Introduction

Nearly two decades ago, Werner Giggenbach (1992a) introduced a N2–Ar–He ternary diagram for studying geothermal and volcanic gases. This diagram is still quite popular because of its simplicity and its ability to distinguish at least three main sources of fluids: atmosphere, sediments, and mantle. The diagram (Fig. 1) does not allow for almost any terrestrial natural gas to be plotted below a mixing line between air-saturated water (ASW) and the He corner, which means that gases with N2/Ar < 38 are very rare (if they exist) in terrestrial fluids. Gases generated by organic-rich sediments (“thermogenic” gases, producing N2 but not Ar), usually have N2/Ar > > 100 (e.g., Jenden et al. 1988), hence, plot close to the He–N2 axis. Because of their relatively high He content with respect to N2 and low N2/Ar (Marty 1994; Marty and Zimmermann 1999), mantle fluids should plot close to the He corner. Taking into account that subduction-type magmatism commonly is induced by fluids released from the subducting oceanic slab carrying organic-rich oceanic sediments, one can suggest that the N2–Ar–He relative contents of such a magmatic fluid should be similar to that of the sediments themselves. When applying this diagram to volcanic gases, Giggenbach (1992a) outlined three main areas on this plot: Air + ASW, “Andesitic” gases and “Basaltic” gases, keeping in mind that “Andesitic” gases are magmatic gases from subduction zones (subduction type or arc type), and “Basaltic” gases are those from rifts and hotspots (MORB–OIB type). This approach provides a convenient way for discriminating the main contributions to a fluid composition in terms of the relative amounts of N2, Ar, and He.

Fig. 1
figure 1

N2–Ar–He ternary plot (Giggenbach 1992a) with data points for Vulcano (Tedesco et al. 1991; Wahrenberger 1997). Intersections of the dotted and solid lines show two apparent endmember compositions

Full size image

A single volcanic gas analysis always shows some mixing with air that occurs either during sampling and/or analysis, or naturally due to a permeable volcanic edifice near fumarolic vents. The volcanic water is always a mixture of magmatic and meteoric water and/or seawater, the latter two being a source of “atmogenic” N2, Ar, He and other noble gases. Traces of air in a volcanic gas can drastically change N2/Ar and N2/He ratios. This is the main reason for fluctuations of N2 and Ar concentrations and N2/He and N2/Ar ratios in a gas sample. Hence to determine and discriminate the non-atmospheric fraction of N2 and Ar for a given volcano, it is necessary to have a statistically reasonable dataset.

Such datasets exist for only a few volcanoes. The largest sets containing N2, Ar, and He have been obtained for the Italian volcanoes Vulcano, La Solfatara, and Vesuvius (Chiodini et al. 1993, 1995; Tedesco et al. 1991; Tedesco 1995; Tedesco and Scarsi 1999; Magro and Pennisi 1991; Wahrenberger 1997; Caliro et al. 2007). The points shown in Fig. 1 are the compositions of gases from Vulcano taken from Wahrenberger (1997) and Tedesco et al. (1991). Chiodini et al. (1995) suggested that the He-enriched gas from Vulcano is more magmatic, and thus, the trend on the ternary diagram (Fig. 1) is a mixing line between a He-rich magmatic endmember and a He-poor endmember which most authors term “hydrothermal”.

Using this ternary diagram, however, it is impossible to distinguish the mantle and crustal (from the old metamorphic crust) gases which both have a high relative He content and a low N2/He ratio (e.g. Hilton et al. 2002). A relative enrichment in He may indicate both crustal and mantle contributions. On the other hand, relatively low He contents and a high N2/He are compatible with subduction-type magmatic fluids derived from subducted sediments (Giggenbach 1996; Hilton et al. 2002) or with fluids derived from organic-rich sediments of the continental crust (Jenden et al. 1988). Isotopic compositions of He and Ar are a powerful additional tool for resolving such kind of problems.

The aim of this contribution is to focus on the N2–Ar–He systematics of volcanic gases from Vulcano Island using a more complicated approach than a simple ternary plot. The published data on N2, Ar, and He are used here to determine their non-atmospheric fractions in volcanic gases. I also attempt to distinguish magmatic, hydrothermal, and crustal contributions to the N2 + Ar + He system using data for other components (HCl, S, CO2) as well as 3He/4He and 40Ar/36Ar isotopic data.

