from Density Functional Theory Modeling - American Chemical Society

Oct 27, 2011 - Haibo Guo* and Amanda S. Barnard. Virtual Nanoscience Laboratory, CSIRO Materials Science and Engineering, Clayton, VIC 3168, Australia...
0 downloads 0 Views 2MB Size

Surface Structure and Environment-Dependent Hydroxylation of the Nonpolar Hematite (100) from Density Functional Theory Modeling Haibo Guo* and Amanda S. Barnard Virtual Nanoscience Laboratory, CSIRO Materials Science and Engineering, Clayton, VIC 3168, Australia

bS Supporting Information ABSTRACT: Hematite (α-Fe2O3) nanoparticles are typically synthesized, stored, or used in hydrous environments, and the mineral/water interfaces are important for the surface stability and reactivity of these nanoparticles. Under such conditions the exposed facets are often passivated by hydroxyl groups. The configurations of surface hydroxylation vary with environmental conditions and affect the morphology and surface chemistry. Among the low-index hematite surfaces, the {100} are the only nonpolar surfaces and are often present on nanorods or nanotubes elongated along the [001] direction. In this paper we explore the relaxation and hydroxylation of this surface using first principles thermodynamics. Our results reveal that depending on the supersaturation of water and oxygen, various extents of hydroxylation may appear. In humid or hydrous environments, undercoordinated subsurface oxygen atoms are hydrogenated. In water singly and doubly coordinated hydroxyl groups coexist with chemisorbed water molecules at the surfaces. In environments where the humidity is reduced, the surface is terminated exclusively by doubly coordinated hydroxyl groups. The clean surface occurs when the humidity is further reduced or when temperature is elevated. On the basis of these findings, we have constructed the surface phase diagrams to describe the thermodynamic stability for two different temperatures. The phase diagrams enable us to predict the density and type of hydroxylation, which is relevant to surface stability, reactivity, and catalytic properties in hydrous or humid environments.

’ INTRODUCTION Iron oxides and oxyhydroxides are widespread in nature in the forms of nanoparticles and ultrathin films, and hematite is one of the most thermodynamically stable iron oxides at ambient conditions.1 Due to its surface reactivity, thermal stability, electronic and magnetic properties, and natural abundance, hematite is being increasingly used in environmental and energy applications.28 Many of the applications are closely related to the properties of the surfaces, as opposed to the bulk, and therefore exhibit an enhancement in performance with decreasing particle size. For example, hematite nanoparticles are much more effective in catalyzing the oxidation of carbon monoxide than larger particles,9 hematite nanostructures can withstand a higher concentration of Li intercalation without phase transformation when used in Li-ion battery anodes,10 and 7 nm hematite nanocrystals catalyze the oxidation of aqueous Mn2+ at rates 1 or 2 orders of magnitude faster than 37 nm nanocrystals.11 Accordingly, research on hematite has tended to focus on engineering the surfaces and morphologies of nanoparticles with these (and other) applications in mind. To date, by varying processes and conditions during synthesis, hematite nanorods,1214 hexagonal prisms,15,16 and nanotubes17 have been produced. In the past years extensive experimental and computational studies have been carried out for the surface termination and r 2011 American Chemical Society

chemistry of the surface (001) (using the three-index scheme)1829 and (012).3040 However, the (012) surfaces enclose rhombohedron shapes that resemble cubes,41 and since the (001) surface do not form an enclosed space, other complementary surfaces always accompany (001) facets and give rise to anisotropic shapes (e.g., nanotubes and nanorods). Morphologically anisotropic particles are usually elongated along the [001] (the trigonal axis) direction,15,17 and expose slow growing (hk0) surfaces parallel to the principal axis. For the cylindrical shapes, the side planes may consist of alternative (100) and (110) surfaces (both having similar surface energies when unpassivated42), while in the case of hexagonal prisms, the side planes may consist exclusively of one of the two competing surfaces. Restricting attention to the (001) or (012) facets cannot describe the stability and reactivity of the anisotropic shapes. It is important to consider the complementary facets if we are to understand the surfacedependent properties of realistic hematite nanostructures. Most surfaces obtained by cleaving bulk hematite (and other iron oxides and oxyhydroxides) are, like the (001) surface, inherently polar. The surface dipole moments, if not compensated, Received: August 8, 2011 Revised: September 28, 2011 Published: October 27, 2011 23023 | J. Phys. Chem. C 2011, 115, 23023–23029

