Introduction

The fine pyroclastic materials produced by explosive eruptions (volcanic ash) are mostly made of iron-bearing silicate glass (40–80 wt%, Blundy et al. 2006) and crystalline materials, both having significant impact on ash physical properties and biological activity (Horwell and Baxter 2006; Langmann et al. 2010; Hamme et al. 2010; Ayris and Delmelle 2012a). The positive role of volcanic ash as a source of bioavailable iron in different natural environments is of growing interest within the scientific community (Duggen et al. 2010; Ayris and Delmelle 2012b; Lindenthal et al. 2013). In fact, iron is a micronutrient that plays an important role in regulating phytoplankton growth in the ocean (Martin and Fitzwater 1988; Boyd et al. 2000; Kustka et al. 2003), and large-scale phytoplankton blooms may cause CO2 drawdown, thus establishing a possible connection between Fe-fertilisation by volcanic ash and climate changes (Sarmiento 1993; Watson 1997). As for human health, Horwell et al. (2010) have shown that the Fe-content in respirable volcanic ash has toxic effects due to its capability to generate free radicals, which are highly reactive molecules with unpaired electrons capable to degrade DNA, resulting in genetic mutations that adversely affect the cell cycle and eventually lead to cancer (Kane 1996).

The iron solubility FeS (i.e., the water-soluble mass fraction of Fe) is a key parameter for assessing the potential of volcanic ash to modify the Fe budget in surface oceanic and lacustrine environments (Duggen et al. 2010; Langmann et al. 2010; Hamme et al. 2010; Olgun et al. 2011). Along the path from magmatic source to lacustrine or oceanic sink, factors that rule the soluble (and thus potentially bioavailable) fraction of Fe in ash remain poorly understood, and among them are the the Si:O and the mineral/glass ratios of the ash, mineral chemistry and Fe topochemistry (i.e., the specific reactions driven by Fe oxidation state as well as Fe coordination environment in both mineral structures and glass). In addition, an important control on iron solubility may be exerted by particle size, possibly as a result of the higher surface area-to-volume ratio in finer particles (e.g., Baker and Jickells 2006; Ooki et al. 2009), by Eh–pH conditions, which rule Fe speciation and Fe2+/Fe3+ equilibria in solution, and by later adsorption of dissolved Fe ions on (hydr)oxides surfaces (e.g., Teutsch et al. 2005; Larese-Casanova and Scherer 2007; Schuth and Mansfeldt 2016).

In volcanic ash, iron is contained in Fe-rich mineral crystallites (e.g., olivine, pyroxene, amphibole, ilmenite, spinel, etc.) and/or in the glass shards. Fe-bearing minerals may contain high amounts of Fe (even more than 50 wt%) and volcanic glass usually contains lower amounts of Fe, typically in the range 2–12 wt% (e.g., Galoisy et al. 2001; Dyar et al. 2006). In addition to Fe-bearing minerals, plagioclase can potentially release significant amount of Fe in solution as it may be the most abundant mineral in some volcanic ash and, despite being a nominally Fe-free mineral, it actually may contain minor amounts of Fe (up to 0.48 wt% FeO is common in plagioclase from dacitic to rhyolitic rocks, Lac 2009). In minerals Fe is strongly bonded in the crystalline structure, whereas in the glass it is expected to be less tightly bonded and then more easily remobilised into an aqueous solution. Dissolution rates for both silicate minerals and glass decrease with increasing Si:O ratio, and at pH ≥ 2 glass dissolution rates are faster than corresponding mineral rates (Stumm and Furrer 1987; Millero and Sotolongo 1989; Hamilton et al. 2000; Journet et al. 2008). As an example, at pH 4, Si-poor basalts (Si:O ~ 0.29) dissolve 0.4 orders of magnitude more quickly than Si-rich dacite (Si:O ~ 0.38), and this latter dissolves 1.5 orders of magnitude more rapidly than minerals with similar Si:O ratio, such as albite (Hamilton et al. 2000; Wolff-Boenisch et al. 2006). Moreover, volcanic ash exposed to acidic leaching with 1 M HNO3 solution for 4 h showed a Fe-release two orders of magnitude higher than that recorded in deionized water (de Hoog et al. 2001). These findings have been crucial to decipher the Fe mobilisation process, as recent studies recognized that an acidic liquid film with pH as low as 0–1 may form on airborne ash particles by interaction with volcanic acids (mainly HCl and H2SO4) in the eruption plume and/or by uptake of anthropogenic acids (e.g.,H2SO4, HNO3) during atmospheric transport (e.g., Delmelle et al. 2007; Cwiertny et al. 2008; Maters et al. 2016).

