Ab initio modeling of Fe(II) adsorption and interfacial electron transfer at goethite (α-FeOOH) surfaces.

View Article Online View Journal

PCCP Accepted Manuscript

This article can be cited before page numbers have been issued, to do this please use: V. Alexandrov and K. Rosso, Phys. Chem. Chem. Phys., 2015, DOI: 10.1039/C5CP00921A.

This is an Accepted Manuscript, which has been through the Royal Society of Chemistry peer review process and has been accepted for publication. Accepted Manuscripts are published online shortly after acceptance, before technical editing, formatting and proof reading. Using this free service, authors can make their results available to the community, in citable form, before we publish the edited article. We will replace this Accepted Manuscript with the edited and formatted Advance Article as soon as it is available. You can find more information about Accepted Manuscripts in the Information for Authors. Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal’s standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains.

www.rsc.org/pccp

Page 1 of 35

Physical Chemistry Chemical Physics View Article Online

Ab Initio Modeling of Fe(II) Adsorption and Interfacial

Published on 06 May 2015. Downloaded by Indiana University on 13/05/2015 00:45:34.

Electron Transfer at Goethite (α-FeOOH) Surfaces VITALY ALEXANDROV†* AND KEVIN M. ROSSO†

†

*

PHYSICAL SCIENCES DIVISION, Pacific Northwest National Laboratory, Richland, Washington 99352, USA

Email: [email protected]

KEYWORDS. Density Functional Theory; Adsorption; Electron Transfer; Goethite Surfaces; Iron Oxyhydroxide.

First-principles study of the mechanism of aqueous Fe(II) adsorption and Fe(II)—Fe(III) interfacial electron transfer at goethite surfaces.

Physical Chemistry Chemical Physics Accepted Manuscript

DOI: 10.1039/C5CP00921A

Physical Chemistry Chemical Physics

Page 2 of 35 View Article Online

DOI: 10.1039/C5CP00921A

interfaces in the

environment, playing an important role in biogeochemical metal cycling and contaminant residence in the subsurface. Fe(II)-catalyzed recrystallization of goethite is a fundamental process in this context, but the proposed Fe(II)aq-Fe(III)goethite electron and iron atom exchange mechanism of recrystallization remains poorly understood at the atomic level. We examine


the adsorption of aqueous Fe(II) and subsequent interfacial electron transfer (ET) between adsorbed Fe(II) and structural Fe(III) at the (110) and (021) goethite surfaces using density functional theory calculations including Hubbard U corrections (DFT+U) aided by ab initio molecular dynamics simulations. We investigate various surface sites for the adsorption of Fe2+(H2O)6 in different coordination environments. Calculated energies for adsorbed complexes at both surfaces favor monodentate complexes with reduced 4- and 5-fold coordination over higher-dentate structures and 6-fold coordination. The hydrolysis of H2O ligands is observed for some pre-ET adsorbed Fe(II) configurations. ET from the adsorbed Fe(II) into the goethite lattice is calculated to be energetically uphill always, but simultaneous proton transfer from H2O ligands of the adsorbed complexes to the surface oxygen species stabilizes post-ET states. We find that surface defects such as oxygen vacancies near the adsorption site also can stabilize post-ET states, enabling the Fe(II)aq-Fe(III)goethite interfacial electron transfer reaction implied from experiments to proceed.

1.

Introduction.

The ferric iron oxyhydroxide mineral goethite (α-FeOOH) is one of the most abundant phases of iron in nature, and its reactivity couples to a wide range of environmental processes including reduction-oxidation reactions due to the presence of redox-active structural Fe(III) [1-3]. For example, goethite is known to be an efficient sorbent for many contaminants, metals, inorganic anions, and organic matter [3-6], it can participate in microbial dissimilatory iron reduction [7, 8], and it can catalyze redox transformations of the metals adsorbed to its surfaces [9-12]. Understanding the interfacial redox reactivity of goethite is thus important for accurate conceptualization of a range of phenomena such as environmental transport and sequestration of contaminants, reductive dissolution and oxidative growth of iron (oxyhydr)oxides, and biogeochemical cycling of various elements.

In recent years, emphasis has been placed on understanding complex redox-driven dynamics taking place at the interface between various Fe(III)-(oxyhydr)oxides and aqueous Fe(II), including sorption, atom exchange and interfacial


ABSTRACT. Goethite (α-FeOOH) surfaces represent one of the most ubiquitous redox-active

Page 3 of 35


DOI: 10.1039/C5CP00921A

electron transfer (ET) reactions [6,9,12-14]. For example, a

57

Fe Mössbauer spectroscopy study showed that ET from

subsequent 57Fe isotope tracer experiments it was shown [14, 16] that this electron exchange process is accompanied by Fe atom exchange between the Fe(II) in the aqueous phase and the Fe(III) in goethite, without change in goethite physical characteristics.

The studies suggest that the redox interaction between Fe(II)aq and Fe(III)oxide catalyzes goethite


recrystallization, enabling exchange of Fe atoms and, as most recently shown, oxygen atoms between solution and goethite phases. Given demonstration of the process of hematite transformation catalyzed by Fe(II) [17], through oxidative addition of Fe from solution coupled by electrical conduction to remote reductive release of Fe from the oxide, this conceptual model appears to explain Fe(II)/goethite interaction as well. However, because of the small size of goethite crystals, most of the experimental evidence for the adsorption of Fe(II), the interfacial ET process, and Fe release from goethite crystallites has been macroscopic or indirect [14,15].

At the atomic scale, the ET properties of Fe(III)-(oxyhydr)oxide such as goethite has been a focus of recent theoretical studies, which have provided significant insights into the factors controlling thermally-activated hopping of electron small polarons [18-26]. This includes reorganization energies and electronic coupling matrix elements that yield the activation energies for polaron hopping. Overall, it is thought that the atomic and electronic structure of the nearestneighbor local environment rather than long-range lattice structure are the major factors affecting ET rates in these materials. Recently, the kinetics of electron transport via small polaron hopping mechanism in the bulk phase of goethite was addressed using DFT+U methodology [25], revealing fast pathways down Fe rows in the double-chains.

Likewise, the electron and atom exchange steps proposed to underpin goethite recrystallization by Fe(II) can be tested for plausibility using computational molecular simulation. The single biggest challenge in doing so is proper construction and sampling of relevant atomic configurations and dynamics at complex goethite/water/Fe(II) interfaces. To date there have been two studies directed at this topic. In a density functional theory (DFT) based study [27] the adsorption of a hypothetical aqueous Fe(II) complex on (110) and (021) goethite facets was considered. Based on the analysis of atomic Bader charges and projected density of states, no evidence for spontaneous interfacial ET was found. Only one Fe(II) adsorption complex type at each facet was examined, despite multiple adsorption site possibilities. Also, each configuration considered was restricted to be 6-fold coordination of adsorbed Fe(II), using a combination of surface OH groups and aqueous H2O ligands, without an overlying water phase present. Secondly, a classical molecular dynamics investigation [18] allowed exploration of a much larger configuration space for adsorbed Fe(II) through potential of mean


adsorbed Fe(II) to structural Fe(III) in a number of Fe oxides including goethite can be spontaneous [15]. Based on



DOI: 10.1039/C5CP00921A

force calculations and explicit treatment of solvating water dynamics. These simulations revealed a wide range of stable

and, because of model limitations, being unable to address the possibility of hydrolysis of water ligands upon adsorption. This disallows for proton transfer coupled to interfacial ET, which could play an important role in the mechanism of electron exchange at Fe-bearing oxide surfaces [24]. Interfacial ET steps were predicted to be predominantly unfavorable,


with large activation energies.

These ground-breaking modeling studies highlighted a number of effects that either have not yet been addressed or have been only partially investigated. A reasonably good level of understanding has been achieved on how various goethite surfaces react with water and what interfacial structures can form, through previous experimental and computational studies of water/goethite interfaces [27-30]. In terms of interaction with aqueous Fe(II), however, there is no comprehensive study of stability of the aqueous Fe(II) complexes across various surface adsorption sites that includes energy minimization of the topology of the adsorbed Fe(II), the nature of the Fe(II) ligands (water molecules and/or hydroxyls) and whether ligand H2O molecules can donate protons to the surface as a means of local charge conservation. For example, while Fe(II) is known to prefer 6-fold coordination with water molecules in solution, this might not be the case for the surface-bound species, and the lower coordinations (four and five) were previously demonstrated to be more probable, e.g., for aqueous Fe(II) on the surfaces of Fe-clay nontronite [24] and for Zn(II) on the (110) surface of TiO2 rutile [31]. In addition to coordination number, the composition of the Fe(II) first coordination sphere can impact stability of the adsorbed Fe complexes, because water ligands can dissociate to donate protons to the surface or undergo hydrolysis to produce OH groups attached to the adsorbed Fe ions. It is also important to understand if there should be any surface exchange reactions such as release of surface protons coupled to uptake of Fe(II) or the release of surface water/hydroxyl species. Finally, it is important to consider the possible presence of structural defects at the goethite surface, such as oxygen vacancies, as possibly more energetically favorable locations for local Fe(III) reduction. Local under-coordination by O should increase the electronegativity on Fe, as well as reduce the reorganization energy required to distort relatively stiff crystal Fe—O bonds to accommodate an electron polaron in the structure at the interface.