The Journal of Physical Chemistry C cause a divergence in the electrostatic potential and surface energies when the surface areas are sufficiently large, or the underlying layers are sufficiently thick.43 Polar surfaces often undergo various stabilization processes, such as reconstruction, defect formation, or passivation by foreign species.19,44,45 The surface polarity plays an important role in surface composition, structure, reactivity, and adsorption. Small particles or thin films can tolerate a certain degree of polar instability without depolarization from the surfaces,19 and many of the observed unusual physical and chemical properties of nano-objects may be associated with the surface dipole moments. The {100} surface, however, is unique as it is the only nonpolar surface, and only one type of termination is possible when the bulk is cleaved in this orientation (all cleavage planes are alike). According to the Tasker classification,43 the (100) surfaces should be electrostatically stable with minimum surface relaxation, and one would expect the surface energy of (100) to be lower than the other polar surfaces. However, our previous calculations showed a clean (100) surface has similar surface energy as other low-index clean surfaces.42 This is not contradictory to the Tasker classification of ionic surface polarity and stability, as it is primarily based on the electrostatic interactions of surface dipole moments. The stability of iron oxide surfaces is dependent on the surface chemical composition and structure, and under ambient atmospheric or hydrous conditions, the surface properties vary largely with adsorbed chemical groups. It has been reported that the hematite (100) surface has both singly and doubly coordinated hydroxyl groups,46 and the adjacent singly coordinated oxygen sites seem to be the primary adsorption sites of SeO32 ions.47 Both studies used a surface model derived from the clean termination by removing the surface Fe atoms. Other types of terminations were not considered in the previous studies but can be energetically stable especially in varying environmental conditions. For this reason the surface chemistry, especially of the interactions between the surfaces and the ambient H2O and O2, is of great importance to our understanding of the surface stability and, ultimately, the shape and structure of the anisotropic nanostructures mentioned above. Surface hydroxylation is common on mineral surfaces in hydrous environments (e.g., liquid water or humid air) and is responsible for a range of variations in surface chemistry. Various studies have been carried out on both kinetic and dynamic aspects of the interactions between mineral surfaces and water, including water dissociation and hydroxylation configurations on the mineral surfaces. Computation has been increasingly used in this respect to verify experimental observations, to calculate key parameters, and to provide deep insight into those important processes. For example, it has been shown in first-principles molecular dynamics simulations that water molecules dissociate via preferred pathways on α-Al2O3 (001) surfaces.48 The energetic stability of the hydroxylated α-Al2O3 (001) surfaces depends on the surface coverage and environmental conditions.49 The structures of hydroxylated surfaces of hematite (012) have been investigated with the collaborative studies of experimental characterization and computational modeling.37 With these wellcharacterized surfaces, researchers have obtained rich data on surface chemistry. However, nanoparticles are usually enclosed by multiple facets which should be accounted for individually to acquire the overall surface chemistry. For the reasons mentioned above we choose to start with hematite (100) surface and encourage experimentalists to perform detailed characterizations on this special surface.


In this study we investigate the surface chemistry and stability of hematite (100) through first-principles electronic structure simulations, which offer the advantages of accurate control of surface structure, chemical composition, and environmental conditions. We will present the surface relaxations of the clean termination, hydroxylation configurations, and the thermodynamic stability of different terminations to predict the surface phase diagrams under different environmental conditions.