In addition to local pH conditions in the liquid film at the ash surface, Fe oxidation state needs to be taken into account when dealing with Fe released from mineral dust and glass (Schroth et al. 2009; Pacella et al. 2014, 2015). Olivine, pyroxene and amphibole mostly host Fe2+ in their octahedrally-coordinated structural sites (although amphiboles may also contain significant amounts of Fe3+), whereas plagioclase, oxides and glass may contain both Fe2+ and Fe3+, either tetrahedrally or octahedrally coordinated (Smith 1974a, b; Annersten 1976; Brantley and Chen 1995; Dyar et al. 2006; Pacella et al. 2012; Nyrow et al. 2014; Perinelli et al. 2014). In marine environment, both Fe redox speciation and release rate rule environmental equilibria and Fe2+–Fe3+ bioavailability. It has been demonstrated that dissolved Fe2+ is more accessible than Fe3+ to marine phytoplankton (Wells 2003; Cwiertny et al. 2008; Gledhill and Buck 2012), and Fe2+ oxidation to Fe3+ may control the dissolved O2 uptake in marine sediments covered by ash deposits (Hembury et al. 2012). The above-mentioned literature studies on Fe-release, based on natural or experimental leaching of mineral dust and volcanic ash in acidic environment, are highly relevant for understanding mineral/glass dissolution pathway in deliquesced mineral aerosols and even in human gastric fluids (Wiederhold et al. 2006; Journet et al. 2008; Oze and Solt 2010; Rozalen et al. 2014; Maters et al. 2017), but are probably too extreme in terms of pH conditions to simulate Fe mobilisation in lacustrine and marine environment.

In the present study, we fully characterized from the mineralogical viewpoint four samples of fresh volcanic ash from Italy and Mexico, and then performed leaching experiments on pristine and washed ash samples in deionized 18.2 MΩ grade water for timescale of hours-to-months to monitor Fe release. Results obtained represent a contribution to understanding the mechanisms of Fe mobilisation from volcanic ash as a function of mineral chemistry and Fe redox speciation, and may help assessing the impact of those mechanisms on Fe bioavailability in marine and lacustrine environments.

Materials and experimental methods

Materials

The fresh volcanic ash samples used in this study were selected from recent tephra deposits in Italy and Mexico. The Etna2011 and Etna2012 samples come from the strong fountaining activity that occurred at Etna volcano (southern Italy) on August 2011 and April 2012, respectively. They were collected in the Giarre area, about 17 km from the vent. Among these pyroclastic materials it was possible to isolate three fractions with different average particle size (0.25, 0.50 and 1.00 mm). The Popo2012 sample comes from the dome explosion activity of April 2012 at Popocatépetl volcano (central Mexico), was sampled at 24 km from the vent and its average particle size is 0.125 mm. In a previous investigation on the same samples, D’Addabbo et al. (2015) showed that the ash particle size has relevant effects on the release of ions in aqueous environment. A clear inverse correlation with the ash particle size was observed for SO42−, Na+, K+, Ca2+ and, to a lesser extent, for Cl−. Release of Mg is almost constant for the different grain sizes, while no regular behaviour was shown by F, Mn, Fe, B and Si (D’Addabbo et al. 2015). Moreover, saturation calculations highlighted that Etna samples were always saturated for F and Si in deionized 18.2 MΩ grade water, while Popocatépetl samples showed saturation levels with respect to some Ca, Si, F, and Fe compounds.

Chemical analysis

Bulk chemical compositions of major and trace elements were obtained by X-ray fluorescence (XRF) and loss on ignition (LOI) analyses. About 7 g of representative samples was powdered in agate mortar, mixed with hydrogen borate and consolidated with Elvacite® acrylic resin before XRF analyses. Further investigations on selected samples were accomplished by scanning electron microscopy (SEM) and energy dispersive X-ray (EDX) analyses. XRF were carried out on lapilli and ash fractions, after they were washed in distilled water and powdered in agate mortar. Major elements were measured using an automatic spectrometer Philips PW 1480 at the Dipartimento di Scienze della Terra e Geoambientali (University of Bari). SEM-EDX analyses of glass shards and glasses from micro-pumice fragments were performed at the Dipartimento di Scienze della Terra (University of Pisa), using an EDAX-DX micro-analyser mounted on a Philips SEM 515 (operating conditions: acceleration voltage = 20 kV, live time = 100 s, beam diameter = 200–500 nm, counting rate = 2100–2400 cps). Raw data were corrected according to the ZAF method (Reed 1993) and employing theoretical K-factors, and normalized to 100 wt%. Analytical precision is ~ 0.5% for abundances higher than 15 wt%; ~ 1% for abundances between 15 and 5 wt%; ~5% for abundances between 5 and 1 wt%; and up to 20% for abundances close to the detection limit (~ 0.5 wt%), and accuracy is ~ 1% (Marianelli and Sbrana 1988).

57Fe Mössbauer spectroscopy

The oxidation state of iron was investigated by 57Fe Mössbauer spectroscopy (MS) at the Dipartimento di Scienze della Terra, Sapienza University of Rome, using a standard Elscint electromechanical spectrometer operated in transmission geometry and constant acceleration mode with a 57Co source of nominal 25 mCi in rhodium matrix. Spectra were recorded at room temperature within a velocity range − 10 to + 10 mm/s, and the signal was transmitted to a multi-channel analyser. Data analysis involved a curve-fitting procedure assuming a Lorentzian peak shape and employing the fitting program Recoil 1.04. The statistical best fit was obtained by using the reduced χ2 method and uncertainties were estimated on the basis of the covariance matrix. The experimental uncertainties were estimated to be about ± 0.02 mm/s for isomer shift (δ) and quadrupole splitting (ΔEQ), ± 1 T for magnetic field (H), and ± 3% for absorption areas.

