Far from equilibrium basaltic glass alteration: The influence of Fe redox state and thermal history on element mobilization

Journal Pre-proofs Far from equilibrium basaltic glass alteration: The influence of Fe redox state and thermal history on element mobilization Marius Stranghoener, Stefan Dultz, Harald Behrens, Axel Schippers PII: DOI: Reference:

S0016-7037(20)30016-8 https://doi.org/10.1016/j.gca.2020.01.005 GCA 11585

To appear in:

Geochimica et Cosmochimica Acta

Received Date: Accepted Date:

23 July 2019 5 January 2020

Please cite this article as: Stranghoener, M., Dultz, S., Behrens, H., Schippers, A., Far from equilibrium basaltic glass alteration: The influence of Fe redox state and thermal history on element mobilization, Geochimica et Cosmochimica Acta (2020), doi: https://doi.org/10.1016/j.gca.2020.01.005

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2020 Elsevier Ltd. All rights reserved.

Far from equilibrium basaltic glass alteration: The influence of Fe redox state and thermal history on element mobilization Marius Stranghoener1*, Stefan Dultz2, Harald Behrens1, Axel Schippersc

*corresponding

1Institute

author: [email protected]

of Mineralogy, Leibniz Universität Hannover, Callinstr. 3, 30167 Hannover,

Germany 2Institute

of Soil Science, Leibniz Universität Hannover, Herrenhäuser Str. 2, 30419 Hannover, Germany 3Geomicrobiology,

Federal Institute for Geosciences and Natural Resources, Stilleweg 2, 30655 Hannover, Germany.

Keywords: Basaltic glass, alteration, element mobilization, Fe redox state, thermal history

Abstract Elemental release from basaltic glasses at far from equilibrium conditions was investigated as a function of the Fe redox state (Fe(II)/Fetot = 0.35 and 0.80) and thermal history (quenched ↔ annealed). A flow-through column setup was used to ensure disequilibrium of basaltic glass and solution during the entire runtime. Percolation experiments were performed at 25 °C for up to 500 h with intermediate sample collection. Two different pH values were adjusted, and the effect of organic matter was tested by adding oxalic acid. Element concentrations in the percolate were measured by Inductively Coupled Plasma Optical Emission Spectroscopy (ICPOES). After an initial high release of elements from fresh glass surface, steady state was achieved after 240-430 h for Si at pH 5-7 and after 170 -240 h for Al at pH 2. At near neutral conditions (pH 5-7) mobilization of Si is relatively high, while Fe and possibly Al are retained in precipitates. Under more acidic conditions (pH 2), the Si concentration of the solutions was very low compared to the other main constituents of the glass. Large amounts of Si-rich residues are formed after glass dissolution at pH 2. On the other hand, Fe (and Mn) is very mobile under acidic conditions, favored by complex formation with oxalic acid and chlorine. In the initial phase of the pH 2 experiments, the element release from reduced glasses is higher than from oxidized glasses. However, this trend is reversed when approaching steady state. Higher dissolution rates for oxidized glasses are predicted due to the progressive replacement of strong Si-O-Si with weaker Fe(III)-O-Si. At pH 5-7 the concentrations of elements in the percolate are too low to establish a systematic difference between oxidized and reduced glasses. Looking at the total amount of mobilized elements, the thermal history of the glasses has no significant effect in the case of oxalate-free solutions, but a noticeable increase of element release in the case of rapidly quenched glasses was observed when using 1mM oxalic acid solution. The strong effect of oxalate on dissolution of quenched glasses is probably related to the more open glass network structure in quenched glasses.

1. Introduction Volcanic glasses, formed by rapid cooling of lava, are widespread among the earth surface and occur in a variety of geological settings. The extent of alteration of these silicate glasses is mainly controlled by their composition and structure and by environmental influences (Amma et al., 2017; Eick et al., 1996; Staudigel and Hart, 1983; Wolff-Boenisch et al., 2004). Because of its widespread occurrence, natural basaltic glasses play an important role for various

(bio)geochemical element cycles including the global carbon cycle (Boyd and Ellwood, 2010; Goudie and Viles, 2012; Van Der Merwe et al., 2015) as well as for subsurface CO2 storage (Wolff-Boenisch et al., 2011; Gysi and Stefánsson, 2012). Basaltic glasses are also considered as natural analogues with regard to the long term stability of vitrified high-level nuclear waste (Techer et al., 2000; Libourel et al., 2011; Michelin et al., 2013; Ducasse et al., 2018). The durability of silicate glasses mainly depends on their SiO2 content and the abundance of other metal cations that either act as network formers (e.g. Al) or network modifiers (e.g. Na, Ca) (Wolff-Boenisch et al., 2004). Fe plays an ambivalent role in natural silicate glasses acting as both, network former (Fe(III)) and modifier (Fe(II)) depending on its redox state (Virgo and Mysen, 1985; Zhang et al., 2016). The impact of the redox state of Fe on glass dissolution was determined by Kumar and Lin (1991) for calcium phosphate iron glasses (dissolution rate was lowered if the glass was melted under oxidizing atmosphere) but was not investigated for silicate glasses so far. In natural magmatic systems the Fe(II)/Fetot ranges between > 0.9 for mid ocean ridge basalts to 0.65 in subduction related settings (Christie et al., 1986). Dissolution of silicate glasses is a complex multi-step process affected not only by solution chemistry but also by microorganisms (Oelkers and Gislason, 2001; Wolff-Boenisch et al., 2004; Conradt, 2008; Parruzot et al., 2015; Stranghöner et al., 2018). Subsequent to the fast removal of mono- and divalent cations from the glass surface, dissolution proceeds by protonation and breaking of T-O-Si bonds (T = Al(III), Fe(III)) and opening of the network due to formation non-bridging Si-OH bonds (Oelkers, 2001). Furthermore, metal-organic surface complexes formed with organic ligands such as oxalate weaken the metal-oxygen bonds, and the detachment of surface metal species is facilitated (Furrer and Stumm, 1986). In addition to chemical composition and redox state of multivalent elements in silicate glasses, the durability is known to be also affected by the thermal history (quenching ↔ annealing) (Pantano, 2001; Angeli et al., 2018). During quenching of natural basaltic glasses the huge temperature differences between the surface (in contact with water or air) and the interior produces compressive (interior) and tensile stresses (surface) that favor fracturing of the glass (Narayanaswamy and Gardon, 1969). The dissolution of silicate minerals and glasses is not uniform and occurred preferentially at microfractures and dislocations (Stumm, 1992). Whereas the effect of thermal history is poorly determined for basaltic glasses, some recent studies exist for industrial glasses. For soda-lime and borosilicate glasses increased hydrogen diffusion and leached layer formation promoted the dissolution of rapidly quenched samples (Amma et al., 2017; Angeli et al., 2018; Stone-