Geochemical models for Vulcano fumaroles

One of the best geochemical datasets comes from fumaroles of La Fossa crater at Vulcano Island (Aeolian Archipelago, south Italy). This dataset is continuously increasing and is used by different groups for geochemical models. One of the first models was the Carapezza et al. (1981) “pressure cooking” model (pressurized geothermal reservoir with a two-phase brine–vapor fluid) based on the available data up to 1979. On the basis of theoretical constraints, Cioni and D’Amore (1984) postulated a “dry” model with fluid chemistry the result of mixing between a gas of magmatic origin and a vapor derived from lateral seepage and evaporation of a brine of marine origin. There also have been debates on the interpretation of the vapor isotopic composition (Panichi and Noto 1992; Capasso et al. 1997; Bolognesi and D’Amore 1993), in particular the nature of magmatic endmember and the importance of the seawater contribution to Vulcano fluids. Chiodini et al. (1993, 1995) suggested three endmembers with one magmatic (with water isotopic composition close to typical subduction-type water in agreement with the Panichi and Noto model) and two hydrothermal sources. These debates were partly the result of interpretative biases because the oxygen isotopic exchange between H2O and CO2 was not considered. According to Chiodini et al. (2000), oxygen isotope exchange between CO2 and steam is fast enough in natural gas phases to allow rapid isotopic reequilibration within a wide temperature range (100–1,000°C). A way to annul or minimize the effects of this process is simply to consider the oxygen isotope balance of the whole H2O + CO2 gas system (Chiodini et al. 2000). After such correction of analytical data, the isotopic composition of the fumaroles of Vulcano clearly shows mixing between hydrothermal fluids whose water has a marine origin, minor contributions from meteoric waters, and a magmatic component with an isotopic composition of H2O typical of subduction zones and with relatively high contents in CO2, N2 and He. The Chiodini et al. (2000) model suggests magmatic source with apparently constant geochemical characteristics in terms of Cl–H2O–S–N2–He–δD.

In contrast, Nuccio et al. (1999), Nuccio and Paonita (2001), Paonita et al. (2002), and Di Liberto et al. (2002) developed a model where a magmatic endmember, highly variable in its chemical and isotopic composition, mixes with a vapor from a hot (380–400°C), saline (24–30% NaCl) hydrothermal brine of marine origin. Leeman et al. (2005) proposed a three endmember model with magmatic, hydrothermal, and mixed magmatic–hydrothermal sources.

In any case, all authors are in agreement with two main sources for the Vulcano fumaroles: magmatic gases and hydrothermal brine. There is also a generally accepted compromise that magmatic fluid of Vulcano has a complex origin, consisting of significant portion of crustal gases and recycled volatiles from the subducted African slab (Tedesco and Scarsi 1999; Capasso et al. 1997, 2001; Leeman et al. 2005).

The main controversy appeared after 1992, when drastic changes in the chemical composition and in the ratios of the main components marked a new stage in the activity of Vulcano. Simple, quasi-linear, two endmember trends such as H2O–HCl and H2O–Stot have broken down, and highly variable and lowered HCl and Stot contents, independent from the H2O concentration or δD values, have became a common behavior for all fumaroles of Vulcano to the present day (Wahrenberger 1997; Di Liberto et al. 2002).

Two stages of Vulcano fumarolic activity revealed from He and other components

Figure 2 shows a difference between “stage 1” (1987–1991) and “stage 2” (after 1992) on plots where H2O, CO2, HCl Stot, and N2 are plotted against He concentration. These plots, based on data from different sources, demonstrate that between 1987 and 1992, a two-endmember “dry” model was reliable, because all components, including acid (HCl, Stot) and uncondensable (CO2, N2), showed a similar strong positive trend when compared with He concentrations. After 1992, HCl and Stot lost their determinant behavior, but the correlation between CO2 and N2 did not change; the same or nearly the same quasi-linear trend continued for these components after 1992. In other words, two endmember mixing is valid for “non-condensable” species N2 and CO2 for the whole period of observations, but not for “acid” species such as HCl and SO2 which after 1992 do not correlate with He. However, a good correlation between HCl and Stot = SO2 + H2S is found during the whole period as shown in Fig. 3 (data from Chiodini et al. 1995; Tedesco et al. 1991, and Wahrenberger 1997). Strong linear correlations between H2O, CO2, N2, He, and δD have already been shown for the datasets obtained before 1991 by Chiodini et al. (1993) and Tedesco (1995).

