and closed (thin lines) system degassing of: (A) H2 O-rich rhyolite at 900³C, (B) H2 O-rich basalt at 1200³C. Initial conditions and melt compositions are as in Figs. 1 and 2. The N2 peak at low CO2 concentration does not depend on the initial N2 concentration, but is mainly due to the large increase in N2 solubility at low pressure and CO2 content (see text).

partitioning into vapor and e¤cient degassing only after the CO2 exhaustion in the melt (see Fig. 1 for rhyolite). On the other hand, CO2 -rich melt shows He solubility higher than CO2 solubility only by a factor three in basalt and ¢ve in rhyolite [41,47]. Moreover, due to the large amounts of dissolved CO2 to be previously degassed, helium will be strongly depleted only to an advanced degassing extent (Fig. 4). As a consequence, CO2 -rich melts exhibit a weak decline in helium content during a large part of their degassing history. At high degassing extents, when the dissolved CO2 is practically exhausted in the system, helium can be largely degassed from the silicate melt (Fig. 4). Therefore, CO2 -rich melts can retain the dissolved helium more e¤ciently than H2 O-rich magmas during protracted degassing. In CO2 -rich basalts, N2 and heavy noble gas solubility is slightly lower than that of CO2 [41,47], so their behavior is exactly the opposite with respect to helium : sharp impoverishment of nitrogen and Ar in the melt occurs during the early degassing (Fig. 4). An intermediate behavior between He and Ar is displayed by neon (not shown in the graphs). The intersection between CO2 and N2 ^Ar curves in Fig. 4, as well as the N2 peak at low CO2 concentration in Fig. 3B, is related to the variable e¡ect of H2 O and CO2 during the inert

gas degassing. In fact, the two settings are mainly caused by the growing solubility of N2 and heavy noble gases throughout the degassing. This occurs because the e¡ect of dissolved H2 O, which increases gas solubility, is much stronger for N2 and heavy gases, and it is no longer bu¡ered by the dissolved CO2 when this latter is practically

Fig. 4. Residual fraction of inert gases and CO2 in the melt throughout closed system degassing of a CO2 -rich MORB at 1200³C. Initial pressure was 700 MPa, whereas initial CO2 molar fraction in the vapor was 0.85. The initial aliquots of inert gases derived from the average ratio between each inert gas and CO2 in MORB [12]. Basalt composition from [60]. Inert gas depletions in degassing CO2 -rich rhyolites are slightly lower.

EPSL 6014 4-12-01

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

473

exhausted in the melt. Indeed, the e¡ect that the main volatiles have on the inert gas degassing is clearly highlighted when comparing our results with those obtained using simpler models which do not take the above e¡ects into account. As shown in Fig. 2 for a H2 O-rich melt, natural melts may retain helium much more e¤ciently with respect to the estimations performed by simpler models. Even more divergent results can be achieved when comparing the behavior of heavier inert gases. 5. Application to an active volcano: the case of Vulcano (Italy) Geochemical studies of the fumarolic exhalations in active volcanoes indicate that the gas emissions are more or less directly fed by a degassing magma [43]. Once the consistency of the hypothesis of magma degassing has been veri¢ed, our model can be usefully applied in order to obtain information about the physico-chemical conditions of magma. In view of the latter, Vulcano Island represents a signi¢cant example for our purpose. 5.1. The chemical composition of the magmatic gases: evidence of magma degassing Vulcano is characterized by an intense fumarolic activity at the crater of La Fossa volcano. It is commonly accepted that the fumarolic emissions are fed by magmatic gases, which mix at depth with vapors coming from a hydrothermal system [31^36]. In order to model the degassing at La Fossa, the magmatic gases must obviously be ¢ltered from the hydrothermal contribution. Nuccio et al. [36] modeled the magmatic^hydrothermal mixing process and they were able to compute the composition of the magmatic endmember (in terms of H2 O, CO2 and He) over time, up to the ¢rst half of 1996. Their calculations were brought up to 1998 by Paonita et al. [49], who also evaluated N2 concentrations in magmatic gases. In this work, we have used the concentrations of CO2 , He and N2 of magmatic gases obtained