Absorption doublets with δ values of about 1.0–1.1 mm/s were assigned to Fe2+ and those with δ values of about 0.3–0.4 mm/s were assigned to Fe3+, according to the abundant existing literature on Fe-bearing silicates and oxides (e.g., Stevens et al. 2005; Fierro et al. 2005; Mansfeldt et al. 2012).

Additionally, a small absorption doublet with δ value of about 0.8 mm/s was assigned to Fe2.5+, representing the inter-valence charge transfer due to electron delocalization between Fe2+ and Fe3+ (Dyar et al. 2006; Andreozzi et al. 2008).

Fe-L2,3 edge ELNES and TEM-EDX analyses

Analyses of individual Fe-bearing phenocrysts of selected volcanic ash samples were performed at the Bayerisches Geoinstitut (University of Bayreuth, Germany) with a Philips CM20FEG transmission electron microscope (TEM) operated at 200 kV. Few milligrams of volcanic ash samples were crushed under ethanol and pipetted on lacey carbon films supported by Cu-grids (300 mesh). Semi-quantitative major elemental compositions were obtained by energy-dispersive X-ray (EDX) spectroscopy using a NORAN 7 system (Thermo Fisher Scientific Inc.) mounting a pure Ge detector. The Cliff–Lorimer k-factors of major elements were experimentally calibrated using a natural pyrope-almandine garnet (van Capellen 1990), except for Na, for which a theoretical k-factor was used. Thickness-absorption correction was performed following the electroneutrality criterion (van Capellen and Doukhan 1994).

The Fe3+/Fetot ratio of Fe-bearing phenocrysts was obtained by electron energy-loss spectroscopy (EELS) using a Gatan 666 parallel electron energy-loss spectrometer (PEELS). Fe-L2,3 edge energy-loss near-edge structure (ELNES) spectra were collected in diffraction mode with collection semi-angle of β = 1.35 mrad, energy-dispersion of 0.1 eV per channel, and 1–3 s integration time per read-out. The energy resolution of the zero loss spectra is 0.8–0.9 eV, measured as the width of the zero-loss peak at half height. Quantification of the Fe3+/FeTot ratios by Fe-L2,3 edge ELNES spectra followed the procedure described by van Aken et al. (1998) and van Aken and Liebscher (2002), using an empirically calibrated universal curve.

X-ray powder diffraction and quantitative phase analyses

The phase assemblage of the volcanic ash samples was determined by X-ray powder diffraction (XRPD) and Rietveld analyses at the Department of Earth and Environmental Sciences of the University of Milano-Bicocca. Samples were grinded in agate mortar, mixed with 10 wt% of α-Al2O3 (grain size 0.3 µm), and back loaded into an Al sample holder. Diffraction spectra were collected in Bragg–Brentano specular (θ–θ) geometry with a PANanalytical X’Pert-Pro PW3060 diffractometer employing a CuKα1,2 radiation and equipped with an X’Celerator position sensitive detector. Diffractometer scans were recorded in the 3°–90° 2θ range with step size of 0.02°, at room temperature and operating conditions of 40 mA and 44 kV. A Ni filter along the diffracted beam path was used to filter out the CuKβ radiation and the sample holder was allowed to spin horizontally during measurement to improve particle statistics.

The identification of major and minor phases was done with the X’Pert High Score software (PANanalytical) using the ICSD PDF2-2004 database. Quantitative phase analyses (QPA) were performed with the fullprof Rietveld method (Hill 1991; Bish and Post 1993) implemented in GSAS/EXPEGUI (Larson and Von Dreele 2004). Starting structure models were adopted from literature. The weight fraction of each crystalline phase (Wα′) in the studied volcanic ash samples was quantified according to the following equation:

$${W^{\prime}_\alpha }={W_\alpha }\frac{{{{W^{\prime}}_{\text{c}}}}}{{{W_{\text{c}}}}}\left( {\frac{1}{{1 - {{W^{\prime}}_{\text{c}}}}}} \right)$$

where Wα and Wc are the refined weight fractions of phase α and of the internal corundum standard, respectively, Wc′ is the actual added weight of the internal standard (10 wt%). The actual weight fraction of the amorphous material (Wg′) is then calculated as Wg′ = 1 – ΣiWαi′. According to Gualtieri (2000), the relative error in glass content quantification is around 10% for fractions of the amorphous phase greater than 10 wt%, and increases with decreasing weight fraction of the glass.

