Real-time measurement of volcanic H₂S and SO₂ concentrations by UV spectroscopy

1. Introduction

[2] The sulphur chemistry of volcanic gases is complex, not least because of the variety of speciation and oxidation state (from −2 to +6): (e.g., SO₂, H₂S, SO₃, OCS, CS₂, S, H₂SO₄). Of these, the first two are generally the most abundant, and can be present in comparable concentrations. Equilibration between SO₂ and H₂S acts as an important gas buffer [Giggenbach, 1996] according to:

with H₂S favoured at higher pressures and lower temperatures [Symonds et al., 1994]. Especially when concentrations of a conservative species such as CO₂ are available, H₂S and SO₂ measurements help to probe mixing of magmatic and hydrothermal fluids, and related physical and chemical processes [Giggenbach, 1996]. Monitoring of both H₂S and SO₂ concentrations is also important in order to assess the risks of exposure of scientists and tourists to these gases when visiting degassing volcanoes.

[3] Volcanic degassing of sulphur compounds is also known to play important roles in atmospheric chemistry, radiation and dynamics [Rose et al., 2000]. A widely quoted figure for the annual source strength of sulphur to the atmosphere from volcanoes is the estimate of 10.4 Tg obtained by Andres and Kasgnoc [1998]. More recently, Halmer et al. [2002] have estimated the global volcanic SO₂ emission to the atmosphere as 15–21 Tg yr⁻¹ for the period 1972–2000, and suggested that the total volcanic sulphur emission, accounting for H₂S could be much higher (9–46 Tg of S) reflecting a very large uncertainty in the H₂S emission (1.4–35 Tg of S). One reason for this is that while open-path spectroscopic determinations of SO₂ columns are comparatively easy due to the position of electronic and rotational-vibrational absorption lines in the ultraviolet (UV) and infrared (IR), H₂S is much more challenging to measure in the IR because its absorption spectrum lies in a region of strong water vapour absorption, and in the UV because its electronic spectrum is at shorter wavelengths at which molecular scattering is much stronger. In the published work on Fourier transform IR spectroscopy of volcanic gases in the field, there are no reports of detection or quantification of H₂S [McGonigle and Oppenheimer, 2003]. The only prior optical detection of volcanic H₂S was achieved (with CO₂ and H₂O) by direct fumarole sampling at Solfatara using an evanescent fibre laser sensor [Willer et al., 2002].

[4] Traditional direct sampling techniques (collection in alkaline solution) followed by laboratory analysis [Giggenbach, 1975] also pose difficulties in resolving sulphur speciation. Typically, total sulphur is measured, and relative proportions of H₂S and SO₂ estimated from the average oxidation state of the sulphur, though Montegrossi et al. [2001] have demonstrated an analytical approach to this problem. Recently an inexpensive electrochemical sensor has been used for airborne in-plume H₂S measurements, and such sensors are also available for SO₂ monitoring [McGee et al., 2001]. However, unlike the approach we present here, these devices are not capable of remote sensing, and in a high temperature fumarole sampling configuration they require substantial cooling of the gases, prior to measurement.

[5] The potential value of H₂S and H₂S/SO₂ surveillance led us to adapt a small UV spectrometer for closed path measurements of volcanic gases. The same spectrometer has recently been applied to measurement of SO₂ fluxes by differential optical absorption spectroscopy [McGonigle et al., 2002; Galle et al., 2003] using skylight as the source radiation but to get from the SO₂ spectrum at 300 nm to the H₂S absorption at 200 nm requires an artificial source (in this case a deuterium lamp). We report here on initial field trials of the instrument at Solfatara and Vulcano (Italy) carried out in November 2002. Compared with the fibre-sensor and FTIR spectrometers and associated retrievals, this approach is much simpler and cheaper, with far greater potential for autonomous, sustained deployment.