Weiss et al., 2018). Heterogeneous dissolution of the basaltic glass surface after acid leaching suggests preferential dissolution along zones of high residual stress (Eick et al., 1996; Gislason and Oelkers, 2003). It is furthermore suspected that the thermal history of the glass affects also microbial colonization by providing reactive sites that are energetically favorable for dissolution and nutrient acquisition (Pederson et al., 1983; Rosso et al., 2003; Stranghöner et al., 2018). Despite the large number of studies investigating basaltic glass dissolution little is known about the impact of Fe redox state and thermal history. In this study, far from equilibrium dissolution experiments with synthetic basaltic glasses were performed using glasses with large difference in Fe redox state and thermal history. A flow-through column setup was used to investigate the influence of these two factors on the mobilization of elements during alteration of synthetic basaltic glasses. Experiments were performed at pH 5-7, which is close to natural meteoric waters but also at pH 2 to simulate more acidic conditions. Effects of organic ligands on glass dissolution and element mobilization was studied by adding oxalic acid in some of the experiments. The focus of our study is on the time-dependent mobilization of the main components from basaltic glasses. Comparison of the release rates of different elements gives insight to the mechanisms of glass dissolution and allows conclusions on specific retainment of elements in relicts of the dissolving glass or in precipitates. 2. Material and Methods 2.1 Preparation and characterization of the synthetic basaltic glasses The synthetic basaltic glass was prepared from high purity powdered oxides and carbonates (SiO2, Al2O3, Fe2O3, CaCO3, MgCO3, Na2CO3, TiO2, K2CO3, (NH4)H2PO4)) according to the composition of natural Hawaiian basalts (Table 1). Melting was done in a platinum crucible at 1600°C for 2 h. The melt was quenched on a brass plate, crushed using a steel mortar and remelted for an additional hour at 1600°C to improve homogeneity. Part of the melt was then poured on a hot (550°C) steel plate, transferred to a chamber furnace at 550°C and slowly (< 1 K/min) cooled down to room temperature to remove internal stresses. This glass is referred to as annealed (B-OA in Table 2). The rest of the melt was rapidly quenched on a cold brass plate and is labelled as quenched (B-OQ). A portion of the quenched glass was crushed to a fine powder (< 50 µm) and reduced at 560°C (≈ 100°C below the glass transition temperature, Cooper et al. (1996)) for 24 h in a Ar/7% H2 gas flow. About 2.5 g of the reduced glass powder was filled into two Au80Pd20 capsules. After welding shut, the capsules were placed in a chamber furnace at 1200°C for 1h to re-melt the

powder. The capsules were either quenched in water (B-RQ) or annealed (B-RA) at 563°C under reducing atmosphere (Ar/H2) and slowly cooled (< 5 K/min) to room temperature.

2.2 Percolation experiments All percolation experiments were performed at 25°C in 5 ml syringes (inner diameter 12.4 mm) using 16 channel peristaltic pumps (Type 205s and 202u; Watson Marlow). The setup used in our experiments is shown in Fig. 1. The syringes were prepared as follows: First a 3 µm glassfibre prefilter (Sartorius) was placed on the outlet side of the syringe and above that a 15 µm black ribbon ash-less filter (Schleicher & Schuell) to retain the solid material. Next, about 1 g of acid-washed 0.5 to 1 mm sized quartz sand was added, followed by 0.2 g basaltic glass powder and another layer of 1 g acid-washed quartz sand on top. This layer packing has been established to ensure that the basaltic glass was always water saturated during the percolation experiment. Quartz is essentially insoluble in the aqueous solution under slightly acidic conditions and, hence, does not affect the dissolution of basaltic glass. The base solutions were either Ultrapure water (UPW; electrical resistivity 18.2 MΩ) or 1 mM oxalic acid (OA, C2H2O4; reagent grade; Alfa Aesar). The concentration of oxalate was close to that in natural environments and close to the minimum concentration at which a dissolution enhancing effect was observed (Drever and Vance, 1994). For the first set of experiments the UPW and OA solutions were adjusted to pH 5 by dropwise addition of 0.1 M HCl and 0.1 M NaOH, respectively. No pH buffer was used to avoid any impact on glass dissolution (Hausrath et al., 2009). The solutions were dripped onto the quartz layer from above with a thin tube with an average flow rate of 0.68 ± 0.03 ml/h (≙ 16.3 ± 0.6 ml/day). The water holding capacity of the syringes was 0.587 ± 0.038 ml. Hence, the residence time of the solution in the syringe was about 52 min in the pH 5-7 series. The pH 5-7 experiments were run in triplicates for a total duration of 504 h. A control experiment containing no sample material was simultaneously run for each of the solutions to check for dissolved elements stemming from the quartz sand and filter materials. The percolated solution was sampled in time intervals (after 24, 48, 72, 96, 144, 192, 240, 288, 360, 432, 504 h), and the pH was controlled in each percolate. The specific flow rate for each time segment was calculated by weighting the amount of percolated solution. After the last sampling, the filling of one of the three parallel columns was washed in ethanol, and the basaltic glass fragments were separated from the quartz sand by sieving for microscopic inspection.

Figure 1: Schematic overview of the experimental setup of the percolation experiment.

The two remaining parallels of each trial were used in follow-up percolation experiments at pH 2. First, the columns were flushed with UPW for 24 h for complete removal of the solution from the previous experiment. Then, the columns percolated with UPW in the first experiment were percolated with UPW adjusted to pH 2 with HCl. The columns percolated with 1 mM oxalic acid in the first experiment were percolated in the second run with 1 mM oxalic acid adjusted to pH 2. Experiments were run in duplicates at 25°C for a total duration of 385 h. Due to the higher reactivity of the more acidic solution the flow rate was doubled compared to the runs at pH 5-7 (average flow rate of 1.26 ± 0.07 ml/h, corresponding to a resistance time of 28 min). The percolated solution was sampled in time intervals (after 7, 24, 48, 71, 97, 169, 241, 313 and 385 h), and the pH was controlled in each percolate. The specific fluid flow rate was calculated as mentioned above. Again, after the last sampling, the column filling was washed in ethanol and separated from quartz sand by sieving for microscopic analysis. The micromorphology and chemical composition of fresh and altered basaltic glass fragments was determined with a SEM JEOL JSM-7610FPlus Energy Dispersive Spectroscopy (EDS) with a 20 kV accelerating voltage and a working distance of 15 mm. Point analysis were made using a focused beam and a counting time of 30 s.

2.3 Chemical analysis

Major element composition and homogeneity of the glass B-OA were analyzed by a Cameca SX100 electron microprobe (see Tab. 1). The Fe redox state of the basaltic glass samples was determined by a wet chemistry method of Wilson, (1960) modified by Schuessler et al. (2008). An advantage of this method is that the total iron is measured on the same solution after reduction of ferric iron with hydroxylamine hydrochloride. Thus, a low analytical error of ±0.02 for Fe(II)/Fetot can be achieved. Values of Fe(II)/Fetot of annealed and quenched glasses agree within error for both the oxidized glasses (0.35 for B-OQ; 0.39 for B-OA) and the reduced glasses (0.78 for B-RQ; 0.80 for B-RA). The redox state of dissolved Fe in the percolate of the pH 2 experiments as well as the total Fe content of the solution were measured spectrophotometrically after complexation with 2,2’bipyridine using the same method. However, it turned out that this approach is only applicable for oxalate-free solutions as the initially formed Fe2+-oxalate complex is very stable hindering complete ligand exchange with 2,2’-bipyridine. This effect was proven by us in tests where oxalic acid was added to FeSO4 solutions before the analysis. On the other hand, it was found that oxalic acid has no influence if added after 2,2’-bipyridine, confirming that this is a kinetic effect. It is noteworthy that the total iron content measured after reduction of the solution with hydroxylamine hydrochloride is not affected by initially present oxalic acid and agrees well with ICP-OES determinations. This indicates that the Fe2+-oxalate complexes are destroyed during the reduction.

Table 1: Chemical composition of the synthetic basaltic glass B-OA and deposits formed in a pH 2 experiment in comparison to the natural basalt chosen as analogue (in wt%).