Fig. 5. CO2 and He concentrations in the magmatic gases of Vulcano Island vs. time, obtained by [36] and [49]. N2 concentrations in magmatic gases displayed synchronous peaks to those of CO2 and He, nevertheless fumarolic data were not enough in order to plot a detailed time curve; anyway, the N2 peak values are shown in the following graphs. The top of each peak in the ¢gure has been regarded as the representative vapor coming from the corresponding degassing event and was adopted for our following computations (see text).

by [36,49], which we have hereafter de¢ned as corrected concentrations in order to distinguish them from those computed by our degassing model. As displayed in Fig. 5 for CO2 and He, the corrected concentrations appear highly variable over time. According to discrete system degassing, we have interpreted each CO2 , He and N2 peak as due to a single event of magma degassing caused by depressurization. The peak amplitude decreases from 1988 until 1995, suggesting a progressive degassing of a small magma body, which has exhausted the less soluble species. The much higher peak of the 1996 crisis (Fig. 5) suggests inputs of new and relatively undegassed magma and starts a new series of degassing events up to 1997. Depressurization is the most likely cause of the depicted sequence of peaks, whereas melt crystallization seems to be less important. In fact, crystallization will be a continuous process during the entire magma cooling history [50], causing a continuous growth of bubbles in the melt. Hence, we can expect that a number of them will always be capable of leaving the magma, making the geochemical signature of

EPSL 6014 4-12-01

474

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

the degassing far more regular. In any case, the crystallization e¡ects on degassing will have to be subject of future investigation. 5.2. Application of the degassing model On the basis of the above statements, two series of degassing events have been identi¢ed: the ¢rst one concerns the time period from 1988 until 1995 and it consists of four events of magma ascent; the second one regards the 1996^97 biennium as having three degassing events. We know the corrected chemical composition of the magmatic vapors exsolved during each event of degassing, in terms of the two main volatiles (H2 O and CO2 ) plus two inert gases (He and N2 ). Therefore, our modeling of the discrete system degassing can be separately applied to each series.

Melt composition and temperature inserted into the model have been taken from chemical and thermometric investigations on the igneous products of Vulcano [51,52]. We have performed simulations for the primitive melts (trachy-basalts at 1150³C), throughout latites (1080³C), up to the most evolved liquids (high silica rhyolites at 1000³C). We insert the corrected CO2 vapor concentrations of all the events of one degassing series, as well as the corrected inert gas concentration of the ¢rst event. By ¢xing an arbitrary pressure value for the ¢rst degassing event, the composition of its gas phase (in terms of H2 O, CO2 and inert gas) allows us to calculate the dissolved concentrations of volatiles for all the volatile components (Eqs. 1^4 in Table 1). Such dissolved quantities become the total amount of volatiles available for the sub-

Fig. 6. (A^C) Fit of the corrected composition of the magmatic gas of Vulcano Island (squares) to our model of discrete system degassing (thick curves). The three degassing events of the series 1996^97, with their H2 O, CO2 , He and N2 concentrations in vapor, were used. Error bars in the concentrations of the inert gases were in the order of the symbol size. Silicate melt was the high silica rhyolite of La Fossa at 1000³C (see text); (D) resulting pressure for each degassing event (circles). Shown uncertainties ( þ 10%) in the computed pressure of each event derived from the e¡ect, on the model ¢tting, due to the error bars in inert gas concentrations. Similar uncertainties in pressure values were obtained when varying melt temperature of þ 50³C. Curves for open (op.) and closed system (cl.) were plotted for comparison.

EPSL 6014 4-12-01

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

sequent degassing event (Eq. 7). For this new event, we can calculate both P and the concentration of the inert gas in the exsolved gas phase by using the total amounts of volatiles and the CO2 concentration in the vapor (Eqs. 1^5). The computed inert gas concentration may be compared with the corresponding corrected concentration of the same event, ascertaining whether the model has suitably predicted the observed values. If this is so, then the inserted initial pressure and that of the subsequent event are the proper magma pressures throughout the degassing series. If not, the initial pressure is modi¢ed as far as reaching the corrected inert gas concentration. We can assess the magma pressures by using the corrected concentrations of one inert gas in two degassing events. When more than two degassing events and one inert gas are available in a single degassing series, we may ¢t all the corrected concentrations to our model, greatly improving con¢dence in the results. 5.3. Results and discussion The above-described procedure has been applied ¢rst to the 1988^95 series of degassing and subsequently to the 1996^97 one. For each series, the pressure of the ¢rst event was modi¢ed until we obtained a modeling curve, which well reproduced the corrected concentrations. The graphic solution of this inverse problem is shown in Fig. 6 for the 1996^97 degassing series, which consists of three events with their corrected He and N2 concentrations. Both helium and nitrogen have been reproduced very well within their error bars, so we have calculated the absolute pressure of each degassing event. As the model is quite sensitive to pressure variations (see Fig. 7), the reliability of our solution is good (see given uncertainties in Fig. 6D), particularly when compared to other assessments of pressure by petrologic or geophysical methods [51^53]. The displayed curves have been computed by using the high silica rhyolite of La Fossa at 1000³C and our solution was relatively insensitive to the selected melt temperature (see caption in Fig. 6). In contrast, we were unable to ¢t the data by any other possible composition of the silicate melt (i.e. see Fig. 7), hence