In this study we provide comprehensive microscopic information on the interactions between aqueous Fe(II) and (110) and (021) goethite surfaces, through the application of density function theory calculations with Hubbard U corrections (DFT+U) aided by ab initio molecular dynamics (AIMD) simulations. We investigate the adsorption of aqueous Fe(II) at different surface sites in the presence of an explicitly represented water phase, and analyze the adsorption


outer- and inner-sphere adsorbed Fe(II) complexes while, however, also restricting the overall Fe(II) coordination to six

Page 5 of 35


DOI: 10.1039/C5CP00921A

energetics and structural characteristics for stable adsorbed Fe(II) complexes. We also estimate the probability for

optimized pre- and post-ET states, along with the energy of the ET transition state. In addition, we investigate the effect of oxygen vacancies at goethite surfaces on the interfacial ET barriers. Because of the predominance of local structural effects, and the preponderance of similar local structural motifs across phases, the obtained information about Fe(II) interactions


with goethite surfaces is likely general, and an important step in building a coherent picture of Fe(II)-catalyzed recrystallization across different Fe(III)-(oxyhydr)oxides.

2.

Computational Details.

All calculations are performed applying density functional theory (DFT) within the generalized-gradient approximation (GGA) functional of Perdew, Burke, and Ernzenhof (PBE) [32] along with the projector augmented wave (PAW) [33] potentials with eight, six, and one valence electrons for Fe, O, and H, respectively, as implemented in the Vienna Ab Initio Simulation Package (VASP) [34-35]. To account for the localization of the Fe 3d electrons, we employ the rotationally invariant form of the DFT+U approach proposed by Dudarev et al. [36] in which the total energy depends on the effective on-site parameter Ueff = U – J rather than on two parameters (U, J) in the Hubbard model. In this study we apply the value of Ueff = 5 eV on the Fe atoms similar to previous DFT+U studies of various iron oxides including goethite [19, 25, 30, 3738]. It was demonstrated that the use of Ueff = 5 eV can provide very reasonable results on structural, electronic including band gap and adsorption properties of goethite [19, 25, 30]. All results presented in this paper correspond to Ueff = 5 eV for consistency if not specified otherwise. In order to drive electron localization to either Fe2+ or Fe3+ state we set the initial magnetic moments on Fe ions that correspond to the desired configuration and in some cases also pre-distorted (contracted) the surrounding Fe-O bonds to slightly change the crystal field. The plane wave cutoff energy was set to 500 eV. The total energy is optimized with respect to atomic positions with cell parameters fixed at the values corresponding to the optimized bulk goethite using a conjugate-gradient algorithm until the forces on the atoms are converged to 0.02 eV/Å and the total energy is converged to 10-5 eV.

The (110) and (021) goethite surfaces (Pbnm setting) are modeled using periodic slabs with a vacuum gap of about 10 Å. To minimize the interactions between periodic images of the adsorbed complexes, the (110) primitive surface cell is enlarged by four times to give the dimensions for the new surface cell 11.04 x 12.16 Å2 and the (021) surface cell is doubled


interfacial ET between adsorbed Fe(II) and structural Fe(III) in the goethite lattice by comparing the energies of fully



DOI: 10.1039/C5CP00921A

the same 2 x 2 k-point mesh in the surface plane. The resulting number of Fe atoms in each slab is thirty two and they are considered to be ferromagnetically coupled in the calculations. The used thickness of slab models for both surfaces was previously shown to be large enough to treat Fe(II) adsorption [16]. The slabs corresponding to the fully hydroxylated


surfaces are shown on Figure 1. The fully hydroxylated surfaces were shown to be much more energetically favorable over the bare surfaces with the (110) and (021) facets exhibiting very similar surface Gibbs free energies [27].

Application of full ab initio molecular dynamics (AIMD) calculations to explore the potential energy surface of Fe(II) adsorption is not currently feasible due to very long simulation times required to sample both rotational and translational degrees of freedom of species diffusion at metal-oxide surfaces (~102-103 ps). This is particularly true for materials which surfaces are characterized by a complex structural topology as in the case of hydroxylated goethite surfaces considered in this study. In addition, rather tight precision settings in the calculations are normally required to achieve sufficient energy conservation thus also limiting the application of full AIMD simulations to the problems of metal adsorption. Instead, in this study we exploit two different strategies. The first is to construct various plausible atomic configurations of the adsorbed complexes in the presence of water in the vacuum region with the average liquid density of about 1.0 g cm-3 and run relatively short (2-3 ps) AIMD simulations at room temperature. In this case, the initial system was pre-equilibrated during about 10 ps at higher temperature of 800 K with the fixed atomic positions of the adsorbed Fe(II) complex and then gradually cooled down to room temperature using simulated annealing. The obtained low-energy configurations are subsequently optimized by DFT to obtain their total energies. The second approach is to run DFT optimizations of the initial adsorption configurations at zero Kelvin prepared without water molecules in the vacuum region. In all calculations the adsorption of Fe(II) complexes is considered as single-sided.

The analysis of atomic charges was done within the Bader topological scheme [39] using a grid-based algorithm developed by Henkelman et al. [40-41]. To evaluate activation barriers of interfacial electron transfer (ET) reaction between adsorbed Fe(II) and structural Fe(III), the small-polaron hopping approach was used in which the barrier is determined by locating an intersection point on the nuclear reaction coordinate between initial- and final-state parabolic potential energy surfaces of the energy profile for the electron hopping path. This approach was shown to be appropriate for describing ET properties in various Fe oxides [19, 24-26].


to give the surface dimensions 9.24 x 11.73 Å2. Similar dimensions of the surface supercells for both facets allows us to use

Page 7 of 35


3.

3.1.

Results and Discussion.

Hydroxylated Surfaces and Adsorption Sites. Goethite (α-FeOOH) is the most thermodynamically stable


polymorph among FeOOH iron oxyhydroxides [42-43]. It crystallizes in an orthorhombic structure with the Pbnm space group and is characterized by parallel double chains of edge-sharing Fe(III)O3(OH)3 octahedra linked to the neighboring double chains by corner sharing, thus forming 2×1 channels. Recently, structural, magnetic and electron transport properties of bulk goethite were examined employing periodic DFT calculations [25, 44-45]. Regarding electron transport properties, it was shown that charge transfer at room temperature should be dominated by the small-polaron hopping type and the electron mobilities along symmetrically inequivalent pathways were calculated to be very similar, being on the order of 10-6 cm2/Vs (using Ueff = 5 eV in the DFT+U calculations) [25].

Typical goethite crystallites have an acicular needle-shape morphology with predominant (110) facets running along the long axis of the needles, while the termini of the needles are made up mostly of (021) surfaces [46-47]. From studies addressing the structure of water/goethite interfaces [27-30], it was demonstrated by applying ab initio thermodynamics and calculating surface Gibbs free energies, γ, that the fully hydroxylated (110) and (021) surfaces are strongly favored over the bare surfaces by about 53 and 34 meV/Å2, respectively (using Ueff = 5.2 eV in the DFT+U calculations) [27]. On the other hand, both (110) and (021) hydroxylated surfaces are found to be equally stable. These hydroxylated surfaces with the various corresponding hydroxyl


DOI: 10.1039/C5CP00921A



Figure 1. Side (a) and top (b) views of the hydroxylated (110) goethite surface (half of the slab is shown).

Figure 2. Side (a) and top (b) views of the hydroxylated (021) goethite surface (half of the slab is shown).

surface site types are shown in Figures 1 and 2. In the case of the (110) face, the available surface sites for Fe(II) binding involve singly (OH1), doubly (OH2), and triply (OH3) coordinated surface hydroxyl groups, as well as triply coordinated oxygen ions (O3) bridging the neighboring double chains of the goethite Fe octahedra through corner sharing. Previously, a different notation for the surface groups has also been used, namely, -OH for singly (equivalent to our OH1), µ-OH for doubly (equivalent to our OH2) and µ3-OH (equivalent to our OH3) for triply coordinated hydroxyl groups [30, 48]. This



DOI: 10.1039/C5CP00921A

Page 9 of 35


DOI: 10.1039/C5CP00921A

hydroxylated structure is obtained by adding water molecules that dissociate to give singly coordinated hydroxyls (OH1)

adsorption sites on the (021) surface involve singly coordinated water molecules (H2O), singly (OH1), and doubly (OH2) coordinated hydroxyl groups. The full hydroxylation of the (021) surface is achieved by adding a water molecule onto the Fe1 site and a water molecule dissociated into the hydroxyl group bound to the Fe2 site and the proton donated to the


bridging oxygen atom. To analyze the energetics of Fe(II) adsorption, we first optimize the outer-sphere Fe(H2O)62+ complex atop each surface, to serve as the reference configuration relative to which we compute the energies of the adsorbed inner-sphere complexes at each surface. This allows us to both compare energies against the same reference for each surface and estimate the driving force for Fe(II) adsorption from solution to a surface. The obtained energies presented below in Tables 1 and 2 are calculated for systems with the same number of species as in the [Fe2+(H2O)6 + hydroxylated surface], i.e., additional water molecules are removed from the vacuum gap after AIMD simulations. An example of the simulation cell used in the AIMD simulations with the outer-sphere Fe2+(H2O)6 complex over the (110) surface and water molecules added to the vacuum gap is shown in Figure 3.

