UV induced mass-independent sulfur isotope fractionation in stratospheric volcanic sulfate



[1] Sulfuric acid aerosols produced in the stratosphere following massive volcanic eruptions possess a mass-independent sulfur isotopic signature, acquired when volcanic SO2 experiences UV photooxidation. The volcanic data are consistent with laboratory SO2 photooxidation experiments using UV light at 248 nm (maximum absorption of ozone), whereas sulfur isotopic anomalies previously observed in Archean samples are consistent with photodissociation at 190–220 nm. A mechanism of SO2 photooxidation, occurring in the early stage of a stratospheric volcanic plume, in the range of 220–320 nm (weak band absorption of SO2), is also proposed. Since mass-independent sulfur isotope anomalies in stratospheric volcanic sulfate appear to depend on the exposure of SO2 to UV radiation, their measurements might therefore offer the possibility to determine the degree of UV penetration in the ozone-absorption window for the present and past atmospheres. They can also be used to determine the stratospheric or tropospheric nature of volcanic eruptions preserved in glaciological records, offering the possibility to reassess the climatic impact of past volcanic eruptions.

1. Introduction

[2] The discovery of mass-independently fractionated sulfur isotopic compositions (sulfur MIF: Δ33S = δ33S − 1000 x [(1 + δ34S/1000)0.515 − 1] ≠ 0 per mil (‰) and Δ36S = δ36S − 1000 x [(1 + δ34S/1000)1.90 − 1] ≠ 0‰) in Archean crustal and sedimentary rocks has significantly advanced our understanding of the Earth's sulfur cycle [Farquhar et al., 2000; 2002]. The simultaneous oxygenation of the Archean atmosphere and the disappearance of anomalous sulfur isotopic compositions lead to the hypothesis that atmospheric photochemistry involving sulfur dioxide in a primitive atmosphere with reduced oxygen and ozone levels was the cause of the observed sulfur isotopic anomaly [Farquhar et al., 2000]. Isotopic results of laboratory experiments where SO2 was subjected to UV radiation below 220 nm support this postulate [Farquhar et al., 2001]. As a result, mass-independent sulfur isotopic composition is a tracer of the UV transparency of the Earth's atmosphere. In today's atmosphere, only SO2 present in the stratosphere at altitudes above 20 km experiences UV photolysis in the 190–210 nm spectral region [Finlayson-Pitts and Pitts, 2000]. Below 20 km, SO2 photolysis does not occur due to the lack of high energy (λ < 300 nm) UV radiation, and consequently, SO2 oxidation proceeds via other thermochemical mechanisms that are mass dependent for sulfur isotopes.

[3] Sulfate deposition on the Antarctic continent is predominantly regulated by the tropospheric oxidation of biogenic marine emissions of reduced sulfur species (dimethylsulfide or DMS) and sea-salt sulfate input, the latter representing generally less than 10% of the total sulfate [Minikin et al., 1998; Wagenbach et al., 1998]. Sporadic, large volcanic eruptions in the mid and low latitudes emit copious amounts of SO2. This sulfur dioxide is oxidized in, and transported through the stratosphere to the polar regions where it may dominate the atmospheric sulfur budget for short time periods with significant climatic consequences [Cole-Dai et al., 2000; Robock, 2000]. Volcanic sulfate is deposited onto the polar ice sheet and can be preserved in snow and ice layers. Large eruptions are generally marked by unusually high sulfuric acid levels in ice [Hammer et al., 1981]. Long and detailed records of past volcanism can be reconstructed from polar ice core sulfate measurements [Cole-Dai et al., 2000; Zielinski et al., 1994].

[4] Here we report nonzero Δ33S and Δ36S values for volcanic sulfates preserved in South Pole glaciological records. Implications of our results are discussed with respect to the characteristics of paleovolcanism recorded in ice cores, the sulfur cycle and the potential use of mass-independent sulfur isotopic compositions to trace the extent and mechanism of modification of UV transmission (i.e., ozone shielding) in the atmosphere.

