Journal of Petrology Advance Access originally published online on July 13, 2005
Journal of Petrology 2005 46(11):2337-2366; doi:10.1093/petrology/egi058
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Volcán Popocatépetl, Mexico. Petrology, Magma Mixing, and Immediate Sources of Volatiles for the 1994–Present Eruption
1 DEPARTMENT OF EARTH AND SPACE SCIENCES, UNIVERSITY OF WASHINGTON, SEATTLE, WA 98195, USA
2 USGS-DEPARTMENT OF EARTH AND SPACE SCIENCES, UNIVERSITY OF WASHINGTON, SEATTLE, WA 98195, USA
RECEIVED APRIL 17, 2005; ACCEPTED MAY 23, 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIXREFERENCESREFERENCES |
|---|
Volcán Popocatépetl has been the site of voluminous degassing accompanied by minor eruptive activity from late 1994 until the time of writing (August 2002). This contribution presents petrological investigations of magma erupted in 1997 and 1998, including major-element and volatile (S, Cl, F, and H2O) data from glass inclusions and matrix glasses. Magma erupted from Popocatépetl is a mixture of dacite (65 wt % SiO2, two-pyroxenes + plagioclase + Fe–Ti oxides + apatite,
3 wt % H2O, P = 1·5 kbar, fO2 = 
NNO + 0·5 log units) and basaltic andesite (53 wt % SiO2, olivine + two-pyroxenes, 3 wt % H2O, P = 1–4 kbar). Magma mixed at 4–6 km depth in proportions between 45:55 and 85:15 wt % silicic:mafic magma. The pre-eruptive volatile content of the basaltic andesite is 1980 ppm S, 1060 ppm Cl, 950 ppm F, and 3·3 wt % H2O. The pre-eruptive volatile content of the dacite is 130 ± 50 ppm S, 880 ± 70 ppm Cl, 570 ± 100 ppm F, and 2·9 ± 0·2 wt % H2O. Degassing from 0·031 km3 of erupted magma accounts for only 0·7 wt % of the observed SO2 emission. Circulation of magma in the volcanic conduit in the presence of a modest bubble phase is a possible mechanism to explain the high rates of degassing and limited magma production at Popocatépetl. KEY WORDS: glass inclusions; igneous petrology; Mexico; Popocatépetl; volatiles
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIXREFERENCESREFERENCES |
|---|
From December 1994 until the time of writing (August 2002), Volcán Popocatépetl (Mexico) has non-explosively degassed an average of 7000 tons/day SO2 into the troposphere [Delgado & Cárdenas-G, 1997
; Delgado-Granados et al., 2001; National Center for Disaster Prevention, Mexico City (CENAPRED), unpublished data, 2000]. The current eruptive period is similar to other eruptions that have occurred in historical time. This is the first time, however, that such massive degassing could be measured quantitatively. Popocatépetl has degassed a total of 20 Mt SO2 since 1994, accompanied by relatively minor eruptive activity, compared with the 17 Mt SO2 (Gerlach et al., 1996) injected into the stratosphere in a single day during the climactic, highly explosive eruption of Pinatubo (Philippines) on June 15, 1991. In this study, we examine various aspects of the process of passive degassing at Popocatépetl. We describe the petrology of recent magmas, and assess the volatile budget during recent volcanic activity. These data are used to estimate magmatic conditions and volatile sources immediately prior to eruption. Based on these constraints we discuss possible processes to liberate the large mass of gases non-explosively emitted by Popocatépetl during recent activity.
Background
Popocatépetl (5452 m) is an andesitic, subduction zone stratovolcano that lies 60 km SE of Mexico City (Fig. 1). Holocene eruptive activity at Popocatépetl included at least three Plinian eruptions (Siebe et al., 1996a, 1996b; Panfil et al., 1999). Since AD 1354, Popocatépetl has experienced at least 12 periods of significant unrest, all of which exhibited volcanic activity similar to the 1994–present eruptive period (De la Cruz-Reyna et al., 1995).
|
, Holocene volcanoes; 
, Mexico City. Map made with GMT (Wessel & Smith, 1991After nearly 70 years of dormancy, Volcán Popocatépetl began a new period of activity in December 1994 (De la Cruz-Reyna & Siebe, 1997
). Elevated levels of degassing, seismicity, and phreatic explosions characterized the initial portion of volcanic unrest at Popocatépetl through early 1996. In March 1996, a dacitic lava dome extruded into the summit crater (GVN, 1996; S. De la Cruz-Reyna et al., unpublished data, 1998; Stimac et al., 1997). Since the onset of the current eruptive period, Popocatépetl has continuously degassed at high rates accompanied by relatively minor eruptions. Eruptive activity includes: (1) vulcanian explosions that eject lithic and juvenile ash and rock debris up to 8 km above the vent; (2) ash-poor ‘exhalations’ of volcanic gas; (3) periodic lava dome growth, subsidence, and explosive destruction (S. De la Cruz-Reyna et al., unpublished data, 1998). Eruptions at Popocatépetl are commonly characterized by mixing and mingling of distinct mafic and silicic magmas (Cantagrel et al., 1984; Robin, 1984; Boudal & Robin, 1988; Kolisnik, 1990). This includes the most recent active phase extending from 1994 to the present (Athanasopoulos, 1997; Straub & Martin-Del Pozzo, 2001).
From March 1996 until December 2000, 3·1 x 107 m3 of magma was extruded, based on aerial photogrammetry (S. De la Cruz-Reyna et al., unpublished data, 2000). During this period the average SO2 flux was 8400 tons/day (CENAPRED, unpublished data, 2000), yielding a total of 14·8 ± 3·0 Mt SO2 (average ±20% error because of uncertainties in measuring windspeed; Delgado-Granados et al., 2001). Assuming 20 vol. % crystals with melt density of 2300 kg/m3, about 5·7 x 1010 kg silicate melt has been erupted during the March 1996 to December 2000 period of activity. For this mass of silicate melt alone to supply the 14·8 Mt observed SO2 emission, the silicate melt phase would have to have initially contained 13 wt % S. This exceeds the experimentally derived sulfur solubility limits in silicic and intermediate magmas by two or more orders of magnitude (Wendlandt, 1982; Carroll & Rutherford, 1985; Luhr, 1990; Scaillet et al., 1998). This apparent paradox has been encountered by other workers and is commonly referred to as the ‘excess sulfur’ problem (Luhr et al., 1984; Sigurdsson et al., 1990; Andres et al., 1991; Westrich & Gerlach, 1992; Gerlach & McGee, 1994; Gerlach et al., 1994, 1996; Kazahaya et al., 1994, 2002; Wallace & Gerlach, 1994; Mandeville et al., 1996; Witter, 1997; Wallace, 2001).
