FIRE and MUD Contents

Preeruption Pressure-Temperature Conditions and Volatiles in the 1991 Dacitic Magma of Mount Pinatubo

By Malcolm J. Rutherford1 and Joseph D. Devine1

1Department of Geological Sciences, Brown University, Providence, RI 02912.


The pumice erupted from Mount Pinatubo June 14-15 1991, is composed of approximately 80 to 90 percent white phenocryst-rich dacite and approximately 10 to 20 percent tan, finer grained, fragmental-looking dacite of the same bulk composition. The phenocryst phase assemblage of plagioclase (An34-66), hornblende, cummingtonite, biotite, quartz, magnetite, anhydrite, ilmenite, and apatite occurs in both pumice types. Although there is both normal and reverse chemical zoning in plagioclase, and slight zoning in some hornblende crystals, the compositions of phenocrysts in contact with matrix glass (melt) are relatively uniform. Microprobe analyses of melt inclusions trapped in plagioclase, hornblende, quartz, and cummingtonite crystals indicate that they are all volatile-rich (H2O=5.1 to 6.4 weight percent by the difference method), high-SiO2 rhyolite glasses similar to the matrix glass on an anhydrous basis. Ion probe analyses confirm the H2O content of 5.5 to 6.4 weight percent for these preeruption melts, and infrared spectroscopic analyses indicate that dissolved CO2 is less than 20 parts per million. The average sulfur content of the melt inclusions ranges from 55 to 77+-28 parts per million, which is 19 to 40 parts per million in excess of the matrix glass concentration.

Conditions in the preeruption white dacitic magma, as deduced from iron-titanium geothermometry and Al-in-hornblende geobarometry, are 780+-10° Celsius, an oxygen fugacity of NNO+3, and a total pressure of 220+-50 megapascals, which is equivalent to a 7 to 11 kilometer depth beneath Mount Pinatubo. This pressure is confirmed by experimentally determined stability limits for cummingtonite and cummingtonite + quartz at 780° Celsius in the Mount Pinatubo dacite composition. The preeruption magma is required to be very H2O rich in order to stabilize cummingtonite, and to explain approximately 6.4 weight percent volatiles in the melt inclusions. The composition of the melt in equilibrium with the natural phenocrysts is reproduced at approximately 200 MPa under H2O-saturated conditions and is not reproduced at higher temperature, or H2O pressure; it is produced in 300-MPa experiments at XH2O in the fluid equal to 0.7. It is concluded that the 1991 dacitic magma of Mount Pinatubo was essentially volatile saturated with an H2O-rich fluid just prior to the eruption.

Note to readers: Figures and tables open in separate windows. To return to the text, close the figure or table's window or bring the text window to the front.


The 1991 eruptions of Mount Pinatubo began with a series of steam and ash emissions in April that were followed by lava dome extrusion in the period of June 7-12. An explosive eruption on June 12 produced a plinian column and a tephra deposit and destroyed parts of the dome (Pallister and others, this volume). These events culminated in a series of plinian and lateral blast eruptions during June 14-15, with a paroxysmal eruption occurring in the early hours of June 15. This was one of the largest explosive eruptions of the past century (Pinatubo Volcano Observatory Team, 1991).

The June 7-11 lava dome and the June 12 plinian eruptions involved magma of overall andesitic composition that was apparently a mixture of basaltic and dacitic components (Pallister and others, 1992; this volume). The June 14-15 eruptions, however, produced mainly dacitic pumice, approximately 80 to 90 percent of which is white and coarsely crystalline (~35 percent crystals with phenocrysts up to 5 mm). The remainder of the erupted pumice (10 to 20 percent) is tan and finer grained but is generally similar in composition and mineralogy to the coarsely crystalline dacite. The tan, fine-grained pumice contains crystals that are both smaller and more angular than those contained in the white pumice and are interpreted to be fragments of larger crystals. Pallister and others (1992) report finding inclusions of the white, phenocryst-rich variety in phenocryst-poor pumice and also find the two types mingled together in banded pumices. Fragments of both pumice types were found in geographically diverse tephra samples supplied to us for study.

The 1991 Pinatubo eruptions are of great interest in the field of petrology and magma dynamics as well as to the atmospheric chemistry and climate dynamics communities. Petrological and volcanological interest stems from the many unique aspects of the magma system and the eruptions. For example, the dacitic magmas are very sulfur rich and oxidized compared to most calc-alkaline dacites (Bernard and others, 1991; Rutherford and Devine, 1991). The dacites also contain a very distinctive phenocryst phase assemblage including plagioclase, hornblende, cummingtonite, quartz, biotite, anhydrite, magnetite, and ilmenite, an assemblage which should contain sufficient petrogenetic information to reveal the preeruptive history of the magma. The presence of mixed basalt and dacite magmas in some samples and the possibility that the tan, fine-grained pumice is from the basalt-dacite thermal boundary layer (Pallister and others, 1992) suggest a rare opportunity for a pressure-temperature-time study of magma interaction in a well-defined, two-magma system. The origin of the crystal fragmentation in the tan pumice must also be explained. Finally, there are interesting questions concerning the origin of the sulfur in the magma and the origin of the huge mass of SO2 injected into the stratosphere during the June 14-15 plinian eruptions. Several alternatives have been suggested for the sulfur in the magma, including build-up through crystal fractionation, vapor transfer from underlying basalt, and assimilation of overlying anhydrite and apatite-bearing hydrothermal vein deposits (Matthews and others, 1992; McKibben and others, 1992; Pallister and others, 1992), but convincing evidence does not yet exist for any of these models.

