Laboratory Testing of Volcanic Gas Sampling Techniques

A series of laboratory experiments were performed designed to calibrate several commonly used methods for field measurement of volcanic gas composition. H₂, CO₂, SO₂ and CHCl₂F gases were mixed through carefully calibrated rotameters to form mixtures representative of the types of volcanic compositions encountered at Kilauea and Showa-Shinzan. Gas mixtures were passed through a horizontal furnace at 700^oC to break down CHCl₂F and form an equilibrium high-temperature mixture. With the exception of Giggenbach bottle samples, all gas sampling was performed adjacent to the furnace exit in order to roughly simulate the air-contaminated samples encountered in Nature. Giggenbach bottle samples were taken from just beyond the hot-spot 10cm down the furnace tube to minimize atmospheric contamination. Alkali-trap measurements were performed by passing gases over or bubbling gases through 6N KOH, NaOH or LiOH solution for 10 minutes. Results were highly variable with errors in measured S/Cl varying from +1600% to -19%. In general reduced Kilauea compositions showed smaller errors than the more oxidized Showa-Shinzan compositions. Results were not resolvably different in experiments where gas was bubbled through the alkaline solution. In a second set of experiments, 25mm circles of Whatman 42 filter paper were impregnated with NaHCO₃or KHCO₃ alkaline solutions stabilized with glycerol. Some filters also included Alizarin (5.6-7.2) and neutral red (6.8-8.0) Ph indicator to provide a visual monitor of gas absorption. Filters were mounted in individual holders and used in stacks of 3. Durations were adjusted to maximize reaction in the first filter in the stack and minimize reaction in the final filter. Errors in filter pack measurements were smaller and more systematic than the alkali trap measurements. S/Cl was overestimated in oxidized gas mixtures and underestimated in reduced mixtures. Alkali-trap methods allow extended unattended monitoring of volcanic gasses, but our results suggest that they are poor recorders of gas composition. Filter pack methods are somewhat better, but are more difficult to interpret than previously recognized. We suggest several refinements to the filter-pack technique that can improve accuracy. Giggenbach bottles remain the best method for volcanic gas sampling, despite the inherent difficulty and danger of obtaining samples in active volcanic environments. Relative merits of different alkali solutions and indicators are discussed.