475

Fig. 7. Discrete system degassing curves N2 vs. CO2 (thick lines) computed for: Vulcano rhyolite at 1000³C and 33 MPa initial pressure (¢tting the 1996^97 series), Vulcano rhyolite at 1000³C and 55 MPa initial pressure, and Vulcano basalt a 1200³C and 33 MPa. Curves for open (gray lines) and closed (thin lines) system are also shown. The three events of the 1996^97 series (squares) have uncertainties in the order of the symbol size.

the magma composition has also been constrained. The ¢t procedure for the 1988^95 degassing series produced very good results only for nitrogen (Fig. 8), whereas this did not occur for helium. However, one minor volatile was enough to estimate the pressure of each degassing event. Also in this case, just the high silica rhyolite was capable of reproducing the corrected data. The helium mis¢t could be in part linked to systematic analytical errors (by about a factor two in excess) in the helium measurements available from the literature for fumarolic gases up to 1995. These errors were highlighted by comparing such measurements, performed by gas chromatography, with some much more accurate helium determinations achieved by mass spectrometry in the same period. Indeed, the routine analytical methodologies have been substantially improved since 1996. However, processes occurring in the natural system and currently neglected in the model, such as convection or He di¡usion from the magma body, could provide alternative explanations and will have to be investigated. The pressure values for the degassing events which occurred during the 10-year period 1988^

EPSL 6014 4-12-01

476

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

sion. The sum of all the ascent events has caused a decrease of the magma pressure from 48 to 28 MPa, implying a ¢nal depth of about 2.5 km (Fig. 9). Our results are in good agreement with the pressure of about 30^50 MPa, obtained for the shallower stage of the magmatic system by the water content in erupted products and the £uid inclusions in partially melted quartzitic xenoliths [51,52]. Moreover, they agree with the depth of 2.5^3.5 km, deduced for the top of the magma body by seismic tomography [53]. 5.3.1. Modi¢cation of magma density On the basis of the concentrations of volatiles exsolved from the melt at each depressurization step of a closed system degassing, we may compute the density of the bubble+melt ensemble, as well as the bubble-free melt density. Starting from the initial pressure and volatile content of the magma of the 1996^97 series, the computed bubble+melt density shows a sharp decrease at about 20 MPa (Fig. 10), which is mainly due to the vapor expansion at low pressure in the bubbles. The e¤cient H2 O degassing subsequent to CO2 exhaustion in the melt is, in fact, less important. In contrast, a slight increase in the density of the bubble-free melt occurs, caused by the progressively lower amount of dissolved H2 O.

Fig. 8. (A) Fit of the corrected magmatic gas composition of Vulcano Island (squares) to our model of discrete system degassing (thick curves). The four degassing events of the 1988^95 series, with their related H2 O, CO2 , N2 concentrations in vapor, were used. The error bars of inert gas concentrations and the silicate melt composition were as in Fig. 6; (B) resulting pressure for each degassing event (circles). The shown uncertainties in the computed pressures were as in Fig. 6. Curves for open (op.) and closed (cl.) system were plotted too.

97 have allowed for the evaluation of the total depressurization of the magma below La Fossa volcano in that period (Fig. 9). It is noteworthy that the last series (1996^97) starts near the ¢nal pressure of the previous series (1988^95), suggesting a re¢lling process of the same magma intru-

Fig. 9. Ascending path of the magma below La Fossa volcano throughout the 10-year period 1988^97. The displayed uncertainties have been given in Figs. 6D and 8B. The deep hydrothermal aquifer at Vulcano Island was plotted on the basis of the pressure estimated by Nuccio et al. [36]. The depth scale was calculated by considering a hydrostatic regime. Indeed, the hydrothermal system suggests that such a regime exists at least up to 2 km.

