Electrolyte effects on surface chemistry of basaltic glass in the initial stages of dissolution

    Electrolyte effects on surface chemistry of basaltic glass in the initial stages of dissolution Stefan Dultz, Harald Behrens, Gundula Helsch, Joachim Deubener PII: DOI: Reference:

S0009-2541(16)30043-2 doi: 10.1016/j.chemgeo.2016.01.027 CHEMGE 17827

To appear in:

Chemical Geology

Received date: Revised date: Accepted date:

5 November 2015 26 January 2016 28 January 2016

Please cite this article as: Dultz, Stefan, Behrens, Harald, Helsch, Gundula, Deubener, Joachim, Electrolyte effects on surface chemistry of basaltic glass in the initial stages of dissolution, Chemical Geology (2016), doi: 10.1016/j.chemgeo.2016.01.027

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. 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.

ACCEPTED MANUSCRIPT

SC R

IP

T

Manuscript prepared for publication in “Chemical Geology”

Electrolyte effects on surface chemistry of basaltic glass in the

MA

NU

initial stages of dissolution

Institute of Mineralogy, Leibniz Universität Hannover, Callinstr. 3, D-30167 Hannover,

Germany

Institute of Non-Metallic Materials, Clausthal University of Technology, Zehntnerstraße 2a,

AC

b

CE P

a

TE

D

Stefan Dultza*, Harald Behrensa, Gundula Helschb, Joachim Deubenerb

D-38678 Clausthal-Zellerfeld, Germany

*Corresponding author. Tel.: +49 511 762 3671 E-mail address: [email protected] (S. Dultz)

ACCEPTED MANUSCRIPT ABSTRACT For understanding of the effect of solution composition on dissolution rate of basaltic glass

T

detailed knowledge of surface chemistry is important. Here the zeta potential (ζ) as a

IP

characteristic parameter of the magnitude of surface charge at the solid-liquid interface was

SC R

used to determine ionic effects on surface chemistry in initial stages of basaltic glass dissolution. In a systematic approach powdered synthetic basaltic glass was dispersed in solutions of different cations (NO3- salts of Na+, K+, Mg2+, Ca2+, Ba2+, Zn2+, and Al3+) and

NU

anions (Na+ salts of F-, Cl-, I-, NO3-, SO42-, C2O42-, HPO42-), each in concentrations of 0.1, 0.5,

MA

1.0, 2.5, and 5.0 mmol/L. ζ was traced in time sequences up to 12000 h at ideally circumneutral pH. Ion affinities to glass surfaces were characterized by sorption isotherms. Change of glass chemical composition by the formation of altered layers was determined by

TE

D

depth profiling using secondary neutral mass spectrometry (SNMS). Dissolution of the glass was quantified by the amount of Si released after 4000 h.

CE P

A marked decrease of ζ in deionized water within the first 3 h reaction time is assigned to desorption of alkali and alkaline earth metal cations from the glass surface and formation of

AC

negatively charged Si-O- sites. Addition of anions resulted in stronger negative initial ζ values in comparison with the experiment in deionized H2O indicating marked anion adsorption on surface sites, most obvious for F-, C2O42- and HPO42-. The initial ζ was increased upon addition of divalent cations indicating neutralization of negatively charged surface sites. Over time a striking shift from negative to positive ζ was obtained, most markedly for Ca2+ and Zn2+. The addition of trivalent Al3+ resulted directly in positive ζ indicating a strong adsorption on glass surfaces. With progress of the experiment the sign of ζ reversed to negative values again. The reason for charge reversal is not fully understood and might be related with cation adsorption exceeding negative surface charge and concentration of Fe oxides at the glass surface. After ~2000 h reaction time ζ adjusted for most electrolytes additions to slightly negative ζ until the end of the experiment, indicating that a final state in

ACCEPTED MANUSCRIPT the composition of surface sites was reached. The presence of monovalent Na+ and K+ in solution suppressed Si release from the glass, whereas it is accelerated by bivalent cations. It

T

appears that neutralization of deprotonated ≡Si-O- sites by monovalent cations - their

IP

preferential binding is also indicated by chemical analysis - favors polymerization resulting in

SC R

slower Si release. Upon addition of Al3+ it is likely that ≡Si-O-Al-O-Si≡ bonds are formed, which can suppress Si release. The presence of F-, C2O42-, and HPO42- clearly enhance glass dissolution, most probably by increasing coordination of network forming cations, weakening

NU

bonds hereby. The observed generation of positive ζ on basaltic glass surfaces is remarkable,

MA

which can improve in natural systems the adsorption capability of the basaltic glass surface for negatively charged compounds from pore solution, anions, dissolved organic matter and

TE

D

also bacterial cell walls.

Keywords: Basaltic glass / Surface chemistry / Electrolyte effects / Zeta potential / Charge

AC

CE P

reversal / Si release

1. Introduction

Basaltic glass is formed during rapid cooling of magma by water or air. Due to its isotropic nature and chemical composition it is rated as a `fast reacting solid´ in soils or aquatic environments (Schott et al., 2009). Based on the large number of volcanic areas and the susceptibility of basaltic glass to weathering, it plays an important role in local and global cycling of elements in many ways. During chemical weathering the release of alkali and earth alkaline cations is linked with buffering of acidification. Released Ca2+, Fe2+ and Mg2+ can be a trap for CO2 (Pierrehumbert, 2010). Microbial colonization of basaltic glass surfaces for nutrient acquisition increases quantity of biomass which is important for carbon storage.

ACCEPTED MANUSCRIPT The secondary product formed during weathering of basaltic glass is called `palagonite´, which can be highly variable in chemical, mineralogical, and structural properties depending

T

on the nature of the glass and environmental conditions. More specific the dissolution rate of

IP

the glass and the preservation of elements in palagonite depends on a number of factors, such

SC R

as glass chemistry, specific surface area, composition of the solution, temperature, and coating on glass surfaces (Stroncik and Schmincke, 2002). Glasses consisting of compounds with different solubilities have the tendency to dissolve incongruently. Mono- and divalent

NU

network modifier cations are released to solution, and a leached glass matrix from network

MA

formers and intermediates remains. Formation of a leached layer directly bond to the glass surface is a widely accepted mechanism of glass alteration (Grambow and Müller, 2001; Techer et al., 2001). However, from experiments in the laboratory (Geisler et al., 2010) and

TE

D

analysis of archaeological (Hasdemir et al., 2013) and natural glasses (Hay and Iijima, 1968; Dultz et al., 2014) there are observations on a distinct separation between glass and secondary

CE P

phases, pointing to a congruent dissolution and reprecipitation mechanism. For understanding of the principal mode of dissolution and turnover of elements the knowledge of surface

AC

chemistry of a dissolving mineral, revealing sorption/desorption reactions and structural changes during dissolution is regarded as a prerequisite (Schott and Berner, 1983). Within the principal changes in surface chemistry of basaltic glass in contact with aqueous solutions the behavior of structurally bond ferrous iron can be exceptional due to its susceptibility for oxidation. Sinks for glass particles deposited with volcanic ash are soils and different aquatic systems, various saline as well as fresh waters being possible with a broad spectrum of solution parameters, which influence element release from the glass. For assessing the effect of solution chemistry on basaltic glass dissolution rate, parameters such as kind and concentration of ions present, pH, and redox potential have to be considered (Schott et al., 2009). Surface adsorbed ions were shown to play an important role for the dissolution rate

ACCEPTED MANUSCRIPT (Chave et al., 2011; Oelkers and Gislason, 2001; Wolff-Boenisch et al., 2004). Studies on ionic effects on glass dissolution rates considering for example divalent SO42- (Flaathen et al.,

T

2010), the organic anion oxalate (Oelkers and Gislason, 2001) and cations such as Ca2+

IP

(Chave et al., 2011) clearly indicate that glass surface sites react with dissolved ionic species.

SC R

Glass dissolution rate appears typically promoted by strongly adsorbing anions such as fluoride and oxalate, whereas addition of certain cations such as Al3+ can suppress glass dissolution (Oelkers and Gislason, 2001; Wolff-Boenisch et al., 2004). In studies on pH

NU

dependent charge properties of mineral surfaces the monovalent cations and anions Na+, K+,

MA

Cl-, Br-, and NO3- are assumed to be largely indifferent to solid surfaces and hence used as background electrolytes (Kosmulski, 2014), whereas the interactions of multivalent ions with mineral surfaces can be stronger.

TE

D

pH dependent protonation and deprotonation of glass surface sites is well known to have a strong impact on far-from-equilibrium dissolution rate (Oelkers and Gislason, 2001). For

CE P

simple oxides the dissolution rate is at its minimum at the point of zero charge (pzc), where the sum of negatively and positively charged sites is zero. It is at pH 2.5 for SiO2, 7.0 for

AC

Fe2O3 and 8.5 for Al2O3 (Kosmulski, 2014). At low and high pH protonation and deprotonation reactions result in charged surface sites, which tend to weaken surface bonds and hereby increase dissolution rate. For basaltic glass with a complex chemical composition the pzc does not necessarily represent the point with lowest dissolution rate, which is due to the presence of surface sites which differ in acidity. Protonation of oxygen bonds with Si will appear first at pH <3 but for those with Al and Fe at pH <7-8.5 already. Hence minimum farfrom equilibrium dissolution rate for the complex basaltic glass will be at the pH where the sum of absolute values of charges at different surface sites is at minimum. Derived from titrations with acid and base the pzc of powdered basaltic glass is at pH 6.8 (Schott, 1990), which is only slightly higher than that of experimental observations and modelling (Flaathen et al., 2010). Hence, in dissolution experiments with focus on ionic effects adjustment of pH

ACCEPTED MANUSCRIPT to circumneutral values is a measure to avoid extensive adsorption of H+ and OH- which can overlay effects from ions introduced. From curves predicted from measured steady-state

T

basaltic glass dissolution rates at 50 °C minimum values in far-from-equilibrium dissolution

IP

rate are between pH 5.3 and 6.2 (Flaathen et al., 2010), which appears at only slightly lower

SC R

pH than the pzc of basaltic glass determined from titrations. From data compilations on pHdependent mineral dissolution rates Drever (1994) derived that for most silicate minerals farfrom-equilibrium dissolution rate is typically at minimum and independent from pH in a

NU

relatively broad circumneutral region from pH 4-5 to ~8. As chemical compounds in basaltic

MA

glass are similar to those in silicate minerals it appears that in our experiments the broad circumneutral pH region is ideally suited for determination of ionic effects on surface chemistry of basaltic glass. The competition between electrolytes and protons/hydroxy ions

TE

D

will be weak and the far-from-equilibrium dissolution rate of basaltic glass at minimum. For dissolution studies basaltic glasses are usually milled. Hereby broken bonds are generated

CE P

at the surfaces, which have a high reactivity in aqueous solution. In natural systems the occurrence of such fresh glass surfaces is limited to initial stages of glass alteration, for

AC

example directly after fracturing due to thermal and mechanical stress evolving from rapid cooling of glass in contact with water or air, during tectonics at mid ocean ridges, transport of glasses by water or ice, and strain release. Sites exposed to the solution on fracture surfaces with unsatisfied bonds differ in chemistry and molecular level 3D-structure to those in the glass interior (Zallen, 2004). A range of surface sites with different energies exists given by the complex chemical composition of basaltic glasses and also by site conformation with plane, edge, and kink positions. Edge and kink sites on a fractured surface will be more reactive than plane sites (Lasaga, 2014). For tracing chemical reactions on glass surfaces such as ion adsorption, element release, precipitation and structural changes we determined ζ, which is a characteristic parameter of the solid-liquid interface in aqueous suspensions (Kirby and Hasselbrink, 2004).