3.2.

Fe(II) Adsorption and ET at the (110) Surface. In this section we discuss aqueous Fe(II) complexes

adsorbed on the (110) surface of goethite. We start by considering the outer-sphere case (Figure 3). Both zero temperature DFT optimization and AIMD calculations at room temperature show that the outer-sphere Fe(II) has 6-fold coordination with aqua ligands, and some of the water ligands in the vicinity of the (110) surface can donate protons to the surface groups (OH1, OH2 and also O3) forming surface water molecules. We do not observe that these surface water molecules, however, dissociate from the surface during our short AIMD simulations and thus they are believed to be relatively strongly bound to the surface Fe atoms.


over the formerly five-fold coordinated Fe atoms and protons binding to the bridging oxygen atoms. The available



Figure 3. General view of the simulation cell involving Fe2+(H2O)6 outer-sphere complex in water solution atop the (110) goethite surface. Iron octahedra are shown in polyhedral representation.

Table 1 summarizes the energies of the adsorbed inner-sphere Fe(II)-aqua complexes calculated relative to the energy of the outer-sphere complex. Atomic configurations of selected complexes are shown in Figure 4.

There are several surface sites that can participate in binding aqueous Fe(II) to the (110) facet. We start by considering adsorption over OH1 and OH2 monodentate sites (see Figure 1) since they are more accessible sites for Fe(II) binding from a steric point of view. Also, the corresponding monodentate adsorption complexes are more probable from the statistical point of view that their presence is a necessary step for forming poly-dentate complexes. From a chemical standpoint, the singly coordinated OH1 site is expected to have a stronger binding interaction with the adsorbed Fe(II) than OH2 and OH3 sites. For example, based on Pauling electrostatic bond strengths, each Fe atom directs a bond strength of +1/2 and each H atom directs a bond strength of +1 to every oxygen atom. As a result, the sum of the bond strengths directed toward each O atom for OH1 group is 1.5, for OH2 is 2, and for OH3 is 2.5. Because the O atom from OH1 is characterized by 0.5 smaller valence than the preferred formal valence of O (+2) and therefore should be a better nucleophile than OH2 and OH3 for binding Fe(II). This is clearly reflected by the calculated adsorption energies presented in Table 1. We see that all examined monodentate complexes over OH1 are more stable than their counterparts over an OH2 site. For example, 6- and 4-fold



DOI: 10.1039/C5CP00921A

Page 11 of 35


DOI: 10.1039/C5CP00921A

coordinated complexes over OH1 are more stable than similar complexes over an OH2 site by about 35 and 20 kJ/mol,

preferred over polydentate complexes, and furthermore there is a clear tendency towards a decrease in Fe(II) coordination at the surface. Among monodentate complexes 4-fold coordinated Fe(II) configurations are the most stable, with the M4-OH1 complex having the largest adsorption energy at the (110) facet (-82 kJ/mol with respect to the reference outer-sphere


complex). In order to make our notation of the adsorption complexes clearer, we write some reactions between aqueous Fe2+ and the goethite surfaces in the following schematic way. For example, the Fe2+ attached to the OH1 group of a goethite surface as a monodentate configuration with 6-fold coordination represented in Table 1 as M6-OH1 and shown on Figure 4a corresponds to the following adsorption reaction: Fe2+(H2O)6 + surf-FeOOH (H2O)5Fe2+–(OH1)OFe3+ + H2O

(1).

Monodentate 5-fold coordinated Fe2+ over OH2 surface group (M5-OH2) can be represented by the reaction Fe2+(H2O)6 + surf-FeOOH (H2O)4Fe2+– (OH2)OFe3+ + H2O

(2)

whereas Fe2+ attached to two surface OH1 groups as a bidentate 4-fold coordinated complex corresponds to Fe2+(H2O)6 + surf-FeOOH (H2O)2Fe2+=(OH1)2OFe3+ + 4H2O

(3).

As seen from Table 1, bidentate and tetradentate adsorption complexes are quite significantly less stable than monodentate complexes, while the initially constructed tridentate configurations converged to bidentate or monodentate complexes during structural optimizations. The most favorable bidentate complex B5-2OH2 is over 50 kJ/mol less stable than the most favorable monodentate M4-OH1 suggesting monodentate attachments as the most probable surface configurations of the adsorbed Fe(II). Tetradentate complex T6 is found to be a little more stable than the most favorable bidentate configuration.

The calculations thus strongly suggest that the total coordination number of inner-sphere adsorbed Fe(II) ion at the (110) surface is lower than 6, and thus loss of water ligands is implied upon going from the hexacoordinate outer-sphere


respectively. By comparing adsorption energies in Table 1 it is apparent that monodentate complexes are energetically



DOI: 10.1039/C5CP00921A

complex to the most stable inner-sphere complexes. Note that during AIMD simulations at room temperature the starting

consistent with this view. The most stable complex at the (110) surface is found to be 4-fold coordinated Fe(II) over monodentate OH1 site (M4-OH1 system). These observations are in excellent agreement with previous studies of adsorption of aqueous metals (Fe2+, Zn2+, U6+) at complex oxide surfaces [24, 31, 49] showing a tendency for the reduction


of metal coordination number at the surface.

We now can compare the obtained structural and energetic information about adsorbed Fe(II) complexes with previous computational insights. In contrast to the present work, in the previous DFT study of Fe(II) adsorption on hydroxylated goethite surfaces the adsorption complex considered on the (110) facet was the bidentate 6-fold coordinated Fe(II) over the two OH1 sites and having 4 water ligands, corresponding to our B6-2OH1 system [27]. According to the adsorption energies presented in Table 1, this is certainly not the most energetically favorable adsorption configuration, even among bidentate complexes. The reason for this is that in the previous investigation [27] only one 6-fold coordinated Fe-water complex with no possibility for the hydrolysis of Fe(II) water ligands was treated, whereas our analysis of various adsorption configurations shows that adsorption complexes prefer to lower Fe coordination upon adsorption and also in some cases water ligands tend to donate protons to the surface oxygen atoms thus further stabilizing adsorption configurations (for example, B5-2OH2 complex). In a recent classical MD study [18] a more comprehensive analysis of a number of 6-fold coordinated Fe(II) adsorption complexes was carried out. A good quantitative agreement between the energy difference between the outer- and the most favorable hexacoordinate inner-sphere complexes is found. Specifically, in the present study this difference is about -82 kJ/mol (M4-OH1 complex) while in the MD study [18] the free energy difference is about -80 kJ/mol for a 6-fold coordinated monodentate complex (corresponds to our M6-OH1 case). Also, both classical MD and the present studies predict monodentate attachments as the most favorable adsorption configurations at the (110) surface. However, the classical MD investigation [18] also was not able to predict both the change of Fe(II) coordination number and the hydrolysis of the Fe(II) water ligands during adsorption due to the methodology limitations, whereas we find here that both processes should take place.

Fe position

System

ΔEpre-ET

ΔEpost-ET


configurations featuring 6-fold coordinated adsorbed Fe(II) complexes rapidly changed their coordination to 5- or 4-fold,

Page 13 of 35


DOI: 10.1039/C5CP00921A

outer-sphere complex (reference)

OS6

0

86

6-fold coordinated Fe over OH1 site

M6-OH1

-35


M5-OH1

-61


M4-OH1

-82 (-62a)


M6-OH2

0.1


M5-OH2

-51


M4-OH2

-47

6-fold coordinated Fe over two OH1 sitesb

B6-2OH1

-23

49

5-fold coordinated Fe over two OH1 sites

B5-2OH1

-4

70


B4-2OH1

-18


B6-2OH2

17

44


B5-2OH2

-29

17

T6

-34

45

Bidentate sites

Tetradentate sites 6-fold coordinated Fe over two O3 sites

Table 1. Energy differences (in kJ/mol) between the reference outer-sphere Fe(II)-aqua complex and inner-sphere adsorption complexes at the (110) goethite surface obtained by DFT optimization of selected adsorption configurations. ΔEpre-ET stands for the energy difference between the outer-sphere complex and the surface-bound Fe(II)-aqua complexes (before electron transfer from the adsorbed Fe(II) to the structural Fe(III)), while ΔEpost-ET corresponds to post electron transfer complexes with the adsorbed Fe(III) and structural Fe(II).

a

Energy calculated for the system with all H2O

molecules in the first coordination sphere of Fe(II) to be compared with the case when one ligand H2O prefers to donate a proton to the surface. b Atomic structure corresponds to Ref. [27]. For explanation of the adsorption complexes notation, see equations (1)-(3) and figures with atomic structures (Figs. 3-5).