EPSL 6014 4-12-01

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

477

magma+bubble ensemble, which in turn could cause a further vesiculation and, possibly, a catastrophic rising of the magma.

Fig. 10. Calculated density of melt+bubble ensemble for the discrete events of magma degassing during the 1996^97 series (circles). Curves display the densities of the bubble-free and bubble bearing Vulcano rhyolite, computed by our model for closed system degassing. Molar volumes of melt and vapor used for the calculation were computed as described in [47]. The path ACBCCCD depicts the variation of magma density during an ascending event. Magma in A moves along the closed system curve during an event of ascent. At the end of such an event, the vesiculated magma will be in B, at lower density. The bubble release causes a sharp decline of the magma density, up to reach bubble-free melt density in C. A further event of ascent moves the magma along the new closed system curve C^D. Such a curve tends always to quickly reach, at lower pressure, the previous closed system curve, so as to cause a much sharper decrease of the magma density.

In Fig. 10 we have also plotted the computed bubble+melt density for each degassing event in the 1996^97 series, taking into account that the exsolved vapor leaves the melt at the end of each event. We clearly see that the magma body below La Fossa is still far from the zone of abrupt variations of density, although it may be reached by future magma rises (see path A^D in Fig. 10). In such a case, a very critical situation could be triggered by any event of ascent: the pressure increment within the magma intrusion, caused by exponential bubble growth, would determine country rock deformation and the opening of fractures [26]. In highly viscous melts, the dramatic increase of the dynamic pressure into bubbles could also cause explosion and magma fragmentation [20,23,54,55]. Moreover, the density decline would determine a relevant ascending spur for the

5.3.2. Forecasting hazardous events The depicted hazardous scenarios at Vulcano, such as magma^water interaction or a catastrophic magma rise by abrupt vesiculation, are strictly linked to the pressure reached by the degassing magma. Equally, the chemical composition of the released gases, for example the CO2 /He and CO2 /N2 ratios in vapor (Fig. 11), depends on the pressure. After we have ¢tted the corrected data to our degassing model, we are now able to evaluate whether the magma pressure is actually reaching the above critical conditions. At Vulcano, this could occur when the CO2 /He ratio in the magmatic endmember moves toward 3U104 (Fig. 11). As the model also provides inferences about the time scale of the magma ascent process, this may help the establishment of a long-term plan for volcanic surveillance.

Fig. 11. Calculated CO2 /He ratio of the gases released during the discrete events of magma degassing of the 1996^97 series (circles). Corresponding curves for open and closed system were plotted in order to constrain the possible CO2 /He ratio of a further degassing event of this series. CO2 /He ratios around 3U104 should indicate that any further ascent event would cause a sharp decrease in magma density.

EPSL 6014 4-12-01

478

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

6. Summary and conclusions By using recent modeling for the solubility of H2 O^CO2 mixtures in silicate liquids [41] and for noble gas partitioning in H2 O^CO2 bearing magmas [47], we have described the magmatic degassing of multicomponent H2 O^CO2 ^noble gas^N2 mixtures, dealing with inert gases as minor or trace components in the system. The developed degassing model is suitable for any silicate melt at magmatic temperatures (900^1400³C) and pressures useful for investigating crustal and pre-eruptive degassing processes (up to 700^800 MPa). At present, this model is the only one that evaluates the e¡ect of H2 O and CO2 on inert gas fractionation by degassing. We have shown that H2 O and CO2 proportions strongly in£uence inert gas degassing. Helium, argon and N2 display a similar behavior during the degassing of H2 O-rich basalt and rhyolite melts, being strongly partitioned into vapor at early degassing extents. Conversely, helium degassing is less e¤cient from degassing CO2 -rich melts, because the more volatile CO2 ¢rst has to be released, whereas nitrogen and heavy noble gases undergo a similar or higher exsolution than CO2 . Due to the e¡ect of H2 O and CO2 , the inert gas solubility varies throughout magma degassing as the aliquot of the main volatiles in the melt progressively decreases. In order to show the applicability of our model to an active volcanic system, we have shown the results for Vulcano Island (Italy), where several geochemical parameters have been regularly acquired in the last 10 years. Although He and N2 were the only available inert gases in the large set of monitored parameters, the degassing model has enabled us to identify two series of exsolution events and to quantitatively assess their absolute pressure. Hence, the process of magma rise towards the surface has been reconstructed. On the basis of our results, the magma seems to have reached a relatively low pressure beneath Vulcano, and interactions with shallow aquifers may become a hazard in the future. In addition, the magma seems to be approaching a critical pressure and depth at which further ascent could trigger feedback mechanisms and a violent