Leaching experiments

Leaching experiments were performed with 2 g sieved, fresh ash samples, which were added to plastic cuvettes containing 20 ml of deionized 18.2 MΩ grade water (Millipore, Milli-Q), i.e. an ash-to-water ratio of 1:10. The grain sizes used were from 0.25 to 1 mm and the experiments were performed at ambient temperature (20–25 °C). During the experiments, the cuvettes were kept plugged for minimizing the oxygen exchange between the solution and the atmosphere. A table shaker stirring at 150 revolutions per minute (rpm) was used in order to ensure efficient mixing of ash and solvent. Before each measurement, a short period of stirring (few minutes) was applied for a number of times depending on the run duration. Leachates were analysed at increasing run durations from 30 min up to 2 months, withdrawing aliquots of solution at defined times. For evaluation of long term Fe-release, i.e. for run durations of 30 and 60 days, samples were washed in deionized water, in order to remove any sublimate salt from the particles surface. Before each measurement, the solution aliquot was filtered with Minisart® cellulose acetate syringe filters (0.45 µm) and treated with HNO3 7 M (prepared from Sigma–Aldrich Suprapur Metal).

Iron concentrations were measured using an inductively coupled plasma optical emission spectrometry (ICP-OES) Perkin Elmer Optimal 2000 DV instrument. The resulting concentration is the average of three different readings at two different wavelengths of 238.204 and 239.562 nm, respectively (six readings in total). The system was calibrated with standards of known concentration of 10, 50, 100, 200, 600 and 1000 µg l−1, prepared from a certified solution of 1000 mg l−1. The calibration was performed at the beginning of each analytical cycle and repeated every 12 measurements. The calibration curves resulted linear with R2 values of 0.999938 for both wavelengths. Operating conditions were as follows: power 1400 W, plasma argon flow rate 15 l min−1, nebulizer argon flow rate 0.55 l min−1, and sample flow rate 2 ml min−1; peak acquisition and integration 3 s; washing time 15 s. The reproducibility of the data was evaluated from repeated analyses of subsamples and is better than 4%.

Results

Composition and mineralogy of volcanic ash

Etna samples

Etna2011 and Etna2012 samples are composed of dark, porphyritic scoriae (Fig. 1a, b) with trachy-basaltic and tephro-basanitic bulk chemistry, respectively (Table 1; Fig. 2). On the basis of trace elements, these samples may be classified as intra-plate alkali basalts according to the Nb–Zr–Y plot of Meschede (1986), and as calk-alkaline basalts according to the Ti–Zr–Sr plot of Pearce and Cann (1973). The glass composition is slightly evolved towards basaltic trachy-andesite and displays hawaiitic character (Fig. 2). Etna2011 sample was further analysed through XRPD and Rietveld analyses (relevant data in Table 2); the glass is quantified at 73.0 wt%, while mineral phases detected at significant level are plagioclase (22.5 wt%), Ca-rich pyroxene (2.7 wt%), and olivine (1.8 wt%). According to SEM–EDX analyses, plagioclase is andesine with average composition Ab52An46Or02, and pyroxene is augite with average composition (Ca0.87Mg0.70Fe0.28Al0.08Ti0.03Mn0.02Na0.01)Σ = 1.99(Si1.91Al0.09)Σ = 2.00O6 (Table 3). Olivine was characterized by TEM–EDX as Fo69Fa30Te01 (Table S1). The presence of unidentified Fe, Ti, Al, Mg oxides, probably mixtures of ilmenite and spinel, was also revealed by SEM–EDX.

Fig. 1
figure 1

SEM–BSE images of the volcanic ash samples investigated in this study: a Scoria fragments from an August 2011 eruption of Etna volcano; b Close view of a scoria fragment from the April 2012 eruption of Etna volcano; c Porphyritic, dense juvenile fragments from the April 2012 eruption of Popocatépetl volcano (Mexico). Mineral phases: cpx = clinopyroxene; opx = orthopyroxene; ol = olivine; ox = oxide; pl = plagioclase

Full size image
Table 1 XRF bulk composition (wt% for oxides and ppm for trace elements) and SEM-EDX glass composition (wt%) of the volcanic ash samples investigated in this study
Full size table
Fig. 2
figure 2

Composition of the studied volcanic ash samples plotted on the total alkali vs. silica (TAS) diagram. Symbols and colours: triangles = Etna samples; crosses = Popo2012 sample; blue = bulk (XRF analyses); red = glass shards (EDX analyses)

Full size image
Table 2 Rietveld quantitative phase analyses (wt%) of selected volcanic ash samples
Full size table
Table 3 SEM-EDX analyses (wt%) of silicate crystallites detected in the Etna2011 and Popo2012 samples
Full size table

Popocatépetl sample

The Popo2012 sample comprises grey porphyritic pumice and scoriae (Fig. 1c) with andesitic bulk composition and dacitic glass shards (Table 1; Fig. 2). Similarly to Etna samples, on the basis of trace elements Popo2012 may be classified as intra-plate alkali basalts according to the Nb–Zr–Y plot of Meschede (1986), and as calk-alkaline basalts according to the Ti–Zr–Sr plot of Pearce and Cann (1973). Popo2012 sample has the lowest glass content (40.2 wt%) and glass composition is more evolved towards dacite (Fig. 2). Mineral phases mainly comprise plagioclase (50.0 wt%), Ca-poor pyroxenes (5.3 wt%) and Ca-rich pyroxenes (1.5 wt%), detected by both XRPD Rietveld analyses and SEM. According to SEM-EDX analyses, plagioclase is labradorite, with average composition Ab53An44Or03, the Ca-poor pyroxene is enstatite, with average composition (Mg1.27Fe0.60Ca0.07Mn0.02Ti0.01)Σ = 1.97Si2.00O6, and the Ca-rich pyroxene is augite, with average composition (Mg0.87Ca0.64Fe0.38Al0.04Ti0.02Mn0.01)Σ = 1.96(Si1.97Al0.03)Σ = 2.00O6 (Table 3).