53.21 ± 0.36

Deposits on B-OQ (EDX) 73.6 ± 3.2

52.07 ± 0.42

2.79 ± 0.02

1.73 ± 0.42

2.81 ± 0.42

12.86 ± 0.21

5.53 ± 1.18

13.05 ± 0.35

Fe2O3

8.22 ± 0.22

-

-

FeO

4.34 ± 0.17

5.18 ± 1.32

11.80 ± 0.30

MnO

0.19 ± 0.07

0.08 ± 0.05

0.18 ± 0.00

6.10 ± 0.12

2.67 ± 0.37

6.26 ± 0.14

CaO

10.63 ± 0.14

6.18 ± 1.29

10.51 ± 0.27

Na2O

2.15 ± 0.07

0.56 ± 0.15

2.23 ± 0.12

K2O

0.44 ± 0.02

0.39 ± 0.05

0.47 ± 0.05

0.26 ± 0.09

1.03 ± 0.18

0.30 ± 0.02

101.19 ± 0.63

96.9 ± 2.0

99.70 ± 0.12

Oxide

B-OA (EMPA)

SiO2 TiO2 Al2O3

MgO

P2O5 Total

Garcia, (1996)

Notes. Composition of the start glass B-OA was measured by EMPA (average of 100 analyses). Deposits on glass surfaces in an experiment with B-OQ for 504 h at B-OQ was analyzed using EDX (average of 10 analyses). For B-OA FeO and Fe2O3 contents were calculated using the total iron content from EMPA and Fe(II)/Fetot determined by the modified Wilson method. For the natural basalt and the deposits all iron is given as FeO. Errors for B-OA and the deposits represent 1 standard deviation.

The pH of all percolates was measured at 25°C immediately after sampling using a VWR SM123 pH gel electrode (uncertainty of < 0.01 pH units). For quantification of Si, Al, Fe, Mn, Mg, Ca, Na, K, and P by Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES; Varian 725-ES) 0.3 M HNO3 was added to the solution. Depending on the element analyzed the detection limit varies between 2 ppb (Mn) up to 30 ppb (K). The measured concentrations, documented in the electronic attachment, were used to calculate element mobilization rates (ri) normalized by the surface area of the glass powder using the following equation (Gudbrandsson et al., 2011): 𝐶𝑖 ∙ 𝐹𝑅

𝑟𝑖 = 𝑆𝑆𝐴

Eq. 1

∙ 𝑚

with ci = measured concentration of element i in the outlet fluid; FR = fluid flow rate; SSA = specific surface area of the glass; m = mass of the glass. This equation was originally used to characterize the dissolution rates of glasses. However, in our experiments the measured concentration in the percolate represents only the mobilized elements, i.e. elements remaining in the columns as residuals of glasses (i.e. leached layers) or precipitates are not included. As proposed by (Conradt, 2008; Stockmann et al., 2011), in the calculation we have used the specific surface area based on grain geometry (SSAGEO) and not the surface area measured with the BET method. The BET surface area was measured and varied between 1.1 and 1.2 m²/g for the fresh ground, unreacted glass powders with no substantial variations between the different glasses. However, these values are strongly influenced by fine-grained components (see Fig. 2a) which are washed out with the solution in the initial phase of the experiment. Thus, the SSA is overestimated by the BET method in late stages of the experiment. According to WolffBoenisch et al. (2004) SSAGEO can be calculated as: 𝑆𝑆𝐴𝐺𝐸𝑂 =

6 𝜌 ∙ 𝑑𝑒𝑓𝑓

Eq. 2

The factor of 6 results from the assumption that grains have a regular and smooth spherical shape, ρ is the glass density (measured to be 2.93 ± 0.06 g/cm³ using the buoyancy method), and deff is the effective particle diameter calculated according to (Tester et al., (1993):

𝑑𝑒𝑓𝑓 =

𝑑𝑚𝑎𝑥 ― 𝑑𝑚𝑖𝑛 𝑑𝑚𝑎𝑥

Eq. 3

𝑙𝑛 𝑑

𝑚𝑖𝑛

where dmax and dmin represent the maximum and minimum particle sizes determined by the mesh size of the applied sieves (50 and 25 µm, respectively). Using this approach SSAGEO equals to 0.062 m2/g for all trials.

3. Results 3.1 Micromorphology of glass surfaces SEM images of starting materials and products of the percolation experiments at pH 5-7 and pH 2 do not differ significantly for the four glasses under investigation and, thus, only images of B-OQ are shown in Fig. 2. The starting materials mainly consisted of particles in the range of 25-50 µm. However, a considerable amount of very fine particles (< 3 µm) adhere to the surfaces of larger grains (Fig.2a). These particles were nearly absent after reaction for 504 h at pH 5-7 (Fig.2b). Turbidity in the percolated solution indicates that these particles were washed out at the beginning of the experiment. Particle size distribution was again very inhomogeneous after the experiment for 388 h at pH 2, with a substantial proportion of fine particles (< 10 µm) (Fig.2c). In this experiment, a large fraction of glass has already reacted, and we suspect that the fine particles originated from the decomposition of leached layers during the preparation of the samples. This is supported by remnants of a nm thick leached layer observed on some glass surfaces (Fig.2g) as well as conglomerates of smaller particles adhering to surfaces of larger grains (Fig.2h). EDX analyses of the deposits are included in Table 1. Strong enrichment of Si and P is evident by comparison to the starting glass. However, significant amounts of all other major and minor elements were also measured by EDX analyses, and it is likely that submicroscopic glass particles in the deposits contribute to the analyses. Thus, the composition of the deposits is not accessible by EDX analyses. Formation of Si-rich leached layers during dissolution of basaltic glass under acidic conditions was also observed in batch experiments of Eick et al. (1996). A closer look on the surfaces of unaltered (Fig.2d) and altered (Fig.2e-f) glasses revealed some notable differences. Surfaces of glass grains, regardless of their Fe redox state and thermal history, showed conchoidal fractures with sharp edges that are typical for glassy materials (Fig.2d; red arrow). During alteration at pH 5-7, the edges of these conchoidal fractures became

rounded and fractures appeared to be deepened (Fig.6e; red arrow). After decreasing pH from 5 to 2, the steps of the fractures became flat and almost disappeared (Fig.2f).

Figure 2: Scanning electron micrographs of the B-OQ starting material (a,d) and products of percolation experiments at pH 5-7 for 504 h (b,e) and subsequently at pH 2 for 388 h (c,f). Details of grain surfaces are shown on bottom with red arrows pointing to characteristic edges. See text for details. Surfaces of products at pH 2 showed nm thick, light shining leached layer (g) and conglomerate of small particles enriched in SiO2 adhering to the surfaces (h).

3.1 pH values of percolates At nominal pH 5, during the first 24 h the pH increased by ~1.5 and ~1.2 for UPW and oxalic acid bearing solutions, respectively (Fig. 3a,b). Afterwards, the pH slowly dropped to pH ~6.2 and ~5.7, respectively, before increasing again to the final value of ~6.7 (Table 2) which was stable for more than 400 h. The high pH is clearly a result of interaction of the solutions with glass powder since in the control experiments (blank) much lower pH was measured (dasheddotted lines in Fig. 3a,b). To account for the pH increase these experiments are referred to as pH5-7 experiments.