ACCEPTED MANUSCRIPT Determination of ζ allows the characterization of the solid-liquid interface in situ, whereas other techniques focusing on the molecular level such as spectroscopic methods work best in

T

the dry state implying the risk of artifacts based on the need of prior drying of the sample. The

IP

use of spectroscopic methods in the wet state is limited by the fact that signals coming from

SC R

surface sites are strongly overlain by signals from the bulk glass and the aqueous solution. Analysis of the electrical state of the interfacial region is a new approach for understanding the complex reactions on glass surfaces in solutions (Snellings, 2015). Here we want to

NU

determine the potential of this method to get new insights in surface reactions of basaltic glass

MA

based on the characteristics of the diffuse double layer. For this purpose we choose a systematic experimental approach considering seven different cations and seven different anions in five different concentrations. Due to the complexity of sites on the basaltic glass

TE

D

surface the ζ signal is an integral value, which can be affected in our experiments by different also overlapping factors such as ion ad-/desorption, structural changes and shifts in surface

CE P

chemistry. Hence complementary data were collected to support findings from ζ measurements. For this purpose, sorption isotherms for quantitative estimates on ion transfer

AC

were established, structural changes were traced, depth profiling of glass chemical composition was performed and Si release from the glass was quantified.

2. Theoretical background Minerals in contact with an aqueous solution establish variable charges on their surfaces. The charged sites establish an electric field, attracting oppositely charged ions, the so-called counter ions, which are located in an electrical double layer. In the outer diffuse region of the double layer concentration profiles of ions exist with highest concentrations of oppositely charged ions at the solid surface. Tangential fluid motion along the surface caused by application of an external electric field, mechanical force or settling of particles induces movement of counter ions in the diffuse layer. As a consequence, a potential (ζ) is generated

ACCEPTED MANUSCRIPT at the plane of shear slightly off the solid surface where slip with respect to bulk solution occurs. ζ indicating the magnitude of surface charge is a fundamental parameter for the

T

detailed chemistry and ion distributions at the diffuse interface between a solid phase and

IP

surrounding solution. The ζ signal as an integral parameter will depend on the abundance of

SC R

different sites on the surface. Thorough reviews on the generation and measurement of ζ are available (Hunter, 2000; Kirby and Hasselbrink, 2004; Delgado et al., 2007). ζ has a broad application in colloid science where it is used for description of dispersion and aggregation

NU

phenomena (Chorom and Rengasamy, 1995). Also for mineral dissolution ζ has been

MA

considered (Walther, 1996).

For characterization of charge properties of different oxides and silicates the pH of the pzc is used to describe pH dependent surface charging by protonation and deprotonation reactions

TE

D

(Kosmulski, 2014). By decreasing pH negatively charged surface sites are protonated, and ζ can get zero and also positive, whereas at high pH deprotonation of surface sites results in an

CE P

increase of negatively charged sites and decrease of ζ. The adsorption of ions affects surface charge strongly depending on the valency, size and concentration of ions in solution. For Fe-

AC

oxides the adsorption of anions such as CO32- or SO42- can drastically increase the number of negatively charged sites on the surface and induce a shift of the pzc to lower pH. On the other hand, the adsorption of divalent cations, for example Ca2+, Mg2+, Cu2+, and Zn2+ tends to increase surface charge by neutralization of negatively charged sites/generation of positively charged sites and shifts the pzc to higher pH (van Geen et al., 1994; Ren and Packman, 2005; Walsch and Dultz, 2010). A higher valency of ions in solution will have a larger influence on ζ than monovalent ones. The generation of charge on surfaces by ion adsorption is selflimiting, as electrostatic repulsion of adsorbed likely charged ions on the surface hinders approach of additional ions towards the surface. The effect of electrolyte concentration on ζ, the so called "concentration effect" depends on the number of counter ions that are adsorbed on the mineral surface changing hereby net

ACCEPTED MANUSCRIPT surface charge density and a change in thickness of the electrical double layer (Gu and Li, 2000; Kirby and Hasselbrink, 2004). At low electrolyte concentrations the thickness of the

T

double layer is large and ζ tends to be higher, whereas high concentrations induce small

IP

thicknesses and ζ will be lower.

SC R

For basaltic glass many different surface reactions in aqueous solutions are likely, whereby all can affect ζ. Detailed descriptions on surface reactions are available for feldspars (Blum and Lasaga, 1991; Chardon et al., 2006), and may serve as a basis for understanding the behavior

NU

of basaltic glasses as well. In addition, oxidation of Fe2+ present in high concentration in

MA

basaltic glasses has to be considered. In the following schematic equations illustrating surface reactions of basaltic glass in solution Si represents for simplification other network former such as Al, Fe, and Mn, whereas Na+ represents also the network modifiers K+, Mg2+ and

TE

D

Ca2+.

As a first reaction in contact with aqueous solution, desorption of Na+, K+, Mg2+, and Ca2+

CE P

from the glass surface is likely (1).

(1)

AC

≡Si-O-Na = Si-O- + Na+

where ≡ indicates three single bonds to bridging and non-bridging oxygens of the glass network. Negatively charged oxygen sites at the surface can be rapidly protonated (2). Due to higher cation charge ≡Si-O- will induce a more acidic surface than ≡Al-O- or ≡Fe-O-.

≡Si-O- + H+ = ≡Si-OH

(2)

At low pH an additional H+ can be adsorbed on ≡Si-O-H, resulting in a positively charged surface (3).

ACCEPTED MANUSCRIPT ≡Si-OH + H+ = ≡Si-OH2+

(3)

T

At the ≡Si-O- site not only H+ but also other cations can be adsorbed. In the case of adsorption

SC R

IP

of a divalent cation a positively charged site is generated (4).

≡Si-O- + Ca2+ = ≡Si-OCa+

(4)

NU

Adsorption reactions of protons and metal cations may also occur at oxygen bridges such as

|

TE

H+

D

surface sites are generated (5a, b).

MA

≡Si-O-Si≡, ≡Al-O-Si≡ and ≡Fe-O-Si≡ at the surface of the glass, whereby positively charged

|

(b)

(5a, b)

≡Al-O-Si≡ + Na+ = ≡Al-O-Si≡

CE P

(a) ≡Al-O-Si≡ + H+ = ≡Al-O-Si≡

Na+

It is likely that cations can adsorb on ≡Al-O-Si≡ much better than on ≡Si-O-Si≡ because of

AC

the need of charge compensation for four-fold coordinated Al. However, the adsorption of cations on oxygen bridges will be small in comparison with negatively charged oxygen sites. The number of adsorbed cations by reactions 5a,b will increase with their concentration in solution. Due to its high ionic field strength, adsorbed protons will polarize the metal-oxygen bonds, hereby weaken the bonding with the underlying lattice and eventually break it up (Stumm and Wieland, 1990). Since structural positions on the glass surface are variable no uniform behavior for the adsorption of ions can be expected as adsorption reactions are affected by adjacent oxygens and also by interactions of adsorbed ion species. Water as a polar molecule will be adsorbed on surface sites, oriented with their electronegative oxygen towards the metal cations Al, Fe, and Mn. Due to a transfer of electron density the length of the bonds can increase and break. In the same way also anions

ACCEPTED MANUSCRIPT derived from solution may react. Network forming Si in ≡Si-O-Si≡ will be less affected by H2O because positive partial charge of Si is too weak, whereas OH- will react, clearly

T

indicated by increased dissolution in alkaline medium (Ehrlich et al., 2010). The dependence

IP

of glass dissolution rate on water activity can be explained by the adsorption of water

SC R

molecules on the glass surface (6).

(≡Si-O-Al≡) + H2O = (Si-O-Al·H2O) → ≡Si-OH + ≡Al-OH

NU

(6)

MA

Due to dissolved O2 in the suspensions, the metal ions Fe2+ and Mn2+ in the glass can be transferred to their higher oxidation state stable in oxic water. In a porous leached layer on the glass surface Fe2+ can be bond with oxygen or hydroxyl. Fe2+ which was released from the

TE

D

glass structure by dissolution of the glass will be bond with hydroxyls. Positively charged surface sites may form upon oxidation of Fe-hydroxo complexes, which are released from the

CE P

glass structure by dissolution (7). Fe2+ oxidation is also related with an increase of volume,

AC

which can change molecular level surface topography (8, 9).

4 Fe(OH)2 + O2 = 4 Fe(OH)2++ 2 O2-

(7)

FeO + 1.5 H2O + 0.25 O2 = Fe(OH)3

(8)

4 FeO + O2 = 2 Fe2O3

(9)

Fe sites on the surface of basaltic glass might be a source for positive charges (Blum and Lasaga (1991), where anions can be adsorbed (10). It is also likely that potentially cationic surface sites are formed by other multivalent cations present in the glass such as Mg2+ and

ACCEPTED MANUSCRIPT Cr3+. Moreover surface hydroxyl groups can be replaced by anions which form strong

T

complexes with iron (11).

≡Fe-OH2+ + CH3COO- = ≡Fe-OCOCH3 + H2O

SC R

IP

(10)

≡Fe-OH + CH3COO- = ≡Fe-OCOCH3 + OH-

(11)

NU

pH is an important factor driving dissolution kinetics. At high pH the increase of dissolution rate is explained by the formation of deprotonated oxygen (-O-) at the surface, where

MA

polarization weakens neighboring ≡Si-O-Si≡ bonds (Brady and Walther, 1990). In addition

TE

D

OH- ions can break bonds directly (12).

(12)

CE P

≡Si-O-Si≡ + OH- = ≡SiOH + -O-Si≡

The adsorption of cations from solution on glass surfaces to bridging oxygens is thought to be

AC

of high significance for the dissolution rate, as the activation energy to break the bridging bond is reduced. On the other hand, cations can also strengthen the surface to resist dissolution when they are adsorbed on deprotonated ≡Si-O- sites and accelerate polymerization (Weres et al., 1981). For example Al3+ in the aqueous phase is known to reduce basaltic glass dissolution at acid and base pH remarkably (Oelkers and Gislason, 2001).