A high temperature silica phase (identified as tridymite, 1.6 wt%) and carbonates (1.3 wt%) were also recognized by XRPD Rietveld analyses (Table 2). The presence of an accessory Fe–Ti-oxide, likely ilmenite, was only detected by SEM-EDX.

Iron speciation of volcanic ash

Mössbauer spectroscopy on bulk samples

The 57Fe Mössbauer absorption spectra of all samples are typical of paramagnetic materials (Fig. 3). No absorption due to magnetic sextets possibly attributed to Fe-rich oxides (e.g., magnetite) was evidenced by spectral analysis. Accordingly, the spectra of all samples are well reproduced by several absorption effects consisting of quadrupole doublets with hyperfine parameters typical of Fe2+ and Fe3+ (Fig. 3; Table 4). The Etna2011 sample needed four doublets to thoroughly model the Fe2+ absorption and two doublets to satisfactorily model the Fe3+ absorption. The complexity of the recorded spectrum reflects the presence of the various paramagnetic ferromagnesian silicate and oxide minerals evidenced by chemical analysis, and is also influenced by the high quantity of glass this sample is made of. Due to the very close values commonly recorded for Mössbauer hyperfine parameters in silicates, oxides and glass (e.g., Stevens et al. 2005; Dyar et al. 2006), it is hard to say if the observed absorption doublets belong to different minerals or better represent distinct short-range environments around Fe ions in the glass, even though the second hypothesis seems to be highly plausible. The spectrum of Popo2012 sample was well reproduced by only one Fe2+ and one Fe3+ doublet, and finally required a small accessory Fe2.5+ doublet to model the absorption by Fe ions at an oxidation state intermediate between Fe2+ and Fe3+, as occasionally observed in both oxides and silicates (Dyar et al. 2006; Andreozzi et al. 2008). In conclusion, the resulting Fe3+/FeTot proportions in bulk samples are quantified at 33% for Etna2011 and 23% for Popo2012 (Table 4). These values are closely comparable with oxidation state of iron in magmatic melts and sensibly lower than values recorded for experimental glasses produced in contact with atmospheric air (Nikolaev et al. 1996; Nyrow et al. 2014; Maters et al. 2016, 2017).

Fig. 3
figure 3

Representative room temperature 57Fe Mössbauer absorption spectra of volcanic ash. Black dots = experimental absorption spectrum; red thick line = calculated absorption spectrum; green thin lines = Fe2+ absorption doublets; blue thin lines = Fe3+ absorption doublets; purple thin line = Fe2.5+ absorption doublet

Full size image
Table 4 57Fe Mössbauer parameters of the volcanic ash bulk samples investigated in this study
Full size table

Fe-L2,3 edge ELNES on specific grains

Fe-L2,3 ELNES analyses were performed on specific grains to determine the Fe-speciation partitioning between minerals and glass, as well as among different ferromagnesian minerals. Results show that in the Etna2011 sample the Fe3+/FeTot ratio measured on Ca-rich pyroxene is 0.23–0.26 (depending on the two different fitting procedures of van Aken et al. 1998 and van Aken and Liebscher 2002), against the 0.33 measured with Mössbauer spectroscopy on bulk materials. Few grains of olivine, which were (hardly) detected by TEM, were not oxidised. Similarly, in the Popo sample, the Fe3+/FeTot ratio measured in Ca-poor pyroxenes resulted comprised between zero and 0.14, which is significantly lower than the value 0.23 measured on the bulk sample (Table S1).

Overall, comparison between Fe-L2,3 edge ELNES results and 57Fe Mössbauer spectroscopy suggests that glass particles are more oxidized than crystallites and phenocrystals. This finding accounts for the observation that glass solidified later and in closer contact with oxidizing atmosphere than crystallites, and is also more permeable to atmospheric O2 than crystals (Zhang et al. 2010). Moreover, H2O is a highly incompatible compound in magmas, and high water abundances in the ash conduit and plume may have triggered in-situ glass oxidation in the studied samples even before the contact with atmosphere.

Iron release of volcanic ash

The Fe released during leaching experiments was monitored on both pristine samples (leaching times from 30 min to 7 days) and washed samples (leaching times of 30 and 60 days). Results revealed a marked Fe release during the first week followed by very low to null values recorded in longer times (Fig. 4; Table 5).