Previous work
During the 1980s various workers pointed out the importance of magma mixing as a petrogenetic process at Popocatépetl and documented giant sector collapse events at the volcano (Cantagrel et al., 1984; Robin, 1984; Robin & Boudal, 1987; Boudal & Robin, 1988, 1989; Kolisnik, 1990). Renewal of eruptive activity at Popocatépetl in 1994 stimulated extensive new research on the eruptive history, petrology, and degassing behavior of the volcano (CENAPRED-UNAM, 1995; Athanasopoulos, 1997; Straub & Martin-Del Pozzo, 2001).
Gas measurements at Popocatépetl have focused on SO2, using correlation spectrometry (COSPEC). Carbon dioxide emissions have also been measured using LI-COR and passive infrared spectroscopy (FTIR). COSPEC measurements began in January 1994 and showed emission rates of 1500 to 12 500 tons/day SO2 (Delgado & Cárdenas-G, 1997; Delgado-Granados et al., 2001). On rare occasions, SO2 emissions have reached 30 000–50 000 tons/day SO2 (Goff et al., 1998). LI-COR measurements of CO2 in June 1995 indicated a flux of between 6400 and 9000 tons/day CO2 (Gerlach et al., 1997). Using the same technique, Delgado et al. (1998) reported average fluxes of >40 000 tons/day CO2 between 1996 and 1998. In February 1998, passive FTIR measured 38 000 tons/day CO2 with excursions that exceeded 100 000 tons/day CO2 (Goff et al., 2001). These measurements establish Popocatépetl as one of the greatest known natural producers of SO2 and CO2.
Relative HCl and HF emission rates were measured using volatile traps from February 1994 until the onset of activity in December 1994 (Goff et al., 1998). Some volatile trap samples were also collected in early 1996. Volatile trap studies, combined with SO2 flux measurements, suggest that between the onset of activity and November 1996, Popocatépetl had emitted 0·75 Mt HCl and 0·075 Mt HF (Goff et al., 1998). This represents an average emission rate of 740 tons/day HCl and 74 tons/day HF, respectively. Emission rates measured by passive FTIR in 1997 and 1998 have the range 190–1700 tons/day HCl and 35–250 tons/day HF (Goff et al., 2001). The uncertainty in these halogen measurements is estimated to be of the order of ±20% (Goff et al., 1998, 2001).
|
ANALYTICAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIXREFERENCESREFERENCES |
|---|
All analyses by electron microprobe (EPMA) were conducted on a four-spectrometer JEOL 733 Superprobe at the University of Washington. Mineral analyses were conducted at 15 kV accelerating voltage,
1–3 µm beam diameter, and 10–25 nA beam current with a maximum count time of 40 s for each element. Elements were calibrated using mineral standards, which were also analyzed at the end of each analytical session to check for instrumental drift. Matrix effects were corrected using the ZAF model of Armstrong (1984).
Major element glass analyses were performed at 15 kV accelerating voltage and 10 nA with a 5 µm beam diameter and maximum count time of 40 s for each element. The combination of analyzing sodium first and a defocused beam minimized Na loss during analysis. Plots of intensity vs time suggest Na loss of <0·2 wt % Na2O over a 40 s count time using an anhydrous glass standard. Na loss during analysis of glass inclusions, however, may be significant because of enhanced Na mobility in hydrous glasses (Devine et al., 1995). No correction was applied for potential alkali loss in hydrous inclusions. Mean standard errors as well as errors based on counting statistics for the major element mineral and glass analyses are shown in the Electronic Appendix (Tables A1 and A2), which may be downloaded from http://www.petrology.oupjournals.org/.
Volatile elements S, Cl, and F were analyzed in glasses, hornblende, and apatite using an improved analytical method originally developed by Thordarson (Thordarson et al., 1996). Analytical conditions were 15 kV accelerating voltage, 100 nA beam current, 5 µm beam diameter, and 400 s total peak count time. To minimize volatile loss, an iterative scheme was employed in which the beam was blanked for 10 s after every 10 s of counting. This procedure minimized heating of the sample and consequent volatile loss. We conducted a test for volatile mobility under the electron beam and found that the X-ray count rates for S, Cl, and F remained stable during the entire period of analysis. Sulfur and chlorine were analyzed using PET diffraction crystals with the spectrometer aperture window set at its widest setting to maximize the number of X-ray counts. Fluorine was analyzed using an LDE1 diffraction crystal and a new empirical correction method that overcomes peak overlap problems with the Fe K
line (Witter & Kuehner, 2004). Because the volatile elements were analyzed in a separate session, we reprocessed the volatile data with major element data (collected previously; see Witter & Kuehner, 2004) to correct for matrix effects using the ZAF model of Armstrong (1984). Prior to each analytical session, volatile elements were calibrated with NiS (sulfide sulfur), celestite (sulfate sulfur), scapolite 17 120 (chlorine), and synthetic F-phlogopite (fluorine). Analytical precision and accuracy were checked using basalt glass standards VG-2 and A-99 (for S, Cl, and F) and NBS soda-lime glasses 620 and 621 (for S and Cl). The 1
S, Cl and F variations for more than 30 analyses are 1·5% at 1500 ppm and 13% at 140 ppm for sulfur, 9% at 200 ppm for chlorine, and 5% at 750 ppm and 12% at 300 ppm for fluorine (Electronic Appendix, Table A3).
H2O, CO2, S, Cl, and F in glass inclusions were analyzed using secondary ion mass spectrometry (SIMS) at the Department of Terrestrial Magnetism, Carnegie Institution of Washington. The analytical methods for this technique have been described by Hauri et al. (2002). SIMS analysis could only be performed on inclusions that were larger than about 40 µm in the smallest dimension.