A sharp initial increase of pH was observed also in the subsequent experiments at nominal pH of 2, but the increase by ~0.2 units was much lower (Fig.3c,d). Between 7 h and 24 h the pH rapidly dropped back to its initial value of pH 2.0 (Fig.3c) and pH 2.1 (Fig.3d), respectively. For UPW solution, the control experiment showed only a very weak initial increase by 0.05 units and then followed the trend of the trials containing basaltic glass samples. With the addition of oxalic acid, the control experiment showed the same trend as the trials containing basaltic glass (Fig.3d).

Figure 3: Development of the pH value over time for the percolation experiments at pH 5-7 without (a) and with (b) 1 mM oxalate and at pH 2 without (c) and with (d) 1 mM oxalate. Initial pH values of the solutions have been: 5.5 (a) and 5.2 (b), 2.0 (c), 2.1 (d). The dashed-dotted lines represent the control experiments (blank). Lines are only guides for the eyes. Error bars are calculated by error propagation. If not visible, error bars are smaller than symbols.

3.2 Initial and steady state Si and Al release rates Significant variations within the experiments run in triplicates were only observed during the initial dissolution and occurred most probably due to high amounts of fine particles at the beginning of the experiments (< 48h). An uncertainty of 25% of the rates during the initial

experimental stage (< 48h) as postulated by Fournier et al. (2016) is also realistic for our initial dissolution rates. Much better reproducibility is achieved at steady state and variations between the triplicates are <10%.

Figure 4: Development of release rates of Si at pH 5-7 (top) and Al at pH 2(bottom) when using deionized H2O (a,c) and 1 mM oxalic acid (b,d). Grey areas indicate conditions close to steady state. Solid lines are only for guidance of the eye. Error bars are calculated by error propagation. If not visible, error bars are smaller than symbols.

In the percolation experiment at pH 5-7 Si, Ca, Mg and Na were clearly detectable in the percolates during the entire runtime. In contrary, K was close to the ICP-OES detection limit of ~20 µg l-1after 72 h, and Al and Fe were only measurable in the solution sampled after 24 h. Thus, to monitor the progress of glass dissolution for the pH 5-7 series we have used the Si content of the percolates (Fig. 4a,b). Glasses with different Fe redox state and thermal history showed similar trends of Si release rates. Within the first three days a strong decrease of Si

mobilization occurs followed by a period of smooth decrease before a steady state is approached. In the case of experiments with UPW, steady state is not well established but is based only on identical Si concentrations in the percolate after 432 h and 504 h. When oxalic acid was added, steady state was reached earlier at slightly higher Si release rate. In the system with UPW it was also possible to determine steady state release rates for Al, whose values are 40% lower than the Si release rates (Table 2). Table 2: Experimental conditions and steady state Si and Al release rates for percolation experiments at pH 5-7 and 2. Sample names indicate their Fe redox state (O = oxidized; R = reduced), their thermal history (Q = quenched; A = annealed) and the experimental conditions. UPW refers to ultra-pure water, OA to 1 mM oxalic acid.

Sample

Fe(II)/Fetot

Solution

Initial pH

Steady state

Flow rate

pH

[ml/h]

Steady state Al and Si release rates Log rSi

Log rAl

[mol/m2/s]

[mol/m2/s]

B-OQ-3

0.35±0.02

UPW

5.5

6.7

0.69

-9.92

-

B-RQ-3

0.78±0.02

UPW

5.5

6.8

0.66

-9.89

-

B-OA-3

0.39±0.02

UPW

5.5

6.8

0.69

-9.88

-

0.71

-9.92

-

B-RA-3

0.80±0.02

UPW

5.5

6.8

B-OQ-4

0.35±0.02

OA

5.3

6.8

0.68

-9.75

-9.95

B-RQ-4

0.78±0.02

OA

5.3

6.7

0.68

-9.80

-9.98

B-OA-4

0.39±0.02

OA

5.3

6.6

0.70

-9.78

-10.01

0.63

-9.78

-10.02

B-RA-4

0.80±0.02

OA

5.3

6.8

B-OQ-1

0.35±0.02

UPW

2.0

2.0

1.29

-9.50

-9.28

B-RQ-1

0.78±0.02

UPW

2.0

2.0

1.26

-9.66

-9.43

B-OA-1

0.39±0.02

UPW

2.0

2.0

1.33

-9.48

-9.28

1.37

-9.68

-9.48

B-RA-1

0.80±0.02

UPW

2.0

2.0

B-OQ-2

0.35±0.02

OA

2.1

2.1

1.20

-

-9.30

B-RQ-2

0.78±0.02

OA

2.1

2.1

1.27

-

-9.43

B-OA-2

0.39±0.02

OA

2.1

2.1

1.23

-

-9.39

2.1

2.1

1.14

-

-9.65

B-RA-2

0.80±0.02

OA

In contrary to the pH 5-7 series, in the pH 2 series dissolved Si was measurable only in the oxalate-free system but dropped below the concentration of the control experiment when 1 mM oxalic acid was added to the solution. Because of the low release rates of Si (due to the formation of Si-rich leached layers, see Fig. 1c,g), we have used the Al content to monitor the progress of glass dissolution and element release for the pH 2 experiments (Fig. 4c,d). When using oxalate-free solutions, initial Al dissolution rates were the same for all glasses and dropped within the first 48 h by 1.2 log units (Fig. 4a). At this point Al release rates for oxidized and reduced glasses started to deviate. Approximately constant release rates were achieved after t ≥ 240 h, with rates for oxidized glasses being 60% higher than those for reduced glasses. These

long-term release rates marked by grey areas in Fig. 4 are assumed to represent conditions close to steady state dissolution. In the presence of 1 mM oxalic acid at pH 2 similar initial Al dissolution rates were observed (Fig. 4b). After 48 h Al release rates of annealed glasses decreased faster than their quenched analogues. Compared to quenched glasses, steady state Al release rates of oxidized and reduced annealed glasses were 20 and 40% lower, respectively. As observed in the pH 5-7 series, steady state is reached earlier in the systems containing oxalic acid. For the UPW experiments steady state release rates could be measured also for Si, whose values are 40% lower than those of Al (Table 2). This finding is consistent with formation of large amount of Si-rich leached layers. 3.4 Cumulative element mobilization To evaluate specific retainment of elements in leached layers or precipitates, it is useful to compare the total amount of mobilized elements normalized to its bulk concentration in the glass (Fig. 5). In the pH 5-7 experiments using UPW the release behavior is similar for the four different glasses (Fig.5a), with element mobilization generally decreasing in the order K > Ca > Mn > P > Si≈Mg≈Na > Fe > Al. A slight trend of higher mobilization of all elements from annealed glasses is visible, however, this is within experimental uncertainty. With the addition of 1 mM oxalic acid, the mobilization of di- and trivalent cations slightly increased (Fig.5b). This is most pronounced for quenched glasses and for Mn with an increase by a factor of 2. The amounts of mobilized elements generally decreased in the order Mn > K > Ca≈Mg≈Fe > Al≈Na≈P≈Si. Furthermore, the observed trend for oxalate-free solutions seemed to be reversed, and quenched glass showed a higher mobilization for all measured elements.

Figure 5: Cumulative mobilization of major elements from oxides in basaltic glasses normalized to their content in the original glass at pH 5-7 and 2 in deionized H2O (a) and with addition of 1 mM oxalic acid (b). Solid lines are only for guidance of the eye. Error bars are calculated by error propagation. If not visible, error bars are smaller than symbols.