Fig. 2
figure 2

Relationships between He and macro-components of the Vulcano gases during 1988–1991 (left) and 1992–2002 (right). Data from Tedesco et al. (1991); Chiodini et al. (1995); Wahrenberger (1997); Paonita et al. (2002)

Full size image
Fig. 3
figure 3

Relationship between Stot and HCl in gases of Vulcano collected in 1988–1996

Full size image

These effects may indicate that the “dry” case, which provided a simple two endmember mixing pattern for all species, has changed to a “wet” situation, when magmatic gas from a deep source passes through a liquid formed by mixing of the magmatic gas condensate with the hydrothermal brine. The poorly soluble CO2, N2, Ar, and He essentially retain their proportions passing through this brine, but vapor concentrations of acid HCl and Stot = SO2 + H2S begin to depend on the vapor/liquid ratio and on the relative solubility of HCl and SO2 in the brine. Solubilities of HCl and SO2, in turn, depend on pH, salinity, and chemical composition of the brine. The SO2 concentrations and SO2/H2S also depend on the redox conditions (Di Liberto et al. 2002). Therefore, the two endmember model, as well as an apparently constant He/Ar, He/N2, and He/CO2 ratios of the magmatic component, still makes sense, with the acid species may varying not because of changes in the magmatic fluid composition, but due to “bubble-liquid” fractionation of acidic gases or gas–brine interaction.

In the two endmember mixing model, the high CO2 (high N2, high He), low H2O endmember was attributed to the magmatic component because of its water isotopic composition, which is typical for other subduction zones (−25‰ < δD < −15‰, Sakai and Matsubaya 1977; Taran et al. 1989; Giggenbach 1992b), although Mediterranean seawater is about 10‰ isotopically heavier that oceanic water (Chiodini et al. 1993, 1995, 2000; Bolognesi and D’Amore 1993; Capasso et al. 1992). Further support for this idea are data on the isotopic composition of the CO2–carbon and 3He/4He reported by Tedesco and Scarsi (1999) for the Vulcano gases sampled during 1986–1993. Together with the interpretation of the δ13C–CO2 correlation by Paonita et al. (2002) based on data by Tedesco (1995) and Capasso et al. (1997, 2001) that the high CO2 endmember has isotopically heavier CO2 (+0.5‰ V–PDB), the data of Tedesco and Scarsi (1999) clearly show that this CO2-rich endmember also has high 3He/4He with maximum values close to 6.2R a (where R a is the atmospheric value of 1.4 × 10−6).

N2–Ar–He in gases of Vulcano

Relative and absolute concentrations of N2, Ar, and He in endmembers

To investigate the N2–He–Ar system, we refer here to published data. The data with simultaneous analyses of N2, Ar, and He have been reported by Magro and Pennisi (1991), Tedesco et al. (1991, 1995) (F5 fumarole, O2 + Ar ≈ Ar) and Wahrenberger (1997). A large data set by Chiodini et al. (1995) does not report Ar concentrations. Furthermore, we refer also to 3He/4He and 40Ar/36Ar data of Vulcano fumaroles published by Tedesco et al. (1995); Magro and Pennisi (1991); Tedesco (1995); Italiano and Nuccio (1997); Tedesco and Nagao (1999), and Tedesco and Scarsi (1999).

A mixing trend for gases of Vulcano does not define the origin and the absolute concentrations of N2, Ar, and He in the endmembers (Fig. 1). The two extreme positions of the data points are characterized by different N2/Ar and N2/He ratios. From Fig. 1, these ratios could be estimated as following: (1) the N2-enriched endmember apparently should have N2/Ar ∼1,000 and N2/He ∼3,000; (2) the N2-depleted endmember may be characterized by N2/Ar ∼400 and N2/He ∼500. It is generally accepted for Vulcano fumarolic gas that the magmatic endmember is characterized by a high CO2 and a low H2O content. This endmember also has the highest N2 and He absolute concentrations (Fig. 2).