Monodentate sites



We further analyze structural information for the hydrated Fe(II) on the (110) surface. Structures of 6- and 4-fold coordinated Fe(II) complexes over monodentate OH1 and OH2 sites are shown on Figure 3, while the structures of bidentate and tetradentate complexes are illustrated on Figures 4 and 5, respectively. It is seen from Table 2 that the distance


between adsorbed Fe(II) and the nearest structural Fe(III) (Fe2+aq—Fe3+goethite) might vary quite considerably (3.1 – 3.9 Å) depending on the system. In general, tetrahedral Fe2+ is located closer to the (110) surface than Fe2+ in octahedral coordination, which is the result of overall stronger binding of 4-fold Fe2+ ion to the surface. We also observe that the deprotonation of either one (e.g., M4-OH2 system) or two (M4-OH1 system) water molecules from Fe2+(H2O)6 complex at the surface gives rise to reduced Fe2+aq—OHaq bonds that turn out to be in the range of 1.9—2.0 Å. Bond distances between the adsorbed Fe(II) ion and surface hydroxyl groups (Fe2+aq—OHsurf) directly attached to it are generally slightly larger, being in the range of 2.0—2.3 Å. Similar range of bond lengths is observed for the adsorbed Fe(II) with its water ligands (Fe2+aq—H2Oaq). Overall, these distances are quite comparable to those obtained previously for the adsorbed Fe(II) complexes at octahedral iron sites of the nontronite clay mineral edges [26], which naturally feature octahedral topology similar to the goethite surfaces.

We should note that one of the important features of the adsorbed Fe(II) complexes at the (110) facet is the deprotonation of some ligating water molecules. As discussed above, this is also observed at times during dynamics simulations of the outer-sphere complexes. For inner-sphere Fe(II) at the (110) surface, protons from water


DOI: 10.1039/C5CP00921A

Page 15 of 35


Figure 4. DFT optimized atomic structures of monodentate Fe(II) complexes adsorbed at the (110) surface in 6- and 4-fold coordinations over OH1 and OH2 sites corresponding to the following systems: (a) M6-OH1, (b) M6-OH2, (c) M4-OH1, (d) M4-OH2 (see Table 1 for their adsorption energetics).



DOI: 10.1039/C5CP00921A



Figure 4. DFT optimized atomic structures of bidentate Fe(II) complexes adsorbed at the (110) surface in 4- and 5-fold coordinations over OH1 and OH2 sites, respectively corresponding to the systems: (a) B4-2OH1, (b) B5-2OH2.

Figure 5. DFT optimized atomic structure of tetradentate Fe(II) complex adsorbed at the (110) surface (T6 system).

molecules are donated to either OH1 surface sites to form surface H2O or to O3 sites to form hydroxyls. The number of OH ligands varies from system to system, being zero for M6-OH2, one for M6-OH1, and two for M4-OH1. These results are in agreement with similar observations made recently, e.g., for Fe2+ adsorption on Fe-bearing clay mineral nontronite [24] and



DOI: 10.1039/C5CP00921A

Page 17 of 35


DOI: 10.1039/C5CP00921A

in a density functional molecular dynamics study of Zn2+ adsorption at the (110) surface of TiO2 rutile [31]. Note also that

than 6-fold Zn2+. It is known that the formation of OH groups instead of H2O ligands due to hydrolysis for aqueous metal cations can drive a change in coordination number of the metal ion from six to five and four in both aqueous solutions and at the metal-oxide surfaces, and it was suggested that the number of OH groups might be a controlling factor driving


coordination number reduction [31, 50-52]. Although this might so, we should point out that the competition between different metal ion coordination at the metal-oxide surfaces might also depend on such factors as unfavorable sterical clashes, specific surface structure and structure of the H-bonding network at the interface. Overall, coordination number reduction and proton transfer to surfaces leaving behind OH ligands with the adsorbed metal appears to be quite general for the adsorption of aqueous cations on metal oxide surfaces.

System

a(Fe2+aq--Fe3+goethite)

b(Fe2+aq--OHsurf)

c(Fe2+aq--OHaq)

d(Fe2+aq--H2Oaq)

M6-OH1

3.78

2.03

2.01

2.19; 2.20; 2.32; 2.65

M4-OH1

3.45

2.14

1.97; 1.99

2.09

M6-OH2

3.86

2.23

M4-OH2

3.48

2.07

B4-2OH1

3.53

1.99; 2.03

B5-2OH2

3.12

2.21; 2.29

2.07; 2.08; 2.15; 2.23; 2.32 1.92

2.09; 2.12 2.03; 2.09

1.94

2.05; 2.26

Table 2. Selected interatomic distances (in Å) for various Fe(II) adsorption complexes on the (110) surface.

We now turn to the discussion of electron transfer (ET) process from adsorbed Fe(II) species to structural Fe(III) at the (110) surface. The previous DFT study [27] addressed the problem of Fe(II) oxidation on goethite through the analysis of the electronic structure (atomic charges, electronic density of states) of the Fe2+(H2O)6, albeit on limited surface complex possibilities. Here, we investigate the ET process for all complexes where stabilization of a post-ET state was found to be possible, where the post-ET state corresponds to a system in which electron hopping from adsorbed Fe(II) to structural Fe(III) has occurred.


Fe2+ and Zn2+ ions have very close ionic radii in 4-fold coordination while 6-fold coordinated Fe2+ is just slightly smaller



DOI: 10.1039/C5CP00921A

Table 1 contains the energies of both the pre-ET and post-ET surface complexes for such cases. First of all, we see

activation barriers for interfacial ET from the adsorbed Fe(II). For the outer-sphere complex we find that the post-ET state with Fe3+aqueous and Fe2+struct is about 86 kJ/mol less stable than the pre-ET state. Including the reorganization energy [18] this should lead to an even larger ET barrier, which along with a relatively long ET distance of about 4.5 Å would result in a


very low ET probability [53]. This suggests that Fe(II) is more stable in aqueous solution than as a polaron in the goethite lattice at the interface across all found stable adsorbed configurations. For monodentate complexes we were unable to stabilize post-ET states using Ueff = 5 eV. To gain some insight into the ET energetics of monodentate complexes, it was possible to stabilize post-ET states in some cases using higher Ueff = 7 eV. Our test calculations comparing the energy differences for bidentate cases obtained applying Ueff = 5 and 7 eV showed that the use of Ueff = 7 eV increases the difference by less than 10 kJ/mol. Among monodentate complexes the energy differences between pre- and post-ET states using Ueff = 7 eV were found to be always more than 40 kJ/mol uphill. Therefore, the lowest energy difference of 27 kJ/mol is calculated for bidentate B6-2OH2 system which is one of the least stable complexes at the (110) surface (see Table 1). Overall, these results on the energetics of ET across the (110) goethite interface are in good conceptual agreement with the reported MD study [18] where the driving force for the forward interfacial ET reaction into goethite at the surface was also found to be energetically uphill. From this analysis we have an indication that the most stable adsorbed Fe(II) configurations are the least powerful ET donors.

Another important observation concerning the structure of pre- and post-ET complexes at the surface is that generally, but depending on the complex, we find that the ET from adsorbed Fe(II) to surface Fe(III) is accompanied by proton transfer from water ligands of the adsorbed complex to surface groups. For example, for B6-2OH2 system, which has the smallest energy difference between pre-and post-ET states, the pre-ET state is characterized by four water ligands of the Fe ion, whereas in the post-ET state, one ligand H2O donates a proton to O3 and the second to OH1 group. For the B62OH1 system, one H2O from the adsorbed complex donates a proton to a surface OH1 site in the post-ET case. Two factors are at play. The first is that as the Fe(II) ion becomes an Fe(III) ion, its electron affinity increases, drawing electron density from ligating water molecules and thus weakening water O-H bonding. Secondly, in the post-ET state a surface Fe ion has become more electronegative, thus attracting protons away from the now weaker association with ligating water molecules. Such proton coupled electron transfer can thus be an important factor for the interfacial ET step, by stabilizing the Fe(III)


that all calculated post-ET states are energetically less favorable than their pre-ET counterparts, implying rather high uphill

Page 19 of 35


DOI: 10.1039/C5CP00921A

ion in the adsorbed complex via formation of ligating OH groups and providing the mechanism of local charge balance

We also tested how deprotonation of the bridging OH ligands between Fe(II) and Fe(III) might affect the ET. Note, however, that such deprotonation seems to be unlikely based on Pauling electrostatic bond strength analysis described


above, and was not observed in our AIMD simulations and DFT optimizations. Nevertheless, we carried out a series of calculations for a few systems by removing protons from bridging hydroxyls to form bridging oxygen ions, and found that ET from adsorbed Fe(II) to structural Fe(III) is still uphill. However, deprotonation of bridging OH groups is well known to lower the ET barrier in complex iron oxides by increasing the magnitude of the electron coupling matrix element [54-55], an aspect not directly treated in this study. In general, the lower stability of Fe(II) in the goethite surface as compared to aqueous solution might be partly related to lower degrees of freedom to adopt a stable geometry among stiffer bonds in a rigid lattice.