Although the total experimental duration was shorter at pH 2, the cumulative mobilization of all di- and trivalent cations and P was at least five times higher compared to pH 5-7 (Fig.5a). This is most pronounced for Fe and Mn for which the concentrations increased 15 and 10 times, respectively. The amounts of mobilized elements generally decreased in the order Fe≈Mn > Ca > Al≈P > Mg > K > Na≈Si. With the addition of 1 mM oxalic acid the mobilization of Ca, Mn

and Na increased compared to oxalate-free conditions whereas P became slightly less mobile (Fig.5b). The general mobilization trend is Mn≈Fe≈Ca > Al≈Mg > Na≈P > K > Si. Effects of the redox state of Fe and thermal history of the glass are relatively weak also at pH 2. The oxidized glasses showed a higher mobilization of all di- and trivalent cations compared to the reduced glasses. Furthermore, in presence of oxalic acid release of elements seems to be enhanced for quenched glasses compared to annealed glasses, similar as indicated for the pH 5-7 experiments (Fig.5b). 3.5. Iron redox state in solution

Figure 6: Redox state of mobilized Fe in aqueous solutions at pH 2 without oxalic acid. Solid and dashed lines show the Fe(II)/Fetot ratio of the quenched (B-OQ; B-RQ) and annealed (B-OA; B-RA) basaltic glass, respectively. Semitransparent blue and red areas indicate the standard deviation. Error bars are calculated by error propagation. If not visible, error bars are smaller than symbols.

The Fe content in the percolated solution was high enough for iron redox determination only in the series at pH 2. Under oxalate-free conditions, the initial Fe(II)/Fetot ratio in solutions containing reduced glasses was ~0.6 and thus much lower than in the fresh basaltic glass (Fe(II)/Fetot ~0.8) (Fig.6a). Solutions containing oxidized glasses showed the same trend (Fe(II)/Fetot ~0.25) but with less difference to the initial Fe redox state of the glass (Fe(II)/Fetot ~0.4). After this initial period, the Fe(II)/Fetot ratio of the solutions slowly increased and reached that of the original basaltic glass after ~300 h and ~250 h for reduced and oxidized glasses, respectively.

4. Discussion

4.1 Mobilization of Fe(II) and Fe(III) from the dissolving basaltic glasses Release rates of Fe are close to the detection limits in the pH 5-7 experiments and much lower than those of Si. This strongly indicates that iron is incorporated in precipitates remaining in the assembly under these conditions. A possible reaction is 𝐹𝑒3 + +2𝐻2𝑂 →𝐹𝑒𝑂𝑂𝐻 + 3𝐻 +

Eq. 4

Indirect evidence for this reaction is given by the low Fe(II)/Fetot ratio and the sharp increase of pH at the beginning of pH 2 experiments, in which the products of the pH 5-7 experiments were used as starting material. The release of Fe(III) from the precipitates dominates over Fe released by glass dissolution in the initial state of the pH 2 experiments. When steady state is reached (> 250 h) the mobilized Fe in the pH 2 experiments corresponds to the Fe released by glass dissolution. The good agreement between Fe(II)/Fetot of the percolated solution and the basaltic glass for both the oxidized and the reduced in case of the UPW (+HCl) solution supports this thesis of an efficient mobilization of iron. 4.2 Influence of the Fe(II)/Fetot ratio on basaltic glass dissolution Considering the different structural roles of Fe(II) and Fe(III) in silicate glasses an effect of the Fe redox state on the dissolution behavior of the other major cations in the glasses such as Na, K, Mg, Ca and Al is expected. Fig.7 compares the cumulative mobilization of major elements (denoted as X) of oxidized and reduced glasses for quenched (XB-OQ/XB-RQ) and annealed (XBOA/XB-RA)

samples at pH 2 (UPW + HCl). Values greater than 1 indicate a higher mobilization

from Fe(III)-rich glasses. Data of K and P show large unsystematic large scatter and are not considered here. The comparison of element mobilization from quenched and annealed glasses can be divided in two parts:

Figure 7: Ratios of mobilized Si, Al, Fe, Ca, Mn, Mg and Na from oxidized and reduced glasses at pH 2 adjusted by addition of 1M HCl. (a) Quenched glasses. (b) Annealed glasses. B-OQ = Oxidized-Quenched; B-RQ = Reduced-Quenched; B-OA = Oxidized-Annealed; B-RA = Reduced-Annealed. The dotted line corresponds to mobilization independent of the Fe redox state. Solid lines are only for guidance of the eye.

(i)

the initial period, which might be affected by dissolution of precipitates stemming from the previous pH 5-7 experiments. Here, di- and trivalent cations (±Al) were preferentially mobilized when using Fe(II)-rich glasses. By contrast, Si and Na were mobilized in higher quantities in the case of Fe(III)-rich glasses.

(ii)

long term experiments (> 100h (Fig. 7a) and 275h (Fig 7b)). All major elements were mobilized in greater quantities from Fe(III)-rich glasses for both quenched and annealed samples than expected from stoichiometric composition. The mobilization of Si and Na from Fe(III)-rich quenched glasses steadily increased over the course of the experiment, but for Na in annealed glasses, the difference was diminished

after 100 h and Na was mobilized in the same proportion regardless of the Fe redox state. The point at which the element mobilization from Fe(III)-rich glasses exceeded that of Fe(II)rich glasses was delayed by ~175 h for annealed glasses compared to quenched glasses. This delay is probably related to structural differences between quenched and annealed glasses, i.e. denser packing in the quenched glasses. The higher mobilization of all major cations from oxidized glasses at timescales > 300 h can be explained by the structural role of Fe(III). X ray absorption near edge structure (XANES) spectroscopy indicate that Fe(III) is in average 5-coordinated by oxygen in basaltic glasses, but it cannot be distinguished whether this results from a mixture of 4- and 6-fold coordination or represents really a 5-fold coordination (Wilke et al. 2004). Nevertheless, it is undisputed that Fe(III) acts as a network former in silicate glasses with similar role as Al, i.e., linking SiO4 tetrahedra by the formation of T-O-Si bonds (T = Al, Fe(III)) (Virgo and Mysen, 1985). Hamilton et al. (2001) showed that the durability of aluminosilicate glasses at acidic and basic pH decreases with increasing Al/Si from albite to jadeite to nepheline glasses due to an increasing abundance of Al-O-Si linkages per SiO4 tetrahedron and lengthening of this bond. This is also supported most recently by the work of Perez et al., (2019) on the dissolution of oligoclase crystals and glass at acidic and basic pH. Protonation of T-O-Si bonds (T = Al, Fe(III)) is faster than for Si-O-Si bonds, and the removal of adjoining TO4- tetrahedra from the glass leads to the formation of Si-OH bonds and, thus, partly liberation of SiO4 tetrahedra (Oelkers, 2001). The reactivity of these partially liberated SiO4 tetrahedral units increases with further detachment of TO4- units (Gautier et al., 2001). With respect to our glass system this would imply that the glass network becomes increasingly vulnerable with increasing Fe(III) contents. The more Fe(III)-O-Si connections are present, the faster is the glass dissolution. Furthermore, this reaction is more effective at high concentrations of H+ (e.g. pH 2). The higher steady state Fe mobilization at pH 2 from Fe(III) dominated compared to Fe(II) dominated glasses with ~3.5·10-11 mol/m²/s and ~2.3·10-11 mol/m²/s, respectively, support this assumption. 4.3. Impact of thermal history on basaltic glass dissolution The combined effects of oxalate and thermal history on element mobilization are displayed in Fig.8. Basaltic glasses altered at pH 5-7 for 504 h have lost in total 0.24 to 0.40% of their cations (Fig.8a). The effect of 1 mM oxalic acid at pH 5-7 on element mobilization was negligible for annealed glasses whereas quenched glasses showed slightly enhanced element release in the

presence of oxalate with the oxidized glass being more affected. Lowering the pH to 2, element release is enhanced by more than a factor of 4, whereby the reaction is more pronounced for oxidized glasses compared to reduced glasses with absolute values of ~1.83±0.06% and 1.55±0.08%, respectively (Fig.8b). The addition of 1 mM oxalic acid further enhances the element mobilization. This effect is particularly strong for the quenched glasses, e.g. the total loss of cations increased for reduced glasses from 1.80±0.04% (annealed) to 2.01±0.06% (quenched) and for reduced glasses from 1.86±0.08% (annealed) to 2.32±0.07% (quenched).