N2 and N2/He are plotted in Fig. 4 vs He. Data of Chiodini et al. (1995) before 1992, Wahrenberger (1997) for 1994–1996 and Paonita et al. (2002) for 1996–1998 are used. The high He, high N2 component has N2/He ratio <1,000. Both linear and hyperbolic trends in Fig. 4 correspond to a situation when variable batches of a high He fluid with N2/He ≈ 500 are mixed with a low He fluid with relatively low N2 but high N2/He ratio (>2,000). However, the lowest He component is characterized by N2 close to 0.02 mol% (200 ppm), much higher than the ASW value of 0.0015% (15 ppm). Such a high concentration of nitrogen in the ASW-derived vapor may occur as the result of partial boiling of meteoric water. Some 3–4 wt.% of steam in the two-phase meteoric steam–water mixture is necessary to produce a batch of steam with ∼400 ppm of N2 (X N2,vapor ≈ X N2,water/Y, where Y is the weight fraction of steam). This amount of N 2 should be accompanied by ∼7–8 ppm of Ar. However, as can be seen in Fig. 5, the low N2 component with ∼500 ppm (data by Wahrenberger 1997) has low Ar (∼0.5 ppm), close to its concentration in ASW of ∼0.3 ppm, hence N2/Ar ∼1,000. This indicates that most N2 in the low He and low N2 endmember has a non-atmospheric origin. Thus, the low N2 (low CO2, high H2O) component which is attributed to the hydrothermal endmember is characterized by high N2/Ar (∼1,000, Fig. 5) and high N2/He (>2,000, Fig. 4) ratios. Apparently, this fluid is associated with a long-living hydrothermal system above an old intrusive body (beneath the Caldera del Piano?). The high He, high N2, and high Ar component is characterized by significantly lower N2/He (∼500) and N2/Ar (∼300; Figs. 4 and 5). The output rate of this component, which is attributed to direct magma degassing, is variable. The fluid may originate from a younger magmatic body beneath the Vulcano crater.

Fig. 4
figure 4

N2 vs. He and N2/He vs. He for Vulcano; data from Wahrenberger (1997), Chiodini et al. (1995) and Paonita et al. (2002). Note that the “zero-He” extrapolation apparently gives non-zero N2

Full size image
Fig. 5
figure 5

N2 vs Ar and N2/Ar vs Ar (data from Wahrenberger 1997). The low Ar gases have Ar ∼0.3 ppm which is close to its concentration in ASW, but high N2 and N2/Ar indicating a high fraction of non-atmospheric nitrogen

Full size image

Air contamination

All estimates above for ratios of N2/Ar and N2/He (and hence, He/Ar) have large uncertainties, more than 50%. These uncertainties include natural fluctuations such as fractionation during magma degassing (Nuccio et al. 1999; Nuccio and Paonita 2001), fractionation during boiling of the deep brine, irregular contributions of meteoric water, and air contamination during sampling and analysis. The air contamination can significantly alter the natural relationships between N2, Ar, and He because of the relatively large Ar excess in air compared with the parent gas of Vulcano. Figure 6a, b shows a series of modeled mixing lines of two endmembers (M, magmatic; H, hydrothermal) with characteristics approximately corresponding to the Vulcano gases with maximum and minimum N2 contents. It is assumed that the M component has 1,400 ppm of N2, 3 ppm of He and 3 ppm of Ar, whereas the H component contains 400 ppm of N2, 0.2 ppm of Ar, and 0.1 ppm of He. Each combination of these endmembers is mixed with a random but small amount of air (as log-normal random numbers with the mean of 0.1 mol%). The resulting compositions are plotted together with the ideal mixing lines. Qualitatively, the modeled and observed plots are similar (Fig. 6a). Moreover, it can be seen from the modeled ternary plot (Fig. 6b) that all modeled points are shifted to the right relative to the ideal mixing line, and the apparent extrapolation to the N2–Ar axis leads to the incorrect conclusion regarding N2/Ar of the magmatic component.

Fig. 6
figure 6

a Modeled compositions of mixtures of H and M gases (see text) disturbed by a random injection of air (log-normal with the mean 0.1%). Note that the most scattered are plots with Ar because of the high relative proportion of Ar in air. b Results of the same modeling plotted on the ternary diagram. Note a large scattering of points with a trend to the air composition. The best-fit line gives a N2/Ar ratio much lower than for the endmember enriched in N2 (see text for details)