To gain additional insight into the redox states of Fe complexes at the (110) facet, we analyzed local atomic charges using the Bader scheme. Table 3 lists calculated Bader charges for Feaq-Fegoethite redox couple before (pre-ET state) and after (post-ET state) the electron transfer from adsorbed Fe(II) complexes into the goethite structure. It is seen that the Bader charges for the 3+ charge state of the Fe ions are about 2.0e and for the 2+ charge state are 1.5-1.6e. These charge states are reversed for the Feaq-Fegoethite ions upon interfacial ET and indicate a charge transfer of about 0.5e. Regarding the Fe charges across different adsorption complexes, we observe a slight increase in the Bader charge for complexes of the same coordination adsorbed over the OH1 site as compared to those adsorbed over the OH2 site (such as M6-OH1 vs. M6OH2). This is consistent with the fact of overall stronger adsorption of the complexes with the same coordination number over OH1 site than OH2 site from electrostatic point of view (see Table 1 for adsorption energies), in agreement with Pauling electrostatic bond strength analysis for surface hydroxyl groups described above. For example, we see that the adsorbed Fe ion in the pre-ET state for M4-OH1 system has the smallest Bader charge of 1.49e and this system is indeed characterized by the largest adsorption energy across all systems. On the other hand, for complexes over the same adsorption site (OH1 or OH2) we observe a small decrease in the Bader charge when going from 6- to 4-fold coordinated Fe complexes. As a result, there is a slightly smaller amount of charge that needs to be transferred across the interface for lower coordination complexes. Overall, when comparing the Bader charges on Fe for adsorbed complexes with the bulk goethite values of 1.99e (for 3+ state) and 1.53e (for 2+ state) we can conclude that there is not any considerable transfer of electron density across the interface indicative of spontaneous ET.


upon interfacial ET.




System

Pre-ET

Post-ET

Fe2+aq

Fe3+goethite

Fe3+aq

Fe2+goethite

Outer-sphere

1.54

2.01

2.02

1.54

M6-OH1

1.53

2.02

M4-OH1

1.49

2.00

M6-OH2

1.58

2.01

M4-OH2

1.54

2.01

B6-2OH1

1.56

2.01

2.03

1.56

B5-2OH2

1.53

1.98

2.01

1.54

T6

1.53

2.03

1.99

1.55

Table 3. Bader atomic charges for Feaq—Fegoethite redox couple at the (110) surface before (pre-ET state) and after (post-ET state) the electron transfer from adsorbed Fe(II) complexes into the goethite structure across different systems. For comparison, the corresponding values in bulk goethite are q(Fe3+) = 1.99e; q(Fe2+) = 1.53e.

3.3.

Fe(II) Adsorption and ET at the (021) Surface. We now consider Fe(II) adsorption and ET at the (021)

surface of goethite. As in the case of (110) surface, water ligands from the outer-sphere Fe2+(H2O)6 complex are observed to donate protons to surface hydroxyls. Table 4 contains the description of adsorption systems and adsorption energetics for Fe(II) complexes at the (021) surface, similar to Table 1 for the (110) surface. First of all, we see that there is a pronounced driving force for surface adsorption as inner-sphere complexes, similar to the (110) surface. Also, the most favorable adsorption configurations are found as monodentate complexes, with a significant preference for adsorption over the OH1 sites. We also observe a clear trend for reducing the coordination number of the Fe(II) upon adsorption, with the lowestenergy complexes in 5- and 4-fold coordination over both OH1 and OH2 sites. Initially constructed as a 6-fold coordinated complex over the OH2 site, aqueous Fe(II) detached from the surface during DFT optimizations suggesting that it is unstable. Interestingly, the lowest-energy complexes at the (021) and (110) surfaces feature very similar adsorption energies


DOI: 10.1039/C5CP00921A

Page 21 of 35


DOI: 10.1039/C5CP00921A

(compare, for example, M5-OH1 and M4-OH1 systems at both facets). This may suggest quite similar adsorption kinetics

acicular needle-like shape of crystallites. The similarity in Fe(II) adsorption energetics might originate from a comparable reactivity of surface OH1 sites of the two faces. We were not able to obtain bidentate complexes over the (021) facet, while one stable tridentate complex is calculated to be less stable than the low-energy monodentate configurations. Thus, we see


that the main conclusions drawn for the Fe(II) adsorption at the (110) surface are the same for the (021) surface.

Fe position

System

ΔEpre-ET

outer-sphere complex (reference)

OS6

0


M6-OH1

19


M5-OH1

-81


M4-OH1

-61


M5-OH2

5


M4-OH2

-21

T5

-7

ΔEpost-ET

Monodentate sites

9

Tridentate sites 5-fold coordinated Fe over

39

Table 4. Energy differences (in kJ/mol) between the reference outer-sphere Fe(II)-aqua complex and inner-sphere adsorption complexes at the (021) goethite surface obtained by DFT optimization of selected adsorption configurations. ΔEpre-ET stands for the energy difference between the outer-sphere complex and the surface-bound Fe(II)-aqua complexes (before electron transfer from the adsorbed Fe(II) to the structural Fe(III)), while ΔEpost-ET corresponds to post electron transfer complexes with the adsorbed Fe(III) and structural Fe(II). For explanation of the adsorption complexes notation, see equations (1)-(3) and figures with atomic structures (Figs. 6-7).


on both surfaces, and therefore might not explain the very different growth rates of goethite crystal faces implied by the



DOI: 10.1039/C5CP00921A

Atomic structures of some systems from Table 4 are presented in Figures 6 and 7, while selected interatomic

structural Fe(III) (Fe2+aqueous—Fe3+struct) are smaller than in the case of Fe(II) adsorption over the (110) surface. The bond lengths between adsorbed Fe(II) and OH groups as well as between adsorbed Fe(II) and its H2O ligands are generally comparable to those obtained for the (110) facet so that Fe2+--OH bonds are slightly less than 2 Å and Fe2+--H2O bonds are


over 2 Å. Thus, the atomic structure and nature of adsorbed aqueous Fe(II) at both surfaces are characterized by similar corresponding interatomic distances and a tendency for Fe(II) to reduce its coordination number upon adsorption.

Figure 6. DFT optimized atomic structures of monodentate Fe(II) complexes adsorbed at the (021) surface in 5- and 4-fold coordinations over OH1 and OH2 sites corresponding to the following systems: (a) M5-OH1, (b) M5-OH2, (c) M4-OH1, (d) M4-OH2 (see Table 4 for their adsorption energetics).


distances are listed in Table 5. We see that the variations in atomic distances between adsorbed Fe(II) and the closest

Page 23 of 35


Figure 7. DFT optimized atomic structure of tridentate Fe(II) complex adsorbed at the (021) surface (T5 system).

System

a(Fe2+aq--Fe3+struct)

b(Fe2+aq--OHsurf)

c(Fe2+aq--OHaq)

M5-OH1

3.76

1.99

2.12; 2.12; 2.16; 2.19

M4-OH1

3.76

1.97

2.04; 2.05; 2.08

M5-OH2

3.80

2.13

1.98; 2.07; 2.28; 2.36

M4-OH2

3.86

2.15

1.95

d(Fe2+aq--H2Oaq)

2.03; 2.19

Table 5. Selected interatomic distances (in Å) for various Fe(II) adsorption complexes on the (021) surface.

Regarding the ET process at the (021) surface, we find that the post-ET states are always less energetically favorable than the pre-ET states, being 90 kJ/mol uphill for the post-ET state of the most stable the M5-OH1 complex. As in the case of (110) surface, we observe the tendency for deprotonation of water ligands upon ET into the goethite lattice. For instance, in the B6-OH1 system one H2O donated a proton to a OH2 site in the post-ET state. The same water hydrolysis is observed for T5 where one H2O donated a proton to an OH1 site upon ET. Therefore, such proton coupled electron transfer seems to be



DOI: 10.1039/C5CP00921A



DOI: 10.1039/C5CP00921A

general for both surfaces under study. Calculated Bader charges of the Fe atoms for the (021) surface listed in Table 6

quite similar for both pre- and post-ET states in which the 3+ charge state is characterized by the Bader charge of about 2.0e and the 2+ charge state by about 1.5e. Therefore, similar to the (110) case, the amount of electron density of about 0.5e should be transferred upon interfacial ET. Our calculations showing only uphill ET barriers from adsorbed Fe(II) to


structural Fe(III) in the goethite lattice are in agreement with the previous DFT investigation [27] suggesting that spontaneous oxidation upon adsorption at defect-free goethite surfaces is unlikely.

System

Pre-ET

Post-ET

Fe2+aq

Fe3+goethite

Fe3+aq

Fe2+goethite

M5-OH1

1.55

2.04

2.01

1.53

M4-OH1

1.54

2.04

M5-OH2

1.52

2.02

M4-OH2

1.52

2.01

T5

1.53

2.01

1.97

1.55

Outer-sphere

Table 6. Bader atomic charges for Feaq—Fegoethite redox couple at the (021) surface before (pre-ET state) and after (post-ET state) the electron transfer from adsorbed Fe(II) complexes into the goethite structure across different systems. For comparison, the corresponding values in bulk goethite are q(Fe3+) = 1.99e; q(Fe2+) = 1.53e.