The June 14-15 Plinian-style eruptions injected an estimated 20 Mt of potentially climate-altering SO2 into the stratosphere (Bluth and others,1992), giving rise to predictions of its atmospheric and climatological importance. The origin of the atmospheric sulfur remains an interesting and unsolved problem, because it is clear that it was not dissolved in the preeruption melt phase of the erupted magma (Rutherford and Devine, 1991; Westrich and Gerlach, 1992). Westrich and Gerlach (1992) have suggested that the preeruption magma contained an excess sulfur-bearing volatile phase, a phase which could have carried the excess SO2. A source involving degassing of associated basalt is suggested by Pallister and others (1992) and Matthews and others (1992). Another alternative is proposed by Baker and Rutherford (1992), who present evidence suggesting that the atmospheric sulfur may have formed by breakdown of anhydrite accompanying H2O degassing of magma during its ascent.

The present study of crystal-melt-fluid equilibria in the phenocryst-rich dacite has been carried out in order to evaluate the various models of magma petrogenesis, magma degassing, and eruption processes proposed for the 1991 Mount Pinatubo eruption. Hydrothermal experiments have been carried out on the natural dacite to simulate phase equilibria in the preeruption magma storage region, and both natural and experimental samples have been analyzed by microbeam and spectroscopic techniques.



Pumice fragments were collected at sites 10 to 15 km south, east, and northeast of the volcanic center by T. Casadevall and E. Endo of the U.S. Geological Survey, who kindly supplied samples for this study. The major element compositions of phenocryst phases, glassy melt inclusions trapped in phenocrysts, and matrix glasses from the 1991 dacites (white and tan) were determined by using established electron microprobe techniques (Rutherford and Devine, 1988). Sodium-loss during microprobe analysis of glasses was monitored and accounted for online by using the method of Nielsen and Sigurdsson (1981). Crystals and coexisting melt (glass) in experimental charges were analyzed by using the same analytical routines.

Sulfur and chloride analyses were also obtained with the electron microprobe. The wavelength of sulfur X-rays produced by electron bombardment of geologic glasses depends on the oxidation state of sulfur in the sample (Carroll and Rutherford, 1988). Extended sulfur peak searches of Pinatubo glasses revealed that the dissolved sulfur exists predominantly in the oxidized state (S6+), as would be expected in a highly oxidized melt (see below) in equilibrium with magmatic anhydrite. Sulfur analyses took into account this wavelength shift. The error bars on the glass sulfur analyses are large because the concentrations are close to the 30-ppm (parts per million) detection limit.

The H2O content of a few plagioclase and quartz melt inclusions was determined by using the ion probe facility at Woods Hole Oceanographic Institution and methods developed by G. Layne. A new set of hydrothermally fused rhyolitic glasses was analyzed by FTIR and used as standards for these ion probe measurements. The small size of Mount Pinatubo melt inclusions generally prevented quantitative H2O and CO2 analyses in our micro-FTIR facility, but we were able to determine that the preeruption CO2 content of the Mount Pinatubo melt was below the detection limit of about 20 ppm.


Hydrothermal experiments have been carried out on a crushed sample of 1991 white pumice at various pressure, temperature, fH2O (water fugacity), and fO2 (oxygen fugacity) conditions that bracket those suggested for the preeruption magma by geothermometry and geobarometry on the natural phenocryst assemblage. Samples consisting of 50 to 100 mg of powder and a weighed quantity of H2O+-CO2 (as AgC2O4) were loaded in Ag or AgPd tubes (one end sealed) and placed in 5-mm sealed Ag tubes along with a second tube containing a solid-phase oxygen buffer. Re-ReO2 (ReO) and MnO-Mn3O4 (MNO) buffers were used. The experiments were carried out in Rene pressure vessels with an H2O pressure medium. Pressures were recorded with a pressure transducer checked against a factory-calibrated Heise gauge and are considered accurate to +-1 MPa. The temperature in the sample position has been checked by melting experiments on NaCl and Au and is within +-5°C of the recorded temperature.



The June 15 phenocryst-rich white dacite, which represents about 80 to 90 percent of the erupted material, contains about 35 volume percent phenocrysts (<5 mm) in a vesiculated glassy matrix. The main phenocryst phases, plagioclase and hornblende, coexist with smaller amounts (<3 percent) of cummingtonite, biotite, quartz, magnetite, ilmenite, anhydrite, apatite, and zircon. The phenocryst phases are generally subhedral to euhedral in appearance, with the exceptions of quartz and biotite, which are more rounded. Many phenocrysts have been fractured (fig. 1), and the fragments have suffered different degrees of separation.

Figure 1. Plane light (A) and crossed-polar (B) photomicrographs of tan pumice from the 1991 eruption of Mount Pinatubo showing fragmental nature and small size of crystals relative to white, phenocryst-rich variety. Field of view in A-D is 10 mm horizontally; field is 1 mm in E-G. Plane light (C) and crossed-polar (D) photomicrographs of a white pumice fragment with crystals up to 5 mm across. Q, quartz; Pg, plagioclase; Hb, hornblende. E, Adjacent plagioclase and hornblende (with cummingtonite rim), both with melt inclusions. F, Photomicrograph showing euhedral green-tan hornblende (lower) without obvious cummingtonite and a similar crystal (upper) with a substantial cummingtonite (C) rim. G, A subhedral anhydrite crystal (A) surrounded by vesicular glass.

Plagioclase in the white pumice is concentrically and variably zoned, with cores as rich in anorthite as An66 and as low as An33. The composition of plagioclase rims in contact with matrix glass averages An41.1+-4.2 (table 1), but values as low as An33 have been measured. Some of the variability on this average is undoubtedly the result of difficulty in analyzing thin rims on crystals with strong chemical zonation. Analyses of plagioclase rims in contact with glass in the tan pumice (An39) show that they are similar in composition to the rims of white pumice plagioclase. In contrast, the average composition of the "cores" of crystal fragments in the tan pumice (An40) are not as anorthite rich on average as the cores of white pumice phenocrysts (~An46). This is considered to be a sampling artifact. Specifically, the "cores" of small, tan pumice plagioclase fragments that have at least one planar face (that is, rim) against glass are themselves near-rim samples of formerly large crystals.