Full size image

As already mentioned, samples with high He (and high N2 and CO2) have the highest 3He/4He (6R a to 6.2R a , Shinohara and Matsuo 1984; Italiano and Nuccio 1997; Tedesco and Scarsi 1999). The low He hydrothermal component is characterized by a high N2/Ar (∼1,000), and according to Magro and Pennisi (1991), by high values of 40Ar/36Ar which are much higher than the atmospheric value of 296. Values of 40Ar/36Ar ∼1,000 correspond to the highest N2/Ar. Tedesco and Scarsi (1999) also reported 40Ar/36Ar ∼1,000 for the least air contaminated samples (He/Ne ∼4,000). The same authors, however, measured the Ne isotope ratios for the least air-contaminated Vulcano samples with low 20Ne/22Ne and high 21Ne/22Ne of 9.4 and 0.04, respectively, indicating a high proportion of crustal gases involved in the degassing of Vulcano. The N2–Ar–He relationships shown above, along with the published isotopic data, may suggest that the main contribution of crustal gases to Vulcano volcanic gases comes through the hydrothermal component. This suggestion is in agreement with data of Magro (1997) on He and Ar isotopes; the highest 40Ar/36Ar ratio in the Vulcano fluids (∼1,700) corresponds to the lowest 3He/4He (∼4.2R a ). However, according to Magro (1997) this high 40Ar/36Ar, low 3He/4He component is derived from an older, more crustal-contaminated magma in contrast to the high 3He/4He fluid derived from a fresh basaltic magma injected into the old magma chamber. This suggestion could be partially argued because volcanic rocks of Vulcano contain Ar with low 40Ar/36Ar, close to the atmospheric value of ∼300 (Magro 1997; Tedesco and Scarsi 1999), indicating that the main source of radiogenic Ar is not magma and is decoupled from the 3He magmatic source. It should also be noted that Ar in the hydrothermal endmember, although it has a concentration close to the ASW value of ∼0.3 ppm, is characterized by a high 40Ar/36Ar ratio, exceeding 1,000. This indicates that at least 2/3 of this argon has a non-atmospheric, radiogenic origin, and that only ∼1/3 of Ar in the hydrothermal endmember may account for the direct atmospheric or mixed atmospheric + magmatic source.

Compositions of magmatic and magmatic–hydrothermal endmembers in fumarolic gases of Vulcano deduced from N2–Ar–He systematics

The estimated N2, Ar, and He concentrations, 40Ar/36Ar and 3He/4He (R/R a ) values for the M and H endmembers are shown in Table 1. The H2O, CO2, and S + HCl + HF concentrations and the δD, δ13C, for the same components of the Vulcano fluids have been reported earlier (Chiodini et al. 1993, 1995; Leeman et al. 2005).

Table 1 Composition of the two endmember components (M magmatic and H hydrothermal) for the Vulcano fumarolic gases including N2, Ar, and He
Full size table

The magmatic endmember (M) is atypical for high-temperature subduction-type volcanic gases (e.g., Shinohara et al. 1993; Symonds et al. 1994; Taran et al. 1995; Giggenbach 1996). The most unusual are low H2O and high CO2 concentrations and relatively high He concentrations; the values are more typical for the hot spot volcanoes such as Kilauea, Hawaii, or Sierra Negra, Galapagos (Symonds et al. 1994; Giggenbach 1996; Goff and McMurtry 2000).

Taking into account the δ13C value of CO2 (+0.5‰) which is typical for limestone and marine carbonates, one can suggest a significant contribution of CO2 to magmatic fluids from a crustal source. The problem is what crust to choose. One of the probable scenarios is that a very high amount of CO2 from the subducting African continental lithosphere (Ionian slab) and Mediterranean carbonate-rich marine sediments participates in the magma generation process within the mantle wedge beneath this part of the Aeolian arc. This amount of CO2 is much higher than in the case of an intra-oceanic setting or at continental margins with subducting oceanic plates. The total amount of CO2 cannot be completely dissolved in magma at relatively deep levels due to its limited solubility; therefore, it forms a separate fluid phase, extracting other magmatic volatiles. This idea is in agreement with the finding of CO2-rich high-pressure fluid inclusions in quartz xenoliths from Vulcano lavas, corresponding to depths of >20 km, at the crust–mantle interface (Zanon and Nikogosian 2006; Peccerillo et al. 2004). Decarbonation and and/or melting of carbonate-rich lithologies from a subducted African lithosphere proposed by Frezzoti et al. (2009) also suggest the presence of high CO2 fluids or a carbonated partially molten CO2-rich mantle at depths of the magma-generation zone. Low CO2 contents in mafic melt inclusions of some Vulcano rocks is not contradictory to the existence of a separate fluid phase and may be caused by low-pressure crystallization of magmas in the shallow magma chamber beneath La Fossa crater (Gioncada et al. 1998; De Astis et al. 1997).