2. Methodology

[6] The system was implemented in three configurations, using a deuterium bulb light source (Hamamatsu) with UV output in the 200–400 nm spectral range. The bulb was powered by a constant current supply that ran from a 24 V input (two 12V lead-acid batteries). The beam was collimated using a 50 mm focal length quartz lens and fibre-coupled into a S2000 spectrometer, manufactured by Ocean Optics, with ∼0.4 nm resolution over the 200–350 nm measurement range. A circular-to-linear converter, 4 × 200 μm, solarisation resistant fibre bundle was used, with the linear fibre end coupled to the spectrometer. A USB cable connected the spectrometer to a laptop computer, in order to power the S2000 and to achieve data transfer. Software written in LabVIEW™ was used to acquire and evaluate spectra (10–500 ms integration times, depending on light levels). Each measured spectrum was divided by a “white” spectrum taken under otherwise equivalent conditions, but with no absorption from H₂S or SO₂. The logarithm of the resulting spectrum was taken to produce an absorbance spectrum, to which a linear least squares fit was applied using laboratory H₂S and SO₂ absorbance spectra of known concentrations (see Figure 1), to retrieve gas concentrations in real time, every ∼1s. It was not necessary to take temperature into account in the fitting routine, as the absorption spectra of H₂S and SO₂ are independent of temperature in the 205–250 nm fitting window, over the 293–600 K range of the measured gases. The fit was constrained to prevent negative concentrations arising from fitting to noise, in which case a concentration of zero was outputted. The evaluation routine could accommodate extinction of up to 99% of the bulb's light in transit through optically thick plumes, without introducing additional errors, by also fitting for broadband losses.

[7] In the first configuration, open paths of up to 40 m separated the lamp and telescope. This provides the safety advantage of working remotely from fumaroles, as well as an increase in sensitivity afforded by the longer pathlength. At the spectrometer end a Newtonian telescope, with a primary aluminium coated mirror of diameter = 150 mm and focal length = 700 mm, coupled the bulb's light into the fibre bundle. The bulb and spectrometer and telescope assembly were mounted on tripods, providing pan, tip and tilt adjustment for alignment.

[8] In the second arrangement the bulb and optical fibre were mounted 550 mm apart on a rigid optical rail. The intensity of the collimated deuterium beam was such that that no focusing lens was required at the optical fibre end, reducing the sensitivity of the system to misalignment and scattering effects. In the final arrangement a titanium tube was placed directly into the fumarole and connected to a 10 mm thick cell with quartz windows, via a quartz dewar. The cell, spectrometer and bulb were mounted in a compact 400 × 160 × 120 cm housing, weighing 2.2 kg. Gas flow through the cell was achieved by attaching a pump (0.6 L/min) to the cell outlet. The bulb and circular fibre tip were mounted at opposite sides of the cell, without collimating or focusing lenses. Realignment of the optics was not required in the field with the latter two systems. The power consumption of the device was 50 W.

[9] The detection limits for the long (10 m) path and the shorter path systems were found to be 2 ppmm, based on the fluctuations in retreived concentrations observed when these systems were left for ∼15 min continuously monitoring clean air samples. By placing calibrated gas mixtures into the instrument's optical path and retrieving H₂S and SO₂ concentrations, the measurement accuracy was determined to be better than 5%. In the same way the retrieved H₂S to SO₂ cross sensitivity was found to be less than 1 part in 20, and that of SO₂ to H₂S less than 1 part in 100. Each H₂S/SO₂ ratio error quoted below was one standard deviation about the mean ratio from the set of spectra taken during that measurement.

3. Results and Discussion

[10] Open-path measurements made ∼20 m downwind of the Bocca Grande fumarole at Solfatara, with a pathlength of 10 m, indicated H₂S concentrations between 4 and 37 ppmm, the variations reflecting variable dispersal of the gases during the 10 min of observation shown in Figure 2. The mean concentration was 13 ppmm (1.3 ppmv). The sustained exposure limit for H₂S in the workplace according to U.S. Occupational Safety and Health Administration standards is 10 ppmv (www.osha.gov). SO₂ was below the detection limit (2 ppmm), which is as expected because the geothermal system at Solfatara scrubs this gas [Chiodini et al., 2001]. Due to high winds it was not possible to make long path measurements at Vulcano during the campaign.