Table 1. Phenocrysts in white dacite from the 1991 eruption of Mount Pinatubo.1

Pleochroic green to tan hornblende in the white dacite is relatively uniform in composition (table 1); early-crystallizing or xenocrystic cores can be recognized by high Al2O3 and MgO contents. Cummingtonite typically occurs as euhedral overgrowths on some hornblende crystal faces, and biotite is commonly present as euhedral to anhedral inclusions. Magnetite (<1 mm) typically contains apatite and rare chalcopyrite inclusions and may be intergrown with ilmenite. All grains analyzed for the purpose of oxygen barometry were pairs in contact. Subhedral anhydrite phenocrysts occur either alone in contact with melt or occasionally in aggregates with apatite. Chalcopyrite occurs exclusively as inclusions in other minerals and was not observed in contact with glass in the groundmass.

Both hornblende and plagioclase contain abundant glassy melt inclusions whose average size is small (<30 lc mum) compared with other tephra we have studied. Overnight leaching of plagioclase mineral separates in HBF4 reveals that many plagioclase melt inclusions are intersected by minute cracks (which are etched by the acid). Melt inclusions in hornblende may also be intersected by cracks or cleavage planes that potentially could have resulted in the loss of some of the original volatile content from melt inclusions during eruption. Melt-inclusion analyses must therefore be interpreted in light of this possibility.


The major and volatile element compositions of melt inclusions and matrix glasses have been determined for samples of the white June 15 dacite pumice from each of the three sites sampled. No differences in the glass (or mineral) chemistry have been observed between the different sites or from sample to sample beyond that which exists in a given sample. When the analyses are normalized to an anhydrous basis, the average hornblende, plagioclase, and quartz melt-inclusion compositions are essentially identical to the high-SiO2 rhyolite matrix glass (table 2). Although inclusions in cummingtonite are rare and small, and therefore difficult to analyze, they also have the same high-SiO2 rhyolite chemical composition.

Table 2. Melt inclusions and matrix glasses in white dacite from the 1991 eruption of Mount Pinatubo.1

The sulfur and chlorine contents of the melt inclusions and matrix glass, also determined by electron microprobe, show very small differences between the different melt- inclusion populations and the matrix glass. In fact, the chlorine content of the matrix glass and melt inclusions is equal (1,200+-120 ppm) within the limits of the electron microprobe analyses, although the thin strands of matrix glass are difficult to analyze. The average sulfur abundance in hornblende melt inclusions (77+-29 ppm) is similar to that determined by Westrich and Gerlach (1992) and about 40 ppm above the average matrix glass composition (36+-28 ppm). The average sulfur content of plagioclase melt inclusions is somewhat lower (55+-23 ppm). A comparison of melt-inclusion sulfur content with variations in the host hornblende composition indicates that a few anomalous inclusions in terms of sulfur abundance are associated with compositionally anomalous hornblende. The cores of some hornblende phenocrysts have higher MgO (and lower K2O) than the rims; melt inclusions in these cores contain higher sulfur contents. The higher MgO hornblendes may be xenocrysts; they chemically resemble hornblendes in basalt inclusions in the dacite (Pallister and others, 1992; this volume). These MgO-rich hornblendes probably crystallized at >780°C on the basis of the higher melt inclusion sulfur abundance (~80 to 100 ppm) and available data on sulfur solubility (see next section). It should be noted, however, that most melt inclusions contain a high-SiO2 melt similar in composition to the matrix glass, and the sulfur abundance in these inclusions is only 20 to 40 ppm in excess of that present in the matrix glass.

The H2O and CO2 contents of melt inclusions were determined as accurately as possible (given their small size) by using a combination of infrared spectroscopy (FTIR), ion probe, and electron microprobe techniques. No doubly polished sections of plagioclase or quartz melt inclusions sufficiently large for quantitative FTIR analyses of H2O and CO2 were obtained. However, infrared spectroscopy of inclusion-bearing quartz and plagioclase showed CO2 abundances below detection (<20 ppm) and H2O contents of >5.0 weight percent. Using a new set of FTIR-calibrated standards, ion probe analyses of melt inclusions in plagioclase and quartz yielded H2O contents of 6.4 to 6.6 and 7.0 weight percent, respectively. The volatile content (H2O+-CO2) of the trapped melts estimated by these techniques is consistent with the electron microprobe "volatiles by difference" method within the error limits of microprobe analyses (table 2); they indicate an H2O abundance of 5.1 to 6.4 weight percent. The lower H2O abundances may reflect some H2O loss during the eruption. It is concluded that samples of melt containing ~6.4 weight percent H2O, ~1,200 ppm chlorine and ~70 ppm sulfur were trapped by growing hornblende, plagioclase, and cummingtonite phenocrysts in the 1991 Mount Pinatubo magma storage region prior to the eruption. The similarity in major and minor element chemistry, particularly K2O, of the melt inclusions and matrix glasses indicates that no significant crystallization occurred after melt entrapment. The apparent lack of chlorine loss from the interstitial melt (matrix glass) as the vapor phase grew during the eruption may be explained, at least partly, by the large decrease in DCl (Cl concentration in melt/Cl concentration in vapor) that occurs with H2O dilution of the vapor phase. Significantly lower amounts (30 percent) of Li, Be, and B and a slightly higher H2O in the quartz melt inclusions, as determined by ion microprobe, suggest that these melts may have been trapped in a separate magmatic stage compared to melts trapped by hornblende and plagioclase. On the basis of the rounded quartz crystal outlines and the slightly lower MgO, FeO and CaO in the trapped melts, the quartz growth took place in a more evolved melt that was part of an earlier more crystal-rich magma.