It is interesting that the H endmember in Vulcano gases is much more similar to a typical arc-type volcanic gas if its water isotopic composition is ignored. According to the above definition, the H endmember consists of the water vapor derived from a boiling-concentrated brine (Nuccio et al. 1999; Nuccio and Paonita 2001; Di Liberto et al. 2002). This brine, in turn, can be considered as a mixture of partially absorbed “magmatic” fluid and significantly evaporated seawater which has evolved due to water–rock interaction. The water–rock interaction is also the main reason for the contribution of a significant fraction of crustal gases (N2, Ar, He) to fumarolic discharges.

Using the volatile balance calculations, it was estimated that the main source of volatiles (except He) in the subduction-type magmas are subducted sediments and altered oceanic basalts (Tatsumi et al. 1986; Plank and Langmuir 1998; Hilton et al. 2002; Jarrard 2003; Wallace 2005; Fischer 2008; Taran 2009). These produce a water-rich fluid with CO2 from subducted carbonates and organic material, sedimentary and seawater sulfur, and Cl predominantly from the subducted seawater. The main feature of the South Italy subduction zone is that the subducting slab is probably not really oceanic, with Mediterranean marine sediments consisting of a large proportion of continental material (Nile sediments and Sahara dust). These sediments are also rich in biogenic material (e.g., Emeis et al. 1996; Bouloubassi et al. 1999). The mechanisms and amounts of this material involved into the magma generation process remains a significant problem, nevertheless, the behavior of H2O and CO2, which are the predominant components of regional magmatic fluids, is drastically different from those of “standard” volcanic arcs. On the other hand, high N2/Ar ratios (>>100) are common for many volcanic arcs, but high He concentrations and He/Ar ratios are a special feature of the South Italy volcanism, as well as relatively low 3He/4He ratios.

Vulcano gas with relatively high 3He/4He of ∼6R a (Shinohara and Matsuo 1984; Italiano and Nuccio 1997; Tedesco et al. 1995) has about 85% of mantle He (7R a is the highest value recorded in Mt. Etna fluids, which may be representative for the mantle beneath the Aeolian arc (Marty et al. 1994; Allard et al. 1997)). This mantle He is transported to the surface from the magma generation zone in the Aeolian mantle wedge together with other volatiles released from the subducting slab and the overlying mantle. The remaining 15% of the He is from the arc crust and the subducted slab. This He, which is mostly radiogenic from the continental crust beneath the volcano and the African lithosphere (Tedesco and Nagao 1999), should be accompanied by some CO2, N2, Ar, Ne, and other gases from the slab and the crustal surrounding magma chambers and conduits. A conclusion based on the analysis of the N2–Ar–He system is that most of the crustal fluids in the Vulcano fumarolic gases are transported by the hydrothermal component.

Comparison with potential source fluids and gases from other subduction-type volcanoes

It is interesting to compare both the magmatic and hydrothermal components of Vulcano gases with the accepted compositions for upper mantle volatiles, some crustal gases and volatile components of oceanic sediments and altered oceanic crust. Oceanic sediments and altered oceanic crust are thought to be the main contributors of almost all species except He to volcanic gases of subduction zones. However, the subducting Ionian slab beneath the Aeolian arc may not be similar to an oceanic plate. According to Emeis et al. (1996), in the Ionian Sea, the Mediterranean sediments near the Calabrian accretionary wedge beneath the Pliocene layer consists of at least 2 km of Messinian evaporites and more than 10 km of a sedimentary complex underlain by Tethyan oceanic crust of Cretaceous age. Bouloubassi et al. (1999) reported the geochemistry of Pliocene (∼3 Ma) organic sediments in cores of Ocean Drilling Project (ODP) site 964 drilled in the Ionian Basin. The organic carbon and nitrogen from this site, and in general from all ODP sites in the Mediterranean, are characterized by lower δ15N (+3‰) compared with the average value of oceanic pelagic sediments (+6‰, Li and Bebout 2005). The composition of other volatiles in Mediterranean sediments is still unknown, as well as the amount of sediments involved in Ionian slab subduction.