Major and trace element compositions of four pumice samples and dome rocks from the current eruptive period were analyzed by X-ray fluorescence at the WSU Geoanalytical Laboratory or by fusion inductively coupled plasma mass spectrometry (ICP-MS) at Activation Laboratories, Ontario, Canada. To augment our dataset of whole-rock analyses we flash-melted selected portions of eight more pumice fragments and dome rocks from the same sample suite. The resulting glass beads were analyzed for major elements by electron microprobe. This method enabled us to obtain a bulk chemical analysis of small, uniform portions of compositionally banded pumice and dome rock. All samples were ground using a mortar and pestle, and then melted in a platinum crucible at 1350°C for 10 min. The resulting glass was reground using a mortar and pestle, remelted at 1350°C for a further 10 min, and then quenched in water.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIXREFERENCESREFERENCES |
|---|
Petrology
Rock samples used in this study were obtained from small explosive events that occurred in June and December 1997 and January, March, and April 1998. All samples represent juvenile magma and were initially studied under the petrographic microscope. We focused our analytical efforts on pumice lapilli from the eruption of June 30, 1997 and dense blocks of dome material from the eruption of January 1, 1998. The mineralogy of the dome rocks from explosive events in December 1997 and January, March, and April 1998 is identical to the mineralogy of the pumiceous samples of the June 30, 1997 eruption. The rapid quenching of lapilli from the June 30, 1997 eruption made these samples more suitable for analysis by electron microprobe.
Whole-rock compositions of the January 1998 dome samples (Table 1, Fig. 2) are silicic andesite to dacite—similar to samples from the March 1996 dome (Athanasopoulos, 1997). Brown and white pumice, erupted in June 1997, are compositionally distinct. Brown pumice is andesitic, whereas white pumice is intermediate in composition between brown pumice and the 1996 and 1998 dome rocks. Whole-rock data suggest no systematic change in major or trace element composition with time.
|
|
, white pumice from the June 30, 1997 eruption; 
, gray dome rock from the January 1, 1998 eruption; +, whole-rock analyses of the March 1996 dome (Athanasopoulos, 1997Recent Popocatépetl eruptive products contain phenocrysts and microphenocrysts of olivine, clinopyroxene, orthopyroxene, plagioclase and hornblende. Microphenocrysts of titanomagnetite, ilmenite, and rare apatite are distributed in a microcrystalline groundmass. Chromian spinel commonly occurs as inclusions in olivine. Titanomagnetite and ilmenite occur as inclusions in pyroxene and rarely in plagioclase. Apatite and rare globular sulfide are also found as inclusions, most commonly in pyroxene. Anhedral xenocrysts of quartz are rare. The groundmass varies from light to dark in color. Dark groundmass consists of a dense network of plagioclase and Fe–Ti oxide microlites in brown glass. Light-colored groundmass contains sparse plagioclase microlites in colorless glass. The relative abundance of mineral phases in Popocatépetl's recent eruptives is: plagioclase > orthopyroxene
clinopyroxene > olivine > ilmenite titanomagnetite hornblende > chromian spinel apatite > sulfide. Modal analyses are shown in Table 2.
|
Olivine (0·1–1·5 mm) comes in two textures. Larger-sized olivine grains (>0·5 mm) are the more abundant, are generally subhedral, and locally glomerocrystic. They contain abundant chromian spinel inclusions (
15 mm diameter) but no glass inclusions and consist of broad, compositionally uniform cores (Fo87–91) with narrow (20–70 µm) normally zoned rims (Fo73–88). NiO content is high (2300–6700 ppm) and varies sympathetically with MgO content in the cores of these olivine grains (Fig. 3 and Electronic Appendix, Table A4). Some of these glomerocrysts as well as some individual grains of olivine are anhedral and display embayed margins that have reacted extensively with the surrounding silicate liquid. We suggest that these large, Cr-spinel-rich olivines are probably xenocrystic.
|
The other type of olivine, microphenocrystic (<0·3 mm), is much less abundant. Some grains have a skeletal (arrow-shaped) habit suggesting rapid growth from the surrounding silicate melt. Olivine microphenocrysts contain few, if any, chromian spinel inclusions but do contain 5–40 µm diameter inclusions of clear brown glass. Glass inclusion-bearing olivine microphenocrysts have uniform cores but have a lower Fo content and more limited compositional range (Fo84–86) than the large-sized Cr-spinel-rich olivines. Glass inclusion-bearing olivine microphenocrysts, however, have much lower NiO contents (1100–1800 ppm) for Fo84–86 (Fig. 3).
Orthopyroxene and clinopyroxene are common as phenocrysts in all samples and reach 1 mm in length. Both clinopyroxene and orthopyroxene phenocrysts contain abundant glass inclusions 5–50 µm in diameter. Rare, globular, sulfide inclusions 5–15 µm in size and lath-shaped apatite 10–50 µm long also occur as inclusions in pyroxene. Microphenocrysts of clinopyroxene and orthopyroxene (30–300 µm) are common, frequently have octagonal or lath-shaped cross-section, and rarely contain melt or mineral inclusions. Most pyroxene core compositions occur as high-Mg-number and low-Mg-number phases [Mg-number = Mg/(Mg + Fe2+); Fig. 4]. Plots of SiO2, Al2O3, and Cr2O3 vs Mg-number help delineate the two groups of clino- and orthopyroxene (Electronic Appendix, Fig. A1).
|
Pyroxenes are characterized by broad, compositionally uniform cores with or without narrow, zoned rims (Electronic Appendix, Table A7). High- and low-Mg-number clinopyroxene cores are augite with compositions En47–49Fs8–10Wo42–44 (Mg-number 83–86 and Cr2O3 0·2–0·7 wt %) and En40–44Fs13–18Wo40–45 (Mg-number 70–78 and Cr2O3 < 0·3 wt %), respectively. High-Mg-number orthopyroxene cores are bronzite with composition En82–83Fs14–15Wo2–4 (Mg-number 84–85 and Cr2O3 0·1–0·4 wt %). Low-Mg-number orthopyroxene cores are hypersthene with composition En61–69Fs29–35Wo2–3 (Mg-number 63–72 and Cr2O3 < 0·15 wt %). Core-to-rim zonation in pyroxene is highly variable. High-Mg-number pyroxenes are generally normally zoned (up to
En = 21 mol %) whereas low-Mg-number pyroxenes are predominantly reversely zoned (up to En = 15 mol %). Many pyroxene grains, however, are unzoned or have weak normal to reverse zoning (En <3 mol %).