2. Sampling and Method

[5] We extracted sulfate from Antarctic snow blocks, ice cores and aerosol filters. Chemical analysis of soluble species revealed that the snow blocks and two ice cores contain the 1991 Pinatubo, the 1991 Cerro Hudson volcanic eruptions and the well-dated 1259 A.D. eruption of an unknown volcano (1259 UE), respectively. Background tropospheric sulfates surrounding these volcanic events were also isolated from snow and ice layers immediately above and below the volcanic events, and were used to remove the background contribution of the volcanic events. Daily aerosol filters, covering the full 1999 year, were combined by season in order to provide sufficient material for isotopic analysis. After extraction of the sulfate from the solution, sulfur isotopic ratios were measured following well-established procedure using SF6 as the working gas (see supplemental materials for details). Based on replica standard and sample measurements, δ33S, δ34S, δ36S, Δ33S and Δ36S uncertainties are estimated to be ±0.05‰, ±0.03‰, ±0.1‰, ±0.05‰, and ±0.5‰ (1 σ), respectively.

3. Results and Discussion

[6] δ34S, δ33S and δ36S values of aerosol and background (i.e., non-volcanic) sulfate are all within small isotopic ranges (Auxiliary Table 1), +13.6 to +18.01‰, +7.01 to +9.22‰, and +26.77 to +33.78‰, respectively, indicating similar isotopic sources. This is consistent with a single source-tropospheric oxidation of marine biogenic sulfur (mainly DMS), which is the predominant source of sulfur over the high Southern latitudes [Arimoto et al., 2001; Minikin et al., 1998]. Sulfur dioxide is believed to be the key intermediate in the tropospheric degradation of DMS [Barnes et al., 1996] and SO2 oxidation to sulfate follows a fast gas-particle conversion process. No UV photolysis is involved, as no high energy UV light is available in the troposphere. Therefore, aerosol and background sulfate are formed through mass dependent processes and no significant sulfur isotope anomalies are expected. This is borne out by the data: all Δ33S and Δ36S are within the 2σ uncertainty range of zero for these samples (Auxiliary Table 1 and Figure 1).

Figure 1.

Plot of Δ33S (A) and Δ36S (B) sulfur isotope anomalies of different sulfate sources. The dashed lines represent the 2σ level of the detection limit. Uncertainty bars include analytical and propagation of error. Volcanic eruptions have been corrected for the background sulfate.

[7] Sulfur isotopes demonstrate a significant range of variations for volcanic sulfate. Sulfates from the three events, Cerro Hudson, Pinatubo and 1259 UE, possess different isotopic values (Auxiliary Table 1). The near-zero Δ33S (−0.09‰) of Cerro Hudson sulfate indicates that the sulfate was formed primarily through a mass dependent process. Observations during the atmospheric distribution and transport of the 1991 Cerro Hudson SO2 clouds indicated that the eruption was a predominantly upper tropospheric/lower stratospheric (11–16 km) event [Cacciani et al., 1993; Schoeberl et al., 1993]; the amount of sulfur emitted (1.5 Tg SO2) was relatively small [Doiron et al., 1991], with a substantial fraction of volcanic debris reaching Antarctica via the middle/upper troposphere (8–12 km) [Deshler et al., 1992; Legrand and Wagenbach, 1999]. Volcanic SO2 in the upper troposphere and lower polar stratosphere is not subjected to photodissociation since little high energy UV light (<220 nm) is available at this altitude, making it similar to the case of tropospheric aerosol and background sulfates. This is consistent with the observed near-zero Δ33S value for Cerro Hudson sulfate. The lack of any substantial fractionation in δ34S value of Cerro Hudson sulfate reinforces the premise of fast gas-particle conversion and transport within the troposphere [Krueger et al., 1995]. In contrast, the Pinatubo and 1259 UE sulfates are characterized by significant sulfur isotope anomalies and fractionations: δ34S of +10.89‰, Δ33S of +0.67‰ and Δ36S of −3.58‰ for Pinatubo and −5.71‰ (average n = 2), −0.50‰ (average n = 2) and +1.7‰ (average n = 2), respectively for 1259 UE (Auxiliary Table 1, Figure 1). These anomalies provide unquestionable, evidence of mass-independent processes for this specific volcanic sulfur debris.

[8] Close comparison of SO2 photolysis experiments and the Archean samples leads to the conclusion that photolysis in the 190 to 220 nm spectral range is responsible for the mass-independent sulfur composition in the Archean atmosphere [Farquhar et al., 2001], with mass-independent compositions of reduced sulfur species (elemental sulfur, pyrite) in opposite direction of the oxidized sulfur species (sulfate) [Farquhar et al., 2000, 2001]. A recent photochemical model [Pavlov and Kasting, 2002] demonstrates that sulfur MIF initiated by the photolysis of SO2 below 220 nm in an anoxic atmosphere can be distributed among oxidized and reduced sulfur species, fulfilling the isotopic mass balance requirement. This is not possible today as reduced sulfur species are oxidized before they arrive at the Earth's surface, and could explain the apparent rarity of sulfur MIF in modern geological deposits. Figure 2 illustrates the linear relationships between Δ33S and δ34S (Figure 2a) and between Δ33S and Δ36S (Figure 2b) for the Archean samples and for laboratory photolysis experiments using a KrF laser at 248 nm. Our volcanic sulfur isotope data appear to fit the slope of the laser photolysis experiment in both figures, indicating that sulfur anomalies of volcanic sulfate in the present-day oxygen-rich stratosphere is controlled mainly through UV photochemistry in the 248 nm wavelength region, and not below 220 nm as it is observed for the Archean samples. Photolysis of SO2 below 220 nm forms SO + O but because of the high content of molecular oxygen in today atmosphere, SO is almost instantaneously converted back to SO2. The net effect of SO2 photodissociation for the sulfur is null and therefore this process cannot introduce any sulfur anomalies in an oxygen-rich atmosphere. Our data are in accordance with this observation. The absorption at 248 nm produces only SO2* (excited triplet and singlet states) [Okabe, 1978]. Usually SO2* species are neglected in chemical stratospheric models as they are efficiently quenched by air molecules [Sidebottom et al., 1972]. However, in environments with high SO2 concentration (e.g., in a photocell of pure SO2 as in Farquhar [2001]), SO2* may self react to an appreciable rate with SO2 to form SO3 and SO [Turco et al., 1982]. In today's atmosphere, SO3 will be quickly converted to H2SO4, whereas SO will give back SO2. Using a 1-D photochemical model (see supplemental materials), we find that this photo-oxidation pathway could represent a significant fraction of the sulfate produced during the first few days following the Pinatubo eruption and even dominate the SO2 oxidation for the first month following the 1259 UE eruption. Depending on the kinetic fractionation, which is unknown, this process may be sufficient to produce the observed sulfur isotopic anomalies. This photooxidation process opens a way to generate sulfur anomalies in a high O2 atmosphere when SO2 is abundant (for instance at an early stage of a volcanic plume), and high levels of 220–340 nm radiation are available (e.g., in the stratosphere), two constraints not filled out by the background and Cerro Hudson events. H2SO4 aerosols and gaseous SO2 formed in this way may carry opposing sign anomalies and be temporally and geographically separated by transport [Castleman et al., 1974]. Recently, sulfur MIF has been observed in tropospheric aerosols from locations with multiple sulfate sources [Romero and Thiemens, 2003]. The authors attribute their sulfur MIF to stratospheric input of sulfate into the troposphere, further supported by 35S specific activity measurements [Romero and Thiemens, 2003]. These observations are therefore not in contradiction with our explanation.

Figure 2.

(A) Plot of Δ33S versus δ34S. The plot shows that Δ33S of stratospheric volcanic sulfate is a function of the δ34S fractionation and coincide with the photolysis experiment at 248 nm rather than the Archean reference line. Plot of (B) Δ36S versus Δ33S. The dashed line represents the reference Archean slope obtained by [Farquhar et al., 2000]. The semi continuous line is the mass independently fractionated line obtained from photolysis experiment of SO2 performed at 248 nm (KrF excimer laser) taken from [Farquhar et al., 2001].