eruptive rise. In the view of latter, our deterministic approach allows us to focus on speci¢c geochemical indicators of the degassing process, helping the long-term planning of volcanic surveillance. Acknowledgements This work constitutes the core of A.P.'s Ph.D. dissertation, supported by the European Social Fund during his Ph.D. course. We thank Ra¡aello Trigila and Daniela Dol¢ for the useful discussions. Critical comments and helpful suggestions by Greg Anderson, Mike Carroll and two anonymous reviewers greatly improved the manuscript.[AH] Appendix A. N2 solubility by the EIP model This appendix reports some details of the procedure that we have applied in this work to include N2 into the EIP model by Nuccio and Paonita [47]. Experimental nitrogen solubility under conditions close to the magmatic ones is poorly known. As physical solubility seems to occur for the inert N2 molecules (see [9] for a review) and the EIP model is based on the concept of physical solubility of non-reactive gas in the silicate melt [47], the latter should be able to describe nitrogen solubility in magmas when nitrogen substantially occurs as molecular N2 . For this aim, we started from the experimental data of N2 solubility performed by Kesson and Holloway [56] in dry melted albite. Such experiments were conducted under redox conditions close to NNO bu¡er, which guarantee the presence of only molecular N2 at the investigated pressures [56]. By the EIP model, we computed the solubilities of the noble gases in the same albitic melt, and we obtained a linear relation between the logarithms of noble gas Henry's constants versus their atomic radii. On the basis of the experimental Henry's constant of N2 in Kesson and Holloway [56], the above relation allowed us to evaluate a sort of N2 radius in the melt. The corresponding radius computed for N

EPSL 6014 4-12-01

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

î , which is in agreement with its covawas 0.77 A î ; Winter [57]). lent radius (W0.75 A The relationship between noble gas solubility and ionic porosity in H2 O^CO2 bearing melts may be expressed as [58]: 3ln…K h † ˆ K IP ‡ L

…B1†

where Kh and IP are the Henry's constant of a noble gas and the ionic porosity of the melt, respectively, and K and L are ¢tting parameters given in [15,58] for all the stable noble gases. It may be shown that parameters K and L show a linear correlation versus the noble gas atomic radii, therefore we were able to estimate K and L values for N2 starting from its computed radius in the melt. This enabled us to use Eq. B1 in order to evaluate N2 Henry's constant in any silicate melt composition. Subsequently, by using the critical constants of molecular nitrogen [59], we included N2 into the EOS in [47], so as to calculate N2 fugacity coe¤cient in H2 O^CO2 -dominated vapors. The computed N2 solubility in natural anhydrous melts has resulted coherent with the available experimental data (around 50^150 ppm per 100 MPa N2 pressure), whereas the H2 O and CO2 e¡ects are very similar to those predicted for Ar solubility. This agrees with the data from natural products, which seem to display negligible fractionations between these two gases [16,17]. References [1] D.L. Hamilton, C.W. Burnham, E.F. Osborn, The solubility of water and e¡ects of oxygen fugacity and water content on crystallization in ma¢c magmas, J. Petrol. 5 (1964) 21^39. [2] P.J. Wyllie, Crustal anatexis: An experimental review, Tectonophysics 43 (1977) 41^77. [3] S. Maaloe, Igneous petrology, Springer, Berlin, 1985, 374 pp. [4] S. Vergniolle, C. Jaupart, Separated two-phases £ow and basaltic eruptions, J. Geophys. Res. 91 (1986) 12842^ 12860. [5] P. Papale, F. Dobran, Magma £ow along the volcanic conduit during the plinian and pyroclastic £ow phases of the May 18, 1980, Mount St. Helens eruption, J. Geophys. Res. 99 (1994) 4355^4373.