Fig. 4
figure 4

Fe release (nmol per gram of ash in solution per litre) vs. time (s). Symbols: orange and green triangles = Etna2011 samples, size 0.25 and 0.50 mm, respectively; red circles = Etna2012, size 1 mm; blue squares = Popo2012 sample, size 0.125 mm. Insets refer to short-term release of Fe within the first week. Solid and dashed lines are guide to the eye

Full size image
Table 5 Fe-release (nmol per gram of ash in solution per litre) in deionized 18.2 MΩ grade water (pH 4.96)
Full size table

For each of the pristine Etna samples, which are well comparable for both composition and mineralogy, the maximum Fe release was found to occur as a function of both particle size and time. In fact, for the Etna2011 with particle size of 0.25 mm, the maximum Fe release (158 nmol g−1 l−1) was recorded after 30 min, for the Etna2011 with particle size of 0.50 mm the maximum (80 nmol g−1 l−1) was reached after 6 h, and for the Etna2012 with particle size 1.00 mm the maximum (74 nmol g−1 l−1) was reached after 7 days (Table 5). Apart from the single values recorded at each leaching time, that are sometimes scattered, the two ash samples with the smallest grain size showed to be mostly active in releasing Fe during the first day of leaching, while the coarsest sample exhibited a sluggish start followed by a sustained activity during the first week of leaching. The pristine Popocatépetl sample showed a very sustained Fe release (up to ten times Etna samples) all along the monitored period, with lowest values never below 400 nmol g−1 l−1 and a maximum of 1672 nmol g−1 l−1 recorded after 5 days.

For the washed Etna samples, Fe release after run duration of 30 days was usually very low (≤ 35 nmol g−1 l−1) and showed a decreasing trend with increasing particle size. After 60 days, values ≤ 1 nmol g−1 l−1 or below the detection limit were recorded. For the washed Popocatépetl sample, values after 30 and 60 days were always below the detection limit.

In summary, the strong difference observed all along the leaching experiments between pristine and washed samples suggests an active role of surfaces, which still contain soluble salts in pristine samples and do not contain such salts in washed samples.

Discussion

The aliquot of iron passing in solution with respect to the total iron content of the volcanic ash, i.e. the fractional Fe solubility (FeS), can be expressed as follows: FeS (%) = (Dissolved Fe/Total Fe) × 100, where the “Dissolved Fe” is obtained from the leachates analyses and the “Total Fe” from XRF measurements (Olgun et al. 2011). Accordingly, for the samples here investigated the maximum FeS values range from 0.004 to 0.011% for the Etna samples and is 0.23% for the Popo2012 sample. These values are comparable with those reported by Olgun et al. (2011) for subduction zone volcanic ash (0.003–0.2%).

Recent studies demonstrated that there is no immediate correlation between Fe release and the Fe content of the bulk ash samples (Jones and Gislason 2008; Olgun et al. 2011; Maters et al. 2017). According to these studies, the trend of Fe release from ash exposed to deionized water/acidic solutions is characterized by transient high initial values followed by lower and almost steady state values, this suggesting the presence of at least two “reservoirs” with distinct Fe solubility, typically active in the short timescales (hours-to-days) and in the long timescales (months-to-years). One of these reservoirs has been individuated in the sublimates of Fe-sulphates and Fe-halides that cover the surface of fresh ash particles and may be relevant when dealing with short timescales. These salts are likely the most soluble components of ash particles and are formed through the interaction of ash particles with volcanic gases (S, H2S, HCl and HF) and atmospheric aerosols in the eruption plume and subsequent cloud (Naughton et al. 1976; Rose 1977; Óskarsson 1980, 1981; Frogner-Kockum et al. 2001; Delmelle et al. 2007; Duggen et al. 2007, 2010; Jones and Gislason 2008; Olgun et al. 2011). The second reservoir can be reasonably associated with dissolution of Fe-bearing glass shards and mineral particles that may account for Fe released in longer timescales (Censi et al. 2010; Ayris and Delmelle 2012b; Maters et al. 2017).

Short timescale Fe release

In the samples here investigated, the first reservoir may be associated with the abundant surface sublimates detected on the same samples by the previous study of D’Addabbo et al. (2015), that evidenced for Popocatépetl the presence of iron sulphides and calcium sulphates together with minor calcium fluoride and sodium chloride (these latter observed in all samples and possibly containing Fe as minor component). Due to rapid dissolution of surface sublimates, all samples showed a significant Fe mobilisation within the first hour, especially after the first 30 min, with increasing Fe quantities released in solution as a function of decreasing particle size (Fig. 5). For Etna samples, that share the same mineralogical composition, the trend is exponential. Notably, the exponential trend is also followed by Popocatépetl sample in spite of its completely different mineralogy, and this suggests that the particle surface area is probably the key factor in controlling both the quantity of Fe-bearing soluble salts adsorbed on ash surfaces and their dissolution in very short times.