[11] The 55 cm path length configuration was placed ∼20 cm from the ground directly over Bocca Grande fumarole, such that the fumarole gas passed up freely into the optical path. The maximum measured H₂S concentration in this case was 220 ppmm. This configuration was also placed ∼50 cm above fumaroles on the crater rim at Vulcano. Figure 3 shows a plot of H₂S against SO₂ concentration for the ‘F4’ fumarole, from which gas column amounts of up to 55 ppmm H₂S and 76 ppmm SO₂ (H₂S/SO₂ ratio of 0.64 ± 0.1) were obtained, demonstrating the ability of this instrument to measure simultaneously both sulphur species. Absorbance spectra for this measurement are shown in Figure 1. The correlation between H₂S and SO₂ suggests that the measurements were not strongly affected by mixing of ‘F4’ emissions with those from neighbouring fumaroles.

[12] The disproportionation reaction:

results in a rapid decrease in H₂S/SO₂ ratio with distance from the fumaroles at Vulcano. In order to overcome this effect, extractive measurements were made at two other fumaroles, designated ‘F11’ (T = 326°C) and ‘F0’ (T = 271°C). Here we measured H₂S/SO₂ molar ratios of 2 ± 0.7 and 2.4 ± 0.7, respectively. The greater proportion of H₂S observed in the F0 fumarole is consistent with its lower temperature (Equation 1). Due to condensation of water vapour within the cell, stable ratio measurements were only possible during the first ∼10 s of pumping gas through the cell. These ratios are within the range of previous H₂S/SO₂ measurements from these fumaroles (i.e., 1–2.4 [Giggenbach et al., 2001], 0.7–6.3 [Chiodini et al., 1995] from F11 and 1.6–3.8 [Giggenbach et al., 2001] for F0).

[13] The data reported here demonstrate the capability of the three configurations of this instrument and its robustness in volcanic environments. The long path configuration where applicable, unlike the shorter path arrangements, provides the possibility for integrated measurements across a number of fumaroles, the safety advantage of remote operation and increased sensitivity. However, it is bulky and requires careful alignment. By using the much more compact extractive sampling configuration, the fumarolic gas can be sampled very close to source, limiting time for potential disproportionation of H₂S to sulphur. However, by virtue of being the most invasive configuration, this approach suffers the greatest likelihood of instrumental corrosion. Due to the low power requirement and simplicity of this configuration, it could be readily adapted to sustained deployments, using an embedded processor, solar power and telemetry in a modest cost package (∼$6,000). The cell could be heated to avoid water vapour condensation, and filled with volcanic gas for a period of a few seconds, during each measurement, then flushed with a N₂ purge to protect the quartz cell from corrosion between measurements.

4. Concluding Remarks

[14] We have demonstrated a simple optical technique for real-time, simultaneous measurement of H₂S and SO₂ concentrations in or close to fumaroles. The approach has applications for monitoring airborne concentrations of H₂S and SO₂ (minimum detection limit of 2 ppmm), and for measuring the ratio of these two species at a time-step as short as 1 second, day or night (given a power supply of 50 W).

[15] The technique shows potential therefore for air quality monitoring at volcanoes, geochemical surveillance programmes, and could be useful for estimating volcanic H₂S fluxes, by combining the estimated H₂S/SO₂ ratio with SO₂ flux determined by traversing beneath or scanning the plume with the same spectrometer configured for open sky observations [McGonigle et al., 2002; Galle et al., 2003]. The high time resolution could also reveal new insights into thermodynamical effects on H₂S/SO₂ ratios that have been hinted at by Montegrossi et al. [2001] and Saito et al. [2002].