479

[6] R.A. Lange, The e¡ect of H2 O, CO2 and F on the density and viscosity of silicate melts, in: M.R. Carroll, J.R. Holloway (Eds.), Volatiles in Magmas, Rev. Mineral. 30, 1994, pp. 331^365. [7] K.U. Hess, D.B. Dingwell, Viscosities of hydrous leucogranitic melts: a non-Arrhenian model, Am. Mineral. 81 (1996) 1297^1300. [8] A. Jambon, H.W. Weber, O. Braun, Solubility of He, Ne, Ar, Kr, Xe in a basalt melt in the range of 1250^1600³C: Geochemical implications, Geochim. Cosmochim. Acta 50 (1986) 401^408. [9] C.J. Allegre, T. Staudacher, P. Sarda, Rare gas systematics: formation of the atmosphere, evolution and structure of the Earth's mantle, Earth Planet. Sci. Lett. 81 (1986) 127^150. [10] B. Marty, A. Jambon, C/3 He in volatile £uxes from the solid Earth: implications for carbon geodynamics, Earth Planet. Sci. Lett. 83 (1987) 16^26. [11] P. Sarda, D. Graham, Mid-oceanic ridge popping rocks: implications for degassing at ridge crests, Earth Planet. Sci. Lett. 97 (1990) 268^289. [12] M. Javoy, F. Pineau, The volatile record of a `popping' rock Mid-Atlantic-Ridge at 14³N: Chemical and isotopic composition of the gas trapped in the vesicles, Earth Planet. Sci. Lett. 107 (1991) 598^611. [13] K. Roselieb, W. Rammensee, H. Buttner, M. Rosenhauer, Solubility and di¡usion of noble gases in vitreous albite, Chem. Geol. 96 (1992) 241^266. [14] A. Jambon, Earth degassing and large scale geochemical cycling of volatile elements, in: M.R. Carroll, J.R. Holloway (Eds.), Volatiles in Magmas, Rev. Mineral. 30, 1994, pp. 479^510. [15] M.R. Carroll, J.D. Webster, Solubilities of sulfur, noble gases, nitrogen, chlorine, and £uorine in magmas, in: M.R. Carroll, J.R. Holloway (Eds.), Volatiles in Magmas, Rev. Mineral. 30, 1994, pp. 231^279. [16] B. Marty, Nitrogen content of the mantle inferred from N2 ^Ar correlation in oceanic basalts, Nature 377 (1995) 326^329. [17] B. Marty, L. Zimmermann, Volatiles (He, C, N, Ar) in mid-ocean ridge basalts: Assessment of shallow-level fractionation and characterization of source composition, Geochim. Cosmochim. Acta 63 (1999) 3619^3633. [18] M. Honda, B.D. Patteson, Systematic elemental fractionation of mantle derived helium, neon, and argon in midoceanic ridge glasses, Geochim. Cosmochim. Acta 61 (1999) 2863^2874. [19] M. Moreira, P. Sarda, Noble gas constraints on degassing processes, Earth Planet. Sci. Lett. 176 (2000) 375^386. [20] R.S.J. Sparks, J. Barclay, C. Jaupart, H.M. Mader, J.C. Phillips, Physical aspects of magmatic degassing I. Experimental and theoretical constrains on vesiculation, in: M.R. Carroll, J.R. Holloway (Eds.), Volatiles in Magmas, Rev. Mineral. 30, 1994, pp. 413^443. [21] Y. Bottinga, M. Javoy, MORB degassing: Bubble growth and ascent, Chem. Geol. 81 (1990) 255^270. [22] A.A. Proussevitch, D.L. Sahagian, Dynamics of coupled

EPSL 6014 4-12-01

480

[23]

[24]

[25]

[26] [27]

[28]

[29]

[30]

[31]

[32]

[33]

[34]

[35]