’ COMPUTATIONAL METHOD In this study we calculate the electronic total energies using the implementation of density functional theory (DFT) in VASP (Vienna ab initio simulation package).50,51 We have chosen Generalized Gradient Approximation (GGA) with the exchangecorrelation functional of Perdew, Burke, and Ernzerhof (PBE).52 The potentials of core electrons and nuclei of Fe, H, and O are reproduced using the projector augmented method (PAW).53 The PAW potentials are generated with valence configurations of d7s1, s1, s2p4, and core radii of 2.3, 1.1, 1.52 Bohr, respectively for Fe, H, and O. Nonlinear core corrections are included for Fe with a partial core radius of 2.0 Bohr. Our convergence tests determined that a plane-wave cutoff of 500 eV is sufficient for all the gas molecules and solid structures to converge the calculated energies to 2 meV/atom. We use a Gaussian distribution function for smearing of electronic occupation with a smearing width of 0.1 eV, and extrapolate the total energies to zero electronic temperature. The surface structures are optimized using the conjugate gradient algorithm until the residual HellmannFeynman forces are smaller than 0.01 eV/Å. The convergence criteria for self-consistent electronic iterations are 105 eV. The k-points are generated using the MonkhorstPack scheme54 and tested for the convergence criteria of 1 meV/atom for both bulk and surface structures. We used a k-point mesh of 11  11  11 for the rhombohedral cell of bulk hematite, and a 4  2  1 mesh for the surface slabs. The strongly correlated 3d electrons are difficult to cope with using either local density approximation (LDA) or GGA, which largely underestimates the band gaps of transition-metal oxides. We have included the on-site Coulomb interactions with the simplified rotationally invariant formulation by Dudarev et al.,55 where only the effective Coulomb repulsion parameter Ueff = U  J (U is the Coulomb parameter and J is the Stoner parameter for exchange field) is significant. Our previous tests showed the value of Ueff = 4.5 eV can reproduce the lattice parameters and energetic stability of five iron oxides and oxyhydroxides.56 A similar value of Ueff = 4.0 eV has been used in the study of FeOOH polymorphs, and it has been shown that the improvements in predicting bulk moduli and electronic structures are substantial.57 The surfaces are represented by periodic slab models. Our convergence tests showed that a vacuum of 10 Å thick is sufficient to reduce the error associated with the interactions with periodic image slabs to 2 mJ m2. All the slab models are built symmetrically so that there is no net surface dipole, and all the atoms are allowed to relax to their equilibrium positions. The thickness of the slabs is tested with the convergence criteria of 5 mJ m2 (see Figure 1). The difference between the calculated surface energies reduces to 0.003 J m2 between a seven-layer slab and a ninelayer slab. Symmetrization is switched off for the slab models, and dipolar corrections are included, so we found the surface net 23024 |J. Phys. Chem. C 2011, 115, 23023–23029

The Journal of Physical Chemistry C


Figure 1. Convergence of calculated surface energies with respect to the thicknesses of slabs.

dipole moments of slabs are negligibly small. A detailed description and visualization of the surface structures is provided in the Supporting Information. The free surface energies of hydroxylated surfaces are calculated using γ ¼ ðEslab  Σi Ni μi Þ=ð2AÞ

Figure 2. Surface relaxations of clean stoichiometric termination of hematite (100). The relaxations Δh and Δd are presented in Table 1. The vertical lines mark the boundaries of the computation box. Viewed along the [010] direction. Oxygen atoms are in red, and iron atoms are in gray.

Table 1. Surface Relaxations of the Clean Stoichiometric Termination of Hematite (100)a