[9] TheΔ33S and Δ36S values are of opposite sign for a given stratospheric volcanic event. This characteristic is also observed in laboratory experiments [Farquhar et al., 2001] and Archean samples [Farquhar et al., 2000]. Pinatubo and 1259 UE possess opposing Δ33S (positive for Pinatubo and negative for 1259 UE) and Δ36S values (negative for Pinatubo and positive for 1259 UE). The observation of opposite sign in sulfur anomalies between Pinatubo and 1259 UE is likely the result of their specific dynamics. The Pinatubo eruption was a large, stratospheric event, with an estimated 17 ± 3 Tg of SO2 directly injected into the stratosphere [Krueger et al., 1995] to altitudes between 20 and 27 km [McCormick and Veiga, 1992]. Since the 1259 eruption is estimated to have emitted approximately 20 times more SO2 (320 ± 80 Tg) than Pinatubo, and is clearly a massive global event [Cole-Dai et al., 2000; Langway et al., 1995], it is very likely that the 1259 UE volcanic plume attained much higher altitudes, exposing volcanic SO2 to higher UV levels. Different dynamics and UV exposure of the SO2 plumes will result in different photochemistry which could explain the opposing mass-independent signatures of these two eruptions. The fact that these two volcanic events possess distinct oxygen isotopic compositions [Savarino et al., 2003] and that the conversion rate of SO2 to H2SO4 is highly dependent on plume altitude [Krueger et al., 1995] further reinforces this interpretation.

[10] The 248 nm wavelength corresponds to the maximum O3 absorption in the Hartley band (230–300 nm). Therefore, the presence of O3 in the stratosphere can influence the sulfur anomalies of H2SO4 by determining the penetration of UV radiation needed for SO2 photolysis. The control of UV transmission by O3 in todays's atmosphere represents a major difference with the Archean atmosphere in which the O2 column is believed to be the regulator of the sulfur anomaly. Our result suggests that, in the post-Archean oxygen-rich atmosphere, the mass-independent isotopic compositions of atmosphere-oxidized sulfur may reflect the degree of UV penetration and hence the level of screening ozone, a hypothesis that must be tested in the future with the improvement of the analytical procedure. Nonetheless, the present results indicate that mass independent sulfur isotope fractionation processes do occur in today's atmosphere via UV photochemistry in and above the ozone screening window, thus providing a unique method of using sulfur isotopes to track this photochemistry.

[11] Polar ice cores are used to study the climatic impact of explosive volcanism. Explosive eruptions produce sulfate spikes in ice cores [Hammer et al., 1981], and Greenland and Antarctica ice cores provide volcanic records of thousands of years [Zielinski et al., 1994]. However, sulfate records usually can not distinguish stratospheric from tropospheric eruptions. The former, in which volcanic gases are directly injected into the stratosphere, is more important climatically since stratospheric aerosols have relatively long residence times and are spread globally. The many volcanic signals from frequent tropospheric eruptions in nearby Icelandic and Aleutian arcs makes it difficult to evaluate the climatic impact with Greenland ice core records. A bi-polar comparison technique may be used to identify large, low latitude stratospheric eruptions, for they produce contemporaneous signals in both polar ice sheets [Langway et al., 1995]. One drawback of the technique is that occasional sub-Antarctic tropospheric eruptions can coincide with some in the Northern Hemisphere [Cole-Dai et al., 1997, 2000]. Our results suggest that sulfur isotopic analysis of ice core volcanic sulfate can be an independent method to identify climatically significant eruptions in the investigation of the climate-volcanism connection.


[12] The 2001 South Pole ice core samples were provided by the US National Ice Core Laboratory in Denver, Colorado. This study is supported in part by NSF Grants 9526725 and 008787151 to J. Cole-Dai. Joel Savarino thanks especially the Balzan Foundation and Claude Lorius for their financial help. Michel Legrand, Bruno Jourdain and the Institut Polaire Paul Emile Victor (IPEV) are deeply acknowledged for providing the aerosol filters. The Institut des Sciences de l'Univers and PNCA are acknowledged for its support.