The average composition of magnetite and ilmenite obtained from analyses of 12 crystal pairs in contact with melt are given in table 1. All analyses cluster tightly around the average, as indicated by the low standard deviations. The compositions of each oxide pair yielded a temperature of 810°C +-10° when the geothermometer calibration of Andersen and Lindsley (1988) and the solution model of Stormer (1983) were used. The use of other solution models (Andersen and Lindsley, 1988) produces less than 10°C variation in the calculated temperatures. The average log fO2 indicated by these oxides is -11.1, three log units above the Ni+NiO oxygen buffer (NNO+3.0). This solution is outside the limits of the Andersen and Lindsley calibration, however, so these temperature and fO2 estimates need to be confirmed. In a study of iron-titanium oxide temperatures and cummingtonite stability, Geschwind and Rutherford (1992) found that in this fO2 range (NNO+3), the Andersen and Lindsley (1988) algorithm yields oxide temperatures that are high by 30°C. This is now confirmed by the results of experiments in which the Re+ReO2 oxygen buffer is used (table 3). This buffer, which is calculated to lie about 2 to 3 log units above NNO (Pownceby and O'Neill, 1991), was used in this study, and the iron-titanium oxides from some experiments have been analyzed. When the Andersen and Lindsley (1988) algorithm is applied, these oxide analyses yield temperatures 30+-5°C above the experimental temperature at a log fO2 of -10 to -11, depending on the temperature (fig. 2). On the basis of these data, it is concluded that the white pumice from Mount Pinatubo was at 780+-10°C just prior to eruption. The oxides of the tan, crystal-poor dacitic magma yield identical temperature-fO2 conditions.

Figure 2. Oxygen fugacity-temperature diagram showing the results from iron-titanium oxide geothermometry (Andersen and Lindsley, 1988, algorithm) for the white pumice (Pin) from the 1991 eruption of Mount Pinatubo. The +'s show the results when the algorithm was used for oxides in the Re+ReO2-buffered experiments at 780, 810, and 840°C. As discussed in the text, a -30°C correction must be applied to algorithm results in order to reproduce experimental temperatures under these oxidizing conditions. ReO is the calculated position of the Re-ReO2 buffer. The 1982 El Chichón (EC) temperature and the FeS-CaSO4 equilibrium are from Carroll and Rutherford (1987). HM, hematite-magnetite; MNO, manganite-hausmanite; NNO, nickel-nickel oxide; QFM, quartz-fayalite-magnetite; WM, wustite-magnetite, (all are standard oxygen buffers).

An independent estimate of magmatic temperature can be obtained from sulfur concentrations in the melt inclusions and experimental sulfur-solubility data if it is assumed that the silicate melt was sulfate saturated when the melt inclusions were trapped. This appears to be a reasonable assumption, given the compositional similarity of the melt inclusion and matrix glasses. Figure 3, modified from Carroll and Rutherford (1987), shows that sulfur solubility increases significantly with temperature and fO2 above 800°C. Experiments at 800°C yield an estimated sulfur-solubility of ~60+-20 ppm in the anhydrite-saturated Mount Pinatubo melts over a range of fO2 from NNO+1 to MNO. If the Mount Pinatubo magma melt inclusions had equilibrated with anhydrite at temperatures above 800°C, they should contain measurably higher sulfur contents than those observed. Melt inclusions with large vapor bubbles were purposely avoided because of the presumed secondary leakage problem; therefore, no variation in the sulfur content was observed that could be attributed to this leakage.

Figure 3. Sulfur solubility versus temperatures for dacitic composition melts modified from Carroll and Rutherford (1987). New experimental data are plotted at 800°C. MNO and NNO oxygen buffer curves as in figure 2; NNO+1, one log unit above the NNO buffer.


The composition of hornblende in the phenocryst-rich dacite of Mount Pinatubo (hereafter referred to as Mount Pinatubo dacite) is slightly variable, with Mg# (100xMg/Mg+Fe atomic) varying from 69 to 62 and Al2O3 contents varying from 7.0 to 11.0 weight percent. The hornblende in contact with melt is relatively uniform in composition however, with an average Mg# of 66+-1 and an Al2O3 content of 7.9+-0.8 weight percent. The variable core compositions are not necessarily representative of earlier high-temperature crystallization; cores have both higher and lower Mg# than the rims. Some of these crystal cores with atypically high Al2O3 contents and irregular zoning may be xenocrystic, as indicated above. In contrast, cummingtonite is present as rims on many hornblendes and as small euhedral crystals in contact with melt, and is quite uniform in composition (table 1).

Using the Johnson and Rutherford (1989) calibration of the Al-in-hornblende geobarometer, the hornblende in equilibrium with melt in the phenocryst-rich 1991 Mount Pinatubo dacite (7.8+-0.9 weight percent Al2O3) indicates a pressure of 220+-50 MPa. This calibration is used because the experiments were done at the same temperature (760 to 780°C) as that determined for the white Mount Pinatubo dacites; the Schmidt (1992) calibration at 650 to 700°C yields a pressure of ~320 MPa. Alkali feldspar, which is rare in the Pinatubo dacite, is required in the geobarometer phenocryst-melt assemblage, however, so these pressure estimates also need to be confirmed. Unfortunately, the experiments performed as part of this project have not yielded new Al-in-hornblende data because of the hornblende composition variability in the natural starting material and because of cummingtonite overgrowth interference with hornblende-melt reequilibration.


Phase equilibrium experiments were performed on samples of the white phenocryst-rich pumice of Mount Pinatubo (hereafter referred to as Mount Pinatubo pumice) to determine the water pressure required to stabilize the natural phenocryst assemblage. Most experiments were done by using the Re+ReO2 oxygen buffer, but MNO was also used to explore the effect of small changes in fO2 under the very oxidizing conditions indicated by the natural iron-titanium oxides. Figure 4 shows the pressure-temperature dacite phase equilibria determined for water-saturated conditions. The iron-titanium oxide temperature and total pressure from Al-in-hornblende geobarometry are indicated. Water-saturated conditions could have existed in the Mount Pinatubo preeruption magma since, at 220 MPa, the rhyolitic composition melt would contain 6.4 weight percent H2O (Fogel, 1989; Silver and others, 1989), equivalent to the H2O content determined from melt-inclusion analyses. Figure 4 indicates that the natural phenocryst phase assemblage is stable between 150 and 320 MPa at 780°C. The stability of cummingtonite is particularly sensitive to water pressure (PH2O) and temperature changes in this general pressure-temperature range; it is replaced by low-CaO pyroxene above 800°C and at pressures above 350 MPa. These are the conditions of maximum cummingtonite stability. Quartz disappears from the Mount Pinatubo dacite phase assemblage at water pressures above 300 MPa at 780°C. Thus, the Mount Pinatubo dacite phenocryst assemblage could not have existed above 800°C under any set of Ptotal or PH2O conditions in the preeruption magma, and the maximum pressure range possible in a 780°C magma reservoir is from 150 to 300 MPa under water-saturated conditions.