Table 2 shows ranges in elemental ratios and isotopic compositions of N2, Ar, and He in the main reservoirs; these are potential contributors of volatiles to volcanic gases of subduction zones. The characteristics of crustal gases are from different sources: Apennines (Collettini et al. 2008), KTB deep well in Germany (Bach et al. 1999; Lippmann et al. 2005), and deep wells of a giant oil field in Kansas, USA (Ballentine and Sherwood-Lollar 2002). The data on He and Ar concentrations and isotopic ratios in altered oceanic crust have been reported only by Staudacher and Allegre (1988). They also reported the abundances of Ar and He in a few samples of pelagic oceanic sediments, isotopic compositions of Ar and He in altered basalts, and Ar isotopes in sediments. The published data on He abundances and isotopes in pelagic sediments (see Ozima and Podosek 2002 for a review) show very high 3He/4He (30–100R a ) due to a constant (though at a very low rate) contribution from cosmic dust. However, Hilton et al. (2002) argued against subducting sediments transporting He into the mantle based on the high mobility of He. Nitrogen in oceanic sediments and altered crust was analyzed by Sadofsky and Bebout (2004), Li and Bebout (2005), and Li et al. (2007).

Table 2 Relative concentrations and isotopic compositions of N2, Ar, and He in different terrestrial reservoirs, potential contributors to the subduction-type magmas and in some subduction-related volcanic gases
Full size table

Distinguishing between various sources of volatiles (mantle wedge, subducted slab, sub-arc crust) is possible by considering the coupled components (i.e., N2, 40Ar, 36Ar, 4He, 3He) together. In this case, the concentration of each component of the parent magmatic gas X i is a weighted sum

$$ {X_i} = \sum {{f_j}{X_{{ij}}},} $$
(1)

where f j is the fraction of fluid from each source (endmember) and X ij is the concentration of each species normalized by the sum N2 + 40Ar + 36Ar + 4He + 3He. If there are four reservoirs (Mt, mantle; S, slab; C, crust; and A, atmosphere (ASW)), then we have five equations for four variables. Taking into account the mass balance equation \( {f_M} + {f_S} + {f_C} + {f_A} = 1 \), we can use three equations for concentrations and one for the mass balance. This approach is considered in detail by Albarede (1995). If all parameters are chosen correctly, the solution (f j ) of the system with the observed or presumed composition of the parent (or any) gas X i is a set of f j values with 0 ≤ f j  ≤ 1. This solution should be the same for any set of three equations for concentrations, for example, (40Ar, 4He, N2), (3He, N2, 4He), (N2, 3He, 40Ar), etc. In practice, all parameters (concentrations) of endmembers are mean values with large standard deviations. The observed or presumed gas compositions at the surface also have certain errors or ranges of values. Therefore, this “ideal” approach sometime does not work, giving either different reasonable solutions for different sets of equations, or solutions that do not make sense (f j negative and/or >1). Therefore, such a formulation may be used for optimization of parameters (concentrations) of endmembers if we know exactly that any observed composition is a mixture of N-1 components from N reservoirs. For the Vulcano gas endmembers “M” and “H”, the set (N2 + 40Ar + 36Ar + 3He + 4He) gives consistent values of f i (Table 3) independent of the chosen set of equations in the system (1).

Table 3 Fractional contributions of MORB (M), slab (S), continental crust (C) and atmosphere (A) to magmatic (M) and hydrothermal (H) components of Vulcano fumarolic gases
Full size table

Two plots with mixing lines between the main potential reservoirs of N2, Ar, and He are shown in Figs. 7 and 8. The ratios N2/Ar, N2/He, and 3He/4He are plotted in Fig. 7 vs. the fraction of the mantle component. Values of 6.2R a and 4.2R a are used for the M (magmatic) and H (hydrothermal) components, respectively. Table 3 shows the calculated fractions Mt, S, C, and A in M and H endmembers for different sets of components. It can be seen from Table 3 that the major contribution to both endmembers comes from the “slab” source. It should be understood that this source, as well as the “crust” source, is not well defined for the Aeolian arc (see above). The crustal CO2-rich gases with low 3He/4He collected by Collettini et al. (2008) from Central Italy, are much more similar in their N2 + Ar + He component to the “slab” component derived from data for oceanic pelagic sediments and altered oceanic crust than to metamorphic, N2-rich gases from the KTB well or oil–gas fields. The least air-contaminated samples from Collettini et al. (2008) have very high N2/Ar and high N2/He, similar to the “slab” gases (Table 2). In other words, the S + C source for Vulcano (and maybe for other Italian volcanoes such as Campi Flegrei or Vesuvius) may be considered as a generalized “crustal” source without distinguishing the continental crust beneath the volcano and the subducting African lithosphere.