Figure 8: Coupled influence of oxalate and thermal history on total release of cations at pH 5-7 (a) and pH 2 (b). Solid and dashed lines are only for guidance of the eye. For pH 5-7 error bars are within the symbol. Error bars are calculated by error propagation. If not visible, error bars are smaller than symbols.

Dissolution of crystals and glasses occurs preferentially on surface sites with abundant defects (kinks/steps) or edges that are more reactive for protonation and especially adsorption of organic complexes (Bunker, 1994; Cama and Ganor, 2006; Briese et al., 2017). Adsorption of organic complexes on reactive surfaces sites is involved in weakening and breaking of M-O framework bonds (Johnson et al., 2004). Surfaces of basaltic glasses used within this study showed conchoidal fractures regardless of their thermal history (Fig.2d). Assuming that quenched glasses comprise higher surface fracture densities (Chemtob and Rossman, 2014), it is expected that even in the absence of oxalate, protonation of these reactive surface sites and therefore dissolution rates should be faster for quenched than for annealed glasses. However, we observed that dissolution rates of quenched and annealed glasses at pH 2 in oxalate-free solutions were the same and conclude that the higher bulk glass dissolution in the presence of oxalate cannot be caused by a higher abundance of surface sites alone.

We suggest that the mechanism by which oxalate enhanced element mobilization from quenched glasses is related to the glass structure, i.e., a more open structure of the rapidly quenched glasses. This is also supported by faster reaction rates of quenched glasses observed in Fig.7. Whether the higher mobilization of divalent cations from quenched glasses is caused by strained bonds as suggested by some authors (Bunker, 1994; Pantano, 2001) is not clear and needs further investigations. 4.4 Leached layer formation and dissolution mechanism at pH 2 In percolation experiments at pH 2 all elements were preferentially mobilized with respect to Si. This was expected since the solubility of Si in aqueous solutions at pH 2 is very low and formation of a Si-rich leached layer is preferred. The fact that we did not observe extensive leached layer formation by SEM (Fig.2) is not in contradiction to this thesis. Intensive mechanical treatment during separation of the basaltic powder from the quartz sand most likely destroys the fragile connections between leached layer and glass grains. This interpretation is consistent with findings of Gislason and Oelkers (2003) who observed grazing and partial detachment of an altered layer from the surface of glassy ash particles after leaching at pH < 3. They suggested that this could be an artifact of grain drying. Similar observations were also made on surfaces of synthetic basaltic glasses (Eick et al., 1996). Further evidence for leached layer formation at pH 2 can be obtained when comparing the content of different elements in the percolate normalized to their contents in the glass (Xaq/Xs) (Fig. 9). Monovalent cations (Na, K) and P showed large scatter and are not considered. For all other elements a strong correlation with released Al is evident over the whole experimental time. Normalized concentrations in solution decrease in the order Mn > Fe > Ca > Mg > Si. Compared to aluminum, Mg and in particular Si are strongly retained, while Fe and Mg are more mobile. It should be noted that the strong mobilization of iron does not only occur at the beginning of the pH 2 run, when iron-rich precipitates from the pH 5-7 experiment are dissolved, but also when steady state is reached. An explanation for relatively high contents of iron and manganese is the high concentration of complexing agents in solution. The concentration of chlorine was 1 - 2 orders of magnitude higher than that of iron even at the beginning of the pH 2 run, explaining also the relatively small effect of oxalic acid. The EDX analyses in Table 1 do not give reliable information about the composition of the leached layers, as glass fragments probably contribute to the analyses. If one assumes that only a negligible proportion of the iron is retained in the trials under steady state conditions at pH 2, the composition of the leached layers can be estimated based on the Xaq/Xs values in Fig. 9.

Following this approach, the leacher layers contain more than 80 wt% SiO2. Minor components are MgO, Al2O3 and, in the case of the pH 2 experiments with 1 mM oxalic acid also some CaO. Considering the EDX data, phosphorus is also enriched in the leached layers and can reach the wt% level.

Figure 9: Comparison of contents of Mg, Ca, Mn, Fe and Si with Al for quenched oxidized glass B-OQ at pH 2. Data were normalized with the element content of the glass. The dashed line represent the X/Al ratio expected if an element X is mobilized by the same rate as Al. The slope m and its standard error (in bracket) were determined by linear regression. Note that high values correspond to the initial stage of glass dissolution and small values to the final state. Error bars are calculated by error propagation. If not visible, error bars are smaller than symbols.

From the slopes of the linear regressions in Fig.9 it can be deduced that the mobilization of divalent cations (Ca, Mg, Mn) as well as Fe is increased relative to Al at pH 2 when oxalic acid is added. On the other hand, the steady state Al release rates from quenched oxidized glasses were not affected by oxalic acid (Fig.3a-b; Table2), and the Si mobilization seems to have been even lowered. Surprisingly, the annealed oxidized glasses showed a noticeable decrease in the steady state Al release rates with Δlog rAl [mol/m²/s] = 0.11 to 0.17 upon addition of oxalate (Fig.3a-b; Table 2). So far, we cannot reasonably explain the inhibition of Al release from annealed glasses in the presence of oxalate at pH 2. Precipitation of Al oxalate is not expected at this low pH (Panias and Paspaliaris, 2001). It was shown by some authors that organic anions inhibit mineral dissolution by blocking active surface sites thereby hindering other dissolution enhancing species (e.g. H+) to interact with these sites (Johnson et al., 2004, 2005). On the other hand, the promoting effect of oxalate on dissolution of quenched basaltic glasses is well studied and accepted (Eick et al., 1996; Oelkers, 2001; Oelkers and Gislason, 2001; Perez et al., 2015; Wolff-Boenisch et al., 2011). 4.5 Glass dissolution rates in the percolation experiments at steady state If one assumes that no silicon is retained in the trial, the glass dissolution rates rglass can be calculated for the pH 5-7 experiments at steady state conditions using the Si release rate rSi in Table 2 as 𝑟𝑆𝑖 ∙ 𝑀𝑆𝑖 ∙ 100

𝑟𝑔𝑙𝑎𝑠𝑠 = (𝑤𝑡% 𝑆𝑖)𝑔𝑙𝑎𝑠𝑠 ∙

𝜌𝑔𝑙𝑎𝑠𝑠

Eq. 5

where MSi is the molar mass of Si and (wt%Si)glass corresponds to the Si content of the glass based on Table 1. For the oxalate-free system the steady state glass dissolution rates is 0.017 nm/h, independent on glass history and redox state of iron. Addition of oxalic acid increases the dissolution rates to 0.24 nm/h for all glasses. Based on reproducibility the error of glass dissolution rates is estimated to be about 10%. As noted above, Si release rates are not suitable to estimate steady state glass dissolution rates at pH 2 due to the formation of Si-rich deposits, but Fe release rates (rFe) may be used to estimate minimum glass dissolution rates. It should be mentioned that rFe follow the trend shown for rAl in Fig. 4c,d with only slightly smaller values compared to Al (raw data are given in the electronic attachment). Compared to the pH 5-7 experiments the dissolution rates are about one order of magnitude higher at pH 2. Consistent with the discussion above, oxidized glasses dissolve 40-50% faster than reduced glasses under these conditions while the thermal history has minor effect. Dissolution rates without oxalate are 0.269 nm/h for B-OQ, 0.192 nm/h for