Also, our AIMD simulations at room temperature and DFT optimizations of both outer- and inner-sphere complexes show that surface H2O molecules formed due to hydroxylation of the (021) surface (Fig. 2a) are rather loosely bound to the surface Fe(III) ions and can easily dissociate from the surface. This lability should in turn create the possibility for almost barrierless attachment of aqueous Fe(II) to the surface via the formation of Fe2+(OH)aq--Fe3+goethite bonds, a process that was not observed for the (110) surface. Given similar Gibbs free energies of hydroxylated (110) and (021) surfaces computed using ab initio thermodynamics [27] and comparable adsorption energies of Fe(II) complexes at both


feature smaller variations across different systems than in the case of (110) facet. The absolute atomic charges, however, are

Page 25 of 35


DOI: 10.1039/C5CP00921A

surfaces calculated in this study, the possibility of nearly barrierless Fe(II) adsorption due to potentially weakly bound

crystal growth by oxidative addition of Fe(II) [56]. Because we observed in the calculations that aqueous species from solution can donate protons to surface OH groups during electron transfer, a steady state supply of labile surface H2O molecules, and thus Fe(II) docking sites, could be formed with each Fe(II) adsorption event. To explore the energetics of


water desorption that would leave behind surface Fe(III) sites reactive for adsorption of Fe(II) complexes, we investigate this process in the following section.

3.4.

Energetics of water desorption.

Desorption of H2O molecules from the hydroxylated (021) surface occurred during both AIMD simulations at room temperature and DFT optimizations at zero temperature. To probe the energetics of H2O desorption from each facet under study we calculate here the energies required to desorb H2O molecule from each surface site. Note that for OH sites on both surfaces we added one proton to each site to create a surface water molecule. The reference state with the desorbed H2O for each site corresponds to an H2O molecule placed in the middle of the vacuum gap to minimize water/surface interactions.

Table 7 lists the calculated energies for the desorption of surface H2O molecules from both goethite facets. It is seen that H2O molecules at the (110) surface are generally much stronger bound to the surface than those at the (021) surface. Moreover, the desorption energy for a water molecule from the (021) H2O site is little bit lower than 2 kJ/mol thus being within the thermal energy at room temperature (~2.5 kJ/mol). This suggests that H2O molecules formed at the surface due to hydroxylation should be much more mobile at the (021) than at the (110) goethite facet. As a result, these outgoing H2O molecules will leave behind under-coordinated Fe(III) surface ions that can saturate their valence by exothermic adsorption of aqueous Fe(II) species from solution. We note here that the most energetically favorable adsorption position for aqueous Fe(II) on the (021) surface is found to be the one over the OH1 site (M5-OH1 complex, see Table 4). This adsorption configuration is characterized by the presence of a very mobile surface H2O molecule in the immediate proximity of the adsorbed Fe(II) complex. This H2O molecule can desorb and leave behind an open surface site for “free” adsorption of the second Fe(II) complex at the nearby site. As a result of this process, two parallel rows of Fe octahedra are expected to grow on the (021) surface thus fully maintaining the underlying atomic stacking sequence. This might help explain larger growth rates of the (021) facet over the (110) and, as a consequence, the experimentally observed needle


surface H2O molecules at the (021) but not at the (110) surface could be a contributing factor to the anisotropy of goethite



DOI: 10.1039/C5CP00921A

shapes of goethite crystals with the sides made mostly of (110) while the caps are formed predominantly by the (021)

Surface

(110)

(021)

Site

OH1

OH2

OH3

OH1

OH2

H 2O

Desorption energy (kJ/mol)

134

151

160

7

--a

2

Table 7. Calculated energies required to desorb an H2O molecule from different surface sites on the (110) and (021) goethite facets. Reference system corresponds to the desorbed H2O molecule placed in the middle of the vacuum gap to minimize H2O/surface interactions. Thermal energy at room temperature is about 2.5 kJ/mol.

3.5.

Effect of oxygen vacancy on interfacial ET. In the above sections we found that although the adsorption of [Fe(H2O)6]2+ complex from outer- to inner-sphere at

the defect-free (110) and (021) goethite surfaces is exothermic, forward interfacial ET reaction into goethite Fe(III) is always characterized by an uphill activation barrier. In this section we probe the influence of structural defects in goethite on the process of interfacial ET, using the example of an oxygen vacancy as a common structural defect in many transition metal oxides. We aim to understand whether the presence of such vacancies might stabilize a post-ET state and thus lower the height of the ET activation barrier or even invert the thermodynamic driving force from uphill to downhill, making the electron hopping from the adsorbed Fe(II) into the lattice more favorable than was found for defect-free surfaces. To maximize the effect on the interfacial ET we remove an oxygen atom from the Fe(III) lattice ion which is accepting an electron from the surface-bound Fe(II) (see Figure 7).

We consider here the most favorable adsorption complex of Fe(II) at the (021) surface (M5-OH1 complex, see Table 4). Figure 7 shows the position of the oxygen vacancy, created by removing an oxygen atom from the electronaccepting Fe octahedron corner-shared with the adsorbed Fe complex. Proton coupling to the electron transfer was again



surface which is the main direction of crystal growth.

Page 27 of 35


DOI: 10.1039/C5CP00921A

found to be favorable; configurations explored include forming one or two OH ligands out of Fe water ligands by

case post-ET states are only 2 kJ/mol (with two OH) and 8 kJ/mol (with one OH) higher in energy than the initial pre-ET state. The more stable post-ET case with two OH ligands on the adsorbed Fe can be directly compared with the situation when the oxygen vacancy is not present (defect-free goethite). In this case the post-ET state is less favorable by about 63


kJ/mol. The oxygen vacancy thus appears to have a very large enhancing effect on the forward interfacial electron transfer reaction.

Figure 7. M5-OH1 adsorption complex with the oxygen vacancy (black box) at the (021) surface of goethite.

It is also instructive to compare Bader charges before and after interfacial electron hopping (Table 8). First of all, we can see that atomic charges of the adsorbed Fe in both pre- and post-ET states are quite similar to those calculated without oxygen vacancy being around 1.5e for +2 (pre-ET) and 2e for +3 (post-ET) states (see Table 6), however, the amount of charge transferred across the interface is smaller in the presence of oxygen vacancy (0.49e vs. 0.43e). This is probably due to the fact that the lattice Fe3+ ions accepting electron density from the adsorbed Fe2+ are already partially reduced because of the presence of oxygen vacancy nearby changing their electrochemical potentials. Such smaller change in Bader charges in the pre- and post-ET states might also reflect a smaller reorganization energy associated with the


transferring one or two protons, respectively, to OH sites nearby on the surface (Figure 7). Our calculations show that in this



DOI: 10.1039/C5CP00921A

electron transfer as compared to the case without oxygen vacancy, because a broader distribution of the transferred charge


two nearly energetically equal pre- and post-ET states (differing by only 2 kJ/mol).

Ion

Pre-ET

Post-ET (one OH ligand)

Post-ET (two OH ligands)

Feaq

1.54

1.97

1.95

Fegoethite(1)

1.34

1.29

1.29

Fegoethite(2)

1.67

1.26

1.28

Table 8. Bader atomic charges for Feaq—Fegoethite redox couple at the (021) surface before (pre-ET state) and after (post-ET state) the electron transfer from the adsorbed Fe(II) complex into the goethite structure in the presence of oxygen vacancy shared between Fegoethite(1) and Fegoethite(2) (see Figure 7). For comparison, the corresponding values in bulk goethite are q(Fe3+) = 1.99e; q(Fe2+) = 1.53e.

3.6

Discussion.

Our modeling results suggest that both (110) and (021) surfaces of goethite are characterized by quite similar qualitative and quantitative features of aqueous Fe(II) adsorption. Specifically, we find that monodentate attachment of aqueous Fe(II) over the OH1 sites at both surfaces with decreased coordination from six to five and four are the most energetically favorable inner-sphere adsorption configurations. These complexes are characterized by rather comparable energies of formation from their outer-sphere counterparts. Also, these monodentate binding configurations should be preferred from a statistical point of view. We also observe the hydrolysis of H2O molecules in the first coordination shell of adsorbed Fe(II). Such hydrolysis as well as the decreased number of H2O ligands around the metal ions upon adsorption were previously observed for other systems such as adsorption of Fe(II) on the Fe-bearing clay mineral surfaces [24] and Zn(II) on the (110) surface of rutile [31].

One of the important differences that we observe when comparing (110) and (210) surfaces is their different ability to exchange surface water molecules with the solution species: H2O molecules at the (021) facet have much lower


requires less nuclear reorganization. This in turn should lead to a lower activation barrier of electron transfer between these

Page 29 of 35


DOI: 10.1039/C5CP00921A

desorption energies than those on the (110) facet and are expected to be very mobile at room temperature. Given quite

interfacial water is expected to have important implications for understanding goethite crystal growth because the fast exchange between surface H2O groups and solution species at the (021) facet can significantly to contribute to the growth


anisotropy of goethite crystals.