3. Material For the experiments a basaltic glass with a chemical composition of average primitive midocean ridge basalt was synthesized. The oxides (MgO, MnO, FeO, Al2O3, TiO2, and SiO2) and carbonates (Na2CO3, K2CO3, and CaCO3), all reagents with grade p. a., were thoroughly

ACCEPTED MANUSCRIPT homogenized in a mortar and melted in a Pt crucible at 1500 °C, poured on a brass plate, ground for achieving optimum homogeneity and re-melted at same conditions. Platelets of

T

basaltic glass (~1  2 cm2) used for depth profiling of chemical composition were stored close

IP

to the glass transition temperature (Tg) at 700 °C for 5 h in order to relax residual tensions and

SC R

subsequently cooled at a rate of 5 K/min to room temperature. Chemical composition, determined by electron microprobe analysis (Cameca, SX100) is 49.67 SiO2, 0.87 TiO2, 16.08 Al2O3, 8.64 FeOtotal, 0.15 MnO, 9.78 MgO, 12.45 CaO, 2.28 Na2O and 0.08 K2O (wt.-%). A

NU

Fe2+/Fetotal ratio of 0.40 ± 0.02 was determined by colorimetric wet-chemical analysis

MA

(Schuessler et al., 2008).

Powdering of basaltic glass for determination of ζ and Si release was performed in a planetary

D

mill (Pulverisette 5, Fritsch). Each 10 g of glass was ground for 18 min at 300 rpm. The

TE

powdered glass was stored with silica gel as desiccant. The glass powder has a mean particle diameter of 1.9 µm (laser light scattering analysis, ZetaPALS, Brookhaven) and a specific

CE P

surface area (SSA) of 1.7 m2/g (N2-adsorption method, Quantachrome NOVA 4000e). Grinding of the basaltic glass for the batch sorption experiments, where relative high

AC

quantities of glass powder were needed, was performed in the mixer mill MM 400 (Retsch). Here the SSA was 0.8 m²/g. Surfaces of glass platelets were polished using a diamond paste in an emulsion of the mineral oil Ballistol and water.

4. Methods 4.1. Ionic effects on surface charge Electrolyte effects on ζ of powdered basaltic glass were determined for solutions of Na+ salts of 7 anions (F-, Cl-, I-, NO3-, SO42-, C2O42-, and HPO42-) and NO3- salts of 7 cations (Na+, K+, Mg2+, Ca2+, Ba2+, Zn2+, and Al3+). Deionized H2O and five different concentrations of the electrolytes, 0.1, 0.5, 1.0, 2.5, and 5.0 mmol/L were mixed with the powdered glass (0.5 g/L) in 250 ml polyethylene bottles. The sample was dispersed directly after addition to the

ACCEPTED MANUSCRIPT solution by an ultrasonic treatment for 15 s. The relative small addition of 0.5 g/L of powdered glass aimed to keep impacts of chemical reactions of the glass on pH and solution

T

chemistry as small as possible. Preliminary tests revealed that this amount of glass powder in

IP

suspension is sufficient for accurate analysis of ζ. Suspensions were stored up to 12000 h at

SC R

20 °C, not exceeding the set point by 2 °C.

ζ was measured using an analyzer with phase-analysis light scattering technique (ZetaPALS, Brookhaven), allowing measurement of very fine particles with low mobility. Particle

NU

movement by a fraction of its own diameter is already sufficient to obtain good

MA

reproducibility of data. ζ is calculated from electrophoretic mobility by the laser-based instrument. After gentle shaking, each 1.6 ml suspension was sampled with a pipette from the PE bottles, transferred in a cuvette and after immersion of the electrode directly introduced in

TE

D

the ζ analyzer. Measurements were performed at 25 °C. The mean ζ value was calculated from 10 runs each partitioned in 20 cycles. The electrical conductivity (EC) of the

CE P

suspensions was determined simultaneously and used to control electrolyte concentration. Subsequent to each ζ measurement the pH of the suspension was determined in the same

AC

cuvette using a glass electrode (Schott, BlueLine 16 pH). The combined ζ, EC, and pH measurements were conducted in time series from 5 min to 12000 h with more frequent measurements at the beginning of the experiment. To diminish the effect of dissolved CO2 (most will be monovalent HCO3-) on ζ, PE bottles were directly closed after sampling.

4.2. Adsorption of ions on glass surfaces For quantification of cations ad-/desorbed at the surface of powdered basaltic glass, batch experiments with seven different cations (see section 4.1) were performed with cation concentrations of 0, 0.025, 0.0625, 0.125, 0.25, 0.625, 1.25, 2.50, 6.25, 12.5 and 25.0 mg/L. Each 10 ml solution was mixed in PE centrifuge tubes with 400 mg of powdered basaltic

ACCEPTED MANUSCRIPT glass. During the reaction time of 14 h at room temperature the centrifuge tubes were horizontally shaken at low speed (IKA, HS 501). The batch experiment was terminated by

T

centrifugation and decantation. Cations in solution were quantified by ICP-OES (Varian, 725-

IP

ES).

SC R

Anion adsorption determined in batch experiments was found to be much weaker than cation adsorption. As it was not possible to detect anion adsorption at the original pH of the suspensions, the initial pH-value was adjusted to 3.8. Concentrations chosen for anion

NU

adsorption studies were 0, 0.025, 0.0625, 0.125, 0.25, 0.625, 1.25, and 2.50 mg/L. Due to the use of HNO3 for adjustment of pH 3.8, NO3- was not considered and batch experiments were

MA

performed only for F-, Cl-, I-, SO42-, C2O42-, and HPO42-. During 14 h reaction time of the batch experiment the pH of the suspensions increased to 8. After terminating the experiment

TE

D

by centrifugation and decantation concentrations of anions were determined by

CE P

chromatography (Dionex, ICS-90).

4.3. Depth profiling of glass surface chemistry

AC

Principal changes of glass chemical composition after contact with a fluid phase were determined for a polished plate, which was exposed to deionized H2O in a PE bottle for 100 h at 20°C. The glass constituents Na, K, Mg, Ca, Al, Fe, and Si were determined simultaneously with a secondary neutral mass spectrometer (INA-X, Specs GmbH), a method which is approved in glass corrosion studies (Frischat et al., 2010). Measurement of the intensities of glass constituents started directly below the surface of the glass sample. High depth resolution was realized by sputtering a relative large sample area (diameter = 5 mm) to achieve sufficient amounts of material for detection. Based on sputter crater depth measurements with a profilometer the mean sputter removal rate of the glass was 0.34 nm/s (80% duty cycle), whereby the fluctuation margin was at 0.05 nm/s.

ACCEPTED MANUSCRIPT 4.4. Ionic effects on glass dissolution To analyze the principal effect of the kind of electrolyte and its concentration on dissolution

T

of the powdered basaltic glass, solutions were separated after 4000 h reaction time from the

IP

suspensions described in section 4.1. The experiments were terminated by filtration of the

SC R

suspension through a 0.45 µm pore-size cellulose acetate filter (Macherey-Nagel). Si in acidified solutions (0.15 M HNO3) was determined by ICP-OES (Varian, 725-ES) in

NU

triplicate.

5. Results

MA

A general issue of evaluation of the ζ signal is the fact that it is especially noisy close to zero zeta potential and in presence of multivalent anions in solution, visible by zigzag courses

TE

D

between the measuring points, making the detection of general trends and concentration effects difficult. At ζ values close to the pzc, repulsive forces between glass particles are at

CE P

minimum and aggregation is likely to occur. The loss in dispersion stability affects the electrophoretic velocity and causes instabilities in ζ. The reason for the stronger noisy ζ signal

AC

in experiments with anions in comparison with that of cations is not fully understood. Here it might be that on the predominantly negatively charged surface repulsive forces to surface adsorbed anions exist, which result in strong molecular level surface roughness, hindering exact determination of ζ.

5.1. Cation effects on surface charge Directly after addition of solutions with mono- and divalent cations to the powdered basaltic glass, marked differences in ζ depending on the kind of cation and its concentration were obtained (Fig. 1). Addition of Na+ in concentrations of 0.5 to 5.0 mmol/L resulted in initial ζvalues relatively close together from -62 to -66 mV (Fig. 1a). In comparison with those obtained for divalent cations (Fig. 1b-c) and deionized H2O, the ones for NaNO3 addition are

ACCEPTED MANUSCRIPT most negative. The observed decrease of the initial ζ with increasing NaNO3 concentration from 0 to 0.1 and ≥0.5 mmol indicates an increase in negatively charged sites on the surface

T

of the glass, which is most probably due to the adsorption of NO3- on the glass surface. Based

IP

on ζ measurements it appears that Na+ has a low affinity to negatively charged surface sites

SC R

and anion adsorption is predominant in contact of the basaltic glass with NaNO3. The same behavior was observed for KNO3 (Fig. 2a). The lowest NaNO3 concentration of 0.1 mmol/L introduced results in a ζ of -52 mV, which is in-between ζ of Na+-concentrations ≥0.5 mmol/L

NU

(-62 to -66 mV) and deionized H2O (-28 mV).

(Fig. 1)

MA

Upon addition of the divalent cations Mg2+, Ca2+, and Zn2+, the initial ζ is markedly higher (38 to -13 mV; Fig. 1b-d) than that for Na+, indicating marked neutralization of negatively

TE

D

charged sites by divalent cations, also formation of positively charged sites on the basaltic glass surface being possible (eq. 4 and 5b). The increase of ζ at the beginning of the

CE P

experiment is most pronounced for Mg2+, Ca2+, and Zn2+ at the highest concentration of 5.0 mmol/L applied. For Mg2+ and Ca2+ the addition of 0.1 and 0.5 mmol solutions resulted in a

AC

decrease of the initial ζ in comparison to deionized H2O, the same as observed for NaNO3 and KNO3, indicating predominant adsorption of NO3- in comparison to cations on the glass surface. NO3- can be adsorbed on positively charged surface sites (eq. 7) and also at network former cations. Towards higher concentrations of 1.0, 2.5, and 5.0 mmol/L ζ clearly increased indicating more extensive adsorption of divalent cations on glass surfaces with increasing concentration, whereby also contribution of the concentration effect is possible (section 2). Within the first hour of reaction a slight decrease of ζ is typically observed for the exposure in deionized H2O and in contact with Mg(NO3)2 and Ca(NO3)2 solutions (Fig. 1b, c), indicating the formation of additional negatively charged sites by the release of alkaline and earth alkaline cations from the glass. For Zn2+ this behavior was obtained for the lowest concentrations of ≤1.0 mmol/L only (Fig. 1d), whereas at higher concentrations of 2.5 and 5.0

ACCEPTED MANUSCRIPT mmol/L a direct increase of ζ was observed indicating faster neutralization of negatively charged sites in contact with Zn2+ than with Mg2+ and Ca2+. The behavior of Ba2+ (Fig. 2b) is

T

similar to that of the other alkaline earth metal cations Mg2+ and Ca2+. For Na+ some increase

IP

of ζ was observed within the first hour of reaction, which is obtained also for K+ (Fig. 2a).