B-RQ, 0.259 for B-OA and 0.174 nm/h B-RA. Oxalic acid has an unsystematic effect on steady state dissolution rates at pH 2, and variations to the oxalate-free system are within the experimental error. When using Al release rates, the calculated dissolution rates are 7% smaller for the oxalate-free system and 18% smaller for the system with oxalic acid, consistent with partial retainment of Al in deposits. The estimated dissolution rates can be used to predict long-term dissolution of basaltic glasses. If fresh solution is provided continuously, about 2 mm of glass is dissolved within 1000 years when neutral solutions are percolating alongside the glass, but about 20 mm when the solution is acidic with pH 2.

5. Conclusions Glass dissolution and mobilization rates of elements in percolation experiment strongly depend on experimental parameters such as the setup of the assembly, the grain size and the amount of glass powder as well as the flow rate and composition of the solution. Nevertheless, important general information can be derived from such experiments by comparing results for different structural states of the glasses (reduced ↔ oxidized; quenched ↔ annealed) and different solutions (variation of pH, addition of oxalic acid). Furthermore, comparison of element concentrations in the percolate allows conclusions about specific retainment of elements in precipitates or residuals. The experimental results of this study demonstrate that basaltic glass dissolution is affected by the Fe redox state and moreover by the thermal history of the glasses. We suggest that both, the Fe redox state and the thermal history, should to be considered for the prediction of the longterm stability of basaltic glasses. Moreover, the thermal history might be a parameter of great importance in microbial alteration as observed in a previous study (Stranghöner et al. 2018), but here more experimental studies are needed to constrain this parameter in modeling of glass dissolution and prediction of durability of glasses. Acknowledgements The project was funded by the German Science Foundation (DFG), priority program ICDP, grant BE 1720/39-1. References Amma S., Luo J., Kim S. H. and Pantano C. G. (2017) Effects of fictive temperature on the

leaching of soda lime silica glass surfaces. J. Am. Ceram. Soc. 100, 1424–1431. Angeli F., Charpentier T., Jollivet P., de Ligny D., Bergler M., Veber A., Gin S. and Li H. (2018) Effect of thermally induced structural disorder on the chemical durability of International Simple Glass. Mater. Degrad. 31, 1–11. Available at: http://dx.doi.org/10.1038/s41529-018-0052-3. Boyd P. W. and Ellwood M. J. (2010) The biogeochemical cycle of iron in the ocean. Nat. Geosci. 3, 675–682. Available at: http://dx.doi.org/10.1038/ngeo964. Briese L., Arvidson R. S. and Luttge A. (2017) The effect of crystal size variation on the rate of dissolution – A kinetic Monte Carlo study. Geochim. Cosmochim. Acta 212, 167–175. Available at: http://dx.doi.org/10.1016/j.gca.2017.06.010. Bunker B. C. (1994) Molecular mechanisms for corrosion of silica and silicate glasses. J. Non-Cryst. Solids 179, 300–308. Cama J. and Ganor J. (2006) The effects of organic acids on the dissolution of silicate minerals : A case study of oxalate catalysis of kaolinite dissolution. Geochim. Cosmochim. Acta 70, 2191–2209. Chemtob S. M. and Rossman G. R. (2014) Timescales and mechanisms of formation of amorphous silica coatings on fresh basalts at Kīlauea Volcano, Hawai’i. J. Volcanol. Geotherm. Res. 286, 41–54. Available at: http://dx.doi.org/10.1016/j.jvolgeores.2014.08.029. Christie D. M., Carmichael I. S. E. and Langmuir C. H. (1986) Oxidation states of mid-ocean ridge basalt glasses. Earth Planet. Sci. Lett. 79, 397–411. Conradt R. (2008) Chemical Durability of Oxide Glasses in Aqueous Solutions: A Review. J. Am. Ceram. Soc. 735, 728–735. Cooper R. F., Fanselow J. B. and Poker D. B. (1996) The mechanism of oxidation of a basaltic glass: Chemical diffusion of network-modifying cations. Geochim. Cosmochim. Acta 60, 3253–3265. Drever J. I. and Vance G. F. (1994) Chapter 6 Role of Soil Organic Acids in Mineral Weathering Processes. In Organic Acids in Geological Processes (ed. E. D. Pittman). Springer. pp. 138–161. Ducasse T., Gourgiotis A., Pringle E., Moynier F., Frugier P., Jollivet P. and Gin S. (2018)

Alteration of synthetic basaltic glass in silica saturated conditions : Analogy with nuclear glass. Appl. Geochemistry 97, 19–31. Available at: https://doi.org/10.1016/j.apgeochem.2018.08.001. Eick M.J., Grossl P. R., Golden D. C., Sparks D. L. and Ming D. W. (1996) Dissolution kinetics of a lunar glass simulant at 25°C: The effect of pH and organic acids. Geochim. Cosmochim. Acta 60, 157–170. Fournier M., Ull A., Nicoleau E., Inagaki Y., Odorico M., Frugier P., Gin S. (2016) Glass dissolution rate measurment and calculation revisited. J. Nuc. Mat. 467, 140–154 Furrer G. and Stumm W. (1986) The coordination chemistry of weathering : I . Dissolution kinetics of Al2O3 and BeO. Geochim. Cosmochim. Acta 50, 1847–1860. Garcia M. O. (1996) Petrography and olivine and glass chemistry of lavas from the Hawaii Scientific Drilling Project. J. Geophys. Res. 101, 11701–11713. Gautier J.-M., Oelkers E. H. and Schott J. (2001) Are quartz dissolution rates proportional to B.E.T. surface areas ? Geochim. Cosmochim. Acta 65, 1059–1070. Gislason S. R. and Oelkers E. H. (2003) Mechanism, rates, and consequences of basaltic glass dissolution: II. An experimental study of the dissolution rates of basaltic glass as a function of pH and temperature. Geochim. Cosmochim. Acta 67, 3817–3832. Goudie A. S. and Viles H. A. (2012) Weathering and the global carbon cycle: Geomorphological perspectives. Earth-Science Rev. 113, 59–71. Available at: http://dx.doi.org/10.1016/j.earscirev.2012.03.005. Gysi A. P. and Stefánsson A. (2012) CO2-water-basalt interaction. Low temperature experiments and implications for CO2 sequestration into basalts. Geochim. Cosmochim. Acta 81, 129–152. Hamilton J. P., Brantley S. L., Pantano C. G., Criscenti L. J. and Kubicki J. D. (2001) Dissolution of nepheline , jadeite and albite glasses : Toward better models for aluminosilicate dissolution. Geochim. Cosmochim. Acta 65, 3683–3702. Hausrath E. M., Neaman A. and Brantley S. L. (2009) Elemental release rates from dissolving basalt and granite with and without organic ligands. Am. J. Sci. 309, 633–660. Johnson S. B., Yoon T. H. and Brown G. E. (2005) Adsorption of organic matter at mineral/water interfaces: 5. Effects of adsorbed natural organic matter analogues on