where Eslab is the ground-state total energy of the slab, Ni represents the numbers of components i (Fe2O3, H2O, and O2), and 2A is the total area of the two free surfaces in the periodic slab model. The surface energy γ apparently depends on the temperature and pressure, because the chemical potentials μi of gas phases are sensitive to these environmental conditions. Here we use bulk hematite, H2O, and O2 as the reservoirs of the excess components Fe, H, and O, respectively. The stoichiometric coefficient of O2 may be negative, which represents O-deficient or Fe-rich terminations. The different reference states in thermodynamics and in firstprinciples calculations are coordinated in the framework of firstprinciples thermodynamics. In this method, the difference in chemical potentials between the two reference states is defined as connection energy, Δμ0, which can be calculated using experimentally measured thermodynamic functions. Concerning the details of the calculations, the readers are referred to ref 42, which contains the calculated connection energies of the gases (H2, O2, and water vapor). The temperature and pressure dependence of solid phases (slabs and bulks) are ignored because they are approximately 1 order of magnitude smaller than the gas phases. By this approximation one can take advantage of vast thermodynamic data for simple chemical systems (e.g., H2O and O2) and at the same time use the accurately calculated ground-state energy differences of complex systems (e.g., slabs with various hydroxylation configurations). By comparing the surface energies calculated using eq 1 we can construct surface phase diagrams for different environmental conditions. The phase diagrams constructed this way correspond to thermodynamic equilibrium and only show those surface terminations with the lowest free energy in a given chemical environment.

’ RESULTS AND DISCUSSIONS Surface Relaxation of the Stoichiometric Termination. Each atomic layer of the (100) plane is stoichiometric and neutral. At the surface, a surface dipole moment is created to prevent the underlying electrons from escaping into the vacuum (see Supporting Information for the surface dipole moment). This electric field of the surface dipole moment drives the near surface ions to adjust their positions accordingly. As a result, the topmost two atom-layers exhibit appreciable relaxations in the

Δh (Å)

Δd (Å)

Δd (%)















Δh is the difference between the average z coordinate of Fe and O atoms in the same layer, and Δd is the difference of interlayer distances compared with that of bulk. a

atomic positions, as shown in Figure 2. The interlayer distance (Δd in Figure 2 and Table 1) decreases (by 0.10 Å or 6.6% relative to the interlayer distance in the bulk) in the first layer, and increases (by 0.06 Å or 3.8%) in the second layer. Within each layer the atomic planes of Fe and O, which overlap in the bulk, become displaced (Δh in Figure 2) during the relaxation. We see in the uppermost layer that the Fe atoms move inward toward the bulk and the O atoms move outward toward the vacuum, while in the second layer the opposite relaxation pattern is observed. In the third layer, both Δd and Δh decrease to almost zero, recovering the bulk structure of hematite, and the Fe and O atoms are almost coplanar. In general, the magnitude of the surface relaxations of (100) is much smaller than those of the (001) surface.21,25,38,5860 This is consistent with the Tasker classification that nonpolar surfaces have minimum surface relaxations.19,43,44 In the present context however, it is the directions of the surface relaxations that are of interest. To balance the dipole moment generated from the distortion of electron clouds at the surface, ions relax so that more electronegative elements (O in the present study) move outward and less electronegative elements (Fe) move inward. This relaxation pattern is similar to that of nonpolar transitionmetal carbide (100) surfaces.6163 Surface Stability and Hydroxylation. As described above, the stoichiometric termination obtained by cleaving bulk hematite along the (100) plane consists of surface Fe atoms. It has been shown that the exposed surface Fe atoms are energetically unstable and likely to be either passivated or chelated away from (001) and (012) surfaces.6,38 To address the incomplete chelation, we remove the surface Fe atoms and create a purely oxygen-terminated surface. At the topmost layer of this surface, two-thirds of the oxygen atoms are singly coordinated (i.e., the oxygen atom is bonded to one Fe atom) and one-third are doubly coordinated. At the second layer, one-third of the oxygen atoms are doubly coordinated and two-thirds are triply coordinated (see Figure 3). This termination has a considerable number of 23025 |J. Phys. Chem. C 2011, 115, 23023–23029

The Journal of Physical Chemistry C


Table 2. Tested Configurations of Hydroxylations on Hematite (100)a OH configuration 1S



OH2 2T


n(H2O) n(O2) γ (J m2)


Figure 3. Atomic structure of hematite (100) with Fe atoms at the topmost layer removed. The coordination of the oxygen atoms in the first two layers is shown with characters “S” for singly, “D” for doubly, and “T” for triply coordinated. The oxygen atoms are in red, and the iron atoms in gray. Projected along the [010] direction.