Regarding the interfacial ET, our calculations, although being more comprehensive than previous theoretical studies, still demonstrate that the forward ET reaction from the adsorbed Fe(II) into goethite Fe(III) is always energetically uphill at the defect-free surfaces. We find that proton transfer from H2O ligands of the adsorbed Fe(II) to the surface should, however, help facilitate the ET into the goethite crystal by stabilizing the post-ET adsorption state of surface-bound Fe(III) with the newly formed OH groups. We also find here that the protonation state of oxygen species mediating electron exchange into the goethite does not have an effect of making ET spontaneous at the surfaces. We point out here that this general conclusion about uphill forward ET reaction into the crystal turns out to be rather general across at least a number of Fe-bearing oxide systems studied until now [18, 27, 53]. This suggests that the energetics of electron hopping into the crystal from a solvated Fe(II) is dictated primarily by the structural environment adsorbed Fe species may or may not be able to attain at the surface, and is less dependent on the nature of the Fe(III)-(oxyhydr)oxide crystal itself. Most Fe-bearing oxides are characterized by similar linkages of octahedral Fe—O topologies at the local motif level. Also, Fe—O crystal bonds that need to distort upon electron transfer into the structural Fe(III) are much stiffer than those for Fe in aqueous solution, and this makes it more difficult to accommodate the localized valence change from 3+ to 2+, relative to Fe in solution. This also means that the forward electron transfer step is characterized by a large barrier due to a large reorganization energy required to distort crystal bonds and polarize the surrounding environment. It is also noteworthy that the higher stability of monodentate Fe(II) complexes over polydentate ones at both surfaces is consistent with the fact that Fe(II) ion is more stable in solution rather than as a charged defect in an otherwise Fe(III) crystal environment. Based on our calculations, the remaining key factors affecting the interfacial electron transfer on nominally perfect crystal terminations appear to be the specific surface structure, the formation of OH groups upon hydrolysis of H2O ligands of the adsorbed Fe(II), and the ability to transfer protons from adsorbed Fe water ligands to surface OH sites.

The general finding of an uphill interfacial electron transfer appears to directly contradict experimental observations for goethite and hematite suggesting the opposite [15, 57]. This discrepancy has led to some speculation about possibly central role of crystal defects in assisting the ET process. In this study, we find that the presence of an oxygen


similar energetics of Fe(II) adsorption on both surfaces as described above, this distinction in the behavior of bound



DOI: 10.1039/C5CP00921A

vacancy at the electron accepting Fe(III) site has indeed decreased the energy difference between pre- and post-ET states to

about 60 kJ/mol uphill. The analysis of atomic Bader charges for both pre- and post-ET states shows that the structural Fe(III) ion with an oxygen vacancy as the nearest neighbor is characterized by a lower positive charge than in the defectfree case. This site is intrinsically more electronegative, is pre-distorted to a larger polyhedral volume consistent with


requirements for structural Fe(II), and less electron density needs to be transferred across the interface to form structural Fe(II). This should result in a decreased activation barrier for ET. Although we observe that the presence of oxygen vacancies should facilitate the injection of electrons into the goethite surfaces, the extent of this enhancement is difficult to evaluate due to the coupling between electron and proton transfer.

3.7

Conclusions. DFT+U calculations aided by ab initio molecular dynamics simulations shows that aqueous

Fe(II) complexes have a strong driving force for surface adsorption from outer- to inner-sphere adsorption configurations on goethite (110) and (021). The most stable adsorption complexes at both surfaces are monodentate over singly-coordinated surface hydroxyl positions (OH1 sites) and these monodentate complexes should dominate the adsorbed fraction of Fe(II) in experiments. Our calculations demonstrate a clear tendency for aqueous Fe2+ to reduce its coordination number from 6 to 5 and 4 upon adsorption. The lowest-energy complexes at the (110) and (021) surfaces are 4- and 5-fold coordinated, respectively. Hydrolysis of water ligands also accompanies adsorption, yielding OH groups attached to the adsorbed Fe(II); the number of the formed OH ligands varies from 0 to 2 depending on the system. Overall, we observe quite similar adsorption energetics for both goethite surfaces for complexes featuring similar bonding environments. Regarding the electron transfer from adsorbed Fe2+ to structural Fe3+ in the goethite lattice, we find that the activation barriers are always energetically uphill rendering relatively low probabilities of electron hopping across both (110) and (021) interfaces. As a general signature of the electron transfer process at both surfaces, we observe stabilization of post-ET states with the adsorbed Fe3+ and structural Fe2+ through proton transfer from the H2O molecules of the adsorbed complexes to the surface oxygen species, a mechanism of local charge balance.

Both room temperature AIMD simulations and zero Kelvin DFT calculations show that surface H2O molecules at the hydroxylated (021) surface are loosely bound and can easily leave the surface. This should in turn create the possibility for nearly barrierless adsorption of aqueous Fe(II) at those sites at the (021) surface. Given a similar stability of (110) and


almost zero, as compared to the case without oxygen vacancy where the driving force for the forward ET is calculated to be

Page 31 of 35


DOI: 10.1039/C5CP00921A

(021) hydroxylated surfaces, as well as similar energies for Fe(II) adsorption at both surfaces, this mechanism of adsorption

of the needles while faster-growing (021) surfaces cap the needles.

Surface imperfections in the form of oxygen vacancies lower the electron transfer barrier to associated


undercoordinated structural Fe(III), by increasing the forward driving force and lowering the reorganization energy required for the transfer to occur. Future studies should consider and examine how solution chemistry might affect the interfacial electron transfer not simply through modifying the speciation of aqueous Fe(II) but simultaneously also through modifying the surface stoichiometry at the goethite/water interface.

AUTHOR INFORMATION Corresponding Author *[email protected]

ACKNOWLEDGMENT This research was supported by the U.S. Department of Energy Office of Science, Office of Basic Energy Sciences from the Chemical Sciences, Geosciences and Biosciences Division through the Geosciences Program at PNNL. The computations were performed using Institutional Computing at Pacific Northwest National Laboratory. The authors gratefully acknowledge Eugene Ilton for fruitful discussions.

REFERENCES (1)

Henderson, M. A. The Interaction of Water with Solid Surfaces: Fundamental Aspects Revisited. Surf. Sci. Rep.

2002, 46, 1-308. (2)

Hiemstra, T.; Van Riemsdijk, W. H. On the Relationship between Surface Structure and Ion Complexation of

Oxide-Solution Interfaces. In Encyclopedia of Colloid and Interface Science; Hubbard, A. T., Ed. Marcel Dekker, Inc.: New York, 2002, pp 3773-3799.


might account for the anisotropy in goethite crystal growth, with slower-growing (110) surfaces comprising dominant faces



DOI: 10.1039/C5CP00921A

(3)

Cornell, R. M.; Schwertmann, U. The Iron Oxides: Structure, Properties, Reactions, Occurrences and Uses. Wiley-

(4)

van der Zee, C.; Roberts, D. R.; Rancourt, D. G.; Slomp, C. P. Nanogoethie is the Dominant Reactive

Oxyhydroxide Phase in Lake and Marine Sediments. Geology 2003, 31, 993-996.


(5)

Mohan, D.; Pittman, C. U. Arsenic Removal from Water/Wastewater Using Adsorbents – a Critical Review. J.

Hazard. Mater. 2007, 142, 1-53. (6)

Latta, D. E.; Bachman, J. E.; Scherer, M. M. Fe Electron Transfer and Atom Exchange in Goethite: Influence of

Al-Substitution and Anion Sorption. Environ. Sci. Technol. 2012, 46, 10614-10623. (7)

Lovley, D. R.; Phillips, E. J. P. Novel Mode of Microbial Energy Metabolism – Organic-Carbon Oxidation

Coupled to Dissimilatory Reduction of Iron or Manganese. Appl. Environ. Microbiol. 1998, 54, 1472-1480. (8)

Schaetzl, R.; Anderson, S. Soils: Genesis and Geomorphology. Cambridge University Press: Cambridge, UK,

2005, p 817. (9)

Sherman, D. M.; Randall, S. R. Surface Complexation of Arsenic(V) to Iron(III) (Hydr)oxides: Structural

Mechanism from Ab Initio Molecular Geometries and EXAFS Spectroscopy. Geochim. Cosmochim. Acta 2003, 67, 42234230. (10)

Waychunas, G. A.; Kim, C. S.; Banfield, J. F. Nanoparticulate Iron Oxide Minerals in Soils and Sediments: Unique

Properties and Contaminant Scavenging Mechanisms. J. Nanopart. Res. 2005, 7, 409-433. (11)

Maher, K.; Bargar, J. R.; Brown, G. E. Environmental Speciation of Actinides. Inorg. Chem. 2013, 52, 3510-3532.

(12)

Pedersen, H. D.; Postma, D.; Jakobsen, R.; Larsen, O. Fast Transformation of Iron Oxyhydroxides by the Catalytic

Action of Aqueous Fe(II). Geochim. Cosmochim. Acta 2005, 69, 3967-3977. (13)

Jones, A. M.; Collins, R. N.; Rose, J.; Waite, T. D. The Effect of Silica and Natural Organic Matter on the Fe(II)-

catalyzed Transformation and Reactivity of Fe(III) Minerals. Geochim. Cosmochim. Acta 2009, 73, 4409-4422.


VCH, Weinheim, Germany, 2003.

Page 33 of 35


DOI: 10.1039/C5CP00921A

(14)

Handler, R. M.; Frierdich, A. J.; Johnson, C. M.; Rosso, K. M.; Beard, B. L.; Wang, C.; Latta, D. E.; Neumann, A.;


Sci. Technol. 2014, 48, 11302-11311. (15)

Williams, A. G. B. and Scherer, M. M. Environ. Sci. Technol. 2004, 38, 4782-4790.