Fig. 7
figure 7

N2–Ar–He characteristics of endmembers (h hydrothermal, m magmatic) as a function of the fraction of the mantle contribution to N2 + Ar + He in gases of Vulcano. The mantle fraction is calculated using Eq. 1. Mixing lines between the mantle endmember (M) and other reservoirs are also shown

Full size image
Fig. 8
figure 8

a N2/He vs N2/Ar for fumarolic gases of Vulcano. Data of Caliro et al. (2007) for La Solfatara and Collettini et al. (2008) for cold CO2-rich gases of Central Italy are also shown for comparison. Points for main reservoirs and mixing lines are calculated using data of Table 2. CC continental crust endmember is assumed to be close to the KTB well gases (see Table 2 and text for details). Except for the atmosphere each endmember is characterized by strong variations in compositions, and the plotted values for endmembers are mean values with large dispersions. b Variations of 40Ar/36Ar as a function of the N2/Ar ratio in Vulcano gases (modified from Magro and Pennisi 1991). The proportion of radiogenic Ar increases with increasing non-atmospheric N2. This indicates that the main source for both non-atmospheric N2 and radiogenic Ar is the continental crust

Full size image

Figure 7 shows that only ∼5% of the mantle N2 + Ar + He is present in the “m” component (“m” and “h” instead of capital “M” and “H” are used in Figs. 7 and 8 to avoid confusion with the mantle source M) yet this mixture contains almost 5% of the mantle N2, more than 80% of the total mantle He and more than 95% of the total 3He. The “hydrothermal” (H) component is characterized by only ∼1% of the mantle N2 + Ar + He mixture and ∼95% of the S + C fraction against ∼85% of the S + C for the “magmatic” (M) component. Both M and H components contain a significant fraction of atmospheric gases (3.5–9.6%).

N2/He and 40Ar/36Ar vs N2/Ar for fumarolic gases of Vulcano are shown in Fig. 8a and b together with the compositions of M and H endmembers (“m” and “h”). The points for main reservoirs and mixing lines are calculated using data of Table 2; the continental crust (CC) area corresponds to the KTB well. On the N2/He vs N2/Ar diagram (Fig. 8a), points for Vulcano and La Solfatara (shown for comparison) form compact clusters close to the mixing line between the mantle and C + S endmembers. Note that the CC (crustal) composition lies close to the M–slab mixing line. Central Italy CO2-rich gases (Collettini et al. 2008) with the highest N2/Ar ratios also plot close to the M–CC–slab mixing line, supporting the idea that some of these gases (with maximum N2/Ar ratios and high N2/He) may represent crustal gases of Central and South Italy.

The data of Magro and Pennisi (1991) plotted in the 40Ar/36Ar vs N2/Ar diagram (Fig. 8b) demonstrate a mixing relationship from a low 40Ar/36Ar, low N2/Ar air-contaminated component, through a low 40Ar/36Ar, moderate N2/Ar magmatic endmember, to a high 40Ar/36Ar, high N2/Ar crustal (hydrothermal) endmember.

Concluding remarks

This study has focused on N2–Ar–He systematics of gases from Vulcano Island. It is an attempt to use the existing database on the geochemistry of Vulcano fluids to consider how the behavior of “inert” components can be used as indicators of primary sources of volcanic gases, and also the mechanism of volcano degassing. This study indicates that the following topics deserve more detailed attention in the near future:

  1. 1.

    Although the He concentrations and isotopic composition have been measured in a number of papers, a statistically reasonable dataset with simultaneous determinations of concentrations of N2, Ar, He, and 3He/4He still does not exist.

  2. 2.

    There remains a wide gap in measured Ar isotopes together with the full set of other components. After the very promising results by Tedesco and Scarsi (1999), more data are also required on Ne isotopes, also accompanied by a full set of the measured components.

  3. 3.

    It is also very important to correlate data on H, C, He, Ar, C, and N isotopes with their abundances and elemental ratios.

  4. 4.

    There are no constraints for the south Italy subduction zones regarding the input parameters. Subduction rate, ages, volatile compositions, and amount of subducting sediments are unknown; this is also the case for the upper, volatile-bearing part of the sub-sediment Ionian plate.