SC R

(Fig. 2)

After the first hours of contact of the solutions with powdered basaltic glass a typical progressive increase of ζ was obtained in all experiments, indicating a decrease of negatively

NU

charged sites with ongoing surface chemical reactions. For the monovalent cations Na+ and K+ the effect of different ion concentrations on ζ is leveled after 100 h reaction time (Fig. 1a,

MA

2a), whereas for the divalent cations Mg2+, Ca2+, Ba2+ and Zn2+ such effects are visible up to 2000 h, most pronounced for Zn2+ (Fig. 1d).

TE

D

Remarkably, in some experiments ζ of basaltic glass surfaces reached the pzc and even positive ζ values. For Zn2+ at all concentrations from 0.1-5.0 mmol/L the pzc was reached,

CE P

whereas for the other divalent cations Mg2+ and Ca2+ the pzc was obtained typically for the higher concentrations only. For Ba2+ the pzc was obtained for the 5 mmol/L concentration

AC

alone. In comparison of the time needed for divalent cations to reach the pzc at 5.0 mmol/L concentration, Zn2+ is at first after 2.5 h reaction time, followed by Ca2+ (75 h), Ba2+ (80 h), and Mg2+ (780 h) (Fig. 3a). At lower concentrations of Mg2+, Ca2+, and Zn2+ time needed to reach the pzc is markedly longer. After reaching the pzc with further progression of the experiment, ongoing increase of ζ indicates generation of more and more positively charged surface sites. Highest ζ value with +50 mV was obtained for the suspension containing 5.0 mmol/L of the transition metal cation Zn2+ (Fig. 1d). Maximum positive ζ values for suspensions containing the alkaline earth metal cations Mg2+, Ca2+, and Ba2+ are up to +20 mV, quite lower than for Zn2+. Time needed for reaching maximum ζ value at a concentration of 5 mmol/L is related to the maximum value of

ACCEPTED MANUSCRIPT ζ. It is shortest for Ba2+ with 150 h followed by Ca2+, Zn2+, and Mg2+ (215, 245, and 780 h respectively; Fig. 3b).

T

(Fig. 3)

IP

Subsequent to reaching maximum ζ values in the experiments with divalent cations, a

SC R

decrease of ζ values was obtained (Fig. 1b-d; Fig. 2b) aligning over time to slightly negative ζ values. For the experiment with 5 mmol/L Ca2+ the pzc is reached the first time after 75 h, maximum positive ζ was obtained after 215 h and, due to the subsequent decrease of ζ, the pzc

NU

is reached again ~1000 h after starting the experiment. After passing the pzc for the second

MA

time ζ is stable at slightly negative ζ values for more than 10000 h until the end of the experiment at 12000 h indicating that a final state in surface structure is reached. The alignment of slightly negative ζ values in advanced stages of the experiment was found to be

TE

D

typical for mono- and divalent cations and all selected anions. Within this principal scheme an exception exists for solutions containing Al3+ (Fig. 2c). Due

CE P

to partial deprotonation of aqua complexes of Al3+ an acidic pH of 3.7-4.8 was obtained in the suspensions (Fig. 10, Appendix Fig. A2c). The decreased pH value accelerates protolysis of

AC

glass and, in consequence, enhances the dissolution rate. Protons can neutralize negatively charged sites (e.g. ≡Si-O- groups) but also generate positively charged sites on glass surfaces upon adsorption (eq. 2 and 5a). Because of acidic pH in experiments with Al3+, data evaluation for ionic effects is difficult. The addition of Al3+ solutions resulted for the first measuring point directly in positive ζ even for the lowest concentration of 0.1 mmol/L (Fig. 2c). For explanation it has to be considered that both, adsorption of protons and Al3+ can generate positively charged surface sites (eq. 3 and 4). Alone at the lowest Al3+ concentration of 0.1 mmol/L a decrease of ζ was obtained over time, which does not occur at higher Al3+ concentrations and indicates neutralization of positively charged sites on the glass surface in advanced stages of the experiment. The fact that ζ at Al3+ concentrations of 2.5 and 5.0

ACCEPTED MANUSCRIPT mmol/L is lower than for concentrations <1.0 mmol/L can most probably be assigned to the

T

compression of the diffuse double layer at high electrolyte concentrations (section 2).

IP

5.2. Anion effects on surface charge

SC R

Initial ζ values upon addition of Na-salts of mono-, di-, and trivalent anions were more negative than those measured in deionized H2O and in presence of divalent cations and Al3+, indicating marked adsorption of anions on the glass surfaces (Fig. 4, 5). For the addition of

NU

the HPO42- anion it has to be considered that at the pH observed in the experiments in the

MA

range of pH 6.8 to 8.7 (Appendix Fig. A3d) the dominant species in solution is H2PO4-. However, upon adsorption on a positively charged site repulsive forces can result in a desorption of protons from H2PO4-. Sorption sites for anions on the overall negatively charged

TE

D

surface will be typical network former cations such as Si in tetrahedral coordination. In comparison of the initial ζ of Cl- and SO42- with HPO42- and C2O42- (Fig. 4), values of the

CE P

latter two anions are much more negative indicating more extensive formation of negatively charged surface sites. For the lowest anion concentration of 0.1 mmol/L ζ is typically between

AC

the value in deionized H2O and that of anion concentrations ≥0.5 mmol/L (Fig. 4, 5) indicating that in the experiments with 0.1 mmol/L concentration anion adsorption sites on the glass surfaces are still available. (Fig. 4, 5) At concentrations from 0.5, 1.0, 2.5 and 5.0 mmol/L, initial ζ values of NO3-, Cl-, I-, SO42-, and C2O42- are each relative close together, for example NO3- (-62 to -66 mV), I- (-74 to -79 mV) and C2O42- (-82 to -97 mV). It appears that the number of adsorption sites for these anions is limited as increasing concentrations do not result in a further decrease of ζ. In contrast, a strong dependence of initial ζ on concentration was observed for F- and HPO42-. For F-, increasing concentrations resulted in a systematic stepwise decrease of the initial ζ (Fig. 5b) indicating an increased number of adsorbed F- on the glass surface towards higher

ACCEPTED MANUSCRIPT concentrations applied. Most probably F- is strongly adsorbed by surface tetrahedral Si, forming a Si-F bond.

T

The same as for F- the initial ζ in experiments with HPO42- decreases with increasing

IP

concentration (Fig. 4c). From the strongly negative ζ values, down to -100 mV obtained it can

SC R

be deduced that extensive adsorption of HPO42- resulted in a high number of negatively charged surface sites. Concerning the ζ trends, HPO42- has stronger similarities with F- than with Cl-, I-, NO3-, and C2O42-. An exception in the series with HPO42- exists at the highest

NU

concentration of 5 mmol/L, where a shift of the initial ζ towards the pzc is observed. This can be assigned to the concentration effect (section 2). The same was observed for SO42-. In

MA

comparison with initial ζ values for cations (section 5.1) the concentration effect appears alone for higher valent anions at high concentrations.

TE

D

A tendency for some increase of pH over time was observed for basaltic glass exposed to deionized H2O, and the electrolytes NaF, NaCl and Na2SO4 (Appendix Fig. A2d and A3b,c),

CE P

indicating marked release of Na+, K+, Mg2+, and Ca2+ from the glass structure which are replaced by protons.

AC

As a general trend for all different anions and independent from concentration, ζ is increasing over time. At the end of the experiment at 12000 h for all concentrations of the monovalent anions and SO42- slightly negative ζ values from -5 to -25 mV were obtained. In experiments with C2O42- and HPO42- some values of ζ were still distinct negative (down to -55 mV) at 12000 h. Here it must be considered that the initial ζ for these anions was the most negative. Despite this fact the typical increase of ζ over time was clearly observable. The tendency for adjustment of a slightly negative ζ over time was obtained even so in the experiments with deionized H2O and mono- and divalent cations (Fig. 1, 2). It appears that independent from kind and concentration of electrolyte in solution, ζ shifts during equilibration in solution towards slightly negative ζ indicating the adjustment of a final state in surface chemistry.

ACCEPTED MANUSCRIPT 5.3. Ad-/desorption of ions at the glass surface The adsorption of cations on glass surfaces can be traced best for Ba2+ and Zn2+ as these two

T

elements are not released from the glass. The sorption isotherms of Ba2+ and Zn2+ show a

IP

steep increase at low applied amounts indicating a high affinity of these cations to the glass

SC R

surface (Fig. 6a). A marked adsorption of Ba2+ and Zn2+ on surface sites was already deduced from measurement of ζ (section 5.1). Maximum adsorbed amounts were obtained at 0.75 mmolc/kg for Ba2+ and 1.25 mmolc/kg for Zn2+ (0.29 Ba2+ and 0.48 Zn2+ cations/nm²).

NU

In the adsorption studies with Na+, K+, Mg2+, Ca2+, and Al3+ indication for a significant

MA

release of these elements from the glass was obtained. For Na+ and K+ the adsorbed amount is decreasing towards higher applied concentrations of these electrolytes suggesting concentration dependent intensification of glass dissolution. Similarly, the addition of highest

TE

D

concentrations for Mg2+, Ca2+, and Al3+ resulted in strong desorption of these elements from the glass. However some uncertainty exists within this result as changes of a certain element

CE P

by ad- or desorption from the powdered glass are at high concentrations not as good detectable as at low concentrations.