Hornblende phenocrysts commonly have an acicular habit and are up to 1 mm in length. More equant hornblende microphenocrysts are generally 0·2 mm x 0·3 mm in size. All hornblendes have dark ‘gabbroic-type’ (Garcia & Jacobson, 1979) breakdown rims 30–100 µm across made of tiny plagioclase, pyroxene, Fe–Ti oxides, and rare sulfides. Rare hornblende phenocrysts and microphenocrysts possess small plagioclase or pyroxene cores. Some hornblendes form partial overgrowths on olivine and pyroxene.
Popocatépetl hornblende has a restricted compositional range (Electronic Appendix, Table A9) with Mg-number = 63–72 [Mg-number = Mg/(Mg + FeT)], 9–13 wt % Al2O3, low concentrations of Cl (200 ppm) and high F (2100–5400 ppm). This contrasts with some subduction zone volcanoes, such as Unzen and Montserrat, that erupt magmas containing two compositionally distinct amphiboles. Popocatépetl hornblende compositions (Electronic Appendix, Fig. A2) are similar to groundmass amphibole (magnesiohastingsite) in mafic inclusions in Montserrat andesite (Murphy et al., 2000) and groundmass amphibole (pargasite) from Unzen volcano (Sato et al., 1999). They plot well outside of the field for amphibole phenocrysts (magnesiohornblende) in Montserrat andesite and Unzen dacite.
Complete descriptions of all samples and analyses of all mineral phases are given in the Electronic Appendix, which can be downloaded from http://www.petrology.oupjournals.org/.
Estimation of intensive variables: T, P, and fO2
Two-pyroxene geothermometry
Magmatic temperatures were calculated using orthopyroxene–clinopyroxene pair compositions and the QUILF program (Andersen et al., 1993). Results are listed in the Electronic Appendix (Table A12). Most pyroxene pairs used in the geothermometry calculation were in direct contact; however, a few pyroxene pairs were separated by up to a few millimeters of groundmass.
Low-Mg-number pyroxene core pairs yield temperatures of 959 ± 17°C (n = 12). High-Mg-number pyroxene core pairs are out of equilibrium and do not yield reliable temperature estimates. Temperatures derived from pyroxene rims occur in low- and high-temperature groups. The low-temperature rims yield 949 ± 32°C (n = 11) and involve unzoned to weakly zoned low-Mg-number pyroxenes or high-Mg-number pyroxenes with strong normal zonation. The temperatures derived from the pyroxene cores and low-Mg-number pyroxene rims are in good agreement. The high-temperature pyroxene rims yield 1055 ± 9°C (n = 3) and involve either low-Mg-number pyroxene with reversely zoned rims or high-Mg-number pyroxenes with weak normally zoned rims. Straub & Martin-Del Pozzo (2001) found one pair of low-Mg-number pyroxenes and determined a temperature of 930–950°C using the geothermometer of Lindsley & Andersen (1983).
Hornblende–plagioclase geothermometry
The composition of Popocatépetl hornblende allows use of the hornblende–plagioclase (Hbl–Plag) geothermometer of Holland & Blundy (1994). However, accurate application of the Hbl–Plag geothermometer is difficult, as the composition of plagioclase in equilibrium with hornblende at the time of crystallization is unknown. Taking a conservative approach, we assume that at hornblende saturation, the coexisting plagioclase had a composition that lies in the range of the measured plagioclase compositions (An28–61). Utilizing the edenite–tremolite geothermometer (for mineral assemblages lacking quartz) over a pressure range of 0–5 kbar, the entire measured plagioclase and hornblende compositional range yields temperatures from 775 to 900°C.
Apatite saturation temperature
Apatite saturation temperatures were calculated using the model equation from Piccoli & Candela (1994):
 | (1) |
is the P2O5 concentration in the initial melt (whole-rock P2O5 content), 
is the SiO2 concentration in the melt at apatite saturation (whole-rock SiO2 content), and X is the crystallinity (in wt %). Using whole-rock compositions of the most silicic clasts, and 35 vol. % (40 wt %) crystallinity (from modal analysis of June 30, 1997 white pumice samples), these calculations suggest apatite saturation at between 945°C and 959°C (n = 4). This is in good agreement with temperatures derived from cores of low-Mg-number pyroxene.
Fe–Ti oxide geothermometry and oxygen barometry
The thermodynamic formulation of Ghiorso & Sack (1991) and QUILF (Andersen et al., 1993) were both used to estimate temperature and fO2 from Fe–Ti oxide pairs (Electronic Appendix, Table A13). Oxide pairs that did not pass the Mg/Mn partition test of Bacon & Hirschmann (1988) were not used. Most of the oxide pairs used in the T–fO2 calculation were in contact; pairs not in contact were less than 100 µm apart.
Fe–Ti oxide compositions cover a wide range (XIlm = 0·63–0·79 and XUsp = 0·30–0·47). Similarly, temperature estimates from both geothermometers cover a broad range [910–1030°C using QUILF and 1040–1190°C using Ghiorso & Sack (1991); n = 27 pairs]. Oxygen fugacity results are comparable using the two methods and have the range NNO + 0·5 to NNO + 1·5 (where NNO is the nickel–nickel-oxide buffer). Athanasopoulos (1997) analyzed several Fe–Ti oxide pairs from the March 1996 lava dome that have a very similar compositional range. Variable degrees of magma mixing and/or incomplete thermal re-equilibration are possible explanations for the wide variation in T and fO2 obtained using each method.
The differing results from the two methods, however, must be reconciled. We suggest that the values presented here using the thermodynamic formulation of Ghiorso & Sack (1991) may significantly overestimate the temperature of Popocatépetl magma. Evans et al. (2001) found that for hydrous, silicic and intermediate volcanic rocks at high oxygen fugacity (>NNO + 1) and 700–900°C, the model of Ghiorso & Sack (1991) overestimates T and fO2. This error may be associated with inadequate characterization of the 
transition in the rhombohedral phase (B. W. Evans & M. S. Ghiorso, personal communication, 2001). New experimental results for the Fe–Ti–O system at 1 bar and 1000–1300°C are consistent with this hypothesis (Lattard et al., 2001). The new data suggest that Popocatépetl Fe–Ti oxides equilibrated at temperatures 
1000°C at an oxygen fugacity between NNO and NNO + 1. This temperature and fugacity estimate is consistent with QUILF calculations. We conclude from this that Fe–Ti oxides record temperatures that probably span the range 900–1000°C and fO2 = NNO to NNO + 1. This range in Fe–Ti oxide results suggests a cooling array that defines a line in T– fO2 space. Where this line intersects the temperature derived from pyroxene thermometry (950°C), oxygen fugacity has the value NNO + 0·5.