Figure 4. Pressure-temperature diagram for the dacite composition of Mount Pinatubo under H2O-saturated conditions. Triangular symbols show the location of experiments (table 3) and point in the direction of approach to the final conditions; filled symbols are used where phenocryst phases found in the natural pumice are stable. The curved phase boundaries indicate the appearance of a phase in this dacitic composition. The shaded symbols at 780°C and 220 MPa mark (with error bars) the conditions in the preeruptive 1991 dacitic magma from iron-titanium oxide geothermometry and Al-in-hornblende geobarometry (see text). Hb, hornblende; Opx, orthopyroxene; Pg, plagioclase; Bi, biotite; Q, quartz; C, cummingtonite. NNO+3, 3 log units above the Ni+NiO oxygen buffer.

The stability of cummingtonite at any total pressure is further restricted if the magma is not water saturated. Figure 5 shows the results of water-undersaturated experiments at a total pressure of 220 MPa. At this pressure and 780°C, cummingtonite is stable in melts only where the XH2O (molar) in the coexisting fluid is >0.75. The results of H2O-undersaturated experiments at 300 MPa (table 3) show that cummingtonite is only stable when the XH2O in the fluid is >0.70.

Figure 5. Temperature-XH2O (mole fraction H2O in the fluid) diagram for the 1991 dacite composition of Mount Pinatubo at 220 MPa and a NNO+3 (Re+ReO2 buffer) oxygen fugacity. Symbols as in figure 4. The shaded area indicates where the Pinatubo phenocryst phase assemblage is stable at 780°C.


The preeruption conditions in the white Mount Pinatubo dacitic magma as indicated by the above methods can be tested by comparing experimental glass compositions with the matrix and melt-inclusion glasses. This is a sensitive test because the melt composition responds rapidly to changes in intensive parameters, coming to equilibrium as indicated by the similar compositions in melting and crystallization experiments. Equilibrium melt-phase compositions appear to be achieved by reaction with crystal rims, even though equilibrium among the crystalline phases is not totally achieved, as indicated by some residual chemical zonation, or is difficult to demonstrate. However, because of the highly variable zoning in the natural Mount Pinatubo plagioclase starting material and because of the slow reequilibration rates of these crystals at 780°C, it has not yet been possible to determine the equilibrium experimental plagioclase compositions with sufficient confidence that they can be used to constrain conditions in the preeruption magma.

The glasses in ReO-buffered experiments carried out under different pressure-temperature and PH2O conditions are compared with Mount Pinatubo melt-inclusion and matrix glasses in figure 6. These plots of Al2O3 versus SiO2 illustrate the fact that an H2O- saturated experiment at ~200 to 220 MPa reproduces a melt composition equivalent to the tight cluster of natural glasses. Glasses produced in higher PH2O experiments are not similar to the natural glasses. Similarly, the experimental glass composition moves away from the natural glass compositions with increases in temperature (constant Ptotal and XH2O) and decreases in XH2O at 220 MPa. The latter is somewhat surprising in the light of previous studies of XH2O on crystallization but is tentatively attributed to destabilization of cummingtonite in our water-undersaturated experiments. The glass produced at 300 MPa, 780°C, and XH2O in the fluid=0.70 is an excellent match with the melt-inclusion glasses, including the dissolved volatile content, although cummingtonite appears incipiently unstable in these experiments. Thus, the glass data suggest that the Mount Pinatubo magma last equilibrated at 220 MPa under H2O-saturated conditions or at a pressure between 220 and 300 MPa; the XH2O in the fluid would decrease as the pressure increases, reaching 0.70 at 300 MPa.

Figure 6. Comparisons of natural and experimental glass compositions on a plot of Al2O3 versus SiO2. Melt inclusion (indicated by host mineral abbreviations as in figure 4) and matrix glass compositions (MG) are from table 2. The experiments, indicated by the experimental temperature in (A), the pressure in MPa (B), or the XH2O in the coexisting fluid (C), are listed in table 3.

Table 3. Hydrothermal experiments on 1991 dacite of Mount Pinatubo.

Table 3 footnotes:
1 Starting material: 1=disaggregated white, crystal-rich pumice from Mount Pinatubo; 2=white pumice annealed under high temperature (860°C) at an H2O pressure of 100-130 MPa for 4-6 days.
2 Product abbreviations are: Pg=plagioclase, Hb=hornblende, C=cummingtonite, Bi=biotite, Opx=low-Ca pyroxene, Q=quartz, M=magnetite, I=ilmenite, and G=glass. Parentheses indicate a phase present only in trace amounts.


The phenocryst-melt assemblage in the white dacitic pumice of Mount Pinatubo is interpreted to be an equilibrium assemblage. The compositions of the phenocryst phases, including the iron-titanium oxides, cummingtonite, and the rims of plagioclase and hornblende, are uniform within a sample and from sample to sample. In addition, the matrix glass is uniform in composition throughout these phenocryst-rich pumice samples and is identical to the melt-inclusion glasses on an anhydrous basis. As a result, it is possible to recognize and determine the equilibrium conditions in the preeruption phenocryst-rich dacite. The temperature of the magma reservoir was 780°C+-10° according to iron-titanium oxide geothermometry (table 1, fig. 2), where the error represents a 1 sigma standard deviation on the oxide analyses. Given that this temperature is also supported by the low sulfur content of melt inclusions compared to experimental sulfur-solubility data (fig. 3), and the thermal limit on cummingtonite stability (Geschwind and Rutherford, 1992; fig. 4), this temperature is considered to be well constrained. The magma was also very oxidized (NNO+3 log units), sufficiently so to stabilize anhydrite.