4/4 2/2 0/2 0/4





B2 B3

2/4 2/2 0/2 0/4 4/4 2/2 2/2 0/4

2/4 0/4

8 8

2 2

1.433 1.420


0/4 2/2 0/2 0/4






4/4 2/2 2/2 4/4






0/2 0/2 0/4





2/2 0/2 0/4





2/2 2/2 0/4





2/2 2/2 4/4









1/2 3/4



Figure 4. Water molecules desorbed from an over-hydrogenated surface. The starting structure has two hydrogen atoms bonded to doubly coordinated oxygen atoms. Hydrogen atoms are in blue, oxygen atoms in red, and iron atoms in gray. Viewed along the [010] direction.

oxygen dangling bonds. When we attempted to optimize the atomic positions of this structure, all the singly coordinated oxygen atoms will combine into oxygen dimers and release from the surface. Only the doubly coordinated oxygen atoms remain bonded to the subsurface iron atoms. This over-oxygen-populated surface is therefore unrealistic, but it can serve as a starting point to construct hydroxylation groups. We also consider configurations that have part of the surface oxygen atoms removed and refer to them as modified configurations (see Supporting Information for details). We also constructed surfaces with chemically adsorbed H2O molecules by adding two hydrogen atoms to one surface oxygen atom, even though the experimental models only considered hydroxyl groups.46 The motivation for this is to examine the stability of chemisorbed water molecules, which have been observed using X-ray diffraction on another hematite surface, (012), under a hydrated environment.31,37 In addition to this we consider hydroxylation to the subsurface undercoordinated (i.e., coordination number is less than 4) oxygen atoms. Our tests showed doubly coordinated water molecules are unstable and desorb in the MD simulations (see Figure 4). Therefore, this type of hydroxylation is excluded from subsequent discussions. The tested configurations of hydroxylations are listed in Table 2 and explained in the Supporting Information. It is noted that the hydroxylated structures can also be constructed using molecule dynamics simulations, where water molecules are gradually added to the system and then interact with the surfaces.40,64 This method is useful to study coveragedependent adsorption energy and dynamic properties of adsorptiondesorption processes. However, the surface stoichiometry is predefined. The method used in the present study, which is hydrogenating the dangling oxygen atoms, allows for variations in the surface stoichiometry such that neutral OH or H groups may also be added to the system in addition to H2O. The

Chemisorbed water (OH2) is only singly coordinated and at the topmost surface. Hydroxyl (OH) can be singly (S), doubly (D), or triply (T) coordinated and can be at the topmost surface (1) and subsurface (2) layer. The notation “1D” means doubly coordinated species at the top surface layer, and other notations are obvious by analogy. The numbers of hydrogenated and available oxygen atoms are separated by a slash in each of the columns. The nonstoichiometric slabs have excess H2O and/or O2, and the amounts are indicated in the columns of n(H2O) and n(O2), respectively. The surface energies are calculated with H2O and O2 at the thermodynamic standard state. The notations of A, B1, etc., refer to hydroxylation configurations, and are described in detail in the Supporting Information.