Glass inclusions
Glass inclusions occur in all mineral phases. We focused analytical efforts on glass inclusions in olivine and pyroxene with the aim of obtaining chemical data on the volatile contents of the pre-eruptive magmatic end-members. Only glass inclusions that were 
10 µm in diameter and without intersecting cracks were considered. Despite taking care in choosing glass inclusions, some analyses displayed elevated Ca and Al contents. These anomalous analyses were interpreted to result from microscopic post-entrapment plagioclase crystallization creating inhomogeneity of the quench products contained in the glass inclusion. Such analyses were excluded from further consideration. The concentration of H2O in glass inclusions was estimated using both SIMS and the difference technique (100 wt % – analytical total of major and volatile elements by EPMA; Anderson, 1974).
Forty-two glass inclusions in olivine were analyzed for major elements (Fig. 5; Electronic Appendix, Table A14). Twenty of these glass inclusions in olivine were analyzed for S, Cl, and F (Fig. 6; Electronic Appendix, Table A16). Sixteen of these were recalculated for post-entrapment crystallization (PEC) of olivine using the thermodynamic model of Kress & Ghiorso (2004). This correction calculation is performed iteratively by incrementally adding back into the liquid (corresponding to the glass composition) instantaneous equilibrium solid in 0·01 wt % increments. Initial glass composition is assumed to have been reached when the inclusion composition most closely approaches equilibrium with the bulk host phenocryst. The method assumes that the inclusions were sealed throughout the crystallization process and that crystallization was rapid relative to diffusive processes in the crystal. S, Cl, and F concentrations are adjusted by assuming that they are totally incompatible. The magnitude of PEC in olivine-hosted inclusions varies between 1 and 10 wt %. Thus, recalculated volatile contents can be up to 10% lower than the measured values. PEC calculations for these samples using PETROLOG v2.1.3 (Danyushevsky, 2001) yielded similar results.
|
|
Twenty-one glass inclusions in orthopyroxene and 30 glass inclusions in clinopyroxene were analyzed for major elements (Fig. 5; Electronic Appendix, Table A14). Of these, 18 glass inclusions in orthopyroxene and 23 glass inclusions in clinopyroxene were analyzed for S, Cl, and F (Fig. 6; Electronic Appendix, Tables A17 and A18). PEC correction calculations were performed on 13 of the orthopyroxene-hosted glass inclusions using the thermodynamic model of Kress & Ghiorso (2004)
. Calculated PEC in orthopyroxene-hosted glass inclusions ranges from 0 to 5 wt %. PEC calculations using PETROLOG v2.1.3 (Danyushevsky, 2001) suggest 1–5 wt % post-entrapment crystallization for the clinopyroxene-hosted glass inclusions.
According to the classification scheme of Le Bas et al. (1986), PEC-corrected glass inclusions in olivine are mostly basaltic andesite. Glass inclusions in clinopyroxene span the range basaltic andesite to rhyolite. Glass inclusions in orthopyroxene are dacite to rhyolite (Fig. 5). With increasing silica, Al2O3, TiO2, MgO, FeO, CaO, and P2O5 decrease and K2O increases. There are no obvious trends for MnO, Cr2O3, and Na2O.
Glass inclusions were chosen for analysis by observation of two-dimensional thin sections. Unseen cracks or re-entrants intersecting a glass inclusion in the third dimension may have escaped detection. These potentially allowed for volatile leakage from the glass inclusion during ascent and eruption. Additionally, diffusive loss of H2O or OH may have influenced the water content preserved in glass inclusions. In an attempt to filter out a possible volatile leakage and/or diffusive bias in our data we used the composition of glass inclusions with the highest sulfur and water concentrations to represent pre-eruptive volatile contents (Fig. 7). Glass inclusions with the highest sulfur and water contents generally have the highest Cl contents.
|
Volatile-rich glass inclusions hosted by olivine are basaltic andesite in composition. The olivine-hosted glass inclusion with the highest volatile content (GI1OL6tsB) contains 1980 ppm S, 1060 ppm Cl, 950 ppm F, and 3·3 wt % H2O. The five glass inclusions in orthopyroxene with the highest volatile contents (MI1opx8tsSCa, MI2opx8tsSCa, MI3opx8tsSCa, MI4opx8tsSCa, MI2opx1tsPumc) are rhyolitic and average 130 ± 50 ppm S, 880 ± 70 ppm Cl, 570 ± 100 ppm F, and 2·9 ± 0·2 wt % H2O (±1
). The five glass inclusions in clinopyroxene with the highest volatile contents (MI1cpx25ts1, MI2cpx25ts1, MI1cpx12ts2, MI3cpx12ts2, MI1cpx14tsPumc) are also rhyolitic and average 80 ± 100 ppm S, 760 ± 110 ppm Cl, 830 ± 270 ppm F, and 3·2 ± 0·6 wt % H2O (±1). The volatile contents of intermediate composition (andesite to dacite) glass inclusions hosted by olivine, orthopyroxene, and clinopyroxene are highly variable.
Matrix glasses
Fifty-seven major element analyses and 13 analyses of S, Cl, and F in matrix glass were performed by EPMA (Figs 5 and 6; Electronic Appendix, Tables A15 and A20). The concentration of H2O in the matrix was calculated using the difference technique (Anderson, 1974). Each analysis represents a single point in the matrix glass. Care was taken to avoid the solid phases in microlite-rich glass by inspecting the region of interest with high-magnification backscattered electron (BSE) imaging. Analyses with unusually high Ca, Al, or Fe contents were presumed to be due to subsurface plagioclase and/or magnetite microlites and were excluded. Our major and volatile element analyses are comparable with those of light and dark matrix glasses found in the March 1996 dome rocks of Popocatépetl reported by Athanasopoulos (1997).
Matrix glasses have compositions ranging from andesite to rhyolite (Fig. 5; Electronic Appendix, Tables A15 and A20) according to the classification scheme of Le Bas et al. (1986). Brown, microlite-rich matrix glasses are more mafic than colorless, microlite-poor matrix glasses. The matrix glass compositional variations mimic that of the glass inclusions: with increasing silica, Al2O3, TiO2, MgO, FeO, CaO, and P2O5 decrease and K2O increases. Within error, no trend exists for MnO, Cr2O3, and Na2O. Matrix glasses from the brown andesite and white dacite pumices of the June 30, 1997 eruption contain 60–68 wt % SiO2 and 70–72 wt % SiO2, respectively. Matrix glass in the dacite dome rock erupted January 1, 1998 is the most felsic with 76–77 wt % SiO2.