The total pressure in the 1991 Mount Pinatubo magma storage zone is estimated to have been 220+-50 MPa (Al-in-hornblende geobarometry), or approximately 7 to 11 km depth. This pressure is supported by cummingtonite stability data (fig. 5), which show that the total pressure had to be in the 150 to 350 MPa range at 780°C, assuming the magma was water saturated, and is also supported by the comparison between natural glass and the experimental glass compositions at 200 to 220 MPa. If the magma was water undersaturated, (PH2O<Ptotal), then the maximum pressure for cummingtonite would be less than 350 MPa. The residual melt compositions produced experimentally indicate that a range of conditions between 220 MPa, H2O saturated, and 280 MPa with XH2O=0.70 are possible; the lower pressure estimate is favored on the basis of the Al-in-hornblende barometry and because there is no detectable CO2 in the melt trapped during the most recent phenocryst growth event (hornblende + plagioclase + cummingtonite growth). Assuming the preeruption magma was near volatile saturation, a volatile species other than SO2 is necessary to lower the XH2O in the fluid if the total pressure in the magma is above 220 MPa; CO2 is the only reasonable candidate, and apparently it was not present. The petrological estimate of magma reservoir pressure/depth compares well with seismic activity following the June 15 eruption, which appears to outline an aseismic zone (fig. 7) below about 7 km (Wolfe, 1992; Mori, Eberhart-Phillips, and Harlow, this volume). The 7 to 11 km depth estimate for the dacite magma storage region also compares well with a large low-velocity seismic zone identified as a magma reservoir by Mori, Eberhart-Phillips, and Harlow (this volume).

Figure 7. A cross section through Mount Pinatubo (modified after Pallister and others, this volume, and Mori, Eberhart-Phillips, and Harlow, this volume) showing seismicity from May 8 to August 19, 1991, projected on to the plane of the east-west cross section. The X marks the 7 to 11 km source depth for the June 14-15 magma storage region; on the basis of the petrology of the eruption products; on the basis of the geobarometry, the horizontally ruled portion of the aseismic magma storage zone appears not to have erupted on June 14-15.

A complete description of conditions in the preeruption magma storage region requires an estimate of the PH2O necessary to stabilize the observed phenocryst-melt assemblage, and if PH2O is less than Ptotal, an indication whether other volatile species are sufficiently abundant to make Pfluid equal to Ptotal is needed. The ion probe analyses of melt inclusions trapped in phenocrysts of the white pumice indicate an H2O content of 6.4 weight percent. As discussed in the previous section, the microprobe "volatiles by difference" estimates of 5.1 to 6.4 weight percent dissolved H2O are consistent with the ion probe data. Interestingly, the Mount Pinatubo plagioclase rim and matrix glass compositions used with the Housh and Luhr (1991) plagioclase-melt model yield H2O estimates of 6.4 and 5.9 weight percent using their AB and AN models, respectively. If the preeruption melt was saturated with H2O at 220 MPa, it would contain ~6.4 weight percent H2O according to water solubility experiments (Fogel, 1989; Silver and others, 1989). This is sufficiently close to the measured concentration to suggest that the magma may have been fluid saturated. Certainly if there was an excess fluid (vapor) phase, it had to be very H2O rich. This conclusion, required by the similarity of measured and calculated H2O-saturation abundances and by the cummingtonite stability at 780°C, is supported by analytical and solubility data for CO2 and SO2. Less than detectable CO2 concentrations in melt inclusions require a PCO2 of <25 MPa (Fogel and Rutherford, 1990). Similarly, experimental data on SO2concentrations in H2O-rich fluids coexisting with anhydrite suggest that the PSO2 would have been ~11 MPa in a 220-MPa pressure fluid (Baker and Rutherford, 1992). Thus, we conclude here that the Mount Pinatubo preeruption dacitic magma was very close to saturation, if not in fact saturated, with an H2O-rich fluid at a depth equivalent to 220 MPa.

The above conclusion that the Mount Pinatubo dacitic magma was volatile saturated suggests that the huge excess of SO2 injected into the stratosphere during the June 14-15 eruptions might be explained by the release of an H2O-rich, sulfur-bearing vapor phase that coexisted with the preeruption magma (Westrich and Gerlach, 1992). The 20 Mt of SO2 injected into the stratosphere (Bluth and others, 1992) is far in excess of the SO2 lost by melt (melt-inclusion sulfur minus matrix-glass sulfur is <40 ppm) in the approximately 5 km3 (DRE) of erupted Mount Pinatubo magma (~0.52 Mt of SO2). While a vapor phase in the preeruption dacitic magma is possible and even likely, it may not be the main source of the excess atmospheric sulfur for several reasons. The excess sulfur injected into the stratosphere is equivalent to ~1,000 ppm sulfur (2,000 ppm SO2) in the 5 km3 of erupted magma. If a magma density of 2.30 g/cm3 and an SO2 vapor density of 0.7 g/cm3 (at 220 MPa) are assumed, 2,000 ppm of SO2 is equivalent to ~0.6 percent of the magma by volume. But the vapor at depth is approximately 99 percent H2O according to recent analyses of the Baker and Rutherford (1992) experiments, which means that the magma would have contained >20 volume percent vapor (vapor bubbles) at depth. It seems likely that this gas-charged magma should have been erupted during the June 7-13 period when there was both extrusion and explosive eruption of andesite formed by the very recent mixing (Rutherford and others, 1993) of dacitic and basaltic magma beneath Mount Pinatubo. It is also difficult to produce the H2O for a water-rich fluid phase of this size. The erupted magma would have had to suffer approximately 15 volume percent crystallization after reaching H2O saturation in order to generate the excess volatiles. Similarly, other magma(s) at depth would only release their H2O as they crystallized beyond the point of H2O saturation. An alternate source for the excess sulfur is suggested by the experiments of Baker and Rutherford (1992), which indicate that anhydrite in an H2O-rich magma would break down during ascent from depth. If degassing in the Mount Pinatubo dacitic magma began more than 3 h prior to the eruption, at least some of the excess sulfur must have come from anhydrite breakdown in the Mount Pinatubo magma during the eruption.