Fig. 5
figure 5

Maximum Fe release during the first hour of leaching vs. average particle size. Data plot refers to Fe values after 30 min. Symbols: filled circles = Etna samples; cross = Popo2012 sample. Solid line is the best fit to Etna data, corresponding to the power function: ln(y) = − 1.20 ln(x) + 3.34 (R2 = 0.98)

Full size image

Assuming a spherical particle shape in the ash, the surface area-to-volume ratio increases linearly with decreasing grain size and, accordingly, solubility should increase linearly with decreasing grain size. However, notwithstanding the key role of particle size, the extremely different values of Fe released in short time by Etna and Popocatépetl samples (Fig. 5), as well as their variable behaviour (Fig. 4), are likely due to a combination of various factors, among which the solution pH is fundamental. During the dissolution experiments it was observed that pH of the leachate solution decreased of ~ 0.4 pH units for Popo2012 sample and increased of ~ 1.4 pH units for the Etna samples with respect to deionized 18.2 MΩ grade water, whose measured pH was 4.96 (D’Addabbo et al. 2015). These variations, which are related to the release of OH− and H+ ions due to hydrolysis reactions of the salts dissolved from ash surfaces, suddenly occurred in the first 30 min of leaching and smoothly fluctuated for longer times, thus possibly justifying the observed fluctuations of Fe release. For Popocatépetl volcanic ash, the pH decrease to ~ 4.6 had the effect of fostering the solid–liquid exchange process, furthermore promoting Fe release. The low pH, in fact, had an active role in favouring partial dissolution of the glass portion of the ash, that produced the Fe scavenging and the subsequent release in solution observed in short timescales. Similar results were obtained by Jones and Gislason (2008) when testing volcanic ash from seven different volcanoes of different ages, eruption type and silica contents, with an ash-to-water ratio similar to that employed in our experiments: for leaching experiments conducted in deionized water at pH of 3.5–4.7, they recorded a Fe release of 10,850 nmol g−1 l−1, even higher than that we obtained. Moreover, pH decrease determines the parallel decrease of the (average) pseudo-first-order kinetic rate constant (k′) of Fe2+–Fe3+ oxidation reaction (Stumm and Lee 1961; Millero et al. 1987). For Popocatépetl sample, on the basis of leaching data after 30 and 60 min, we calculated k′ = 4.25−5 s−1, much lower than values calculated for Etna samples, that range from 4.56−4 to 2.85−4 s−1. Accordingly, in the solution of Popocatépetl sample the ferrous iron oxidation state was favoured, whereas for Etna samples the ferric iron oxidation state was relatively advantaged. As a consequence, the high Fe amounts in solution exhibited by Popocatépetl sample are reasonably due to both the rapid dissolution of abundant iron sulphide (and other Fe-bearing sublimates) occurring on the surfaces of ash particles and Fe residence time in the aqueous phase, this latter due to the high solubility of Fe2+ ions. For Etna samples, on the contrary, not only the Fe released is much lower but its residence time in solution is contrasted by the very low solubility of Fe3+ at neutral-to-basic pH, and therefore it is highly probable that most of the Fe is removed by precipitation of amorphous/nanocrystalline Fe3+-(hydr)oxides. In addition, Fe3+-(hydr)oxides may progressively aggregate and effectively adsorb on their surfaces the dissolved Fe ions, further decreasing in the medium-to-long timescale the amounts of Fe dissolved in the solution (e.g., Teutsch et al. 2005; Larese-Casanova and Scherer 2007; Schuth and Mansfeldt 2016).

The released Fe measured for the investigated samples is comparable with the released Fe retrieved from literature for dry deposition of pristine ash samples in seawater (Duggen et al. 2007; Jones and Gislason 2008; Olgun et al. 2011). The amounts of Fe released in seawater in short timescale are retained of main importance for fertilization of surface ocean, because this timescale is comparable with the residence time of ash particles in the euphotic zone, where Fe is available to marine phytoplankton, while sinking towards the sea floor (Olgun et al. 2011). Noteworthy, such a residence time is estimated of few minutes for coarse (2–0.5 mm), 1–2 h for medium size (250–150 µm), and 1–2 days for fine (< 50 µm) ash particles (Duggen et al. 2007). Accordingly, in our case the maximum contribution to surface water fertilization would be assured by Popo2012 sample which has grain size of 125 µm and an estimated residence time roughly around 2 h, a period in which its Fe release is very high (up to 751 nmol g−1 l−1). Although downscaled, almost the same mechanism applies for the Etna2011 sample with the smallest grain size (250 µm), whose maximum Fe release (up to 158 nmol g−1 l−1) occurs within the first hour, a time lapse comparable with its residence time. Both these values are relevant since mesoscale Fe fertilization experiments showed that an increase by only 2 nmol g−1 l−1 of Fe levels are sufficient to stimulate massive diatom blooms in Fe-limited oceanic regions (Wells 2003). Conversely, the coarsest Etna2011 and the Etna12 samples showed the highest Fe-releases (~ 80 nmol g−1 l−1) for run durations of 6 h and 7 days, respectively (Table 5), thus well beyond their residence time within the euphotic zone (estimated at few minutes). In spite of that, their Fe release within the first 30 min of leaching was not negligible (~ 57 and ~ 30 nmol g−1 l−1, respectively), this suggesting also for these samples a possible contribution to the biogeochemical Fe cycle.