The averages of all matrix glass volatile analyses are 30 ± 30 ppm S, 600 ± 110 ppm Cl, 720 ± 200 ppm F, and 1·3 ± 1·0 wt % H2O by difference (Fig. 7). Our data suggest a relatively high retention of H2O in the matrix glasses. Typically, degassed unaltered glasses at the surface will contain <0·3 wt % H2O (Newman et al., 1988). The estimated minimum detection limit using the H2O by difference technique is 1 wt % H2O. We believe the anomalously high H2O contents reported here for matrix glass reflect cumulative errors in the ‘H2O by difference’ technique. There appears to be no systematic variation in volatile content of the matrix glass, regardless of the matrix glass color, microlite content, or silica content.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIXREFERENCESREFERENCES |
|---|
Evidence for magma mixing
Evidence for magma mixing is abundant in the pumice and dome rocks from Popocatépetl on the macroscopic, petrographic, and chemical level. Hand samples of pumice and dome rocks contain bands and lenses of light and dark material suggesting the mechanical mingling of magmas of differing composition. Disequilibrium textures and mineralogy include olivine (Fo90) in a dacitic to rhyolitic matrix glass (some with orthopyroxene reaction rims), sieve-textured and resorbed plagioclase, reverse zonation in the rims of low-Mg-number pyroxene, and normal zonation in the rims of olivine and high-Mg-number pyroxene. Our observations are consistent with other studies of Popocatépetl that suggest magma mixing has been a common process during the 40 kyr history of the modern Popocatépetl cone (Cantagrel et al., 1984
; Robin, 1984; Boudal & Robin, 1988; Kolisnik, 1990; Athanasopoulos, 1997).
Definition of the end-member magmas: MELTS modeling
It is likely that one or both of the presumed magmatic end-members may themselves be the product of mixing processes. Nevertheless, variation in the most recent eruption products can be well described by mixing of two distinct and relatively homogeneous compositions immediately prior to eruption, and this final mixing event is likely to be the most relevant in terms of unraveling details of the eruption process. We used the MELTS algorithm (Ghiorso & Sack, 1995) to help define the compositions and intensive parameters of the mixing end-members in the following manner.
We first assumed that end-member compositions lay along the linearly regressed mixing line (Fig. 2). Compositions along extrapolations of this mixing line were explored to find potential end-member compositions that could have been in equilibrium with phenocryst phases observed in the rock samples. The mineral compositions determined by MELTS must be consistent with phase compositions determined by electron microprobe. Reasonable pressure conditions were iteratively tested using MELTS. The magmatic conditions were constrained with temperatures, H2O contents, and fO2 determined from the methods described previously. This approach allows simultaneous application of a large number of thermodynamic constraints on potential conditions and compositions. In practice, this places tight constraints on permissible end-member compositions. Estimated bulk compositions and magmatic conditions of the silicic and mafic end-members are shown in Table 3.
|
It is generally recognized that MELTS is better calibrated and more accurately represents phase equilibria in mafic compositions. Nevertheless, there is currently no tool that allows simultaneous application of as many phase equilibria constraints in compositions relevant to nature.
The MELTS-modeled silicic end-member is a dacite (65 wt % SiO2) that crystallized predominantly at low pressures (<1·5 kbar or <6 km) with 0·5–1·5 wt % H2O and fO2 = NNO + 0·3 log units. A crystallization interval of 1000–900°C produces a phase assemblage of plagioclase + clinopyroxene + orthopyroxene + titanomagnetite ± ilmenite. MELTS predicts 30–40 wt % feldspar, 6–8 wt % pyroxene, and 0·5–1 wt % titanomagnetite (at 0·5–1·5 wt % H2O, 0·1–1 kbar, 950°C, and fO2 = NNO + 0·3 log units), which is consistent with modal observations (Table 2). The MELTS-predicted silicate liquid compositions at these conditions also match the microprobe analyses for matrix glasses in the samples of June 30, 1997 white dacite pumice (Fig. 5).
The MELTS-modeled mafic end-member is a basaltic andesite (53 wt % SiO2) with 3 wt % H2O and fO2 = NNO + 1 log units (Table 3). The pressure–temperature evolution of the mafic end-member consists of olivine fractionation at moderate pressures (2–4 kbar, 1150–1300°C) followed by low-pressure (1–2 kbar, 1050–1150°C) pyroxene fractionation.
MELTS modeling and experimental determinations of the olivine/silicate melt distribution coefficient for Ni (Seifert et al., 1988; Snyder & Carmichael, 1992; Gaetani & Grove, 1997) support a xenocrystic origin of the large, Fo87–91, Ni-rich and Cr-spinel-bearing olivine (Fig. 3). Using experimental values for the Ni distribution coefficient, the high Ni content in these olivines required large degrees of fractionation to produce the observed Ni contents in the mafic liquid. Such high degrees of fractionation cannot be reconciled with the lower degrees of Fe/Mg fractionation implied in the same mafic liquid. The low-Ni, Fo84–86 olivines that contain glass inclusions probably did crystallize from the mafic end-member. We speculate that the xenocrystic Fo87–91, Ni-rich olivines and their included Cr-spinel were plucked from the mantle and/or crystallized from a primitive Mg-, Ni-, and Cr-rich mantle melt (Clynne & Borg, 1997) which were subsequently incorporated into Popocatépetl's mafic end-member magma.
Hornblende: a product of mixing
Hornblende phenocrysts and microphenocrysts are found evenly disseminated, surrounded by both the dark and light matrix glasses, in all pumices and lava blocks. Petrographic observations are inconsistent with hornblende originating from only the silicic or mafic end-members because hornblende is found as partial overgrowths on both low-Mg-number pyroxene and olivine. If Popocatépetl hornblende is derived neither from the mafic nor the silicic end-member, it must have crystallized during mixing. Plagioclase and pyroxene microphenocrysts form cores of some hornblende, suggesting that hornblende saturated very late in the crystallization sequence. In addition, the acicular habit of hornblende suggests that crystallization was rapid. Compositionally, Popocatépetl hornblende is very similar to groundmass pargasite in Unzen dacite (Sato et al., 1999) and groundmass amphibole (magnesiohastingsite) found in mafic inclusions in Montserrat andesite (Murphy et al., 2000; Electronic Appendix, Fig. A2).