The origin of the sulfur in the Mount Pinatubo magma system is a problem not directly addressed by this experimental study, but it may be related to the highly oxidized nature of the preeruption magma. The iron-titanium oxide phenocrysts indicate that the magma is among the most oxidized (NNO+3 log units) of calc-alkaline magmas that have been analyzed (for example see Carmichael, 1991). In addition, experimental studies have shown that sulfur solubility may be high in an oxidized melt (fig. 3), depending on the temperature. Luhr (1990) developed a model for the sulfur in the El Chichón magma system that evoked differentiation of an equally oxidized high-temperature (>1,000°C) magma that would have been capable of carrying the sulfur in solution in the melt. A similar origin for the high amount of sulfur carried by the Mount Pinatubo magma is proposed by Bernard and others (1991) and Westrich and Gerlach (1992) to produce anhydrite and an excess sulfur-rich vapor. The problem with this model is that highly oxidized basaltic and andesitic magmas would be required in order to carry the observed sulfur abundances in Mount Pinatubo dacites (fig. 3), and these oxidized magmas have not erupted at Mount Pinatubo (Pallister and others, this volume) or elsewhere (Carmichael, 1991). High-temperature calc-alkaline magmas are generally more reduced than associated low-temperature silicic magmas, in keeping with their mantle origin.

Another possible source of the Mount Pinatubo sulfur is the associated basalt that appeared in the mixed magma of the June 7-11 dome. Pallister and others (1992) and Matthews and others (1992) suggest that the anhydrite and sulfur-rich vapor, respectively, resulted from degassing of basalt that intruded the dacitic magma storage region. Although it is certainly possible that cooling and crystallization of the basalt could have produced a sulfur-rich vapor that was released during the June 14-15 eruption, it is unlikely that this vapor was the source of the anhydrite in the dacite. The ubiquitous presence of anhydrite in the dacite, the presence of anhydrite inclusions in hornblende, and the uniform sulfur abundance (anhydrite saturated) in melt inclusions suggest that the sulfur-rich character of the dacite was developed significantly prior to the eruption. At least some of the anhydrite may have originated from an adjacent hydrothermal zone containing anhydrite and apatite, according to the range of sulfur isotopic compositions found in individual crystals of the June 12 mixed magma (McKibben and others, 1992; this volume).


The main conclusions reached as a result of this analytical and phase equilibrium study of the 1991 Mount Pinatubo eruption products are the following:

1. Although there is evidence of disequilibrium in the form of zoned and cored crystals in the phenocryst-rich dacite pumice, possibly resulting from magma mixing some significant time before the eruption, most phenocrysts are uniform in composition from sample to sample and appear to be in equilibrium with the matrix glass. The rims of chemically zoned plagioclase and hornblende phenocrysts are uniform in composition. In contrast, clear evidence of disequilibrium and magma mixing is present in the andesitic magmas erupted immediately before the June 15 dacites (Pallister and others, 1992; this volume).

2. Melt inclusions trapped in growing phenocrysts in the dacite are H2O-rich (~6.4 weight percent), high-SiO2 rhyolite glasses similar (on an anhydrous basis) to the matrix glass regardless of the trapping phase. This similarity suggests one major melt-entrapment event in the magma, which is surprising, given the complex compositional zoning of some phenocrysts. The sulfur content of the melt inclusions indicates entrapment at ~800°C and a PSO2 of 1 MPa; the H2O content indicates a PH2O of ~220 MPa.

3. Phenocryst geothermometry and geobarometry using iron-titanium oxides and the Al-in-hornblende method indicate preeruption equilibration conditions in the dacite of 780°C +-10°, an oxygen fugacity that equals NNO+3 log units, and a total pressure of 220 +-50 MPa. Similarity of the PH2O and Ptotal estimates suggests that the preeruption magma was volatile saturated.

4. The highly oxidized character of the dacite probably reflects both the H2O-rich nature of the magma and the episodic long-lived character of the storage region (Pallister and others, 1992).

5. Phase equilibrium studies of the white pumice of Mount Pinatubo confirm the geothermometry and geobarometry estimates of conditions in the preeruption magma. In particular, the pressure must have been in the 150- to 320-MPa range for H2O saturation at 780°C in order to stabilize cummingtonite plus quartz in this bulk composition, and the 320- MPa upper limit decreases for H2O-undersaturated conditions.

6. The preeruption June 14-15 Mount Pinatubo magma was H2O rich and essentially volatile saturated, according to analytical data and phase equilibrium constraints, and could have contained an excess vapor phase. However, a vapor phase of sufficiently large volume to explain the excess SO2 injected into the stratosphere on June 14-15 is considered unlikely because of the physical problems involved in building up this H2O-rich volatile phase and in maintaining it during the June 7-13 eruptions.


The authors thank John Pallister, Terry Gerlach, and Chris Newhall for thoughtful reviews that significantly improved the manuscript and thank Elliot Endo and Tom Casadevall for collecting the samples. This research was supported by NSF grant EAR-9105110.


Andersen, D.J., and Lindsley, D.H., 1988, Internally consistent solution model for Fe-Mg-Mn-Ti oxides: Fe-Ti oxides: American Mineralogist, v. 73, p. 714-726.

Baker, L., and Rutherford, M.J., 1992, Anhydrite breakdown as a possible source of excess sulfur in the 1991 Mount Pinatubo eruption: Eos, Transactions, American Geophysical Union, v. 73, no. 43, p. 625.

Bernard, A., Demaiffe, D., Mattielli, N., and Punongbayan, R.S., 1991, Anhydrite-bearing pumices from Mount Pinatubo: Further evidence for the existence of sulfur-rich silicic magmas: Nature, v. 354, p. 139-140.