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481 di¡usive and decompressive bubble growth in magmatic systems, J. Geophys. Res. 101 (1996) 17447^17456. A.A. Proussevitch, D.L. Sahagian, Dynamics and energetics of bubble growth in magmas: analytical formulation and numerical modeling, J. Geophys. Res. 103 (1998) 18223^18251. J.E. Gardner, M. Hilton, M.R. Carroll, Experimental constrains on degassing of magma: isothermal bubble growth during continuous decompression from high pressure, Earth Planet. Sci. Lett. 168 (1999) 201^218. J.E. Gardner, M. Hilton, M.R. Carroll, Bubble growth in highly viscous silicate melts during continuous decompression from high pressure, Geochim. Cosmochim. Acta 64 (2000) 1473^1483. S. Tait, C. Jaupart, S. Vergniolle, Pressure, gas content and eruption periodicity of a shallow crystallizing magma chamber, Earth Planet. Sci. Lett. 92 (1989) 107^123. S. Newman, S. Epstein, E.M. Stolper, Water, carbon dioxide and hydrogen isotopes in glasses from the CA. 1340 A.D. eruption of the Mono Craters, California: constraints on degassing phenomena and initial volatile content, J. Volcanol. Geotherm. Res. 35 (1988) 75^96. J.E. Dixon, D.A. Clague, P. Wallace, R. Poreda, Volatile in alkalic basalts from the North Arch Volcanic Field, Hawaii: Extensive degassing of deep submarine-erupted alkalic series lavas, J. Petrol. 38 (1997) 911^939. P.M. Nuccio, M. Valenza, Magma degassing and geochemical detection of its ascent, in: G.B. Arehart, J.R. Hulston (Eds.), Water^Rock Interaction, Balkema, 1998, pp. 475^478. B. Badalamenti, G. Chiodini, R. Cioni, R. Favara, S. Francofonte, S. Gurrieri, S. Hauser, S. Inguaggiato, F. Italiano, G. Magro, P.M. Nuccio, F. Parello, M. Pennisi, L. Romeo, M. Russo, F. Sortino, M. Valenza, F. Vurro, Special Field Workshop at Vulcano (Aeolian Islands) during summer 1988: geochemical results, Acta Vulcanol. 1 (1991) 223^227. G. Chiodini, R. Cioni, L. Marini, Reactions governing the chemistry of crater fumaroles from Vulcano Island, Italy, and implications for volcanic surveillance, Appl. Geochem. 8 (1993) 357^371. G. Chiodini, R. Cioni, L. Marini, C. Panichi, Origin of fumarolic £uids of Vulcano Island, Italy and implications for volcanic surveillance, Bull. Volcanol. 57 (1995) 99^ 110. G. Capasso, R. Favara, S. Inguaggiato, Chemical features and isotopic gaseous manifestation on Vulcano Island (Aeolian Island): an interpretative model of £uid circulation, Geochim. Cosmochim. Acta 61 (1997) 3425^ 3442. L. La Volpe, P. Dellino, P.M. Nuccio, E. Privitera, A. Sbrana, C.N.R.-G.N.V., Progetto Vulcano ^ Risultati dell'attivita© di ricerca 1993^1995, Felici Editore, Pisa, 1997, 284 pp. G. Capasso, R. Favara, S. Francofonte, S. Inguaggiato, Chemical and isotopic variations in fumarolic discharge and thermal waters at Vulcano Island (Aeolian Island,

[36]

[37] [38] [39]

[40] [41] [42] [43]

[44] [45]

[46] [47]

[48]

[49]

[50] [51]

[52]