Long timescale Fe release

For the 1 and 2 months experiments, Fe release was found to be very low or below detection limit for all samples (Table 5). As suggested above, it is very probable that Fe released during long timescale runs was higher and part of it was oxidized and precipitated as Fe3+ (hydr)oxides, determining a low Fe concentration in solution. Notwithstanding, the two Etna2011 samples do show some Fe in solution after 1 month (and even after 2 months). This Fe cannot be ascribed to the sublimates covering particles surface, because they were removed by carefully washing the samples with deionized water before leaching experiments. In this case, only partial dissolution of glass shards and mineral particles may have played a role. In marine environment this process may occur both in solution and by alteration and upwelling of volcanic ash particles deposited at the sea floor (Olgun et al. 2011). The number of factors influencing the Fe solubility of glass shards and mineral particles for long timescales can be reasonably summarized as follows: (i) acidic solutions favour ash leaching and Fe-release (de Hoog et al. 2001); (ii) high silica contents contrast ash dissolution rate (Hamilton et al. 2000); (iii) glass dissolves faster than crystalline phases of comparable composition (Wolff-Boenisch et al. 2006); (iv) the Fe2+-bearing silicates are more soluble than Fe3+-bearing oxides (Stumm and Furrer 1987; Millero and Sotolongo 1989; Journet et al. 2008); (v) fine particles dissolve faster than coarse particles (Baker and Jickells 2006; Ooki et al. 2009). As for the various Etna samples, all mineralogical characteristics being equal, a limited dissolution due to size effect may account for Fe release of Etna2011 samples (with particle size of 0.25 and 0.50 mm), whereas a negligible (or slower) dissolution may justify the apparently negligible (or just slower) Fe release from the coarse Etna 2012 (with particle size of 1.00 mm). As for Popo2012, on one hand the size effect favours its dissolution (and concomitant Fe release) with respect to Etna samples; on the other hand this sample has the highest silica content (61.7 vs. 45.6–47.6 wt%) and the lowest glass fraction (40.2 vs. 73.0 wt%), two factors that contrast (rather than promote) its dissolution and, together with the lowest total Fe content (Fe2O3 = 5.8 vs. 11.6–13.2 wt%), account for the apparently negligible (or just slower) Fe release.

Finally, one can speculate on the effects of ash aging on Fe released in aqueous environment. Old volcanic ash from “fossil tephra” deposited in the mainland can in principle be remobilized by windstorms and dispersed on the ocean surface. Irrespective of particle chemistry and size distribution, old volcanic ash has been exposed since a long time to meteoritic water, which quickly removed Fe-bearing salts from the surface, and experienced for very long time alteration processes typical of Earth surface, becoming significantly more oxidized than fresh samples (e.g., Bell III et al. 1993; Bakker et al. 1996). As an example, fossil tephra of 3000–600,000 years old showed Fe3+/FeTot ratios from 0.24 to 0.68, and a linear correlation with aging was established (Capitani et al., unpublished data). On these bases, old volcanic ash likely represents a less effective source of bioavailable Fe for marine phytoplankton. Notably, from human health perspectives, inhalation of mineral dust coming from remobilization and air dispersion of an old volcanic ash by windstorms or any kind of human activity (mining, tunnel, pavement and road construction, etc.) also represents a less effective source of bioavailable Fe, which however means a lower hazard than fresh ash as in physiologic conditions Fe3+ is less effective than Fe2+ in generating free radicals (Fubini and Fenoglio 2007; Horwell et al. 2010; Pacella et al. 2012; Andreozzi et al. 2017).

Conclusions

Results obtained in the present study are in line with previous studies on volcanic ash and confirm that, when attempting to explain (or even to predict) the rate and extent of volcanic ash dissolution and subsequent Fe release in aqueous environment, several mineralogical factors (including ash composition, particle size, surface chemistry, Fe oxidation state and structural constraints) need to be carefully considered in addition to environmental physico-chemical parameters (e.g., pH conditions).

The fine ash coming from tephra deposits of the Mexican Popocatépetl volcano accumulated large quantities of Fe-bearing sublimate salts, which quickly dissolved during early dissolution steps and determined both Fe release and pH decrease in very short times (< 30 min). Local pH decrease, in turn, had an active role in favouring partial leaching of the glass portion of the ash, that produced the Fe scavenging and subsequent release in solution observed in short times (hours-to-days). The coarser ash samples coming from tephra deposits of the Italian Etna volcano probably accumulated less sublimates on their surfaces and consequently released less Fe in the early steps. However, their glass component is almost double than that of Popocatépetl ash and contains a double amount of Fe, so that glass partial dissolution is considered to be responsible for the Fe release recorded in the medium-to-long timescale. For all the investigated samples, Fe in solution is also modulated by particle size and by Fe precipitation in form of Fe3+-(hydr)oxides, that may aggregate and later adsorb on their surfaces part of (or all) the dissolved Fe ions. Results obtained suggest that volcanic ash chemistry, mineralogy and particle size assume a relevant role on Fe release mostly in the medium-to-long timescale, while Fe release in the short timescale is dominated by dissolution of surface sublimates, formed by physicochemical processes occurring within the eruption plume and volcanic cloud, and by local pH conditions.

For all samples here investigated, a moderate to sustained Fe release occurred for leaching times comparable with their residence time within the euphotic zone of marine and lacustrine environments (that is variable from few minutes to few hours as an inverse function of particle size), this revealing their possible contribution to increase Fe bioavailability.