Bluth, G.J.S., Doiron, S.D., Schnetzler, C.C., Krueger, A.J., and Walter, L.S., 1992, Global tracking of the SO2 clouds from the June 1991 Mount Pinatubo eruptions: Geophysical Research Letters, v. 19, no. 2, p. 151-154.

Carmichael, I.S.E., 1991, The redox states of basic and silicic magmas: A reflection of their source regions: Contributions to Mineralogy and Petrology, v. 106, p. 129-141.

Carroll, M.R., and Rutherford, M.J., 1987, The stability of igneous anhydrite: Experimental results and implications for sulfur behavior in the 1982 El Chichón trachyandesite and other evolved magmas: Journal of Petrology, v. 28, p. 781-801.

------1988, Sulfur speciation in hydrous experimental glasses of varying oxidation state: Results from measured wave-length shifts of sulfur X-rays: American Mineralogist, v. 73, p. 845-849.

Fogel, R.A., 1989, The role of C-O-H-Si volatiles in planetary igneous and metamorphic processes: Providence, R.I., Brown University, Ph.D. thesis, 200 p.

Fogel, R.A., and Rutherford M.J., 1990, The solubility of carbon dioxide in rhyolitic melts: A quantitative FTIR study: American Mineralogist, v. 75, p. 1311-1326.

Geschwind, C.H.G., and Rutherford, M.J., 1992, Cummingtonite and the evolution of the Mount St. Helens magma system: An experimental study: Geology, v. 20, p. 1011-1014.

Housh, T.B., and Luhr, J.F., 1991, Plagioclase-melt equilibria in hydrous systems: American Mineralogist, v. 76, p. 477-492.

Johnson, M.C., and Rutherford, M.J., 1989, Experimental calibration of the Al-in-hornblende geobarometer with application to Long Valley Caldera, Calif., volcanic rocks: Geology, v. 17, p. 837-841.

Luhr, J.F., 1990, Experimental phase relations of water- and sulfur-saturated arc magmas and the 1982 eruptions of El Chichón volcano: Journal of Petrology, v. 31, p. 1071-1114.

Matthews, S.J., Jones, A.P., and Bristow, C.S., 1992, A simple magma-mixing model for sulfur behaviour in calc-alkaline volcanic rocks: Mineralogical evidence from Mount Pinatubo 1991 eruption: Journal of the Geological Society of London, v. 149, p. 863-866.

McKibben, M.A., Eldridge, C.S., and Reyes, A.G., 1992, Multiple origins of anhydrite in Mt. Pinatubo pumice: Eos, Transactions, American Geophysical Union, v. 73, no. 43, p. 633.

------this volume, Sulfur isotopic systematics of the June 1991 Mount Pinatubo eruptions: A SHRIMP ion microprobe study.

Mori, J., Eberhart-Phillips, D., and Harlow, D.H., this volume, Three-dimensional velocity structure at Mount Pinatubo, Philippines: Resolving magma bodies and earthquakes hypocenters.

Nielson, C.H., and Sigurdsson, H.R., 1981, Quantitative methods of electron microprobe analysis of sodium in natural and synthetic glasses: American Mineralogist v. 66, p. 547-552.

Pallister, J.S., Hoblitt, R.P., Meeker, G.P., Knight, R.J., and Siems, D.F., this volume, Magma mixing at Mount Pinatubo: Petrographic and chemical evidence from the 1991 deposits.

Pallister, J.S., Hoblitt, R.P., and Reyes, A.G., 1992, A basalt trigger for the 1991 eruptions of Pinatubo volcano?: Nature, v. 356, p. 426-428.

Pinatubo Volcano Observatory Team, 1991, Lessons from a major eruption: Mt. Pinatubo, Philippines, Eos, Transactions, American Geophysical Union, v. 72, no. 49, p. 545-555.

Pownceby, M., and O'Neill, H., 1991, The NiPd redox sensor: Calibration from EMF measurements and application to the Re-ReO2 oxygen buffer. Eos, Transactions, American Geophysical Union, v. 72, no. 44, p. 522.

Rutherford, M.J., and Devine, J.D., 1988, The May 1, 1980 eruption of Mount St. Helens: 3, Stability and chemistry of amphibole in the magma chamber: Journal of Geophysical Research, v. 93, no. 11, p. 949-959.

------1991, Pre-eruption conditions and volatiles in the 1991 Pinatubo magma: Eos, Transactions, American Geophysical Union, v. 72, no. 44, p. 62.

Rutherford, M.J., Baker, L., and Pallister, J.S., 1993, Petrologic constraints on timing of magmatic processes in the 1991 Mount Pinatubo volcanic system: Eos, Transactions, American Geophysical Union, v. 74, no. 43, p. 671.

Schmidt, M.W., 1992, Amphibole composition in tonalite as a function of pressure: An experimental calibration of the Al-in-hornblende barometer: Contributions to Mineralogy and Petrology, v. 110, p. 304-310.

Silver, L.A., Ihinger, P.D., and Stolper, E., 1989, The influence of bulk composition on the speciation of water in silicate glasses: Contributions to Mineralogy and Petrology, v. 104, p. 142-162.

Stormer, J.C., 1983, The effects of recalculation on estimates of temperature and oxygen fugacity from analyses of multi-component iron-titanium oxides: American Mineralogist, v. 68, p. 586-594.

Westrich, H.R., and Gerlach, T.M., 1992, Magmatic gas source for the stratospheric SO2 cloud from the June 15, 1991 eruption of Mount Pinatubo: Geology, v. 20, p. 867-870.

Wolfe, E.W., 1992, The 1991 eruptions of Mount Pinatubo, Philippines: Earthquakes and Volcanoes, v. 23, no. 1, p. 5-37.

FIRE and MUD Contents

PHIVOLCS | University of Washington Press | U.S.Geological Survey

This page is <>
Contact: Chris Newhall
Last updated 06.11.99