(16)

Handler, R. M.; Beard, B. L.; Johnson, C. M.; Scherer, M. M. Atom Exchange between Aqueous Fe(II) and

Goethite: An Fe Isotope Tracer Study. Environ. Sci. Technol. 2014, 43, 1102-1107. (17)

Yanina, S. V.; Rosso, K. M. Linked Reactivity at Mineral-Water Interfaces through Bulk Crystal Conduction.

Science 2008, 320. 218-222. (18)

Zarzycki, P.; Kerisit, S.; Rosso, K. M. Molecular Dynamics Study of Fe(II) Adsorption, Electron Exchange and

Mobility at Goethite (a-FeOOH) Surfaces. J. Phys. Chem. C 2015, DOI: 10.1021/jp511086r (19)

Alexandrov, V., Neumann, A., Scherer, M. M. and Rosso, K. M. Electron Exchange and Conduction in Nontronite

from First-Principles. J. Phys. Chem. C 2013, 117, 2032-2040. (20)

Rosso, K. M., Smith, D. M. A. and Dupuis, M. J. Chem. Phys. 2003, 118, 6455-6466.

(21)

Iordanova, N., Dupuis, M., and Rosso, K. M. Charge Transport in Metal Oxides: A Theoretical Study of Hematite

alpha-Fe2O3. J. Chem. Phys. 2005, 122, 144305-. (22)

Kerisit, S. and Rosso, K. M. Computer Simulation of Electron Transfer at Hematite Surfaces. Geochim.

Cosmochim. Acta 2006, 70, 1888-1903. (23)

Kerisit, S. and Rosso, K. M. Kinetic Monte Carlo Model of Charge Transport in Hematite (alpha-Fe2O3). J. Chem.

Phys. 2007, 127, 124706. (24)

Alexandrov, V. and Rosso, K. M. Insights into the Mechanism of Fe(II) Adsorption and Oxidation at Fe-Clay

Mineral Surfaces from First-Principles Calculations. J. Phys. Chem. C 2013, 117, 22880-22886. (25)

Alexandrov, V. and Rosso, K. M. J. Chem. Phys. 2014, 140, 234701-234708.

(26)

Adelstein, N., Neaton, J. B., Asta, M. and Jonghe L. C. De. Density Functional Theory Based Calculation of

Small-Polaron Mobility in Hematite. Phys. Rev. B 2014, 89, 245115-245123.


Pasakarnis, T.; Premaratne, W. A. P. J.; Scherer, M. M. Fe(II)-Catalyzed Recrystallization of Goethite Revisited. Environ.



(27)

Russell, B. Payne, M., and Ciacchi, L. C. Phys. Rev. B 2009, 79. 165101-165113.

(28)

Kubicki, J. D., Paul, K. W., and Sparks, D. L. Geochem. Trans. 2008, 9, 4-19.

(29)

Ghose, S. K., Waychunas, G. A., Trainor, T. P., and Eng, P. Geochim. Cosmochim. Acta 2010, 74, 1943-1953.

(30)

Otte, K., Schmahl, W. W., and Pentcheva, R. Surf. Sci. 2012, 606, 1623-1632.

(31)

Bandura, A. V., Sofo, J. O. and Kubicki, J. D. Adsorption of Zn2+ on the (110) Surface of TiO2 (Rutile): A Density

Functional Molecular Dynamics Study. J. Phys. Chem. C 2011, 115, 9608-9614. (32)

Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865-3868.

(33)

Blöchl, P. E. Phys. Rev. B 1994, 50, 17953-17979.

(34)

Kresse, G.; Hafner, J. Phys. Rev. B 1993, 47, 558–561.

(35)

Kresse, G.; Hafner, J. Phys. Rev. B 1994, 49, 14251–14269.

(36)

Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.; Sutton, A. P. Phys. Rev. B 1998, 57, 1505-1509.

(37)

Leonov, I., Yaresko, A. N., Antonov, V. N., Korotin, M. A. and Anisimov, V. I. Phys. Rev. Lett. 2004, 93, 146404.

(38)

Rollmann, G., Rohrbach, A., Entel, P. and Hafner, J. Phys. Rev. B 2004, 69, 165107.

(39)

Bader, R. Atoms in Molecules: A Quantum Theory. Oxford University Press, New York, 1990.

(40)

Henkelman, G., Arnaldsson, A. and Jónsson, H. Comput. Mater. Sci. 2006, 36, 354.

(41)

Sanville, E., Kenny, S. D., Smith, R. and Henkelman, G. J. Comput. Chem. 2007, 28, 899.

(42)

Laberty, C. and Navrotsky, A. Geochim. Comochim. Acta 1998, 62, 2905.

(43)

van der Zee, C., Roberts, D. R., Rancourt, D. G. and Slomp, C. P. Geology 2003, 31, 993.

(44)

Guo, H. B. and Barnard, A. S. Phys. Rev. B 2011, 83, 094112.

(45)

Tunega, D. J. Phys. Chem. C 2012, 116, 6703.



DOI: 10.1039/C5CP00921A

Page 35 of 35


(46)

Schwertmann, U. Clay Miner. 1984, 19, 9.

(47)

Randall, S. R., Sherman, D. M., Ragnarsdottir, K. V. and Collins, C. R. Geochim. Cosmochim. Acta 1999, 63,


2971. (48)

Boily, J.-F. and Felmy, A. Geochim. Cosmochim. Acta 2008, 72, 3338-3357.

(49)

Kremleva, A.; Martorell, B.; Krueger, S.; Roesch, N. Uranyl Adsorption on Solvated Edge Surfaces of

Pyrophyllite: A DFT Model Study. Phys. Chem. Chem. Phys. 2012, 14, 5815-5823. (50)

Kosmulski, M. and Maczka, E. Langmuir 2004, 20, 2320.

(51)

Trainor, Th. P., Brown, G. E., Jr., Parks, G. A. L. Colloid Interface Sci. 2000, 231, 359.

(52)

Zhu, M. and Pan, G. J. Phys. Chem. A 2005, 109, 7648.

(53)

Kerisit, S.; Rosso, K. M. J. Phys. Chem. C (submitted).

(54)

Kerisit, K.; Rosso, K. M. Charge Transfer in FeO: A Combined Molecular-Dynamics and Ab Initio Study. J.

Chem. Phys. 2005, 123, 224712. (55)

Zarzycki, P.; Kerisit, K.; Rosso, K. M. Computational Methods for Intramolecular Electron Transfer in a Ferrous-

Ferric Iron Complex. J. Coll. Inter. Sci. 2011, 361, 293-306. (56)

Chun, C. L.; Baer, D. R.; Matson, D. W.; Amonette, J. E.; Penn, R. L. Characterization and Reactivity of Iron

Nanoparticles Prepared with Added Cu, Pd, and Ni. Environ. Sci. Technol. 2010, 44, 5079-5085. (57)

Rosso, K. M.; Yanina, S. V.; Gorski, C. A.; Larese-Casanova, P.; Scherer, M. M. Connecting Observations of

Hematite (α-Fe2O3) Growth Catalyzed by Fe(II). Environ. Sci. Technol. 2010, 44, 61-67.


DOI: 10.1039/C5CP00921A

Optical spectroscopy of the bulk and interfacial hydrated electron from ab initio calculations.

Ab initio modeling of the bonding of benzotriazole corrosion inhibitor to reduced and oxidized copper surfaces.

Modeling adsorption and reactions of organic molecules at metal surfaces.

Hydrated Electron Transfer to Nucleobases in Aqueous Solutions Revealed by Ab Initio Molecular Dynamics Simulations.

Electron crystallography at atomic resolution: ab initio structure analysis of copper perchlorophthalocyanine.

Theory of plasmon enhanced interfacial electron transfer.

Role of the dispersion force in modeling the interfacial properties of molecule-metal interfaces: adsorption of thiophene on copper surfaces.

Ab initio RNA folding.

Al(111).

Defects are needed for fast photo-induced electron transfer from a nanocrystal to a molecule: time-domain ab initio analysis.

Ab initio atomistic thermodynamics study on the oxidation mechanism of binary and ternary alloy surfaces.

Ab initio study of the adsorption of CO(2) on functionalized benzenes.

Ab Initio Computational Study of Chromate Molecular Anion Adsorption on the Surfaces of Pristine and B- or N-Doped Carbon Nanotubes and Graphene.

Li adsorption, hydrogen storage and dissociation using monolayer MoS2: an ab initio random structure searching approach.

Photo-driven oxidation of water on α-Fe2O3 surfaces: an ab initio study.

In situ imaging of interfacial precipitation of phosphate on Goethite.

Ab initio modeling approach towards establishing the structure and docking orientation of the Porphyromonas gingivalis FimA.

Nature of charge transport and p-electron ferromagnetism in nitrogen-doped ZrO2: an ab initio perspective.

CH3 NH3 PbI3 Heterojunction.

Ab initio and all-atom modeling of detergent organization around Aquaporin-0 based on SAXS data.

Dynamical scattering and electron crystallography--Ab initio structure analysis of copper perbromophthalocyanine.

Photoluminescent SBA-16 Rhombic Dodecahedral Particles: Assembly, Characterization, and ab Initio Modeling.

Synthesized heterogeneous Fenton-like goethite (FeOOH) catalyst for degradation of p-chloronitrobenzene.

Ab initio molecular dynamics simulation study of dissociative electron attachment to dialanine conformers.