AC

For all Mg2+ concentrations applied a release of this element from the glass was obtained, whereas for Ca2+ small applied amounts resulted in measurable adsorption shifting with increasing Ca2+ amounts applied towards desorption from the glass. The same as for Ca2+, also for Al3+ ad- or desorption from the glass strongly depends on the concentration applied. At very low applied amounts of Al3+ some release of Al3+ from the glass was obtained. At medium applied amounts (1.74 mmolc/kg Al3+ applied) adsorbed Al3+ has a maximum value with 0.98 mmolc/kg. Hereafter with increase of applied Al3+ the amount of adsorbed Al3+ continuously decreased resulting at the highest applied Al3+ concentration in desorption. Due to hydrolysis of Al3+ there is a strong effect of pH, which has to be considered in evaluation (section 5.1). (Fig. 6)

ACCEPTED MANUSCRIPT In contrast to the cations, anions introduced in adsorption experiments are not present in the glass. In the batch experiments with different anions no adsorption could be detected at the

T

original weakly acidic to neutral pH of the suspensions. This finding was surprising as in ζ

IP

measurements (section 5.2, Fig. 4, 5) anion adsorption could clearly be detected, however,

SC R

without any quantitative information. In order to get information on anion affinities to basaltic glass surfaces with the batch experimental approach, the pH of the suspensions was decreased in order to generate additional sites for anion adsorption. This decrease of pH implies that the

NU

experiment was performed under non-ideal conditions. After adjustment of pH 3.8 directly

MA

before the start of the experiment, anion adsorption could be traced (Fig. 6b). At the end of the batch experiment after 14 h the pH of the suspensions was found to be at 8, indicating marked proton buffering by the glass. Apparently the processes initiated by the addition of

TE

D

acid generated adsorption sites for anions, which are effective also at the final pH of 8. In comparison of the different anions highest adsorbed amounts were determined with 0.062

CE P

mmolc/kg for SO42-, 0.052 for HPO42-, and 0.040 for C2O42- (0.024 SO42-, 0.020 HPO42-, and 0.016 for C2O42- anions/nm²). The maximal adsorption capacity of these three anions is only

AC

3-8 % of that of Ba2+ and Zn2+ (Fig. 7a), indicating that adsorptions sites for cations on glass surfaces are much more frequent than those for anions. At low anion concentrations applied a steep increase of the adsorbed amounts of F-, C2O42-, SO42-, and HPO42- was obtained indicating marked affinity to binding sites. Towards higher concentrations the adsorption capacity for anions was reached. The shape of the sorption isotherms for F-, C2O42-, SO42-, and HPO42- resembles the Langmuir isotherm with a limited number of equivalent adsorption sites. Multiple site-types available for adsorption on the glass surface, where some parameters decisive for adsorption vary from site to site account most probably for some deviations from the Langmuir form. For Cl- strong scattering of adsorption data was obtained (not shown), which can most probably assigned to impurities in the synthetic glass. The adsorbed amount of the relatively big monovalent anion I- was markedly

ACCEPTED MANUSCRIPT lower than the ones of F- and di- and trivalent anions. Here the shape of the sorption isotherm

T

points to non-specific adsorption of I- on the glass surface.

IP

5.4. Depth profiling of surface chemical composition

SC R

Marked concentration gradients of elements were determined in the depth profile of the basaltic glass sample exposed to deionized H2O for 100 h (Fig. 7). The depth profiles of the intensities of elements reveal the presence of an altered layer which has a thickness of ~60

NU

nm. In the altered layer the intensity of Si and the monovalent cations Na+ and K+ is

MA

increased, whereas losses of the divalent ions Mg2+ and Ca2+ occurred, strongest at the interface to the solution. For Al and Fe only minor changes of intensity with depth were observed. There seems to be slightly lower Al contents close to the surface, but that could be

TE

D

an artifact of measurement. Count rates for Fe are very low and, hence, scatter of data is too large to enable a clear conclusion.

CE P

(Fig. 7)

For the observed increase of Na+ and K+ in the altered layer in comparison with the fresh

AC

basaltic glass the question arises about the source of these cations as the glass sample was exposed to deionized H2O. Here it might be that Na+ and K+ stem from complete dissolution of the glass and are readsorbed in the altered layer or precipitated in secondary phases. Thus, the actual surface visible in Fig. 8 does not represent the initial surface of the glass. The marked Na+ and K+ contents indicate a high chemical potential of these components in the altered layer when compared to dissolved species in the solution. On the other hand, for divalent cations dissolved species in aqueous solution are energetically preferred to species bond in the altered layer of basaltic glasses.

5.5. Effect of electrolytes on glass dissolution

ACCEPTED MANUSCRIPT Results of Si release after 4000 h from basaltic glass powder to different aqueous solutions are shown in Fig. 8. Si release from solutions containing NaNO3, KNO3, NaCl, NaI and Na2SO4

T

is similar to deionized water and rather insensitive to the concentration of dissolved salts.

IP

Slightly higher amounts of Si were found for nitrates of alkaline earth elements with a smooth

SC R

trend of increasing Si with increasing cation concentration for Mg2+ and Ca2+. The largest effect of cations on Si dissolution was observed for Al3+ and in particular for Zn2+. In 5 mmol/L salt solution 0.32 mmol Si/g glass was released to Zn2+ bearing solutions and 0.22

NU

mmol Si/g glass to Al3+-bearing solutions. While a continous increase of Si content with salt concentration can be noted for Zn2+, there appears to be a minimum in the trend for Al3+.

MA

Most probably the observed trend for Zn2+ and Al3+ addition is related to decrease of pH (Appendix Fig. A2b,c). It is likely that upon Al3+ addition ≡Si-O-Al-O-Si≡ bonds are formed,

TE

D

which can suppress Si release i.e. at pH 4,5-5. (Fig. 8)

CE P

A large increase in release of Si occurred also when the anions F-, C2O42- or HPO42- are present in the solution. In solutions containing F- and C2O42- Si contents increase continuously

AC

with salt concentration reaching 0.46 mmol Si/g and 0.28 in 5 mmol/L, respectively. In the case of phosphate the Si release is insensitive to the concentration in the range 0.1 to 0.5 mmol/L.

6. Discussion 6.1. Evaluation of surface charge data The use of ζ for determination of electrolyte effects on surface chemistry of basaltic glass in initial stages of dissolution revealed surface parameters which can be directly interpreted: (i) Systematic dependence of initial ζ on nature and concentration of anions and cations in solution; information on the affinity of ions to surface sites

ACCEPTED MANUSCRIPT (ii) Identification of the potential of different ions and their concentration for generation of negative or positive ζ and different charge densities including the condition where it is zero

T

(iii) Effect of time of exposure to solution on changes of charge densities and charge reversal

IP

The ζ signal itself in our experiments used for analysis of ionic effects and exposure time was

SC R

found to be vulnerable by the following factors. Despite these limitations it was possible to extract consistent trends from ζ data. During the experiment dissolution of the basaltic glass changes the chemical composition of the solution. Glass compounds released increase

NU

electrolyte concentration and can be readsorbed on the glass surfaces. The pH can be affected by the release of Na+, K+, Mg2+ and Ca2+ from the glass where negatively charged oxygen

MA

sites are protonated (eq. 1 and 2). Fe3+ and Al3+ released from the glass form aquacomplexes which undergo hydrolysis and contribute to acidification of the solution. The adjustment of

TE

D

constant solution composition in experiments for determining ζ of powdered basaltic glass in time series is not easy to realize. In preliminary experiments exchange of solution by

CE P

centrifugation and decantation in time intervals resulted in marked scattering of ζ probably because strong physical contact of particle surfaces during centrifugation changes the surface

AC

structure. The adjustment of pH by the addition of acids or bases to suspensions with basaltic glass is problematic as friction of particles during mixing of the suspension might affect surface properties. Dropwise addition of acid or base can have undesired local effects on surface chemistry of basaltic glass. The suspension is diluted by the titration solution and KCl filling solution from the pH electrode can be released. Adjustment of circumneutral pH in experiments with Zn2+ and Al3+ might pose the risk of precipitation of Zn(OH)2 and Al(OH)3. In most of the experiments the strategy to keep pH at circumneutral values by using a wide solution/solid ratio approved and extensive adsorption of H+ and OH- overlaying effects from ions introduced was avoided. From pH data measured during the runtime of the experiment (Appendix Fig. A1-4) it was derived that pH upon addition of mono- and divalent cations Na+, K+, Mg2+, Ca2+, and Ba2+ was in the range from 5.7-7.4 and upon addition of the anions

ACCEPTED MANUSCRIPT NO3-, Cl-, I-, F-, C2O42-, and SO42- in the range from pH 5.3-7.8. However experiments with the acidic cations Zn2+ and Al3+ resulted in low pH (up to pH 5.2 and 3.8 respectively) and in

T

experiments with HPO42- due to protonation of the anion in more alkaline pH (down to pH

IP

8.6). For these experiments the increase in glass far-from-equilibrium dissolution rate by pH

SC R

values outside the circumneutral region was considered for data evaluation. The effect of different pH values inherited by addition of electrolyte solutions on ζ signal is demonstrated for the initial period in Fig. 9. To isolate concentration effects, ζ was

NU

normalized according to Kirby and Hasselbrink (2004) by the negative logarithm of the molar

MA

counter ion concentration (pC). In a strict sense this method is only valid for ions without specific adsorption. In the resulting graph (Fig. 9) broad scattering around the regression line is observed, which is based on the one hand on the analytical uncertainty of ζ determinations,

TE

D

derivable from the data points shown for silica glass in Na+ or K+ solutions. On the other hand pH dependent protonation and deprotonation reactions on glass surfaces can be overlain by

CE P

effects resulting from selective sorption of ions and surface chemical reactions affected by solution composition. However from the data it can be revealed that, specific for each glass,

AC

the relation between pH and ζ follows the theoretical linear trend. Derived from the regression of these data sets the pzc of the basaltic glass surface is observed at higher pH (pH 5.2) than that for silica glass with only acidic ≡SiOH groups on the surface (pH 2.6). (Fig. 9)

6.2 Ionic effects on surface charge Higher initial ζ values were obtained for divalent cations than monovalent ones, indicating the effect of valency but also dependencies from the ion radius were observed. In comparison of initial ζ values of divalent cations at 5 mmol/L concentration, that of Mg2+, the one with the smallest ion radius (65 pm) is most negative. For the monovalent cations Na+ and K+ more negative ζ were obtained for the bigger cation K+ (radius 133 pm) in comparison to Na+ (98

ACCEPTED MANUSCRIPT pm). For the observed weak effect of adsorbed K+ on ζ it has to be considered that the K+ ion is known to fit well with a hexagonal depression in the siloxane surface of silica, increasing

T

strength of adsorption.

IP

The addition of Al3+ showed the strongest effect as even at the lowest Al3+ concentration

SC R

applied directly charge reversal and positive ζ were obtained. Al3+ is most probably adsorbed on the deprotonated Si-O- sites at pH 4-5. At higher Al3+ concentrations (1.0-5.0 mmol/L) a weaker effect of Al3+ in increasing ζ was obtained. The binding of Al3+ to multiple Si-O- sites

NU

can result in some kind of sealing of the surface. It is also likely that in the experiments where

MA

the pH is more close to 4, competition between Al3+ and H+ for binding on Si-O- sites is increased.

For the anions initial ζ values are stronger negative for the multivalent C2O42- and HPO42-

TE

D

(similar ion radii at 238 pm) than the smaller monovalent anions NO3-, Cl-, and I- (ion radii between 181-216 pm). An exception within this scheme is the small monovalent anion F- (ion

CE P

radius at 136 pm), where increasing additions of F- resulted in a systematic decrease of ζ. It is likely that F- is adsorbed due to its specific affinity to Si, whereas the other indifferent anions

AC

are adsorbed by Coulomb forces only (Delgado et al., 2007). For determination of ionic effects counter ions were chosen, which are known to have a weak affinity to surface sites of minerals. However in experiments on cation adsorption evidence for NO3- adsorption was obtained as ζ decreased upon addition of Na+ and K+ indicating that adsorption of NO3- dominates the ζ signal. Here the complementary methods, batch experiments on cation adsorption and depth profiling of chemical composition revealed that in addition to NO3- (detected by ζ measurements) also fair amounts of cations were adsorbed (Fig. 6a and 7). The observed uptake of Na+ and K+ in the leached layer is described also in other studies on glass alteration (Frischat et al., 2010). Based on the ionic radii of NO3- (189 pm) and Na+ and K+ (98 and 133 pm respectively) it is likely that the smaller cations are

ACCEPTED MANUSCRIPT located in molecular level cavities on the surface where their charge is shielded, whereas the

T

relative big NO3- might protrude on the surface and hereby dominate ζ.

IP

By tracing ζ over time strong shifts and remarkable charge reversals were obtained especially

SC R

in the initial period of the experiments up to 2000 h, indicating marked changes of the surface charging mechanism. In the time period >2000 h a slightly negative ζ value adjusted which was stable until the end of the experiment indicating a final state, broadly independent from

NU

kind and concentration of ions introduced. It appears that the initial ζ is modulated over time by the formation of a more or less uniform surface layer masking the charge properties of the

MA

original glass. For changes of ζ over time different surface reactions have to be considered: (i) Chemical reactions in the surface layer of the glass where the more mobile elements are

TE

D

preferentially lost whereas less mobile elements are preserved in the glass (ii) The presence of broken bonds on fresh glass surfaces results in condensation reactions,

CE P

where OH-groups can be transformed by dehydroxylation to bridging oxygens (eq. 6) (iii) Rearrangement of surface and near surface atoms (surface relaxation), which reduces the

AC

surface free energy

(iv) Adsorption of ions from solution modifies charge properties and can facilitate but also suppress surface reactions (v) Diffusion of Fe2+ through a leached layer and subsequent oxidation results in precipitation of Fe-oxides on the glass surface.

6.3. Reasons for surface charge reversal The increase of ζ over time, reaching of the pzc and charge reversal/generation of positive ζ on basaltic glass surfaces is a striking effect indicating strong changes of the original surface chemistry. As mentioned above changes by glass dissolution have to be considered, which might result in the enrichment of Fe- and Al-oxides on the glass surface. If strongly acidic

ACCEPTED MANUSCRIPT ≡SiOH groups are depleted from the glass other network former cations with weakly acidic ≡AlOH, ≡FeOH and ≡TiOH groups get more dominant on the surface (Berger et al., 1987)

T

and in consequence ζ is assumed to increase. For Fe2+ besides oxidation on the surface also

IP

volume oxidation within a leached layer might be significant. Templeton et al. (2009)

SC R

observed Fe- and Mn-oxides on the glassy rim of young pillow basalts at Loihi seamount, Hawaii. However the authors found only little evidence for glass dissolution and assumed that most of these oxides precipitated from sea water on glass surfaces.

NU

Another issue for promotion of charge reversal might also be changes in surface roughness.

MA

Leached layer formation during glass dissolution results in a porous network which might get more developed over time. Here site shielding in a porous surface layer might occur, which is different from a flat surface. The general theory of electrokinetic phenomena strictly applies

TE

D

to ideal, nonporous and rigid surfaces (Delgado et al., 2007) whereas rough surfaces will make description of hydrodynamics problematic. Morphological changes are known to

CE P

influence electrokinetic properties of fibres (Stana-Kleinschek et al., 1999). The effect is also comprehensible by molecular dynamics simulations (Qiao, 2007). Increase of volume by

AC

oxidation of Fe2+ (eq. 8) might result in the formation of Fe-oxides protruding on the surface into solution.

Surface charge inversion can be caused also by the adsorption of certain cations (van der Heyden et al., 2006; Snellings, 2015). In presence of Ca2+ and Zn2+ ζ increased over time to positive charges, indicating the generation of adsorption sites for these cations with time. Addition of Al3+ resulted directly in charge reversal. Monte Carlo simulations show that concentrations needed to obtain charge reversal depend on ion size (Martín-Molina et al., 2009). Monovalent cations can suppress or even cancel charge inversion (van der Heyden et al., 2006). Electrokinetic behavior might also be affected by the penetration of ions and water in the leached layer during progression of dissolution.

ACCEPTED MANUSCRIPT The interpretation of ζ measurements for such heterogeneous and porous surfaces is problematic, as type and magnitude of molecular scale non-idealities, are not fully known. We

T

do not have data on changes of surface roughness during the experiment and hence cannot

IP

estimate its relevance. The observation of positively charged surfaces of basaltic glass is

SC R

important as negatively charged compounds, anions, dissolved organic matter and also bacterial cells can be stabilized on the surface. Bacterial attachment by Coulomb interactions on the glass surface might be the first step for nutrient mining from the basaltic glass and

NU

might result in advanced stages in microbiological imprints on glass surfaces, which are

MA

typically observed in the fresh glass of basaltic rocks (Dultz et al., 2014; Johnson et al., 2006; Banerjee et al., 2011).

TE

D

6.4. Surface charge and linkage to Si release Quantities of Si released in presence of Na+, K+, Mg2+, Ca2+, Ba2+, NO3-, Cl-, I-, and SO42-,

CE P

show only some differences to those released in deionized H2O. It appears that the interactions of these ions with basaltic glass surfaces, visible from changes of ζ, has only

AC

weak effects on glass far-from-equilibrium dissolution rate at the pH values in the experiments. In contrast strong effects on Si release were obtained in presence of the ions F-, C2O42-, HPO42-, Zn2+, and Al3+, showing both, intensification of glass dissolution but also diminishing of Si release from the glass, highly depending on pH. The anions F-, C2O42- and HPO42- which strongly increased Si release had most negative ζ values in the first hours of the experiment of all anions introduced suggesting strong adsorption on glass surface sites. A relation between Si release and ζ is also visible for cations, where for Zn2+ and Al3+, which had most marked effects on Si release, highest ζ values were obtained. The cations Mg2+, Ca2+ and Ba2+ showed a distinct increase of Si release with concentration and markedly higher ζ values than for Na+ and K+. From the principal increase of Si release with increasing concentrations, typically observed for Mg2+,

ACCEPTED MANUSCRIPT Ca2+, F- and C2O42-, it can be derived that the rate of surface chemical reaction strongly depends on the surface concentration of the adsorbed ion and here ζ indicated more extensive

IP

T

binding of ions too, best visible for F-.

SC R

6.5. Reactions on the glass surface

Neutralization of deprotonated Si-O- sites by cations can accelerate polymerization, whereby the formation of such Si-O-Si bonds is thought to be rate-limiting for dissolution (Bickmore et

NU

al., 2006). A preferential adsorption of Na+ and K+ on the glass surface accelerating polymerization can explain a stronger effect in decelerating Si release over experiments with

MA

Mg2+ and Ca2+.

Al3+ showed a marked potential to decrease Si release. Batch experiments revealed that Si

TE

D

released at Al3+ concentrations of 0.1 to 1.0 mmol/L were quite lower than in deionized H2O indicating that adsorbed Al3+ acts as a prohibitor for glass dissolution. It is likely that Al3+ is

CE P

structurally associated with Si forming ≡Si-O-Al-O-Si≡ bonds and is not surface-adsorbed (Dixit and Van Cappellen, 2002). Al3+ in fourfold coordination results in a negative charges in

AC

O atoms of the tetrahedral, which can prevent adsorption of anions to Si in tetrahedra and thus decreases glass dissolution. In our experiments the effect is lost at higher Al3+ concentrations where pH is decreased and binding of Al3+ is in competition with adsorption of protons. Based on the observation that OH- has a strong effect in increasing glass dissolution rate it can be assumed that other anions might have similar effects, highly depending on their size, field strength and their tendency to form a bond with Si. Whereas Si release in solutions containing F–, C2O42-, and HPO42- was markedly higher than in deionized H2O, some decrease of Si release was obtained for NO3-, Cl-, I- and SO42-. For the latter four anions it might be that their adsorption on glass surface sites hinders OH- attack. The strongly anionic additives F-, C2O42-, and HPO42- react with the glass network: F- has a strong affinity to Si, C2O42- will have the tendency to adsorb on bonds with a stronger ionic character, most probably at network

ACCEPTED MANUSCRIPT modifier cations and PO43- itself is a network former. In case OH- forms a ≡Si-OH bond with surface network former cations, a ≡Si-O-Si≡ bond will break and in subsequent hydrolysis

T

steps a monomer, also a dimer being possible, is dissolved out of the structure. Considering

IP

various possible surface sites with different network former cations (Si, Al, Fe, Mn), the

SC R

properties of these complexes are difficult to understand. The effectiveness of ions depends on their structure, and the kind of surface sites and thermodynamic stability they form. Adsorption reactions of ions on basaltic glass surfaces, clearly observable for Ca2+, Ba2+,

NU

Zn2+, and Al3+ (Fig. 6a), raise the question on the possibility of secondary minerals

MA

precipitation in our long-term batch experiments (Figs. 1, 2, 4, and 5). At circumneutral pH realized in these experiments the formation of Fe-oxides and smectites, favored by the release of Fe from the basaltic glass, is likely to occur. Gysi and Stefánsson (2012) observed in CO2-

TE

D

water-basaltic glass batch experiments at 40°C, runtimes up to 260 days, and 100 g glass/L the formation of poorly crystalline Ca-Mg-Fe carbonates, Fe-oxides and Ca-Mg-Fe clays.

CE P

However, in our batch experiments with powdered basaltic glass performed at a more narrow solid:solution ratio (0.5 g/L), temperature of 20 °C, and 500 days runtime, the detailed

AC

inspection of powdered glass samples by IR-spectroscopy and XRD (not reported in this manuscript) did not deliver any hints for secondary mineral formation. Microscopic techniques for secondary mineral identification used by Gysi and Stefánsson (2012) could not be applied for our samples as basaltic glass was intensively powdered for determination of ζ, hindering identification of small-sized secondary phases. In future work the predictability of secondary phase formation has to be improved by a more intensive investigation of solution chemistry including definition of far-from-equilibrium conditions and calculation of saturation indices for expected alteration products.

7. Concluding remarks

ACCEPTED MANUSCRIPT The zeta potential of fresh basaltic glass surfaces was analyzed at different solution compositions and times as a new approach for a more detailed understanding of surface

T

chemistry which is essential for the far-from equilibrium dissolution rate and identification of

IP

underlying mechanisms. Systematic differences of ζ in initial stages and for changes over time

SC R

were obtained, which can be interpreted for the abundance of surface sites. Interface properties of basaltic glass were shown to be dynamic, highly depending on solution chemistry and time span of exposure. Strongly adsorbing ions in solution were identified,

NU

which have with intensification of negative charge, reaching of the point of zero charge and

MA

generation of positively charged surfaces a strong effect on the environmental effect of glass surfaces. Surface charge properties are not only important for glass dissolution but also biogeochemical processes, particle aggregation, solute adsorption, and organic matter

TE

D

retention. Here analysis of ζ is an additional tool for better understanding of interfacial reactions of basaltic glass. Potentially striking biological impact has the observation that

CE P

under certain conditions charge neutrality and positive charges arise on basaltic glass surfaces, providing adsorption sites for negatively charged cell walls of bacteria and dissolved organic

AC

matter. This can favor microbial colonization and can explain extended formation of microbial imprints in basaltic glass surfaces. For more detailed clarification of charge reversal other methods should be used for measuring surface roughness and surface charge components.

APPENDIX A. Electronic Annex Supplementary data associated with this article can be found in the online version.

Acknowledgements We thank Ulrike Pieper and Heiko Steinke (Leibniz Universität Hannover) for their assistance in laboratory work. We are grateful to Thomas Peter (Clausthal University of Technology) for

ACCEPTED MANUSCRIPT performing Secondary Neutral Mass Spectrometry. This study was supported by the Deutsche Forschungsgemeinschaft (DFG) within the International Continental Scientific Drilling

T

Program (ICDP) under contract number Be 1720/29-1, 2 and the Niedersächsische

SC R

IP

Technische Hochschule (NTH).

References

Banerjee, N.R., Izawa, M.R.M., Sapersa, H.M., Whitehouse, M.J., 2011. Geochemical

NU

biosignatures preserved in microbially altered basaltic glass. Surface and Interface

MA

Analysis 43, 452-457.

Berger, G., Schott, J., Loubet, M. (1987) Fundamental processes controlling the first stage of alteration of a basalt glass by seawater: an experimental study between 200° and

TE

D

300°C. Earth and Planetary Science Letters 84, 431-445. Bickmore, B.R., Nagy, K.L., Gray, A.K., Brinkerhoff, A.R., 2006. The effect of Al(OH)4- on

CE P

the dissolution rate of quartz. Geochimica et Cosmochimica Acta 70, 290-305. Blum, A.E., Lasaga, A.C., 1991. The role of surface speciation in the dissolution of albite.

AC

Geochimica et Cosmochimica Acta 55, 2193-2201. Brady, P.V., Walther, J.V., 1990. Kinetics of quartz dissolution at low temperatures. Chemical Geology 82, 253–264. Chardon, E.S., Livens, F.R., Vaughan, D.J., 2006. Reactions of feldspars with aqueous solutions. Eart-Science Reviews 78, 1-26. Chave, T., Frugier, P., Gin, S., Ayral, A., 2011. Glass-water interphase reactivity with calcium rich solutions. Geochimica et Cosmochimica Acta 75, 4125-4139. Chorom, M., Rengasamy, P., 1995. Dispersion and zeta potential of pure clays as related to net particle charge under varying pH, electrolyte concentration and cation type. European Journal of Soil Science 46, 657-665.

ACCEPTED MANUSCRIPT Delgado, A.V., González-Caballero, F., Hunter, R.J., Koopal, L.K., Lyklema, J., 2007. Measurement and interpretation of electrokinetic phenomena. Journal of Colloid and

T

Interface Science 309, 194-224.

SC R

Geochimica et Cosmochimica Acta 66, 2559-2568.

IP

Dixit, S., Van Cappellen, P., 2002. Surface chemistry and reactivity of biogenic silica.

Drever, J.I. (1994) The effect of land plants on weathering rates of silicate minerals. Geochimica et Cosmochimica Acta 58, 2325-2332.

NU

Dultz, S., Boy, J., Dupont, C., Halisch, M., Behrens, H., Welsch, A.-M., Erdmann, M.,

MA

Cramm, S., Helsch, G., Deubener, J., 2014. Alteration of a submarine basaltic glass under environmental conditions conducive for microorganisms: Growth patterns of the

TE

Journal 31, 813-834.

D

microbial community and mechanism of palagonite formation. Geomicrobiology

Ehrlich, H., Demadis, K. D., Pokrovsky, O. S., Koutsoukos, P.G., 2010. Modern views on

CE P

desilicification: Biosilica and abiotic silica dissolution in natural and artificial environments. Chemical Reviews 110, 4656-4689.

AC

Flaathen, T.K., Oelkers, E.H., Gislason, S.R., 2010. The effect of aqueous sulphate on basaltic glass dissolution rates. Chemical Geology 277, 345-354. Frischat, G.H., Leschik, M., Heide, G., Behrens, H., Wiedenbeck, M., 2010. Hydration and corrosion of polymerised alumosilicate glasses. European Journal of Glass Science and Technology B, Physics and Chemistry of Glasses 51, 107-116. Geisler, T., Janssen, A., Scheiter, D., Stephan, T., Berndt, J., Putnis, A., 2010. Aqueous corrosion of borosilicate glass under acidic conditions: A new corrosion mechanism. Journal of Non-Crystalline Solids 356, 1458-1465. Grambow, B., Müller, R. 2001. First order dissolution rate law and the role of surface layers in glass performance assessment. Journal of Nuclear Materials 298,112-124. Gu, Y., Li, D. 2000. The ζ-potential of glass surface in contact with aqueous solutions.

ACCEPTED MANUSCRIPT Journal of Colloid and Interface Science 226, 328-339. Gysi, A.P., Stefánsson, A., 2012. CO2-water-basalt interaction. Low temperature experiments

T

and implications for CO2 sequestration into basalts. Geochimica et Cosmochimica

IP

Acta 81, 129-152.

SC R

Hasdemir, I., Striepe, S., Deubener, J., Schmidt, B.C., 2013. Micromechanical properties of banded alterations of archaeological glass fragments. Journal of Non-Crystalline Solids 376, 126-132.

NU

Hay, R.L., Iijima, A., 1968. Nature and origin of palagonite tuffs of the Honolulu Group on

MA

Oahu, Hawaii. In: Studies in volcanology – a memoir in honor of Howel Williams. Geological Society of America, Boulder. 331-376. Hunter, R.J., 2000. Foundations of Colloid Science. Oxford University Press, 2nd edition, 816

TE

D

p..

Johnson, H.P., Baross, J.A., Bjorklund, T.A., 2006. On sampling the upper crustal reservoir of

CE P

the NE Pacific Ocean. Geofluids 6, 251-271. Kirby, B.J., Hasselbrink, E.F., 2004. Zeta potential of microfluidic substrates: 1. Theory,

AC

experimental techniques, and effects on separations. Electrophoresis 25, 187-202. Kosmulski, M., 2014. The pH dependent surface charging and points of zero charge. VI. Update. Journal of Colloid and Interface Science 426, 209-212. Lasaga, A.C., 2014. Theory of crystal growth and dissolution, Chap. 7. In: Kinetic theory in the Earth sciences. Princeton Legacy Library. Martín-Molina, A., Hidalgo-Álvarez, R., Quesada-Pérez, M., 2009. Additional considerations about the role of ion size in charge reversal. Journal of Physics: Condensed Matter 2, 424105 (7pp) Oelkers, E.H., 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.

ACCEPTED MANUSCRIPT Geochimica et Cosmochimca Acta 65, 3671-3681. Pierrehumbert, R.T., 2010. Principles of planetary climate. Cambridge University Press, 652

T

p..

IP

Qiao, R., 2007. Effects of molecular level surface roughness on electroosmotic flow.

SC R

Microfluid Nanofluid 3, 33-38.

Ren, J., Packman, A.I., 2005. Coupled stream-subsurface exchange of colloidal hematite and dissolved zinc, copper and phosphate. Environmental Science and Technology, 39,

NU

6387-6394.

MA

Schott, J., Berner, R.A., 1983. X-ray photoelectron studies of the mechanism of iron silicate dissolution during weathering. Geochimica et Cosmochimca Acta 47, 2333-2340. Schott, J., 1990. Modelling of the dissolution of strained and unstrained multiple oxides: The

TE

D

surface speciation approach. Chapter 12. 337-365. In: W. Stumm (ed.), Aquatic chemical kinetics, Wiley. J.,

Pokrovsky,

O.S.,

Oelkers,

E.H.,

2009.

The

link

between

mineral

CE P

Schott,

dissolution/precipitation kinetics and solution chemistry. Reviews in Mineralogy &

AC

Geochemistry 70, 207-258. Snellings, R., 2015. Surface chemistry of calium aluminosilicate glasses. Journal American Ceramic Society 98, 303-314. Schuessler, J.A., Botcharnikov, R.E., Behrens, H., Misiti, V., Freda, C., 2008. Oxidation state of iron in hydrous phono-tephritic melts. American Mineralogist 93, 1493-1504. Stana-Kleinschek, K., Strnad, S., Ribitsch, V., 1999. The influence of structural and morphological changes on the electrokinetic properties of PA 6 fibres. Colloids and Surfaces A: Physicochemical and Engineering Aspects 159, 321-330. Stroncik, N.A., Schmincke, H.-U., 2002. Palagonite - a review. International Journal of Earth Sciences 91, 680-697.

ACCEPTED MANUSCRIPT Stumm, W., Wieland, E., 1990. Dissolution of oxide and silicate minerals: rates depend on surface speciation. In: Aquatic chimica kinetics: Reaction rates of processes in natural

T

waters. Environmental Science and Technology Series. Wiley, New York, 367-400.

IP

Techer, I., Advocat, T., Lancelot, J., Liotard, J.-M., 2001. Dissolution kinetics of basaltic

SC R

glasses: control by solution chemistry and protective effect of the alteration film. Chemical Geology 176, 235-263.

Templeton, A.S., Knowles, E.J., Eldridge, D.L., Arey, B.W., Dohnalkova, A.C., Webb, S.M.,

NU

Bailey, B.E., Tebo, B.M., Staudigel, H., 2009. A seafloor microbial biome hosted

MA

within incipient ferromanganese crusts. Nature Geoscience 2, 872-876. van der Heyden, F.H.J., Stein, D., Besteman, K., Lemay, S.G., Dekker, C., 2006. Charge inversion at high ionic strength studied by streaming currents. Physical Review Letters

TE

D

96, 224502(4).

van Geen, A., Robertson, A.P., Leckie, J.O., 1994. Complexation of carbonate species at the

CE P

goethite surface: Implications for adsorption of metal ions in natural waters. Geochimica et Cosmochimica Acta 58, 2073-2086.

AC

Walsch, J., Dultz, S., 2010. Effects of pH, Ca- and SO4-concentration on surface charge and colloidal stability of goethite and hematite. Clay Minerals 45, 1-13. Walther, J.V., 1996. Relation between rates of aluminosilicate mineral dissolution, pH, temperature, and surface charge. American Journal of Science 296, 693-728. Weres, O., Yee, A., Tsao, L., 1981. Kinetics of silica polymerization. Journal of Colloid and Interface Science 84, 379-402. Wolff-Boenisch, D., Gislason, S.R., Oelkers, E.H., 2004. The effect of fluoride on the dissolution rates of natural glasses at pH 4 and 25 degrees C. Geochimica et Cosmochimica Acta 68, 4571-4582. Zallen,R., 2004. The physics of amorphous solids. Wiley, 318 p..

ACCEPTED MANUSCRIPT Figure captions Fig. 1a-d. Effect of Na+ (a), Mg2+ (b), Ca2+ (c) and Zn2+ (d) on the surface charge of powdered

T

basaltic glass in the initial stages of dissolution determined in batch experiments. Zeta

IP

potential (ζ) of suspensions measured as a function of cation concentration (0-5.0 mmol/L) for

SC R

tracing cation effects on the principal course of ζ in a time period from 5 min to 12000 h. Fig. 2a-c. Effects of K+ (a), Ba2+ (b), and Al3+ (c) on the surface charge of powdered basaltic glass in the initial stages of dissolution determined in batch experiments. Zeta potential (ζ) of

NU

suspensions measured as a function of cation concentration (0-5.0 mmol/L) for tracing cation

MA

effects t on the principal course of ζ in a time period from 5 min to 12000 h. Fig. 3. Reversal of surface charge of powdered basaltic glass over time in suspension containing the divalent cations Mg2+, Ca2+, Ba2+, and Zn2+. Time needed for reaching point of

TE

D

zero charge (a) and maximum value of zeta potential (b) of powdered basaltic glass in contact with 0.1-5.0 mmol/L solutions. Data extracted from Figs. 1b, c, d and 2b. Fig. 4a-d. Effects

CE P

Cl- (a), SO42- (b), HPO42- (c) and C2O42- (d) on the surface charge of powdered basaltic glass in the initial stages of dissolution determined in batch experiments. Zeta potential (ζ) of

AC

suspensions measured as a function of anion concentration (0-5.0 mmol/L) for tracing anion effectson the principal course of ζ in a time period from 5 min to 12000 h. Fig. 5a,b. Effects of I- (a) and F- (b) on surface charge of powdered basaltic glass in the initial stages of dissolution determined in batch experiments. Zeta potential (ζ) of suspensions measured as a function of anion concentration (0-5 mmol/L) for tracing anion effects on the principal course of ζ in a time period from 5 min to 12000 h. Fig. 6. Adsorption isotherms describing the binding/release of ions at the surface of powdered basaltic glass in batch experiments with a reaction time of 14 h. Adsorbed and desorbed amounts of seven different cations (a) and five different anions (b) as a function of the applied amount of ions present in the solution at the beginning of the experiment.

ACCEPTED MANUSCRIPT Fig. 7. Surface chemical composition of polished basaltic glass exposed for 100 h to deionized H2O by depth profiling using SNMS . Intensities are shown for Na, K, Mg, Ca, Al,

T

Fe, and Si.

IP

Fig. 8a,b. Electrolyte effects on the dissolution of powdered basaltic glass determined by the

Zn2+, and Al3+ (a) and anions (b) on release of Si..

SC R

release of Si after 4000 h in batch experiments. Effect of cations Na+, K+, Mg2+, Ca2+, Ba2+,

Fig. 9. Effect of pH on charging of basaltic glass surfaces. Zeta potential (ζ) as a function of

NU

pH of powdered basaltic glass for all electrolyte-glass combinations shown in Figs. 1, 2, 4, and 5 at the initial period of the experiment (~5 min runtime). To isolate concentration effects,

MA

ζ was normalized by the negative logarithm of the molar counter ion concentration (pC). For comparison data for silica glass in Na+ or K+ solutions based on a data collection of Kirby and

TE

D

Hasselbrink (2004) are given. Lines are trends for basaltic (short dash) and silica (long dash)

AC

CE P

glasses. Note the much stronger dependence of ζ on pH for the basaltic glass.

ACCEPTED MANUSCRIPT 10

10

-10

-10

-20

-20

-30

-30

-40

-40 -50

MA

-50

-60

-60

-20

AC

Zeta potential (mV)

-10

-70 50 40

d) Zn(NO3)2

30 20 10

Zetapotential (mV)

CE P

0

TE

c) Ca(NO3)2

D

-70 10

-30

IP

0

SC R

0

T

b) Mg(NO3)2

NU

Zeta potential (mV)

a) NaNO3

c) Ca(NO3)2

0 -10 -20

-40 -30 -40

-50

-50 -60 -60 -70 0.1

Fig. 1

1

10 100 Time (h)

-70 1000 10000

deionized H2O Salt concentration (mmol/L): 0.1 0.5 1.0 5.0 2.5

0.1

1

10 100 1000 10000 Time (h)

ACCEPTED MANUSCRIPT 20

a) KNO3

90

b) Ba(NO3)2

10

c) Al(NO3)3

80 70

0

60

-20

IP

40

T

50

30

-30

20

-40

SC R

Zeta potential (mV)

-10

10

-50

0

-10

-60 deionized H2O

-20

-80 1

10 100 Time (h)

1000 10000

0.1

1

10 100 Time (h)

MA

0.1

NU

Salt concentration (mmol/L): 0.1 0.5 1.0 2.5 5.0

-70

AC

CE P

TE

D

Fig. 2

-30

-40 1000 10000

0.1

1

10 100 1000 10000 Time (h)

ACCEPTED MANUSCRIPT

IP

T

1000

SC R

Time (h)

100

NU

10

Concentration (mmol/L):

0.1 2.5

1.0

0.5 5.0

CE P

AC

Time (h)

TE

D

1 1000

MA

a) Time for reaching pzc

100

b) Time for reaching maximum positive charge Mg2+

Ca2+

Ba2+

Cation in solution

Fig. 3

Zn2+

ACCEPTED MANUSCRIPT 0

b) Na2SO4

a) NaCl -10

T

-20

IP SC R

-40 -50 -60

NU

Zeta potential (mV)

-30

-70

deionized H2O

-90

Salt concentration (mmol/L): 0.1 0.5 1.0 5.0 2.5

D

-100

TE

0 -10

c) Na2HPO4

d) Na2C2O4

CE P

-20 -30 -40

AC

Zeta potential (mV)

MA

-80

-50 -60 -70 -80 -90 -100

0.1

Fig. 4

1

10 100 1000 10000 0.1 Time (h)

1

10

100 1000 10000 Time (h)

ACCEPTED MANUSCRIPT 0

a) NaI

b) NaF

T

-10

IP SC R

-30 -40

NU

-50 -60

MA

Zeta potential (mV)

-20

-70

D

H2O

-80 -90

10 100 Time (h)

CE P

Fig. 5

1

AC

0.1

TE

Salt concentration (mmol/L):

0.1 0.5 1.0 2.5 5.0

1000 10000 0.1

1

10 100 Time (h)

1000 10000

ACCEPTED MANUSCRIPT

Zn2+

a) cations +

K 2+

Ba

Ba

Mg

Zn2+

2+

SC R

IP

0.02 K+

0.00

Na+ Ca2+

Al3+

NU

-0.02

Na

Ca

-0.04

-0.06 0.5

Ba

Mg2+

1.0 1.5 2.0 Applied amount (mmolc/g)

2.5

3.0

D

0.0

TE

0.0025

PO43-

CE P

0.0020

SO42-

C2O42-

0.0015

AC

Adsorbed amount (mmolc/g)

Na+ Ca2+ Al3+

2+

T

0.04

MA

Adsorbed/desorbed amount (mmolc/g)

0.06

F-

0.0010

b) anions F-

0.0005 I-

0.0000 0.00

Fig. 6

0.01

C2O42-

0.02 0.03 0.04 0.05 Applied amount (mmolc/g)

SO42-

IPO43-

0.06

0.07

ACCEPTED MANUSCRIPT

25 50

K

SC R

Fe

100 125

Al

NU

Depth (nm)

75

150

MA

175 200

250

AC

CE P

10

TE

D

225

Fig. 7

Na

IP

Ca

Si

T

Mg

0

100

1000

Intensity (cps)

10000

ACCEPTED MANUSCRIPT Cations in solution

0.35

a) cations

T

5.0

IP

0.25

Concentration (mmol/L):

SC R

0.20 0.15

0.05

0.1 0

Na+

|

K+

|

| | | Mg2+ Ca2+ Ba2+ Cations in solution

Zn2+

|

Al3+

TE

D

0.00

0.5

MA

0.10

2.5

1.0

NU

Si-release (mmol/g)

0.30

0.50

0.35 0.30 0.25

CE P

0.40

b) anions

5.0

AC

Si-release (mmol/g)

0.45

Concentration (mmol/L):

2.5

0.20

0.5 1.0

0.15

0.1

0.10

0

0.05 0.00

NO3-

|

F-

|

Cl-

|

I-

|

SO42-

Anions in solution

Fig. 8

|

C2O42-

|

PO43-

ACCEPTED MANUSCRIPT

Basaltic glass: KNO3

Mg(NO3)2 Zn(NO3)2

T

NaNO3

Ca(NO3)2 Al(NO3)3

Ba(NO3)2

IP

NaJ NaCl Na2SO4 Na2C2O4 Na2HPO4 deionized H2O

SC R

NaF

Silica glass: 38 40

NU

0

MA

-20

39 63 102

D

-40

TE

Normalized zeta potential /pC (mV)

20

Fig. 9

AC

2

CE P

-60

4

6

8

pH

10

12

ACCEPTED MANUSCRIPT Highlights:  Surface reactions of powdered basaltic glass were determined in a time period of 500 days

T

Response of basaltic glass surfaces on solution composition was studied by zeta

IP



potential (ζ)

Over time a shift from negative to positive ζ was observed, most markedly in presence

SC R



of Ca2+ and Zn2+

Positive ζ values on basaltic glass surfaces foster adsorption of anionic compounds

NU



and can have biological impact

MA

Neutralization of deprotonated Si-O- sites by monovalent cations can favor

CE P

TE

D

polymerization resulting in smaller Si release

AC