Italy) during 1996: evidence of resumed volcanic activity, J. Volcanol. Geotherm. Res. 88 (1999) 167^175. P.M. Nuccio, A. Paonita, F. Sortino, Geochemical model of mixing between magmatic and hydrothermal gases: the case of Vulcano Island (Italy), Earth Planet. Sci. Lett. 167 (1999) 321^333. T.M. Gerlach, Exsolution of H2 O, CO2 , and S during eruptive episodes at Kilawea Volcano, Haway, J. Geophys. Res. 91 (1986) 12177^12185. S. Cardoso, A.W. Woods, On convection in a volatile saturated magma, Earth Planet. Sci. Lett. 168 (1999) 301^310. J.E. Dixon, E.M. Stolper, J.R. Holloway, An experimental study of water and carbon dioxide solubilities in MidOcean Ridge basaltic liquids. Part II Application to Degassing, J. Petrol. 36 (1995) 1633^1646. J.E. Dixon, Degassing of alkalic basalts, Am. Mineral. 82 (1997) 368^378. P. Papale, Modeling of the solubility of a two component H2 O+CO2 £uid in silicate liquids, Am. Mineral. 84 (1999) 477^492. B.E. Taylor, Magmatic volatiles: isotopic variations of C, H, and S, Rev. Mineral. 16 (1986) 185^225. W.F. Giggenbach, Chemical composition of volcanic gases, in: R. Scarpa, R.I. Tilling (Eds.), Monitoring and Mitigation of Volcanic Hazards, Springer, 1996, pp. 221^ 256. P. Burnard, The bubble-by-bubble volatile evolution of two mid-ocean ridge basalts, Earth Planet. Sci. Lett. 174 (1999) 199^211. A. Paonita, G. Gigli, D. Gozzi, P.M. Nuccio, R. Trigila, Investigation of the He solubility in H2 O^CO2 bearing silicate liquids at moderate pressure: a new experimental method, Earth Planet. Sci. Lett. 181 (2000) 595^604. T.M. Gerlach, B.E. Nordlie, The C^O^H^S gaseous system I: composition limits and trends in basaltic gases, Am. J. Sci. 275 (1975) 353^376. P.M. Nuccio, A. Paonita, Investigation of the noble gas solubility in H2 O^CO2 bearing silicate liquids at moderate pressure II: the Extended Ionic Porosity (EIP) model, Earth Planet. Sci. Lett. 183 (2000) 499^512. J.R. Holloway, J.G. Blank, Experimental results applied to C^O^H in natural melts, in: M.R. Carroll, J.R. Holloway (Eds.), Volatiles in Magmas, Rev. Mineral. 30, 1994, pp. 231^279. A. Paonita, R. Favara, P.M. Nuccio, F. Sortino, Genesis of £uids involved in fumarolic emissions inferred by isotope mass balances: CO2 and water at Vulcano Island, Italy, Geochim. Cosmochim. Acta, in press. B.D. Marsh, On the mechanics of igneous diapirism, stoping and zone melting, Am. J. Sci. 282 (1982) 808^855. R. Clocchiatti, A. Del Moro, A. Gioncada, J.L. Joron, M. Mosbah, L. Pinarelli, A. Sbrana, Assessment of shallow magmatic system: the 1888^90 eruption, Vulcano Island, Italy, Bull. Volcanol. 56 (1994) 466^486. A. Sbrana, Il sistema di magmatico La Fossa di Vulcano, in: L. La Volpe, P. Dellino, P.M. Nuccio, E. Privitera, A.

EPSL 6014 4-12-01

P.M. Nuccio, A. Paonita / Earth and Planetary Science Letters 193 (2001) 467^481

[53]

[54] [55]

[56]

Sbrana (Eds.), Progetto Vulcano: Risultati dell'attivita© di ricerca 1993^95, Felici Editore, Pisa, 1997, pp. 22^36. F. Ferrucci, G. Gaudiosi, G. Milano, A. Nercessian, G. Vilardo, G. Luongo, Seismological exploration of Vulcano (Aeolian Islands, southern Tyrrhenian sea): Case history, Acta Vulcanol. 1 (1991) 143^152. R.S.J. Sparks, The dynamics of bubble formation and growth in magmas: a review and analysis, J. Volcanol. Geotherm. Res. 3 (1978) 1^37. O. Navon, A. Chekhmir, V. Lyakhovsky, Bubble growth in highly viscous melts: theory, experiments, and autoexplosivity of dome lavas, Earth Planet. Sci. Lett. 160 (1998) 763^776. S.E. Kesson, J.R. Holloway, The generation of N2 ^CO2 ^ H2 O £uids for use in hydrothermal experimentation II.

[57] [58]

[59] [60]

481

Melting of albite in a multispecies £uid, Am. Mineral. 39 (1974) 598^603. M. Winter, The Periodic Table on the WWW: WebElements, 1993^1999, http://www.webelements.com/. M.R. Carroll, E.M. Stolper, Noble gas solubilities in silicate melts and glasses: new experimental results for Ar and the relationship between solubility and ionic porosity, Geochim. Cosmochim. Acta 57 (1993) 5039^5051. R.C. Weast, J.A. Melvin, W.H. Beyer, CRC Handbook of Chemistry and Physics, CRC Press, Boca Raton, FL, 1985. A.R. Pawley, J.R. Holloway, P.F. McMillan, The e¡ect of oxygen fugacity on the solubility of carbon^oxygen £uids in basaltic melts, Earth Planet. Sci. Lett. 110 (1992) 213^ 225.

EPSL 6014 4-12-01