mineral dissolution. Langmuir 21, 2811–2821. Johnson Stephen B., Yoon T. H., Kocar B. D. and Brown G. E. (2004) Adsorption of organic matter at mineral/water interfaces. 2. Outer-sphere adsorption of maleate and implications for dissolution processes. Langmuir 20, 4996–5006. Johnson Stephen B., Yoon T. H., Slowey A. J. and Brown G. E. (2004) Adsorption of organic matter at mineral / water interfaces. 3. Implications of surface dissolution for adsorption of oxalate. Langmuir 20, 11480–11492. Kumar, B. and Lin, S. (1991) Redox state of iron and its related effects in the CaO-P2O5-Fe2O3 glasses. J. Am. Ceram. Soc. 74, 226-228. Libourel G., Verney-carron A., Morlok A., Michelin A., Neff D. and Dillmann P. (2011) The use of natural and archeological analogues for understanding the long-term behavior of nuclear glasses. Comptes Rendus - Geosci. 343, 237–245. Van Der Merwe P., Bowie A. R., Quéroue F., Armand L., Blain S., Chever F., Davies D., Dehairs F., Planchon F., Sarthou G., Townsend A. T. and Trull T. W. (2015) Sourcing the iron in the naturally fertilised bloom around the Kerguelen Plateau: Particulate trace metal dynamics. Biogeosciences 12, 739–755. Michelin A., Burger E., Rebiscoul D., Bruguier F., Drouet E., Dillmann P. and Gin S. (2013) Silicate Glass Alteration Enhanced by Iron: Origin and Long-Term Implications. Environ. Sci. Technol. 47, 750–756. Narayanaswamy O. S. and Gardon R. (1969) Calculation of Residual Stresses in Glass. J. Am. Ceram. Soc. 52, 554–558. Oelkers E. H. (2001) General kinetic description of multioxide silicate mineral and glass dissolution. Geochim. Cosmochim. Acta 65, 3703–3719. Oelkers E. H. and Gislason S. R. (2001) The mechanism, rates and consequences of basaltic glass dissolution: I. An experimental study of the dissolution rates of basaltic glass as a function of aqueous Al, Si and oxalic acid concentration at 25°C and pH 3 ‫ ؍‬and 11. Geochim. Cosmochim. Acta 65, 3671–3681. Panias D. and Paspaliaris I. (2001) Calculated Solubility of Trivalent Iron and Aluminum in Oxalic Acid Solutions at 25 ° C . Can. Metall. Q. 40, 421–432. Pantano C. G. (2001) What do we know about glass surfaces? ed. C. H. Drummond, The

American Ceramic Society. Parruzot B., Jollivet P., Rébiscoul D. and Gin S. (2015) Long-term alteration of basaltic glass: Mechanisms and rates. Geochim. Cosmochim. Acta 154, 28–48. Pederson L. R., Buckwalter C. Q. and Mcvay G. L. (1983) The Effects of Surface Area to Solution Volume on Waste Glass Leaching. Nucl. Technol. 62, 151–158. Perez A., Daval D., Fournier M., Vital M., Delaye J.-M. and Gin S. (2019) Comparing the reactivity of glasses with their crystalline equivalents : The case study of plagioclase feldspar. Geochim. Cosmochim. Acta 254, 122–141. Perez A., Rossano S., Trcera N., Verney-Carron A., Huguenot D., van Hullebusch E. D., Catillon G., Razafitianamaharavo A. and Guyot F. (2015) Impact of iron chelators on short-term dissolution of basaltic glass. Geochim. Cosmochim. Acta 162, 83–98. Rosso K. M., Zachara J. M., Fredrickson J. K., Gorby Y. A. and Smith S. C. (2003) Nonlocal bacterial electron transfer to hematite surfaces. Geochim. Cosmochim. Acta 67, 1081– 1087. Schuessler J. A., Botcharnikov R. E., Behrens H., Misiti V. and Freda C. (2008) Oxidation state of iron in hydrous phono-tephritic melts. Am. Mineral. 93, 1493–1504. Staudigel H. and Hart S. R. (1983) Alteration of basaltic glass: Mechanisms and significance for the oceanic crust-seawater budget. Geochim. Cosmochim. Acta 47, 337–350. Stockmann G. J., Wolff-boenisch D., Gislason S. R. and Oelkers E. H. (2011) Do carbonate precipitates affect dissolution kinetics ? 1 : Basaltic glass. Chem. Geol. 284, 306–316. Available at: http://dx.doi.org/10.1016/j.chemgeo.2011.03.010. Stone-Weiss N., Pierce E. M., Youngman R. E., Gulbiten O., Smith N. J., Du J. and Goel A. (2018) Understanding the structural drivers governing glass – water interactions in borosilicate based model bioactive glasses. Acta Biomater. 65, 436–449. Stranghöner M., Schippers A., Dultz S. and Behrens H. (2018) Experimental microbial alteration and Fe mobilization from basaltic rocks of the ICDP HSDP2 drill core, Hilo, Hawaii. Front. Microbiol. 9, 1252. Available at: https://www.frontiersin.org/articles/10.3389/fmicb.2018.01252/abstract. Stumm W. (1992) Chemistry of the Solid-Water Interface: Processes at the Mineral-Water and Particle-Water Interface in Natural Systems., John Wiley & Sons.

Techer I., Advocat T., Lancelot J. and Liotard J. (2000) Dissolution kinetics of basaltic glasses: control by solution chemistry and protective effect of the alteration film. Chem. Geol. 176, 235–263. Tester J. W., Worley G., Robinson B. A., Grigsby C. O. and Feerer J. L. (1993) Correlating dissolution kinetics in pure water from 25 to 625°C. Geochim. Cosmochim. Acta 58, 2407–2420. Virgo D. and Mysen B. O. (1985) The Structural State of Iron in Oxidized vs. Reduced Glasses at 1 Atm: A 57Fe Mössbauer Study. Phys. Chem. Miner. 12, 65–76. Wilke M., Partzsch G. M., Bernhardt R. and Lattard D. (2004) Determination of the iron oxidation state in basaltic glasses using XANES at the K-edge. Chem. Geol. 213, 71–87. Wilson A. D. (1960) The Micro-determination of Ferrous Iron in Silicate Minerals by a Volumetric and a Colorimetric Method. Analyst 85, 823–827. Wolff-Boenisch D., Gislason S. R., Oelkers E. H. and Putnis C. V. (2004) The dissolution rates of natural glasses as a function of their composition at pH 4 and 10.6, and temperatures from 25 to 74°C. Geochim. Cosmochim. Acta 68, 4843–4858. Wolff-Boenisch D., Wenau S., Gislason S. R. and Oelkers E. H. (2011) Dissolution of basalts and peridotite in seawater, in the presence of ligands, and CO2: Implications for mineral sequestration of carbon dioxide. Geochim. Cosmochim. Acta 75, 5510–5525. Available at: http://dx.doi.org/10.1016/j.gca.2011.07.004. Zhang H. L., Hirschmann M. M., Cottrell E., Newville M. and Lanzirotti A. (2016) Structural environment of iron and accurate determination of Fe3+/ΣFe ratios in andesitic glasses by XANES and Mössbauer spectroscopy. Chem. Geol. 428, 48–58. Available at: http://dx.doi.org/10.1016/j.chemgeo.2016.02.022.

Declaration of interests

☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: