Impact of surface complexation and electrostatic interactions ...

General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

Users may download and print one copy of any publication from the public portal for the purpose of private study or research.

You may not further distribute the material or use it for any profit-making activity or commercial gain

You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from orbit.dtu.dk on: Aug 28, 2022

Impact of surface complexation and electrostatic interactions on pH front propagationin silica porous media

Stolze, Lucien; Wagner, Jakob Birkedal; Damsgaard, Christian Danvad; Rolle, Massimo

Published in:Geochimica et Cosmochimica Acta

Link to article, DOI:10.1016/j.gca.2020.03.016

Publication date:2020

Document VersionPublisher's PDF, also known as Version of record

Link back to DTU Orbit

Citation (APA):Stolze, L., Wagner, J. B., Damsgaard, C. D., & Rolle, M. (2020). Impact of surface complexation andelectrostatic interactions on pH front propagation in silica porous media. Geochimica et Cosmochimica Acta,277, 132-149. https://doi.org/10.1016/j.gca.2020.03.016

https://doi.org/10.1016/j.gca.2020.03.016

https://orbit.dtu.dk/en/publications/37ce25ce-d2a1-47d8-8e7a-f004aff7c4a3


Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 277 (2020) 132–149

Impact of surface complexation and electrostatic interactionson pH front propagation in silica porous media

Lucien Stolze a, Jakob B. Wagner b, Christian D. Damsgaard b,c, Massimo Rolle a,⇑

aDepartment of Environmental Engineering, Technical University of Denmark, Miljøvej, Building 115, 2800 Kgs. Lyngby, DenmarkbTechnical University of Denmark, DTU Nanolab, Fysikvej, Building 307, 2800 Kgs. Lyngby, DenmarkcTechnical University of Denmark, DTU Physics, Fysikvej, Building 311, 2800 Kgs. Lyngby, Denmark

Received 6 November 2019; accepted in revised form 12 March 2020; available online 20 March 2020

Abstract

The coupled effects of pH and ionic strength impact a variety of geochemical processes in the subsurface. In this study, weinvestigate the interactions of H+ and major ions at the surface-solution interface of silica porous media under advection-dominated flow-through conditions. A series of 21 column experiments were performed by systematically injecting solutionsof different pH and ionic strengths. Three types of porous media (i.e., two natural sands and quartz beads) were considered inorder to explore differences in surface/solution interactions among quartz materials. Multiple lines of evidence were used tocharacterize the geochemical processes taking place during the flow-through experiments: (i) breakthrough curves of pH andmajor ions were measured at the column outlets; (ii) the natural sand surfaces were characterized by chemical extractions andscanning electron microscopy and the quartz structure was analyzed by XRD; (iii) reactive transport modeling was performedto quantitatively interpret the experimental results.

We observed strong reactivity of the quartz surface characterized by significant release of protons when interacting with themajor ions. The results also show significant differences in the quantity of protons emitted from the surfaces of the three silicaporous media, as well as in the shape of the pH breakthrough curves. Reactive transport modeling was based on a multicom-ponent ionic transport formulation and on surface complexation description of the solid/solution interactions. The surfacecomplexation model included the individual contributions of quartz and Fe/Al oxides present in the sand coatings withthe component additive approach. For each medium, a single set of surface complexation parameters capable of reproducingthe experimental dataset (i.e., 7 columns) was calibrated by parallelizing the simulations of the flow-through experiments. Thisapproach allowed us to capture the pH and ionic species behavior within a large range of ionic strengths (i.e., [0–100] mM).The SCMs quantitatively show that H+ is released from quartz upon adsorption of Na+ and that the protonation of the oxi-des surfaces retards the pH front. Differences in acidity behavior between the silica surfaces seem to be primarily controlled bydifferences in surface topology, crystal structure of quartz and/or various presence of aluminosilicates. This study demon-strates that the reactivity and diversity of quartz surfaces in silica porous media and the complex interplay with oxide surfacesin the coatings control the transport of aqueous charged species in flow-through systems.� 2020 Elsevier Ltd. All rights reserved.

Keywords: Silica sand; Surface complexation; Flow-through experiments; Scanning electron microscopy; Reactive transport modeling


0016-7037/� 2020 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: [email protected] (M. Rolle).

1. INTRODUCTION

Groundwater pH and background electrolytes have fun-damental and coupled implications in a variety of geochem-ical processes and influence the transport of charged species


mailto:[email protected]


http://crossmark.crossref.org/dialog/?doi=10.1016/j.gca.2020.03.016&domain=pdf

L. Stolze et al. /Geochimica et Cosmochimica Acta 277 (2020) 132–149 133

in porous media (Appelo and Postma, 2005; Sposito 2008;Bretzler et al., 2019). In particular, these hydrochemical fac-tors strongly affect dissolution, precipitation and transfor-mation mechanisms (Cornell and Schwertmann, 2003;Hansel et al., 2005; Battistel et al., 2019), redox reactions(Kocar and Fendorf, 2009; Fakhreddine et al., 2016) sorp-tion processes (Davis and Kent, 1990; Kent, et al., 2007;Stachowicz et al., 2008; Prigiobbe et al., 2012) and diffu-sive/dispersive processes (Appelo and Wersin, 2007;Muniruzzaman et al., 2014; Muniruzzaman and Rolle,2015, 2019). Therefore, it is of fundamental importance tounderstand the mechanisms controlling the displacementand distribution of pH and major ions in the subsurface.

Despite the quantitative importance of silica minerals inthe environment, with silica sand representing themost abun-dant material in the composition of groundwater aquifers(Iler, 1979; Kent et al., 2007), their influence on the mobilityof aqueous charged species through sorption processes inenvironmental systems is commonly disregarded. In fact,sorption processes are traditionally interpreted with focuson iron and/or aluminium oxides (Fuller et al., 1996; Daviset al., 1998; Akai et al., 2004; Fukushi and Sverjensky,2007; Jessen et al., 2012; Johannesson and Neumann, 2013;Sø et al., 2018) or on clay surfaces (Appelo and Postma,2005; Tournassat and Steefel, 2015; Tournassat et al.,2018). However, molecular interfacial chemistry studies haveshown that silica surfaces exhibit a complex acid-base behav-ior controlled by the presence of multiple silanol surfacegroups with varying acidic properties (Ong et al., 1992;Sulpizi et al., 2012; Pfeiffer-Laplaud et al., 2015; Darlingtonand Gibbs-Davis, 2015). To date, the impact of this surfacebehavior on the mobility of aqueous species and, in particu-lar, pH fronts propagation in flow-through silica porousmedia has not been adequately assessed.

Surface complexation models (SCMs) are instrumentalfor the quantitative interpretation and prediction of sorp-tion processes (Davis and Kent, 1990; Koretsky, 2000) asthey mathematically describe the chemical and electrostaticinteractions at the liquid-mineral interface by explicit con-sideration of the electrochemical composition of the aque-ous phase and the solid surface. Whereas SCMs of metaloxides allow the successful macroscopic description of sorp-tion processes in multi-component batch and flow-throughsystems under variable chemical conditions (Stachowiczet al., 2008; Prigiobbe and Bryant, 2014; Hiemstra, 2018;Stolze et al., 2019a), the description of natural quartz sur-faces with SCMs remains challenging. In fact, recent studieshave shown that existing SCMs for quartz can only qualita-tively predict the effects of the ionic strength on the proto-nation of natural silica surfaces but fail at reproducingquantitatively their acid-base behavior under flow-throughconditions McNeece and Hesse, 2017; McNeece et al.,2018). Furthermore, natural quartz sand is generally coatedby a complex mineral assemblage of Fe-, Al- and Si-bearing-oxides representing a small fraction of the bulksolid (Coston et al., 1995; Fuller et al., 1996; Davis et al.,1998; Zhang et al., 2011; Penn et al., 2001). For the descrip-tion and understanding of sorption processes onto suchheterogeneous sorbent, the component additive approach(CA) is of particular interest as it aims at explicitly consid-

ering the different mineral reactive surfaces that affect themobility of the aqueous species under scrutiny (Daviset al., 1998; Lund et al., 2008; Landry et al., 2009; Alessiand Fein, 2010). The implementation of the CA approachdirectly based on experiments performed with naturallycoated sand under well-controlled laboratory conditionsrepresents an important intermediate step for the descrip-tion of sorption processes in natural systems (Hiemstra,2010). However, such application remains rare and limitedto batch systems and the contribution of the quartz surfacehas not been rigorously assessed. Furthermore, previousimplementations of SCMs with a CA approach did notexplicitly account for the dependency of the adsorptionprocesses on the ionic strength (Davis et al., 1998; Logueet al., 2004).

This study aims at investigating and comparing the reac-tivity of different silica porous media (i.e., natural sandsand synthetic quartz beads) by analyzing the propagationof pH fronts and the influence of major ions under flow-through conditions. A series of column experiments wasperformed by injecting salt and acid solutions with differentelectrolyte (NaBr) concentrations to explore the electro-static and chemical interactions between H+, major ions(Na+, Br� and Cl�), the quartz surface and its coating.Measurements of pH and ion’s breakthrough at the columnoutlets, combined with chemical and microscopy surfacecharacterization techniques were used to infer the interac-tions between aqueous charged species and the surface ofthe silica porous media. Reactive transport modeling wasperformed to quantitatively interpret the experimentalobservations and to disentangle the complex interplaybetween pH, major ions, quartz and metal oxide surfacescontrolling the transport and retardation of charged speciesin the silica porous media.

2. LABORATORY EXPERIMENTS

The experiments were conducted with quartz beads (Sili-beads, Type Q, Sigmund Lindner) and two types of naturalquartz sand originating from two different spatial locations:Dorsten, Germany (Aquagran – dry, Euroquartz) andBrædstrup, Denmark (Dava Kvarts, Dana Kvarts IndustriA/S). The three media have similar grain sizes (�1 mm) andare referred to as quartz beads (QB), sand Germany (SG),and sand Denmark (SD) throughout the text.

All experiments were carried out in a temperature con-trolled room (T = 20 �C) and laboratory solutions wereprepared using Milli-Q water. Prior to use, the silica mediawere cleaned according to the method proposed byMcNeece and Hesse (2016): the sand and quartz beads werefirst washed in a pH 4 HCl solution, thoroughly rinsed withMilli-Q water until reaching pH � 5.9 and EC < 0.1 lS/cm.Subsequently, the silica material was washed in a NaOHsolution of pH 10, similarly rinsed and dried at 100 �C.The surface areas of SG and SD indicated in Table 1 weredetermined by single- and multi-point BET sorption withN2 at �195 �C and are in agreement with previous reportedvalues (Burgisser et al., 1994; McNeece and Hesse, 2016).The surface area of QB was found below the detection limitof the BET.

Table 1Material properties of the silica quartz media used in the experiments.

Sand Germany (SG) Sand Denmark (SD) Quartz beads (QB)

Diameter [mm] 1.3 0.8 1Particle density qs [kg/L] 2.65 2.65 2.2Surface area (calculated) [m2/g] 1.7 � 10�3 2.8 � 10�3 2.7 � 10�3

Surface area (BET) [m2/g] 0.14* 0.03* BD

*Average values of duplicate measurements; BD: below detection.

134 L. Stolze et al. /Geochimica et Cosmochimica Acta 277 (2020) 132–149

2.1. Column experiments

Proton propagation fronts through silica porous mediawere investigated by performing column experiments incylindrical glass columns (length: 14.4 cm, inner diameter:1.75 cm) that were uniformly filled by following a wet-packing procedure to avoid entrapment of air in the pores(Haberer et al., 2012; Haberer et al., 2015a). All experimentswere performed with continuous vertical injection from bot-tom to top using 2 ports at each end of the columns. One ofthe outlet ports was connected to a 10 mL flow-through vialfor in-line measurement of pH breakthrough with a timeinterval of 5 minutes. The remaining port was used to sam-ple the outlet solution at 1-hour interval for monitoring thebreakthrough of ions. The average seepage velocity for allcolumn experiments was set to 1 m/day by connecting theinlet and outlet ports to a high-precision peristaltic pump(IPC-N24, ColeParmer, Vernon Hills, United States). Theapplied flow rate was measured at the end of each columnexperiment (Table EA1) and the porosity was estimatedbased on the breakthrough of conservative anions measuredat the outlet of the columns (Table EA2).

The column experiments were initiated by flushing thedomain during 24 h (i.e., 6.5 pore volumes) with a pH 2HCl solution in order to remove easily dissolvable mineralphases which can affect the transport of H+. Subsequently,the columns were reequilibrated with Milli-Q water untilpH � 5.5 was reached before starting the flow-throughexperiments. The interplay between H+, major ions and sil-ica porous media was investigated by injecting acidic (i.e.,pH � 2.8 HCl) and pure salt solutions (i.e., pH � 5.9) withvarying concentrations of NaBr electrolyte [0 – 100] mM.The solutions were injected as a step change from the bot-tom of the column setup as illustrated in Fig. 1 (seeTable EA1 for operational details about the 21 flow-through experiments performed). The specific surface areaindicated in Table 1 and the results of the chemical extrac-tions are representative of the media used during the totaltime of the column experiments. This was ensured byrepeating the BET measurements and chemical extractionsfor solid samples collected (i) after conducting pH 2 HClblank injections and (ii) at the end of the column experi-ments. Samples of natural sands used for SEM and XRDanalyses were prepared with the same pre-treatment ofthe porous media used in the column experiments. Theeffluent H+ concentrations Cout were calculated from thecontinuously measured pH in the flow-through vial at theoutlet of the columns using the continuous stirred tankreactor (CSTR) equation for removing dilution effects (seeSection EA-1 of the Electronic Annex).

2.2. Analytical methods

In-line pH measurements in column experiments wereperformed using Hach Intellical PHC281 electrodes. Theelectrodes were calibrated before each experiment withstandard solutions of pH 4, 7 and 10 with an accuracy of0.02 pH units. Na, Mg, Ca, K, Fe, Al concentrations weremeasured from 2% HNO3 acidified samples and analyzedwith inductively-coupled-plasma optical emissionspectroscopy (ICP-OES, PerkinElmer Avio 200, in axialand/or radial mode). Chloride and bromide concentrationswere determined from unacidified samples using ionchromatography (IC, Dionex ICS-5000, ThermoScientific).

2.3. Characterization of the natural sand coatings

According to the vendor product technical informa-tion, the natural silica sands contain minor amounts ofAl, Fe, Na, K, Mg and Ca oxides (<4% and <1% ofthe total mass of SG and SD, respectively). In order toquantify the amount and to differentiate the crystallinityof minerals present in the sand coatings, three methodsof partial-chemical extraction were applied: (i) hydroxy-lamine hydrochloride extraction targeting the poorly crys-talline oxide (Procedure of Fuller et al., 1996: 10 g ofmedium were put in contact with 20 mL of solution(0.25 M NH2OH-HCl in 0.25 M HCl) and continuouslyshaken at 60 �C for 1 h); (ii) dithionite-ammonium-citrate (DC) extraction (Procedure of Coston et al.,1995: a 0.08 M Na2S2O4 in 0.2 M ammonium citrate solu-tion was prepared and pH was adjusted to 8.4 with NH4-OH. 35 mL of solution were brought in contact with 10 gof samples; thereafter, 0.5 g Na2S2O4 were added in eachof the centrifuge tubes that were shaken for 24 h); (iii) Ti-citrate EDTA carbonate extraction reductively dissolvingthe amorphous and crystalline Fe and Al oxides (Heronet al., 1994) (Procedure of Tucillo et al., 1999: a 0.05 MTi3+ solution in 0.05 M Na4-EDTA and 0.05 M Na citratewas prepared from TiCl3 (15% w/v, in 15% HCl); the pHof the resulting solution was adjusted to 7 with NaOH;35 mL of the Ti-citrate EDTA were brought in contactwith the 10 g of samples before adding 3 mL of 1 MNaHCO3 solution; finally, the samples were sonicatedfor 5 minutes and shaken for 24 h). For the three typesof extraction, blanks as well as duplicate samples wereprepared and analyzed. Polyethylene centrifuge tubes wereused to avoid silica contamination. At the end of theextractions, the samples were centrifuged for 15 minutesat 3500 rpm and aliquots were filtered with 0.45 lm filtersprior to analyses.

Fig. 1. Diagram summarizing the 7 flow-through experiments performed for each of the three considered porous media (SG, SD, QB) and thesimultaneous calibration approach of the model-based analysis through parallelization of the simulations for each specific porous medium.


Scanning electron micrososcopy (SEM) was performedwith a Quanta FEG 250 operated at 15 kV to analyze thesurface textures and the presence of coatings on the naturalsands SG and SD. We both inspected carbon-coated sam-ples of full sand grains and flat epoxy-embedded cross-sections of sand grains (polished with 1 lm diamond abra-sive film). For cross-section analyses, the samples werecopper-taped to avoid charging effects. While the sand sur-face morphology was inspected with secondary electrons,back-scattered electrons were used to inspect the mineralassemblage of the sand coatings. Elemental distribution (ex-pressed as atomic percentage [at%]) of the natural sand sur-face was analyzed by energy dispersive X-Ray spectroscopy(EDS) with an X-Max silicon drift detector (Oxford Instru-ments). EDS-SEM maps of coatings cross sections werealso acquired.

In order to gain insight on the natural sand crystal struc-ture, powder X-ray diffraction (XRD) analyses of fine-milled samples of the natural sands SG and SD were carriedout with a PANanalytical empyrian diffractometer in aBragg-Brentano Theta-Theta geometry. The catalysts wereanalyzed in a 10� < 2b < 90� range using a monochromaticCu-Ka radiation (k � 1.5418 A). Mineral phases identifica-tion was performed by profile fitting with the XRD soft-ware HighScore Plus.

3. MODELING APPROACH

Modeling of the column experiments was performed toreproduce the pH and ionic breakthrough measured atthe outlet of the column and to quantitatively interpretthe sorption processes taking place at the surface of thequartz porous media. The model simulates the advectiveand dispersive transport of aqueous species through thehomogeneous porous media and accounts for geochemicalprocesses including aqueous speciation and interaction ofthe aqueous species with the porous matrix via surface com-plexation mechanisms.

3.1. Transport of charged species

The transport of charged solutes in the column setupwas described with the 1-D multicomponent ionic transportequations, accounting for compound-specific diffusive/dis-persive properties and Coulombic interactions betweenthe charged species. Such formulation has been developedfor diffusive transport (e.g., Lasaga, 1998; Boudreauet al., 2004; Appelo and Wersin, 2007) but has been shownto be important to capture cross-coupling interactions dur-ing transport of charged species also in advection-dominated flow-through porous media (Rolle et al., 2013;Muniruzzaman et al., 2014; Rolle et al., 2018). The multi-component ionic transport equation is expressed as:

h@Ci

@tþ 1� hð Þqs

@Si

@t¼ �q

@Ci

@xþ @

@x

XNj¼1

hDijL

@Ci

@x

!ð1Þ

where Ci and Si are the aqueous and adsorbed concentra-tions of a species i, q is the specific discharge, qs is the mass

density of the solids, h is the porosity and DijL is the matrix

of longitudinal cross-coupled dispersion coefficients of thecharged species in the pore water defined as(Muniruzzaman and Rolle, 2017):

DijL ¼ dijDi

L �zizjDi

LDjLCiPN

k¼1ðz2kDkLCkÞ

ð2Þ

where DiL is the longitudinal hydrodynamic dispersion coef-

ficient of species i, dij is the Kronecker index which is equalto 1 when i = j and 0 when i–j, zi is the charge of species i,

N is the number of mobile species. For each specie i, DiL was

calculated with a linear parameterization (Guedes deCarvalho and Delgado, 2005; Kurotori et al., 2019):

DiL ¼ hDi þ 0:5dv ð3Þ

where Di, is the aqueous self-diffusion coefficient reportedby (Lasaga, 1998) at T = 18 �C and corrected for tempera-ture and viscosity at T = 20 �C (Table EA6 of the Elec-


tronic Annex), d is the average grain size and v is the seep-age velocity.

3.2. Surface complexation model

We initially performed forward simulations using sur-face complexation databases for well-characterized miner-als (i.e., quartz, ferrihydrite, hematite, goethite,corundum). These simulations were conducted by usingthe mass of oxides determined by chemical extractionsand by scaling the specific surface area of these mineralsto the total measured BET surface area (Tables EA3 andEA4 in the Electronic Annex provide a complete overviewof the model parameters). The comparison between the sim-ulations outcomes and the measured pH was used, togetherwith the experimental characterization of the natural sandssurfaces, to identify the mineral phases affecting transportof charged species in the columns.

The natural sand surface was described using the com-

ponent additive approach (Davis et al., 1998), consideringtwo types of reactive surfaces exposed to the pore water:quartz and a generic metal oxide. The metal oxideaccounted for both iron and aluminium oxides presentin the coatings as these phases have similar surface behav-ior. The CA approach is particularly suited for thedescription of the sorption processes at the surface of nat-ural silica sand as it can distinctively account for the verydifferent surface charge behavior of quartz and Fe/Al oxi-des (Hiemstra et al., 1989). The sets of surface reactionsfor the considered mineral phases were defined similarlyin the three SCMs developed for the two types of naturalsand and for the glass beads; however, the affinity con-stants were calibrated independently for the differentmedia. A parameter calibration was performed in orderto retrieve a single set of surface complexation parameterscapable of reproducing the 7 column experiments per-formed for each medium under a wide range of ionicstrengths [0–100 mM] and pH [2.5–5.6]. For the quartzbeads, the SCM did not include the metal oxidecomponent.

The main difference between the SCMs that can befound in the literature resides in the interfacial structuresthat have been proposed to describe the electrostatic inter-actions as well as the positions of the ions present at themineral-solution interface (Koretsky, 2000). Here, adsorp-tion processes on quartz and metal oxides were simulatedby applying the Basic Stern model (BS) and the triple layermodel (TLM), respectively. These models were chosenbased on the following reasons:

� They can capture the influence of major ions on the sur-face protonation/deprotonation over a large range ofionic strength through the explicit consideration ofouter-sphere complexes formation (Hayes and Leckie,1987; Hayes et al., 1990; Venema et al., 1996);

� They account for the effects of the electrostatic potentialfield on the sorption reactions by including an electro-static correction term in the equilibrium reactionconstants;

� The TLM for metal oxide does not require structuralinformation at the molecular level (Hiemstra, 2018) thatmight be complicated to retrieve from complex mineralassemblages;

� The BS and TLM have been widely used for modelingsorption onto quartz (Stumpf et al., 2008; McNeeceand Hesse, 2016, 2017; McNeece et al., 2018; Garcıaet al., 2019) and Fe/Al oxides (Koretsky, 2000;Villalobos and Leckie, 2001; Peacock and Sherman,2004), respectively.

The description of quartz surface protonation behaviorwith SCMs is generally performed with the definition of asingle surface site, using a simple 1-pKa approach andneglecting the protonation of silica leading to positivelycharged surface groups (Riese, 1982; Hiemstra et al.,1989; Sverjensky and Sahai, 1996; Stumpf et al., 2008;Wang and Revil, 2010). However, recent studies havepointed out the inability of this approach to capture thesorption processes at the quartz-liquid interface in flow-through systems (McNeece and Hesse, 2016, 2017;McNeece et al., 2018). This disagreement between modelingoutcomes and observations suggests that additional surfacemechanisms need to be included in SCM in order to simu-late complex surface behavior of quartz. In particular,recent findings have highlighted the complex bimodal acid-ity behavior of silica resulting from the presence of bothpositively and negatively charged silanol sites even thoughthe overall silica surface remains negatively charged withinenvironmentally relevant pH ranges (i.e., pH > 2) (Onget al., 1992; Duval et al., 2002; Leung et al., 2009; Sulpiziet al., 2012; Pfeiffer-Laplaud et al., 2015). To account forthe effects of the different types of silanol sites, the quartzsurface was represented with an amphoteric behaviorthrough the definition of a single type of surface site thatcan undergo deprotonation and protonation reactions, aswell as the formation of Na, Cl and Br outer-sphere com-plexes that compensate the 0- plane charge deficiencyand/or surplus. The total site density was fixed to 4.6sites/nm2 (Hiemstra et al., 1989; McNeece and Hesse,2016; Garcıa et al., 2019) and the affinity constants control-ling the surface reactions were part of the calibrationapproach. Assuming that the reactive surface area of sandwas equal to the total measured BET surface area, the sur-face area of quartz was defined as:

Aquartz ¼ ABET � Aoxide ð4Þwhere ABET is the total BET measured surface area andAoxide is the calibrated surface area of the metal oxide.For quartz beads, we used the calculated surface areaassuming a perfect spherical shape as it was not possibleto determine this value with BET. Furthermore, the valueof the Stern layer capacitance (C1,quartz) for quartz beadswas fixed to 3.3 F/m2 (Hiemstra et al., 1989) whereas, forthe natural sands, C1,quartz was optimized to capture theobserved behavior of H+. The mass of oxide in the modelwas directly calibrated within the range determined fromthe extraction experiments whereas the site density, the sur-face area and the inner-layer capacitance C1,oxide were cali-


brated within range of values typical for iron/aluminiumoxide surfaces found in the literature. The outer-layercapacitance C2,oxide was fixed to 0.2 F/m2 (Hayes et al.,1990). The set of surface complexation reactions comprisedprotonation/deprotonation reactions and the formation ofNa, Cl, Br and carbonate outer-sphere complexes. The car-bonate reaction stoichiometry involving a monodentatecarbonate surface complex based on the work ofVillalobos and Leckie (2001) was included. In order to fur-ther constrain the calibration based on known quartz andmetal oxides surface properties, the affinity constants ofthe deprotonation and protonation reactions of thesephases were linked through their point of zero charge bydefining the relationship (Wang and Revil, 2010):

pHPZC ¼ 1

2logK1- logK2ð Þ ð4Þ

where pHPZC is the pH at which PZC is achieved, logK1

and logK2 the affinity constants of the protonation anddeprotonation reactions, respectively. For metal oxideand quartz, pHPZC was calibrated within range of literaturevalues. An overview of the surface reactions as well as thecalibration range and calibrated values of the surface com-plexation parameters for the three implemented SCMs isshown below in the Results section (Table 3).

3.3. Numerical implementation and calibration strategy

The 1-D reactive transport model was implemented inthe geochemical code PHREEQC-3 (Parkhurst andAppelo, 2013) coupled with Matlab� by using the IPhreeqcmodule (Charlton and Parkhurst, 2011). Coupling of thesecodes enables to combine the transport and geochemicalcapabilities of PHREEQC with the computational methodsof Matlab� (e.g., Haberer et al., 2015b; Muniruzzaman andRolle, 2016). This allows the parallelization of multipledatasets simulations as well as comprehensive model cali-bration (Stolze et al., 2019a). The 1-D domain representingthe column setup was discretized into 50 grid-cells(Dx = 2.88 mm) and the total time of the simulation wasdivided into a number of time steps (Dt = [215–272] sec-onds) based on the measured flow rates and estimatedporosities (Tables EA1 and EA2). The multicomponentionic transport capabilities of PHREEQC (Appelo andWersin, 2007) were used to couple the hydrodynamic dis-persive fluxes. The Basic Stern model was implemented inPHREEQC by setting the second capacitance for the con-sidered surfaces to a large number (i.e., 1000 F/m2,McNeece and Hesse, 2016). The thermodynamic databaseWATEQ4f was used to calculate the aqueous speciationand reactions between dissolved species. Activity coeffi-cients of aqueous species were calculated with the Debye-Huckel and Davis equations as specified in the WATEQ4fdatabase.

The laboratory dataset collected from the 21 columnexperiments was divided with respect to the considered por-ous media (i.e., 7 experiments for SG, SD and QB, respec-tively). Modeling of the flow-through experiments andcalibration of the surface complexation parameters for agiven medium was performed simultaneously by running

in parallel the reactive transport simulations of the corre-sponding seven column experiments (Fig. 1). This paral-lelization approach was applied in order to test andimplement single sets of SCM parameters for each solidmatrix. A single simulation of a column experiment con-sisted of the following steps:

� Speciation calculation to determine the surface and porewater composition in the 1-D domain using the initialpH measured at the outlet of the column;

� Definition of the inlet solution based on the measure-ments of pH and ion concentrations;

� Simulation of the advection and multicomponent ionicdispersion of aqueous species;

� Surface complexation modeling assuming local equilib-rium after each transport step.

The surface complexation parameters (i.e., affinity con-stants, capacitances, surface area and mass of accessibleiron oxides for aqueous solutes) were calibrated by mini-mizing the sum of the normalized root mean square errors(NRMSEi) calculated between the simulated and measuredconcentrations of charged species i (i.e., H+, Na+, Br� andCl�) at the outlet of the column. Normalization was per-formed by dividing the RMSE of a given species by themaximum measured concentration of the correspondingdataset. Search for optimum solution was performed usingthe particle swarm optimization (PSO) method imple-mented in the MATLAB� environment (Rawson et al.,2016; Stolze et al., 2019b). This global-search algorithmwas chosen due to the high number of parameters and thenonlinearity of the problem resulting from the interdepen-dence of surface reactions.

4. RESULTS

4.1. Experimental results

4.1.1. Column experiments

During the flow-through experiments, pH was continu-ously monitored at the outlet of the columns. Fig. 2 dis-plays the protons breakthrough curves for the salt(Fig. 2a-f) and acid injection experiments (Fig. 2g-i). Tofacilitate the comparison between breakthrough curves col-lected among the silica porous media, we present the pH(Fig. 2a-c) as well as the H+ concentrations normalized tothe total surface area (Fig. 2d-f) of the pure NaBrinjections.

As shown in Fig. 2a-c, the pH dropped sharply after onepore volume (PV) for all silica media upon pure salt (NaBr)injections. This drop was then followed by a sluggish recov-ery towards the pH of the injected solution and the intensityof the H+ emitted from the column is proportional to theinjected salt concentration in agreement with the experi-mental results of McNeece and Hesse (2017). However,the amount of emitted protons significantly differs amongthe silica media. In particular, it increases by about twoorders of magnitude in the order QB < SD < SG. The low-est pH was found for the 100 mM NaBr injection and val-ues of 2.75, 3.82 and 4.70 were measured for SG, SD and

Fig. 2. Measured breakthrough curves of protons at the column outlets. (a)-(c) pH and (d)-(f) H+ concentrations normalized to the totalsurface area of the exposed medium during salt (NaBr) injections; (g)-(i) normalized H+ concentration during acid (HCl) injection. Note thatthe acid injection of the lowest ionic strength was performed without NaBr while the salt injection of lowest ionic strength was performed with0.1 mM NaBr background as specified in the legend.

Table 2Solid concentrations in [lmol/g] measured from the chemicalextractions performed on the natural sands. Values are average ofduplicate samples.

Fe Al Si

Hydroxylamine hydrochloride

SG 0.13 0.56 NDSD 0.28 0.25 NDTitanium chloride

SG 0.86 0.46 0.37SD 1.20 0.30 0.39Dithionite ammonium citrate

SG 1.91 0.65 0.55SD 2.21 0.49 0.67


QB, respectively. The concentrations of released protonsnormalized to the total exposed surface area of the mediaconsidered here (Fig. 2d-f) indicate that the release of H+

cannot be linearly related to the measured surface area asillustrated by the significant stronger release for SG com-pared to SD and QB. Furthermore, there seem to be a max-imal capacity of proton release from the media asillustrated by the similar H+ breakthrough in the case ofthe 100 and 10 mM NaBr injections for SG and SD (Fig. 2-a-b, d-e).

Differences in the H+ breakthrough curves between thesilica porous media were also observed for the acid injectionexperiments and interesting H+ transport behaviorsoccurred for natural sands. For SG, in the case of the acidinjection with 100 mMNaBr (Fig. 2g), a sharp wave of pro-tons was visible at 1 PV, before the pH stabilized to thevalue of the injected solution (i.e., 2.94). In this case, thepH dropped to 2.78 which is similar to the minimum pHmeasured upon injection of the 100 mM NaBr at neutralpH. Thus, this fast proton wave can be directly attributedto the presence of NaBr in the injected solution. The effectof 100 mM NaBr was very different for SD during the acidinjection (Fig. 2h). A fast traveling proton wave can be seenat 1 PV before the arrival of the main acidic front whichwas significantly retarded compared to the other acid injec-tions. When lower concentrations of NaBr were used in theinjected acid solutions, the fast-traveling proton wave wasnot observed (dotted lines in Fig. 2g-h). In particular, forthe injections of pure HCl, the proton front was monotonicand showed a more sluggish increase for both SD and SG.In contrast to the natural sands, the H+ breakthrough forQB during acid injections occurred at 1 PV and suggestsconservative transport of HCl for all ionic strengths(Fig. 2i).

4.1.2. Chemical extraction of natural sand coatings

The concentrations of elements relative to the mass ofsand measured from chemical extraction are presented inTable 2. The three extractions show that the metal oxidescontained in the coatings represent a very small fractionof the natural sand mass: <0.02% and <0.007% for Feand Al minerals, respectively (i.e., assuming an oxide molarmass of 100 g/mol) in agreement with the results of Costonet al. (1995) and Fuller et al. (1996). Although the results ofthe extraction are very similar for both sands, SG containsslightly less Fe and more Al than SD. Furthermore, thehigher amount of Fe obtained during more aggressiveextractions (i.e., titanium chloride and dithionite ammo-nium citrate) suggest that the Fe phases are predominantlycrystalline (93.2% and 87% of the total Fe for SG and SD,respectively). Differently, poorly crystalline aluminium oxi-des seem to dominate in the coatings (i.e., 86% and 51% forthe total Al for SG and SD, respectively). These oxidesmight be associated with silica as aluminosilicates accord-


ing to the similar amount of extracted Si and Al. We did notfind any evident correlation between extracted Al and Fe(R2 = 0.14) contrary to the observations of Fuller et al.(1996) suggesting that mixed Fe-Al oxides coatings do notdominate. Ca, Mg and K were measured in very lowamount (<0.07 lmol/g).

4.1.3. Microscopy analysis and crystal structure of the

natural sands surfaces

The SEM inspection of the grains (Fig. 3) and the SEM-EDS analyses (Fig. 4) provide insights on the surface tex-ture of the samples and on the composition of the mineralphases exposed to the pore water and therefore accessiblefor the aqueous solutes during the column experiments.

SEM images of the natural sand surface acquired at dif-ferent magnifications suggest that quartz is directly exposedto the pore water for both natural sands. Presence of coat-ings was shown by SEM backscattered electrons images inthe form of agglomerates/flakes with a patchy distributionover the surface of the sands (Fig. 3b-c for SG and Fig. 3-f-g for SD). However, the secondary and backscatteredelectrons images reveal higher surface roughness and coat-ings having larger crystal size for SG (Fig. 3a-c). In compar-ison, the surface of SD is smoother and identifiable coatingsaggregates are smaller (Fig. 3e-g).

Selected areas and point EDS-SEM measurements per-formed at several locations similarly indicate the direct expo-sure of quartz to the pore water (e.g., spectra SG1, SG7 andSD1, SD7 shown in Fig. 3d and h). The spectra also show thewidespread presence of small amounts of Al over the surfacealthough it was not possible to distinguish any clear phasewith the backscattered electron images. This corroboratesthe results of the chemical extractions suggesting that Alexists in the form of alumino-silicates distributed over thenatural sand surfaces. In contrast, iron was detected (i.e.,spectra SG3, SG9 and SD4, SD12 shown in Fig. 3d and h)at specific locations where agglomerates were visible on thebackscattered electron images Fig. 3b-c e-f). Small presenceof Mg and K could only be detected in the coatings of SG(e.g., spectra SG3 and SG9 shown in Fig. 3d).

SEM images and SEM-EDS mapping acquired at theedge of the natural sand cross-sections provide furtherdetails regarding the composition and the thickness of thecoatings identified at the surface of the sand grains. As anexample, Fig. 4 shows one location of SG at which thisanalysis was performed. Results of similar analyses per-formed at other spatial locations of SG and SD are shownand discussed in Section EA-2 of the Electronic Annex.Although preparation of the samples could have removedpart of the coatings, tenth of micrometers thick cavitiesfilled with oxides can be observed at the surface of the nat-ural sands (Fig. 4). Backscattered electron images andSEM-EDS maps revealed that the coatings consist in amix of Si, Al, Fe, Mg, K and Ti mineral phases (Fig. 4d-n). These images also indicate that the coatings have aheterogeneous composition. In particular, Al, Fe and Mgseem to be present over the entire coating areas with a pat-chy spatial distribution. In contrast, spatial variabilities ofK and Ti are more pronounced with contents clearly spik-ing at specific locations (Fig. 4g and Fig. 4l-n).

SEM-EDS data were used to quantify the relative pres-ence of the identified elements and their spatial distributionat the scales of the sand grain [mm] and of the coatings [lm](Section EA-2 of the Electronic Annex). These measure-ments indicate that, for both types of natural sands, thecoatings are mostly composed of Si ([21–25] at%) and Al([9–14] at%) with a 1:2 Al:Si ratio. In comparison, Fe con-centrations were globally lower and showed significant spa-tial variability at the mm scale with SD (average of 3.01[0.36–6.81] at%) exhibiting higher concentrations than SG(average of 0.92 [0.57–1.71] at%). Within individual bodiesof identified coatings (as shown for instance in Fig. 4), boththe Al and Fe contents were found to be heterogeneouslydistributed at the lm scales (i.e., relative standard deviation– RSD - for Al was 38 [18–69]% and 31 [21–49]% for SG andSD, respectively; RSD for Fe was 57 [42–74]% and 54 [33–73]% for SG and SD, respectively). The high variability ofthe calculated Fe/Al ratios (RSD 59% and 125% for SGand SD, respectively) suggest that Fe and Al are not linearlyassociated. This result is in agreement with the chemicalextractions and suggests that the iron oxides are embeddedwithin Al-, Si- rich phases as reported by Penn et al. (2001).However, contrary to the chemical extractions, the SEM-EDS quantifications indicate a higher Al molar content thanFe. This discrepancy suggests either that: (i) the main bodyof the thick coatings did not dissolve during the extractionwith the latter only able to dissolve a minor reactive fractiondominated by iron oxides, (ii) these thick coatings dissolvedduring the extraction but are minor compared to other sur-face coatings that are not visible on the cross-section images(i.e., nano-thick coatings), or (iii) possible heterogeneitieswere not captured by the SEM-EDS analysis. At mm scale(i.e., size of the grain), K ([0–2.27] at%), Mg ([0–1.38] at%), Cl ([0–0.29] at%) and Ti ([0–1.4] at%) molar contentswere low although Ti and K molar concentration at lmscale reached 15.6 at% and 7.57 at% in zones of high concen-trations, respectively (e.g., Fig. 4g, l and n).

Regarding the crystal structure of the sand grains, XRDanalysis of the two natural sands showed that a-quartz (i.e.,low-temperature ordinary polymorph quartz) with a hexag-onal crystal structure was the dominant constituent of bothnatural sands SG and SD with minor amount of alumino-silicate minerals (i.e., zeolite). XRD also indicated thatSG, contrary to SD, includes additional quartz crystalstructures having higher densities (i.e., [2.91–4.12] g/cm3)than typical quartz low crystal (i.e., 2.64 g/cm3). Fe oxidescould not be detected in any of the sands suggesting thattheir presence is limited to the sand surface in very lowamount and in the forms of coatings in agreement withthe chemical extraction and the microscopy characteriza-tion. Results of the XRD analysis are provided inSection EA-2 of the Electronic Annex.

4.2. Modeling results

4.2.1. Reactive transport simulations of the column

experiments

In order to identify the mineral phases controlling theinteractions between protons, major ions and the naturalsand surface, we performed forward simulations applying

Fig. 3. Characterization of the natural sands surfaces and coatings by SEM-EDS images, location of the point/area SEM-EDS measurementsand corresponding selected spectra for SG (a)-(d) and SD (e)-(h). The remaining spectra are shown in Fig. EA6-EA7 of the Electronic Annex.


existing SCMs for quartz and well-characterized oxides.Results of the simulations are provided in Section EA-4of the Electronic Annex. These forward simulations indi-cated that the proton release observed during salt injection(Fig. 2a-c) is primarily controlled by quartz surfacesalthough the quartz SCM could not capture the magnitudeof the released protons concentrations in agreement with

McNeece and Hesse (2017). Furthermore, these simulationsshowed that the oxides present in the coatings of naturalsand lead to the tailing observed in the measured H+ break-through curves during acid injections performed in theabsence of an electrolyte background (Fig. 2g-h). However,none of the existing SCMs could quantitatively reproducethe measured breakthrough curves during salt and acid

Fig. 4. Characterization of the natural sand surface and coatings by backscattered electron images (a)-(c) and SEM-EDS mapping of SGcross-section samples (d)-(n). The zones delimited by the white rectangles correspond to the areas over which SEM-EDS maps were acquired.Relative X-ray intensity for a specific element increases from cold to warm colors; black is non-detected.


injections. Based on the experimental characterization ofthe silica porous media and the outcomes of these initialsimulations, we implemented a reactive transport modelaccounting for both the surfaces of quartz and metal oxidesusing the component additive approach as described in Sec-tion 3 and schematically illustrated in Fig. 5.

Transport of aqueous solutes was satisfactorily simu-lated with the proposed reactive transport model. Fig. 6 dis-plays the comparison between the measured pH andsimulation outcomes obtained with the optimized set ofSCM parameters shown in Table 3. For the sake of clarity,Fig. 6 only shows the results for selected relevant injections:(i) 100 mM NaBr solution, (ii) pure HCl pH 3 solution, and(iii) HCl pH 3 with a 100 mM NaBr electrolyte backgroundperformed in the three considered porous media (SG, SD

Fig. 5. Schematic illustration of the surfaces, sorption sites andsorption mechanisms considered in the reactive transport model,using the component additive approach.

and QB in Fig. 6a-c, d-f and g-i, respectively). Simulationresults for the 12 remaining experiments are shown inSection EA-4 of the Electronic Annex.

Regarding the natural sands, two SCM versions wereimplemented and compared in order to appreciate theimpact of the coatings on the transport of protons. The firstoptimized model only accounted for the contribution ofquartz (shown in blue), the second model version includedadditionally the contribution of the Fe/Al coatings (shownin red). Calibration results indicate that the definition of ametal oxide surface was necessary to achieve a satisfyingagreement with the H+ breakthroughs measured at the out-let of columns packed with natural sand (Fig. 6a-b, d-e andg-h). For SG, the agreement between model results anddata was improved when defining Fe/Al coatings (Fig. 6a,d and g). This is particularly clear in the case of the pureHCl injection (Fig. 6d). Results for SD were different asthe SCM that only included the quartz surface failed atreproducing the observed H+ behavior for all injections(Fig. 6b, e and h). The quartz SCM could, however, par-tially simulate the initial fast appearance of H+ precedingthe retarded front for the injection of HCl with a high elec-trolyte background. This result suggests that the observedH+ transport behavior was primarily controlled by thequartz surface. Thus, the differences between the reactivitiesof the quartz surfaces of SG and SD not only controlled theamount of H+ released but also the pattern of the H+

breakthrough curve. The definition of a metal oxide surfaceresulted in major improvements in the agreement with theexperimental dataset of SD. As expected, the dataset ofquartz beads was satisfyingly reproduced by only defining

Fig. 6. Comparison between measured and simulated H+ breakthrough curves for the three silica porous media considering differentinjections in the column setup: (a)-(c) NaBr 100 mM, (d)-(f) pure HCl and (g)-(i) HCl with a NaBr 100 mM electrolyte background.


a quartz surface in the SCM (Fig. 6c, f and i) in contrast tothe natural sands which require to account for the surfaceof metal oxides present in the coatings.

Interactions between the aqueous solutes and the surfaceof natural sand were also visible from the measured break-through curves of the major cations and anions. Fig. 7shows the measured and simulated breakthroughs of Na+,Br� and Cl� at the outlet of the columns packed withSG, as well as the simulated H+ breakthrough curves.While the transport of anions Cl� and Br� was conserva-tive, retardation of Na+ was visible for the salt injectionsperformed with low NaBr concentrations ([0.1–10] mM).As illustrated by Fig. 7a-c, the breakthroughs of Na+ mir-ror the decrease in H+ concentration following the acidicpH front. Differently, the retardation of Na+ was not visi-ble for the highest injected concentration of NaBr(100 mM, Fig. 7d) although H+ release was highest(Fig. 2a), as in this case the amount of Na+ injected largelyexceeds the amount adsorbed. Sodium retardation was alsonot observed upon the injection of HCl with a 10 mMNaBrelectrolyte background (Fig. 7f) suggesting that acidic con-ditions hamper the adsorption of Na+ when concentrationof NaBr is relatively low. Adsorption of Na+, triggering therelease of H+ from quartz, remains important at high NaBrconcentration as illustrated by the surplus of H+ observedduring injection of HCl with a 100 mM NaBr electrolytebackground (Fig. 2g). Retardation of Na+ was lower dur-ing transport through the other silica porous media (SDand QB) in agreement with the lower amount of protonsreleased (Figs. EA18 and EA19).

4.2.2. Surface composition of the silica porous media

A quantitative interpretation of the propagation of H+

and major ions through the columns was performed withthe implemented reactive transport model. Here, we focus

on the injection of a 100 mM NaBr solution in a columnpacked with SG as this experiment led to major interactionsbetween NaBr and quartz. Fig. 8 displays simulated spatialprofiles of pH (Fig. 8a-b) as well as the change in surfacecomposition (Fig. 8c-f) and surface charge of quartz(Fig. 8g-i) relative to initial conditions of the experimentsalong the length of the column. Similar results for injectionsof HCl solutions with and without an electrolyte back-ground are shown in Section EA-4 of the Electronic Annexand briefly summarized below.

Fig. 8a-b shows that the H+ released from the sand sur-face increases gradually along the flowpath, as the plume ofNaBr propagates through the pore domain and, thus, con-tacts an increasing amount of quartz surface. The simulatedchange in surface composition of quartz indicates that therelease of H+ primarily results from the formation of Naouter-sphere complexes and concomitant decrease in SiOH

(initial amount2.85 mmol/L) and, to a lower extent, SiOHþ2

(initial amount 0.34 mmol/L) surface species that compen-sates the positive charge brought to the surface by Na+

(Fig. 8b-e). In comparison, the change in SiO� (initialamount 0.37 mmol/L) concentration was predicted to besignificantly lower (i.e., <4 � 10�2 mmol/L). The effect ofNa+ on quartz protonation is also illustrated by Fig. 8gand h showing that the decrease in the charge of the 0-plane mirrors the increase in surface charge of the surface1-plane. In comparison, the charge of the 2-plane is lessaffected (Fig. 8i).

Integration of the DHþaq curves expressing

the net increase of H+ in the column pore water (e.g.,6.91, 21.64, 36.26 mmol/L at 1, 2 and 3 h,respectively) was found slightly lower than the decrease insurface protons (i.e., 8.12, 25.55 and 42.84 mmol/Lat 1, 2 and 3 h, respectively; calculated as

C ¼ R L

0DSiOHþ 2 DSiOHþ

2 þ��

DSiOH2BrÞÞdx). The

Table 3Quartz and metal oxide surface reactions and optimized surface complexation parameters used to simulate the interaction between theaqueous charged solutes and the silica porous media.

Quartz (Basic Stern)

Calibration range SG SD QB

Surface area Aquartz [m2/g sand] Atotal-Aoxide 0.129 0.014 0.003

Site density Nquartz [sites/nm2] 4.6 4.6 4.6 4.6

Capacitance Cquartz [F/m2] [1 ; 50] 8.55 12 3.3 (fixed)

pHPZC, quartz [–] [1 ; 5](a�d) 1.80 4.98 3 (fixed)

Reactions, affinity constants and charge allocations (Dz0 and Dz1)

Calibration range logK

(SG)

logK

(SD)

logK (QB) Dz0 Dz1

SiOHþHþ ¼ SiOHþ2 (logK1, quartz = pKa1, quartz) 2pHPZC + logK2, quartz 0.90 6.49 0.34 1 0

SiOH ¼ SiO- þ Hþ(logK2, quartz = -pKa2, quartz) [-10.3 ; -2.1](e) -2.7 -3.46 -5.66 -1 0SiO- þ Naþ ¼ SiONa [-1.9(f) ; 5.4(g)] 0.76 2.89 1.83 0 1SiOHþ

2 þ Cl� ¼ SiOH2Cl [-5 ; 5] -1.90 0.96 -1.9 0 -1SiOHþ

2 þBr� ¼ SiOH2Br [-5 ; 5] 2.15 -0.34 5.00 0 -1

Metal oxide (Triple layer model)

Calibration range SG SD

Mass moxide [mg/g sand] Based on extraction 12 260Surface area Aoxide [m

2/g oxide] [10(h) ; 1260(i)](j�l) 968 61Site density Noxide [sites/nm

2] [2(m) ; 30.5(b)](j, m�o) 8.70 30.50Capacitance inner-Stern layer C1, oxide [F/m

2] [0.6(b) ; 1.9(n)](j, m, o) 1.90 1.38Capacitance inner-Stern layer C2, oxide [F/m

2] 0.2 0.2 0.2pHPZC, oxide [-] [7 ; 10](h�p) 9.03 7.18

Reactions, affinity constants and charge allocations (Dz0 and Dz1)

Calibration range logK

(SG)

logK

(SD)

Dz0(SG/SD)

Dz1(SG/SD)

MeOHþHþ ¼ MeOHþ2 (logK1, oxide) 2pHPZC + logK2, oxide 9.31 5.50 1 0

MeOH ¼ MeO- þ Hþ(logK2, oxide) [-12.3(n) ; -8.69(j)](m�o) -8.74 -8.86 -1 0MeO- þ Naþ ¼ MeONa [-0.22(j) ; 3.7(n)](m�o) -0.22 -0.21 -1 1MeOHþ

2 þ Cl- ¼ MeOH2Cl [1.7(b) ; 4.5(o)] 2.20 4.5 1 -1MeOHþ

2 þ Br- ¼ MeOH2Br [-5 ; 5] 1.46 5 1 -1MeOHþHþ þ CO2-

3 ¼ MeOCO-2þ H2O [5 ; 20] 17.40 17.69 -0.13/-0.27 -0.87/-0.73

(a)Sverjensky and Sahai (1996); (b)Koretsky (2000); (c)Duval et al. (2002); (d)Sulpizi et al. (2012);(e)Pfeiffer-Laplaud et al. (2015); (f)Hiemstraet al. (1989); (g)Garcıa et al. (2019); (h)Weerasooriya et al. (2001); (i)Hiemstra (2010); (j)Peacock and Sherman (2004); (k)Dzombak and Morel(1990), (l)Han and Katz (2019) (m)Sahai and Sverjensky (1997a,b); (n)Sverjensky (2005); (o)Fukushi and Sverjensky (2007); (p)Kosmulski (2009).References provided outside of the brackets [ ] supported the choice of the implemented calibration interval.


increasing discrepancy with time between protons in solu-tion and protons released from quartz results from the pro-tonation of the metal oxides present in the coatings.

In comparison to the effects of NaBr, injection of anHCl acidic solution has a minor effect on the surface com-position of quartz (Fig. EA20). In contrast, the effects ofNaBr on the surface composition of quartz during the acidinjection with a 100 mM NaBr background electrolyte areshown to remain significant although the amount of H+

emitted from the surface is predicted to be lower than forthe case of the 100 mM NaBr injected alone (i.e., 4.71,12.54 and 20.59 mmol/L released from the surface at 1, 2and 3 h after the start of the injection, respectively,Fig. EA21). The lower release of protons from the surfacedirectly results from an increase stability of SiOH groupand a reduction of formation of Na outer-sphere complexesdue to the more acidic conditions induced by HCl.

However, the release of protons remains important relativeto the amount of HCl injected as shown by the significantsurplus of H+ measured at 1 PV (Fig. 2g).

5. DISCUSSION

The column experiments and simulations performedhave shown that silica surfaces have a macroscopic impacton the pore water hydrochemistry and on the breakthroughof protons and other ions. In particular the uptake of Na+

by quartz can lead to release of H+ resulting in strong acidicconditions. The difficulty of existing SCMs and thermody-namic database in describing the coupling between ionicstrength and proton surface charge at the quartz-waterinterface for natural porous silica sand (McNeece andHesse, 2016, 2017; McNeece et al., 2018) needs to beaddressed. This limitation might primarily be ascribed to

Fig. 7. Comparison between measured and simulated breakthrough curves of protons and ions for the 7 column experiments performed withSG.

Fig. 8. Simulated spatial profiles of pH, composition (expressed in mmol/L water) and charge distribution of the quartz surface during100 mM NaBr injection in a column packed with SG. D values express the change in concentrations with respect to conditions at t = 0.


the limited attention that silica sand surface properties havereceived compared to iron and aluminium oxides in surfacecomplexation modeling. In particular, the surface represen-tation of the interactions occurring at the silica-water inter-face is still debated, although the need to account for theamphoteric behavior of quartz is increasingly apparent(Pfeiffer-Laplaud et al., 2015; McNeece et al., 2018;Garcıa et al., 2019). Our comparison between different silicaporous media reveals that quartz surfaces can exhibit signif-

icant differences in their surface acidity behavior. In fact,significantly more protons are released from natural sandsthan from quartz beads upon contact with cations despitethe capacity of Fe/Al oxides present in the coatings to buf-fer the decrease in pH. By accounting for the amphotericbehavior of quartz, the implemented reactive transportmodel could reproduce the propagation of H+ observedin our column experiments. As a result of the very differentsurface behavior, implementation of distinct sets of SCM


parameters was necessary to reproduce the experimentaldatasets collected from the different silica porous media(Table 3). In particular, the protons transport in quartzbeads was modeled with parameter values similar to previ-ously reported literature values for quartz (Hiemstra et al.,1989). In contrast, such parameters for quartz in naturalsands significantly differ from classical quartz SCM param-eters. The differences can be summarized as:

� Significantly lower equilibrium constants for the depro-tonation reactions (i.e., pKa2, SG = 2.7 and pKa2,SD = 3.46) reflect a higher acidity behavior than previ-ously reported values (i.e., pKa2 = [6.9–8.8]; Riese,1982; Hiemstra et al., 1989; Sahai and Sverjensky,1997b; Koretsky, 2000; Wang and Revil, 2010).

� While the protonation reaction equilibrium constants ofSG (pKa1,SG = 0.9) is similar to values of classical SCMfor quartz, the calibrated protonation constant of SD issignificantly higher (pKa1, SD = 6.49).

� The calibrated inner-layer capacitances C1 were foundhigher (i.e., C1,SG = 8.55 and C2,SD = 12 F/m2) thancommonly used values (i.e., [2–3.3] F/m2; Hiemstraet al., 1989; McNeece and Hesse, 2016; Garcıa et al.,2019).

These discrepancies can be interpreted in the light ofinterfacial chemistry studies focusing on the quartz-liquidinterfaces. In particular, molecular dynamics simulationsas well as second harmonic generation and other spec-troscopy techniques have revealed that the macroscopicamphoteric behavior of quartz is controlled by a varietyof silanol surface groups exhibiting different acidic charac-ter (Ong et al., 1992; Duval et al., 2002; Leung et al., 2009;Sulpizi et al., 2012). The relative abundance and acid-baseproperties of these surface sites directly result from the localsurface environment (i.e., the surface roughness and thehydrophobicity/hydrophilicity properties of the silanolgroups; Leung et al., 2009). For instance, Pfeiffer-Laplaudet al. (2015) showed that convex geminals groups presentin surface defects and hydrophilic vicinals groups exhibitan acidic behavior (i.e., lower pKa values) whereas, concavegeminals, isolated and hydrophobic surface groups have ahigher tendency for protonation (i.e., higher pKa). Interest-ingly, the low values of calibrated deprotonation equilib-rium constants for natural sand quartz are similar to pKavalues of acidic silanol sites determined from moleculardynamic simulations: [2.1–2.9] (Pfeiffer-Laplaud et al.,

Fig. 9. Surface charge r0 of quartz surfaces as a function of pH simulconsidered in this study: SG (a), SD (b) and QB (c). These simulati(concentrations indicated in the legend) and the model parameters reportthe column experiments.

2015), 3.8 (Darlington and Gibbs-Davis, 2015). Therefore,the observed acidic behavior likely results from the highersurface roughness (i.e., higher surface curvature) of the nat-ural quartz surface with respect to the quartz beads thathave significantly lower surface defects. This surface reac-tivity property is inverse to the effect of crystal face contri-butions of metal oxides, for which more surface roughnessfavors proton accumulation (Rustad and Felmy, 2005;Gaigeot et al., 2012; Han and Katz, 2019). The presenceof different quartz crystal structures among natural sandsas shown by the XRD analysis and/or impurities in quartz(e.g., in the form of aluminosilicates) might further influ-ence the quartz surface protonation behavior and con-tribute to surface differences such as those observed inthis study between SG and SD. In fact, for quartz of higherdensities, silicon is coordinated with a higher number ofoxygen atoms (Iler, 1979) and likely leads to different sur-face sites configurations than for quartz of lower density.The high inner layer capacitance (C1) values for naturalsands might be more difficult to interpret. However, highervalues of C1 imply shorter distances between outer-spherecomplexes and the solid surface which corroborate recentobservations of inner-shell interfacial complexes (i.e., chem-ical binding) of alkali ions onto the quartz surface (Pfeiffer-Laplaud and Gaigeot, 2016a, 2016b).

In order to illustrate the different surface charge behav-ior of quartz, we performed series of batch calculations thatconsisted in modeling the change in surface charge ofquartz when varying the pH. This was performed usingthe SCMs of the three silica porous media considered inthe flow-through column experiments and by extrapolatingtheir predictions to a broader pH range. The simulationresults (Fig. 9) show that the shapes of the r0-pH curvessignificantly differ although a 2-step surface charge behav-ior with varying pH can be identified for all quartz surfacesin agreement with the work of Garcıa et al. (2019). Thecapability of the SG quartz to release protons under highionic strength conditions is shown by the faster decreasein surface charge with increasing pH compared to the twoother types of quartz. The simulated surface charge of theSD quartz reaches significantly higher values at lower pHdue to the higher protonation reaction equilibrium con-stants. In contrast the surface charge curves for quartzbeads exhibit smaller changes under acidic conditions, sim-ilarly to previously reported results (Bolt, 1957; Riese,1982).

ated with the SCMs calibrated for the three types of silica mediaons were performed considering a NaCl electrolyte backgrounded in Table 3. The grey areas show the range of pH encompassed in


Notice that these modeling results can only be consid-ered as a preliminary attempt of extrapolating the differ-ences of the quartz acid-base behavior observed in ourflow-through experiments to batch systems. In fact, silanolsites with less acidic behavior might significantly impact thecharging curve under basic conditions considering their sig-nificantly higher pKa (3–7 pH units; Sulpizi et al., 2012).The limitation of single-site model for describing the charg-ing behavior of silica has been reported by McNeece et al.(2018) who noted a substantial degradation of the modelquality under alkaline conditions when applying parame-ters implemented for acidic conditions. Therefore, multi-sites models such as the one proposed by Garcıa et al.(2019) might be more suited for the mechanistic descriptionof the adsorption processes occurring at the quartz surfaceover broad pH ranges.

6. CONCLUSION

We investigated the propagation of pH fronts in silicaporous media and the electrostatic interactions between pro-tons, major ions and silica surfaces. Flow-through columnexperiments were performed with different types of silicaporous media to assess differences in the reactivity of thequartz surfaces. The experimental datasets consisted of mea-sured breakthrough curves of pH and ions, chemical extrac-tions, SEM inspection of the silica surfaces, and X-Raypowder diffraction characterization of the crystal structure.A reactive transport model was implemented to interpret theexperimental datasets. The model included a surface com-plexation formulation accounting for the distinct contribu-tion of the quartz and the Al/Fe coatings surfaces presentin natural sands based on the component additive approach.The CA approach was shown to be particularly importantfor reproducing the transport of aqueous charged species,taking into account the very different protonation/deproto-nation behavior of quartz and metal oxides.

Our experimental and modeling findings reveal strongsurface interactions between the quartz surface, protonsand sodium. In particular, large amounts of protons can bereleased from the quartz surfaces upon adsorption of Na+.The acidic behavior of quartz significantly differs amongthe silica porousmedia since quartz surfaces of natural sandswere found to be significantly more reactive than quartz sur-faces. The higher capacity in releasing protons may resultfrom higher surface roughness, from differences in the crystalstructures and/or from the presence of impurities (e.g., alu-minosilicates such as zeolite detected by the XRD measure-ments) that control the electrochemical properties of thesurfaces. This highlights the large variability in reactivity ofnatural silica surfaces and the challenges in estimating sorp-tion parameters from independent observations. Further-more, the low amount of iron oxides spatially distributedover the sand grain surface was found to hinder the releaseof H+ in the pore water from the quartz surface due to theiropposite surface charge behavior (i.e., higher pKa). Thistwo-stepmechanism was shown to strongly affect the surfacecharge of the oxides and, thus, might ultimately have animpact on the mobility of trace contaminants.

The strong effects of quartz on macroscopic hydrochem-ical conditions point towards the need for explicit consider-ation of the chemical and electrostatic processes takingplace at the quartz surface when exposed to the pore waterin order to accurately describe solutes transport and con-taminants mobility in the subsurface. This may be particu-larly important when sudden change in the ionic strengthconditions occur such as during managed aquifer recharge(MAR) or salt water intrusion. Our experiments were per-formed with natural porous media and under-well con-trolled laboratory conditions. We think that experimentaland modeling studies should further investigate the surfacebehavior of complex mineral assemblages under naturalgroundwater composition. In these conditions, non-linearsorption behavior resulting from the variety of aqueouscharged species (Stolze et al., 2019a) will overlap with theinterplay between mineral phases having different surfacebehavior. Furthermore, the change in pH conditionsinduced by the quartz surface might trigger the dissolutionof minerals (e.g., calcite) leading to further complexity inthe transport behavior of aqueous charged species. Futureresearch should also explore the effects of dimensionalityon surface processes since the coupled electrochemical dis-persion of charged species in multidimensional systems isexpected to result in significantly different mass transferand more complex and spatially-dependent surface/solutioninteractions than in 1-D column setups (Ye et al., 2015a,2015b; Rolle et al., 2018). Furthermore, additional workis required to understand the underlying parameters leadingto the variability of quartz surface properties. This wouldsupport the formulation of a single and flexible descriptionof quartz surfaces similar to ongoing work focusing on ironoxides (Han and Katz, 2019).

Declaration of Competing Interest

The authors declare that they have no known competingfinancial interests or personal relationships that could haveappeared to influence the work reported in this paper.

ACKNOWLEDGMENTS

This work was funded by the Sino-Danish Center forEducation and Research (SDC) with a PhD scholarshipgranted to Lucien Stolze. Massimo Rolle also acknowl-edges the support of the Villum Block Fellowship (VillumFonden, Denmark). We are grateful for the technical assis-tance of Berit Wenzell (DTU Nanolab), Sinh Nguyen andMikael Olsson (DTU Environment). We would like tothank two anonymous reviewers for their helpful commentsand constructive feedback.

In memory of our colleague Takeshi Kasama who wasinvolved in this study.

APPENDIX A. SUPPLEMENTARY MATERIAL

Supplementary data to this article can be found online athttps://doi.org/10.1016/j.gca.2020.03.016.



REFERENCES

Akai J., IzumiK., FukuharaH.,MasudaH., Nakano S., YoshimuraT., Ohfuji H., Md Anawar H. and Akai K. (2004) Mineralogicaland geomicrobiological investigations on groundwater arsenicenrichment in Bengladesh. App. Geochim. 19, 215–230.

Alessi D. S. and Fein J. B. (2010) Cadmium adsorption to mixturesof soil components: Testing the component additivityapproach. Chem. Geol. 270, 186–195.

Appelo C. A. J. and Postma D. (2005) Geochemistry, groundwater

and pollution, second ed. CRC Press, London.Appelo C. A. J. and Wersin P. (2007) Multicomponent diffusion

modeling in clay systems with application to the diffusion oftritium, iodide and sodium in Opalinus clay. Environ. Sci.

Technol. 41, 5002–5007.Battistel M., Muniruzzaman M., Onses F., Lee J. and Rolle M.

(2019) Reactive fronts in chemically heterogeneous porousmedia: Experimental and modeling investigation of pyriteoxidation. Appl. Geochem. 11, 77–89.

Bolt G. H. (1957) Determination of the charge density of silica sols.J. Phys. Chem. 61, 1166–1169.

Boudreau B. P., Meysman F. J. R. and Middelburg J. J. (2004)Multicomponent ionic diffusion in porewaters: Coulombiceffects revisited. Earth Planet Sci. Lett. 222(2), 653–666.

Bretzler A., Stolze L., Nikiema J., Lalanne F., Ghadiri E.,Brennwald M. S., Rolle M. and Schirmer M. (2019) Hydro-geochemical and multi-tracer investigations of arsenic-affectedaquifers in semi-arid West Africa. Geo. Front. 10, 1–17.

Burgisser C. S., Scheidegger A. M., Borkovec M. and Sticher H.(1994) Chromatographic charge density determination ofmaterials with low surface area. Langmuir 10(855), 860.

Charlton S. and Parkhurst D. (2011) Modules based on thegeochemical model PHREEQC for use in scripting andprogramming languages. Comput. Geosci. 37, 1653–1663.

Cornell R. and Schwertmann U. (2003) The Iron Oxides: Structure,

Properties, Reactions, Occurrence and Uses, second ed. VCH,Weinheim, Germany.

Coston J. A., Fuller C. C. and Davis J. A. (1995) Pb2+ and Zn2+

adsorption by a natural aluminium- and iron-bearing surfacecoating on an aquifer sand. Geochim. Cosmochim. Acta 59(17),3535–3547.

Darlington A. M. and Gibbs-Davis J. M. (2015) Bimodal ortrimodal? The influence of starting pH on site density anddistribution at the low salt aqueous/silica interface. J. Phys.Chem. C 119, 16560–16567.

Davis J. A. and Kent D. (1990) Surface complexation modeling inaqueous geochemistry. Rev. Mineral. 23, 177–260.

Davis J. A., Coston J. A., Kent D. and Fuller C. C. (1998)Application of the surface complexation concept to complexmineral assemblages. Environ. Sci. Technol. 32(19), 2820–2828.

Duval Y., Mielczarski J. A., Pokrovsky O. S., Mielczarski E. andEhrhardt J. J. (2002) Evidence of the existence of three types ofspecies at the quartz-aqueous solution interface at pH 0–10:XPS surface group quantification and surface complexationmodeling. J. Phys. Chem. 106, 2937–2945.

Dzombak D. and Morel F. (1990) Surface Complexation Modeling-

Hydrous Ferric Oxide. Wiley, New York.Fakhreddine S., Lee J., Kitanidis P. K., Fendorf S. and Rolle M.

(2016) Imaging geochemical heterogeneities using inverse reac-tive transport modeling: An example relevant for characterizingarsenic mobilization and distribution. Adv. Water Resour. 88,186–197.

Fukushi K. and Sverjensky D. A. (2007) A predictive model(ETLM) for arsenate adsorption and surface speciation onoxides consistent with spectroscopic and theoretical molecularevidence. Geochim. Cosmochim. Acta 71, 3717–3745.

Fuller C. C., Davis J. A., Coston J. A. and Dixon E. (1996)Characterization of metal adsorption variability in a sand andgravel aquifer, Cape Cod, Massachussetts, U.S.A. J. Contam.

Hydrol. 22, 165–187.Gaigeot M., Sprik M. and Sulpizi M. (2012) Oxide/water

interfaces: how the surface chemistry modifies interfacial waterproperties. J. Phys.; Condens. Matter 124106, 1–11.

Garcıa D., Lutzenkirchen J., Petrov V., Siebentritt M., Schild D.,Lefevre G., Rabung T., Altmaier M., Kalmykov S., Duro L.and Geckeis H. (2019) Sorption of Eu(III) on quartz at high saltconcentrations. Colloids Surf. A 578, 123610.

Guedes de Carvalho J. and Delgado J. (2005) Overall map andcorrelation of dispersion data for flow through granular packedbeds. Chem. Eng. Sci. 60, 365–375.

Haberer C. M., Rolle M., Cirpka O. A. and Grathwohl P. (2012)Oxygen transfer in a fluctuating capillary fringe. Vadose Zone

J., 11.Haberer C. M., Rolle M., Cirpka O. A. and Grathwohl P. (2015a)

Impact of heterogeneity on oxygen transfer in a fluctuatingcapillary fringe. Ground Water 53(1), 57–70.

Haberer C. M., Muniruzzaman M., Grathwohl P. and Rolle M.(2015b) Diffuse-dispersive and reactive fronts in porous media:Iron(II) oxidation at the unsaturated-saturated interface.Vadoze Zone J. 14(5).

Han J. and Katz L. E. (2019) Capturing the variable reactivity ofgoethites in surface complexation modeling by correlatingmodel parameters with specific surface area. Geochim. Cos-

mochim. Acta 244, 248–263.Hansel C. M., Benner S. G. and Fendorf S. (2005) Competing Fe

(II)-induced mineralization pathways of ferrihydrite. Environ.Sci. Technol. 39(18), 7147–7153.

Hayes K. F., Redden G., Wendell E. and Leckie J. O. (1990)Surface complexation models: An evaluation of model param-eter estimation using FITQL and oxide mineral titration data.J. Colloid Interf. Sci. 142(2), 448–469.

Hayes K. F. and Leckie J. O. (1987) Modeling ionic-strength effectson cation adsorption at hydrous oxide-solution interfaces. J.Colloid Interf. Sci. 115(2), 564–572.

Heron G., Christensen T. H. and Tjell J. C. (1994) Oxidationcapacity of aquifer sediments. Environ. Sci. Technol. 28, 153–158.

Hiemstra T. (2010). Surface complexation at mineral interfaces:Multisite and charge distribution approach. Wageningen NL:Thesis Wageningen University.

Hiemstra T. (2018) Ferrihydrite interaction with silicate andcompeting oxyanions; Geometry and hydrogen bonding ofsurface species. Geochim. Cosmochim. Acta 238, 453–576.

Hiemstra T., De Wit J. C. M. and van Riemsdijk W. H. (1989)Multisite proton adsorption modeling at the solid/solutioninterface of (hydr)oxides: a new approach. II. Application tovarious important (hydr)oxides. J. Colloids Interface Sci. 133

(1), 105–117.Iler R. (1979) Chemistry of Silica - Solubility, Polymerization,

Colloid and Surface Properties and Biochemistry. John Wiley &Sons.

Jessen S., Postma D., Larsen F., Nhan P. Q., Hoa L. Q., Trang P.T. K., Long T. V., Viet P. H. and Jakobsen R. (2012) Surfacecomplexation modeling of groundwater arsenic mobility:Results of a forced gradient experiment in a Red River floodplain aquifer, Vietnam. Geochim. Cosmochim. Acta 98, 186–201.

Johannesson K. H. and Neumann K. (2013) Geochemical cysclingof mercury in a deep, confined aquifer: Insights from biogeo-chemical reactive transport modeling. Geochim. Cosmochim.

Acta 106, 25–43.Kent D. B., Wilkie J. A. and Davis J. A. (2007) Modeling the

movement of a pH perturbation and its impact on adsorbed

http://refhub.elsevier.com/S0016-7037(20)30187-3/h0005

































































































































zinc and phosphate in a wastewater-contaminated aquifer.Water Resour. Res. 43(7), 1–17.

Kocar B. D. and Fendorf S. (2009) Thermodynamic constraints onreductive reactions influencing the biogeochemistry of arsenic insoils and sediments. Environ. Sci. Technol. 43, 4871–4877.

Koretsky C. (2000) The significance of surface complexationreactions in hydrologic systems: a geochemist’s perspective. J.Hydrol. 230, 127–171.

Kosmulski M. (2009) pH-dependent surface charging and points ofzero charge. IV. Update and new approach. J. Colloids

Interface Sci. 337, 439–448.Kurotori T., Zahasky C., Hosseinzadeh Hejazi S. A., Shah S.,

Benson S. M. and Pini R. (2019) Measuring, imaging andmodelling solute transport in a microporous limestone. Chem.

Eng. Sci. 196, 366–383.Landry C., Koretsky C. M., Lund T. J., Schaller M. and Das S.

(2009) Surface complexation modeling of Co(II) adsorption onmixtures of hydrous ferric oxide, quartz and kaolinite. Geochim.

Cosmochim. Acta 73, 3723–3737.Lasaga A. (1998) Kinetic Theory in the Earth Sciences. I N.J.

(Red.). Princeton Univ. Press, Princeton.Leung K., Nielsen I. M. B. and Criscenti L. J. (2009) Elucidating

the bimodal acid-base behavior of the water-silica interfacefrom first principles. J. Am. Chem. Soc. 131, 18358–18365.

Logue B. A., Smith R. W. and Westall J. C. (2004) U(VI)adsorption on natural iron-coated sand: comparison ofapproaches for modeling adsorption on heterogeneous envi-ronmental materials. Appl. Geochim. 19, 1937–1951.

Lund T. J., Koretsky C. M., Landry C. J., Schaller M. S. and DasS. (2008) Surface complexation modeling of Cu(II) adsorptionon mixtures of hydrous ferric oxide and kaolinite. Geochem.

Trans. 9(9), 1–16.McNeece C. J. and Hesse M. (2016) Reactive transport of aqueous

protons in porous media. Adv. Water Resour. 97, 314–325.McNeece C. J. and Hesse M. (2017) Challenges in coupling acidity

and salinity transport in porous media. Environ. Sci. Technol.51, 11799–11808.

McNeece C. J., Lutzenkirchen J. and Hesse M. (2018) Chromato-graphic analysis of the acidity-salinity transport system. J.

Contam. Hydrol. 216, 27–37.Muniruzzaman M. and Rolle M. (2015) Impact of multicomponent

ionic transport on pH fronts propagation in saturated porousmedia. Water Resour. Res. 51(8), 6739–6755.

Muniruzzaman M. and Rolle M. (2016) Modeling multicomponentionic transport in groundwater with IPhreeqc coupling: Elec-trostatic interactions and geochemical reactions in homoge-neous and heterogeneous domains.Adv. Water Resour. 98, 1–15.

Muniruzzaman M. and Rolle M. (2017) Experimental investigationof the impact of compound-specific dispersion and electrostaticinteractions on transient transport and solute breakthrough.Water Resour. Res. 53(2), 1189–1209.

Muniruzzaman M. and Rolle M. (2019) Multicomponent ionictransport modeling in physically and electrostatically heteroge-nous porous media with PhreeqcRM coupling for geochemicalreactions. Water Resour. Res. 55(12), 11121–11143.

Muniruzzaman M., Haberer C. M., Grathwohl P. and Rolle M.(2014) Multicomponent ionic dispersion during transport ofelectrolytes in heterogeneous porous media: Experiments andmodel-based interpretation. Geochim. Cosmochim. Acta 141,656–669.

Ong S., Zhao X. and Eisenthal K. B. (1992) Polarization of watermolecules at a charged interface: second harmonic studies of thesilica/water interface. Chem. Phys. Letters 191(3–4), 327–335.

Parkhurst D. L. and Appelo C. A. J. (2013). Description of inputand examples for PHREEQC version 3 - A computer programfro speciation, batch-reaction, one-dimensional transport, and

inverse geochemical calculations. U.S. Geological SurveyTechniques and Methods, book 6, chap. A43, 497 p., availableonly at http://pubs.usgs.gov/tm/06/a43.

Peacock C. L. and Sherman D. M. (2004) Copper(II) sorption ontogoethite, hematite and lepidocrocite: a surface complexationmodel based on ab initio molecular geometries and EXAFSspectroscopy. Geochim. Cosmochim. Acta 68(12), 2623–2637.

Penn R. L., Zhu C., Xu H. and Veblen D. R. (2001) Iron oxidecoatings on sand grains from the Atlantic coastal plain: High-resolution transmission electron microscopy characterization.Geology 29(9), 843–846.

Pfeiffer-Laplaud M. and Gaigeot M. (2016a) Adsorption of singlycharged ions at the hydroxylated (0001) alpha-quartz/waterinterface. J. Phys. Chem. C 120, 4866–4880.

Pfeiffer-Laplaud M. and Gaigeot M. (2016b) Electrolytes at thehydroxylated (0001) alpha-quartz/water interface: location andstructural effects on interfacial silanols by DFT-based MD. J.Phys. Chem. C 120, 14034–14047.

Pfeiffer-Laplaud M., Costa D., Tielens F., Gaigeot M.-P. andSulpizi M. (2015) Bimodal acidity at the amorphous silica-waterinterface. J. Phys. Chem. C 119, 27354–27362.

Prigiobbe V. and Bryant S. (2014) pH-dependent transport ofmetal cations in porous media. Environ. Sci. Technol. 48, 3752–3759.

Prigiobbe V., Hesse M. and Bryant S. (2012) Fast strontiumtransport induced by hydrodynamic dispersion and pH-depen-dent sorption. Geophys. Res. Lett. 39, 1–5.

Rawson J., Prommer H., Siade A., Carr J., Berg M., Davis J. andFendorf S. (2016) Numerical modeling of arsenic mobilityduring reductive iron-mineral transformations. Environ. Sci.

Technol. 50(5), 2459–2467.Riese A. (1982) Adsorption of Radium and Thorium onto Quartz and

Kaolinite: A Comparison of Solution/Surface Equilibria Models.Colorado School of Mines, Golden.

Rolle M., Muniruzzaman M., Haberer C. M. and Grathwohl P.(2013) Coulombic effects in advection-dominated transport ofelectrolytes in porous media: Multicomponent ionic dispersion.Geochim. Cosmochim. Acta 120, 195–205.

Rolle M., Sprocati R., Masi M., Jin B. and Muniruzzaman M.(2018) Nernst-Planck based description of transport, coulombicinteractions and geochemical reactions in porous media:modeling approach and benchmark experiments.Water Resour.

Res. 54(4), 3176–3195.Rustad J. R. and Felmy A. R. (2005) The influence of edge sites on

the development of surface charge on goethite nanoparticles: amolecular dynamics investigation. Geochim. Cosmochim. Acta

69(6), 1405–1411.Sahai N. and Sverjensky D. A. (1997a) Solvation and electrostatic

model for specific electrolyte adsorption. Geochim. Cosmochim.

Acta 61(14), 2827–2848.Sahai N. and Sverjensky D. A. (1997b) Evaluation of internally

consistent parameters for the triple-layer model by the system-atic amalysis of oxide surface titration data. Geochim. Cos-

mochim. Acta 61(14), 2801–2826.Sø H. U., Postma D., Hoang V. H., Mai L. V., Kim T. P. T., Hung

V. P. and Jakobsen R. (2018) Arsenite adsorption controlled bythe iron oxide content of Holocene Red River aquifer sediment.Geochim. Cosmochim. Acta. 239, 61–73.

Sposito G. (2008) The chemistry of soils, second ed. OxfordUniversity Press, New York.

Stachowicz M., Hiemstra T. and van Riemsdijk W. H. (2008)Multi-competitive interactin of As(III) and As(V) oxyanionswith Ca2+, Mg2+, and CO32- ions on goethite. J. Colloid

Interf. Sci. 320, 400–414.Stolze L., Zhang D., Guo H. and Rolle M. (2019a) Surface

complexation modeling of arsenic mobilization from goethite:
































































































































Interpretation of an in-situ experiment. Geochim. Cosmochim.

Acta 248, 274–288.Stolze L., Zhang D., Guo H. and Rolle M. (2019b) Model-based

interpretation of groundwater arsenic mobility during in situreductive transformation of ferrihydrite. Environ. Sci. Technol53, 6845–6854.

Stumpf S., Stumpf T., Lutzenkirchen J., Walther C. and FanghanelT. (2008) Immobilization of trivalent actinides by sorption ontoquartz and incorporation into siliceous bulk: investigations byTRLFS. J. Colloids. Interf. Sci. 318, 5–14.

Sulpizi M., Gaigeot M. and Sprik M. (2012) Silica-water interface:how the silanols determine the surface acidity and modulate thewater properties. J. Chem. Theory Comput. 8, 1037–1047.

Sverjensky D. A. (2005) Prediction of surface charge on oxides insalt solutions: Revisions for 1:1 (M+L-) electrolytes. Geochim.

Cosmochim. Acta 69(2), 225–257.Sverjensky D. A. and Sahai N. (1996) Theoretical prediction of

single-site surface-protonation equilibrium constants for oxidesand silicates in water. Geochim. Cosmochim. Acta 60(20), 3773–3797.

Tournassat C. and Steefel C. I. (2015) Ionic transport in nano-porous clays with consideration of electrostatic effects. Rev.

Mineral. Geochem. 80, 287–329.Tournassat C., Tinnacher R. M., Grangeon S. and Davis J. A.

(2018) Modeling uranium(VI) adsorption onto montmorilloniteunder varying carbonate concentrations: A surface complexa-tion model accounting for the spillover effect on surfacepotential. Geochim. Cosmochim. Acta 220, 291–308.

Tucillo M. E., Cozzarelli I. M. and Herman J. S. (1999) Ironreduction in the sediments of a hydrocarbon-contaminatedaquifer. Appl. Geochem. 14, 655–667.

Venema P., Hiemstra T. and van Riemsdijk W. H. (1996) Multisiteadsorption of dacmium on goethite. J. Colloid Interf. Sci. 183

(0575), 515–527.Villalobos M. and Leckie J. O. (2001) Surface complexation

modeling and FTIR study of carbonate adsorption to goethite.J. Colloid Interf. Sci. 235, 15–32.

Wang M. and Revil A. (2010) Electrochemical charge of silicasurfaces at high ionic strength in narrow channels. J. ColloidInterf. Sci. 343, 381–386.

Weerasooriya R., Aluthpatabendi D. and Tobschall H. (2001)Charge distribution multi-site complexation (CD-MUSIC)modeling of Pb(II) adsorption on gibbsite. Colloids Surf., A

189, 131–144.Ye Y., Chiogna G., Cirpka O. A., Grathwohl P. and Rolle M.

(2015a) Experimental investigation of compound-specific dilu-tion of solute plumes in saturated porous media: 2-D vs. 3-Dflow-through systems. J. Cont. Hydrol. 172, 33–47.

Ye Y., Chiogna G., Grathwohl P. and Rolle M. (2015b) Exper-imental evidence of helical flow in porous media. Phys. Rev.Lett. 115(19).

Zhang S., Kent D. B., Elbert D. C., Shi Z., Davis J. A. and VeblenD. R. (2011) Mineralogy, morphology, and textural relation-ships in coatings on quartz grains in sediments in a quartz-sandaquifer. J. Contam. Hydrol. 124, 57–67.

Associate editor: Carl Steefel
























































Impact of surface complexation and electrostatic interactions ...

Documents

Transcript of Impact of surface complexation and electrostatic interactions ...