Volcanic ash ice nucleation activity is variably reduced by aging in water and sulfuric acid: the effects of leaching, dissolution, and precipitation

Volcanic ash nucleates ice when immersed in supercooled water droplets, giving it the potential to influence weather and climate from local to global scales. This ice nucleation activity (INA) is likely derived from a subset of the crystalline mineral phases in the ash. The INA of other mineral-based dusts can change when exposed to various gaseous and aqueous chemical species, many of which also interact with volcanic ash in the eruption plume and atmosphere. However, the effects of aqueous chemical aging on the INA of volcanic ash have not been explored. We show that the INA of two mineralogically distinct ash samples from Fuego and Astroni volcanoes is variably reduced following immersion in water or aqueous sulfuric acid for minutes to days. Aging in water decreases the INA of both ash samples by up to two orders of magnitude, possibly due to a reduction in surface crystallinity and cation availability accompanying leaching. Aging in sulfuric acid leads to minimal loss of INA for Fuego ash, which is proposed to reflect a quasi-equilibrium between leaching that removes ice-active sites and dissolution that reveals or creates new sites on the pyroxene phases present. Conversely, exposure to sulfuric acid reduces the INA of Astroni ash by one to two orders of magnitude, potentially through selective dissolution of ice-active sites associated with surface microtextures on some K-feldspar phases. Analysis of dissolved element concentrations in the aged ash leachates shows supersaturation of certain mineral species which could have precipitated and altered the INA of the ash. These results highlight the key role that leaching, dissolution, and precipitation likely play in the aqueous aging of volcanic ash with respect to its INA. Finally, we discuss the implications for understanding the nature and reactivity of ice-active sites on volcanic ash and its role in influencing cloud properties in the atmosphere.

. Bulk chemical composition as oxide wt.%, normalized to 100 wt.% (excluding loss on ignition), and specific surface area (SSABET ) of the ash samples studied. Background subtraction and statistical treatment of ice nucleation data The sample data were first binned by temperature with a 0.5 °C bin width allowing the number of droplets that freeze in each bin to be resampled 100 times from a Poisson distribution. Each resampling across all temperature bins is then treated as a simulated experiment with its own freezing spectrum. The 2.5 th and 97.5 th quantiles of these simulated spectra experiments are then used to approximate the 95% confidence interval around each binned datapoint. Pure water background freezing spectra are subtracted from each sample through the differential spectrum with 2-sided propagation of error to account for both sets of confidence intervals. The differential spectra are then numerically re-integrated with propagation of error to produce the background subtracted INA spectra. Finally, data from experiments examining the same sample at different suspension concentrations were combined by averaging the differential ns spectra over each temperature bin with propagation of error. The cumulative ns spectra were then recalculated by numerical integration as above.

Regression model between dissolved element signatures and descriptors of ice nucleation activity
An epsilon-nonlinear support vector regression model with a radial basis function kernel, a degree of three, a regularization parameter of six and an epsilon of 0.   Figure S1: Ice nucleation active site density normalized to surface area (ns) versus temperature spectra of FUE ash non-aged or aged for different durations in a) H2O or b) pH 1.75 H2SO4, and AST ash non-aged or aged for different durations in c) H2O or d) pH 1.75 H2SO4. The 95% confidence intervals were approximated as the 2.5 th and 97.5 th quantiles using Monte Carlo simulations based on a Poisson distribution of droplet freezing events. Each spectrum is background-subtracted and is a combination of three experiments on 1, 0.2, and 0.04 wt.% suspensions with error and activity propagated through the differential ice nucleation site density spectra.
b) Acid-aged FUE ash c) Water-aged AST ash d) Acid-aged AST ash a) Water-aged FUE ash Figure S2: Correlation coefficient heatmap showing elementwise relationships between normalized dissolved element signatures and coefficients of the Chebyshev polynomial representing the quotient of FUE ash aged in H2O (water-aged) with non-aged FUE ash at each time point. Red indicates a correlation, while blue indicates anticorrelation. Blank spaces appear where these elements were not detected in the aging solution at any timestep. Acid-aged FUE ash Figure S4: Correlation coefficient heatmap showing elementwise relationships between normalized dissolved element signatures and coefficients of the Chebyshev polynomial representing the quotient of AST ash aged in H2O (water-aged) with non-aged AST ash at each time point. Red indicates a correlation, while blue indicates anticorrelation. Blank spaces appear where these elements were not detected in the aging solution at any timestep.

Water-aged FUE ash
Water-aged AST ash Figure S5: Correlation coefficient heatmap showing elementwise relationships between normalized dissolved element signatures and coefficients of the Chebyshev polynomial representing the quotient of AST ash aged in pH 1.75 H2SO4 (acid-aged) with non-aged AST ash at each time point. Red indicates a correlation, while blue indicates anticorrelation.
Acid-aged AST ash