The concentration of Cl in Popocatépetl hornblende is consistent with a low-pressure origin. Sato et al. (1999) found that magnesiohornblende phenocrysts in Unzen lavas contained much higher concentrations of Cl than did the groundmass pargasite. They concluded that the low Cl content of the Unzen groundmass pargasite (100–300 ppm Cl) is a result of crystallization from a partially degassed magma. Popocatépetl hornblende also has low Cl contents (<300 ppm Cl). We infer that Popocatépetl hornblende also crystallized from a partially degassed magma and that partial degassing may have served to enhance hornblende saturation (Ghiorso, 1999).
Proportions and depth of magma mixing
With the compositions of the mixing end-members estimated, we can calculate the proportions of mixing needed to derive the compositions of individual pumice or lava clasts (Table 3). Compositions of the most mafic samples (58 wt % SiO2), represented by brown andesite pumice erupted on June 30, 1997, are consistent with mixing of 55 wt % mafic end-member and 45 wt % silicic end-member. The most silicic clasts (63 wt % SiO2) are represented by white dacite pumice erupted on June 30, 1997 and gray dacite dome rocks erupted on January 1, 1998. These are inferred to be made up of 85 wt % silicic end-member and 15 wt % mafic end-member.
Based on the minimum pressure estimates for the mafic end-member (1 kbar) and the maximum pressure estimate for the silicic end-member (1·5 kbar) derived from the MELTS simulations presented above, we infer that magma mixing occurred at pressures between 1 and 1·5 kbar (4–6 km depth). This result is toward the shallow end of the 4–13 km range of mixing depths proposed by Straub & Martin-Del Pozzo (2001) for the recent eruptives of Popocatépetl.
Immediate sources of volatiles
There are several models that could explain the observed gas emissions at Popocatépetl. These include: (1) degassing from erupted silicate melt; (2) degassing from large bodies of unerupted magma intruded at shallow levels into the volcanic edifice; (3) fumarolic remobilization; (4) breakdown of sulfur-bearing magmatic phases; (5) assimilation of carbonate and evaporite country rock; (6) a separate volatile phase in the erupted magma. We also present convection of magma in the volcanic conduit supplemented by a separate volatile phase in the convecting magma as a viable scenario to explain the degassing process.
Degassing from erupted silicate melt
Volatile contents of glass inclusions and coexisting degassed matrix glasses can be compared to estimate the extent of SO2 degassed from silicate melt during eruption. This calculation is performed using the equation (Gerlach et al., 1996)
 | (2) |
melt is the density of the silicate melt phase, S is the difference between sulfur concentration in glass inclusions and matrix glass, and Fmelt is the volume fraction of melt (i.e. 1 – crystallinity). The amount of degassing from the silicate melt portion of Popocatépetl's erupted magma is governed by the relative contribution of volatiles from the melts of the two end-member magmas. As the bulk proportion of mixing between the end-member magmas prior to eruption is unknown and/or may vary with time, we investigate the entire range of mixing proportions represented by individual clasts, i.e. 45:55 wt % and 85:15 wt % silicic: mafic magma. The amount of sulfur each magmatic end-member contributes to degassing can be estimated by taking the difference in sulfur content between glass inclusions and the matrix glass for mafic and silicic end-members. Equation (2) yields 0·13 Mt SO2 for the 45:55 mixture and 0·04 Mt SO2 for the 85:15 mixture. Regardless of mixing proportion, sulfur dioxide emitted from the melt portion of 3·1 x 107 m3 of erupted magma is over two orders of magnitude lower than the 14·8 Mt SO2 emission measured by COSPEC during the magmatic period under study. Even if 100% of the exsolved gas came from the melt phase of an aphyric mafic end-member with a volume of 3·1 x 107 m3, SO2 emission increases to only 0·27 Mt.
HCl and HF emissions can be estimated using expressions analogous to equation (2). Chlorine emission calculations utilizing petrological data yield a total emission of 0·02 Mt HCl for the recent 4 year 10 month magmatic period. Similar to the case for SO2 described above, the calculated HCl emission is approximately two orders of magnitude less than the 1·8 Mt HCl total emission measured by passive FTIR (Goff et al., 2001). The amount of fluorine that has exsolved from the melt portion of erupted magma is difficult to assess because the matrix glass F content significantly overlaps the F contents in glass inclusions in phenocrysts. As a result, the mass of degassed F cannot be reliably calculated.
Degassing from large bodies of unerupted magma intruded to shallow levels
Degassing of magma emplaced at shallow levels within the volcanic edifice has been proposed as a source of ‘excess’ sulfur and other volatiles (Rose et al., 1982; Andres et al., 1991). To produce 14·8 Mt SO2, melt degassing from a hypothetical silicic end-member composition (S = 100 ppm, melt = 2340 kg/m3, Fmelt = 0·65) would require complete degassing of 50 km3 magma. Melt degassing from a mafic end-member composition (S = 1950 ppm, melt = 2470 kg/m3, Fmelt = 0·9) would require a volume of 2 km3 magma to completely degas.
Efficient degassing of silicate melt requires either high degrees of crystallization or low pressure. Our analyses of glass inclusions suggest Popocatépetl mafic and silicic melts contain 3 wt % H2O prior to eruption. According to the H2O solubility model of Moore et al. (1998), the mafic and silicic magma compositions of Popocatépetl are saturated with 3 wt % H2O at a pressure of 700 bars (3 km depth). CO2 will probably saturate at greater depth, but the volume fraction of this CO2-rich vapor will be small until depths at which appreciable H2O comes out of solution (Newman & Lowenstern, 2002).
If we assume complete degassing of a 2 km3 magma body intruded at 3 km depth as a result of crystallization, we can employ modified Mogi modeling (Okada, 1992) to estimate the surface deformation associated with intrusion of such a magma body. Calculations using the Mogi.m code (P. Cervelli, personal communication, 2000) predict 57 m of vertical displacement. Incomplete degassing would require more magma, leading to greater surface deformation. Likewise, emplacement at depths shallower than 3 km would also lead to greater vertical deformation than the predicted 57 m. Continuous GPS deformation monitoring of Popocatépetl has so far shown negligible deformation (Cabral-Cano et al., 1999). In conclusion, degassing from a shallow magma body recently intruded into the volcanic edifice is an unlikely source of ‘excess’ sulfur at Popocatépetl.
Fumarolic remobilization
Some sulfur emissions may be derived by fumarolic remobilization of pre-existing sulfur deposits within the volcanic edifice (M. O. Garcia, personal communication, 2000). Such sulfur deposits may take the form of fumarolic precipitates of native sulfur and a variety of other sulfur-bearing minerals filling cracks in wall rocks. Intrusion of fresh magma could provide the heat necessary to liberate sulfur from these deposits.
To produce the 14·8 Mt SO2 would require remobilization of 7·4 Mt of native sulfur. Such a mass of native sulfur would occupy 3·6 x 106 m3 ( = 2070 kg/m3 for native sulfur). This volume is equivalent to a 15 m wide, 1000 m long annulus of pure sulfur around a hypothetical 30 m radius cylindrical conduit. The existence of a pure sulfur deposit of this size seems unlikely. Additionally, remobilization of much smaller volume sulfur deposits that are widely disseminated in narrow cracks would require heating of an unreasonably large volume of wall rock to produce the observed sulfur emissions. The flux of sulfur derived from fumarolic remobilization is probably negligible compared with the total sulfur output.
Breakdown of sulfur-bearing magmatic phases: anhydrite and sulfide
The volume of silicate melt required to produce the observed SO2 emission can be reduced if melt degassing is supplemented by significant breakdown of sulfur-bearing phases such as anhydrite or sulfide. Primary anhydrite has not been observed in Popocatépetl magmas. This is consistent with our estimate of redox state for these lavas (NNO 0·5), which is not sufficiently oxidized to stabilize anhydrite (Carroll & Rutherford, 1987; Luhr, 1990).
Fe-rich sulfide globules are found as inclusions in pyroxene. In addition, Fe–Ni- and Fe–Cu-rich sulfide globules are found among the breakdown products in hornblende reaction rims and on the edges of phenocrysts in contact with silicate melt (Witter, 2003). These sulfides could contribute to sulfur output at Popocatépetl. Larocque et al. (2000) proposed that immiscible sulfide liquids break down during depressurization and degassing to form spongy Fe-oxides and sulfur gas. Rare spongy Fe-oxide grains are found on vesicle walls of recent pumices and lava blocks of Popocatépetl (Witter, 2003).
To produce the entire 14·7 Mt of SO2 gas not accounted for by melt degassing would require breakdown of 15 vol. % sulfide liquid in the erupted volume. If this occurred, we would expect to find a sub-equal amount of spongy Fe-oxide breakdown products in the recent Popocatépetl eruptives. As such a large concentration of sulfide liquid is unlikely and we find only trace amounts of spongy Fe-oxides, breakdown of sulfide liquid is unlikely to have produced more than a small portion of the measured SO2 emissions.
Assimilation of carbonate and evaporite country rock
Cretaceous marine carbonate, shale, and evaporite form outcrops near Popocatépetl and are presumed to compose part of the basement rocks that underlie the volcano. Early Tertiary intrusive bodies cut these marine rocks and it seems plausible that dikes feeding Popocatépetl would also intersect these rocks (Goff et al., 2001). Volatile-bearing carbonate and evaporite country rocks could be assimilated into Popocatépetl magma and provide a rich non-magmatic source for the emitted volatiles.
Goff et al. (2001) hypothesized that episodic ingestion and assimilation of carbonate country rocks may explain huge excursions in the CO2 flux from Popocatépetl. Large, short-term increases in the CO2/SO2 mass ratio, the presence of metamorphosed carbonate xenoliths in eruptives from the last 14 kyr, and the lack of an adequate alternate source of CO2 were provided as support for their hypothesis.
Although we do not have CO2 data from glass inclusions in phenocrysts from the recent eruptives of Popocatépetl to address this hypothesis, S, Cl, and F emissions appear to be decoupled from CO2 emissions. During large increases in the CO2/SO2 mass ratio, SO2/HCl and SO2/HF ratios remain stable (Goff et al., 2001). If the ‘excess’ CO2 is caused by carbonate assimilation then this suggests that the assimilated country rock does not contain appreciable quantities of S-, Cl-, and F-bearing minerals.
Sulfur isotope data for Popocatépetl gases indicate 
34S = +1·5 to +6·5
with an average of +3·4 (Goff et al., 1998). Gypsum from a regional evaporite deposit yielded a 34S value of 16·4. The sulfur isotopic composition of Popocatépetl gases falls within the range for gases from continental margins and island arcs (34S = 0 to 12; Taylor, 1986). Sulfur originating from the primitive mantle has 34S 1. Goff et al. (1998) suggested that isotopic enrichment of the gas phase over mantle values reflects assimilation of evaporite country rocks. Alternatively, if a more typical arc 34S magmatic signature for Popocatépetl magmas is assumed, then little to no contamination is required to produce the observed 34S values. Isotopic fractionation of sulfur between various phases and oxidation states can be expected to increase 34S without requiring assimilation (Ohmoto & Rye, 1979). Continued degassing of the magma reservoir can be expected to increase 34S still further, as described by McKibben et al. (1996). Both acid gas ratios and sulfur isotope systematics provide evidence that the contribution of evaporite to the volatile budget at Popocatépetl is probably relatively minor.
Separate volatile phase in erupted magma
Recent work has suggested that arc magmas are commonly volatile-saturated at depth. Many of these magma bodies may also harbor a substantial coexisting volatile-rich fluid phase prior to eruption (Luhr et al., 1984; Anderson et al., 1989; Lowenstern et al., 1991; Westrich & Gerlach, 1992; Lowenstern, 1993; Gerlach & McGee, 1994; Gerlach et al., 1994, 1996; Wallace & Gerlach, 1994; Wallace et al., 1995; Wallace, 2001). A separate volatile phase that exists at depth and ascends with the silicate liquid could be a rich source of magmatic gases to account for the large-magnitude gas emissions at Popocatépetl. The low dissolved Cl content measured in glass inclusions (<1000 ppm) is insufficient for the formation of a hydrosaline brine in equilibrium with Popocatépetl silicate melt (Webster et al., 1999). A separate volatile phase at Popocatépetl is more likely to be dominated by H2O and CO2. The composition and relative proportion of such a volatile-rich fluid can be calculated (Table 4).
|