latter method can construct terminations corresponding to alkaline or acid conditions. As mentioned above, the clean (100) surface is nonpolar and has the minimum electrostatic dipolar interactions. The converged surface energy is 1.369 J m2, as reported in ref 42 and listed in Table 2. Although it is known that local or semilocal functionals tend to underestimate surface energies,65 our result is close to 1.46 J m2, which is calculated using semiempirical interacting bond methods,66 but is approximately 60% of 2.251 J m2, calculated using classical interatomic potentials.67 However, we have confidence in this value as a frame of reference, as our results of the other low-index clean surfaces of hematite and goethite (using the same computational settings and methodologies) are consistently in good agreement with those in the literature.42 Unlike the clean surface, the surface energies of nonstoichiometric configurations depend on the chemical potential of the excess species, which we have calculated within the framework of first-principles thermodynamics. The values of the surface energies are provided in Table 2, with H2O and O2 at the thermodynamic standard state. However, in natural environmental settings, the partial pressures (or supersaturation) of water vapor and oxygen are often much lower than the standard pressure, such as in dry or anaerobic environments. To account for the variable chemical potentials under realistic conditions, we have drawn the relative stability of the surface configurations in Figure 5 for a range of chemical potentials of H2O and O2 at room temperature (298.15 K), and Figure 6 for temperature T = 373.15 K (the boiling temperature of water under standard pressure). The chemical potential of liquid water is largely insensitive to external pressures but dependent on the solution conditions.68 23026 |J. Phys. Chem. C 2011, 115, 23023–23029

The Journal of Physical Chemistry C

Figure 5. Relative stability of the surface configurations of hematite (100) at 298.15 K. The vertial dashed lines, from left to right, correspond to the chemical potential of hematite reduction to form bulk Fe, hematite reduction to form magnetite, O2 in atmospheric air, and the formation of ozone (O3) at thermodynamic standard state. The chemical potential of liquid water is marked with a horizontal line. Refer to the texts and ESI for the notations about the surface terminations.

Figure 6. Relative stability of the surface configurations of hematite (100) at 373.15 K. Refer to the caption of Figure 5 for the notations and marks.

The chemical potentials of the gas phases (water vapor and oxygen) are variable with their partial pressures. First, we find that several surface hydroxylation configurations are energetically unstable compared with those shown in Figure 5 and do not appear on this equilibrium phase diagram. Among those that do appear, Figure 5 shows that in most hydrous environments at room temperature the surface is predicted to have both doubly and singly coordinated surface hydroxyl groups and chemisorbed water molecules (configuration C5 in Table 2 and the Supporting Information). The number density of hydroxyl groups of this modified configuration is 7.1 nm2 (including both surface and subsurface hydroxyl groups). Hydrogen bonds are formed where the singly coordinated surface O atoms or the triply coordinated subsurface O atoms are acceptors (see Figure 7). It is worthy noting that not all the hydrogen atoms form hydrogen bonds, and these sites may attract electronegative


Figure 7. Top view of hm(100) surface with doubly coordinated hydroxyl groups (configuration C5 in Table 2). The oxygen atoms in the topmost layer are in purple, other oxygen atoms in red, hydrogen atoms in blue, and iron atoms in gray. The hydrogen bonds are highlighted with dashed lines and distances (unit, Å). The coordination numbers are marked with S for singly, D for doubly, and T for triply coordinated hydroxyl groups. Dashed blue lines mark the cell boundaries.

species that are close to the surface. This surface structure contains contiguous singly coordinated groups, different from that proposed in ref 46 which corresponds the configuration B1 in this paper. The hydroxylation configuration B5 has larger number densities of hydroxyl groups than configuration C5 but requires higher chemical potential of water. Second, under reductive conditions (which correspond to the large negative chemical potential of oxygen) the oxygen-deficient termination C4 is thermodynamically stable (see Figure 5 and Figure 6). In this termination all the undersaturated oxygen atoms are hydrogenated, forming half doubly coordinated and half triply coordinated hydroxyl groups. According to the phase diagram, this termination should occur before the reduction of hematite into magnetite under solution conditions. Under dry conditions and especially at elevated temperatures the clean surface is the most stable termination before reduction to magnetite (see Figure 6). We also examined using the phase diagrams the highly oxidative conditions, under which the stable termination may be B4, C1, or C2. Among these terminations, C1 contains no hydroxyl groups and is only stable under highly dry and oxidative conditions. The phase boundaries among the thermodynamically stable terminations (configurations C5, B5, C3, and A) are independent of the chemical potential of O2, because the chemical compositions of the stable terminations can be written as xFe2O3 3 yH2O. Those terminations with excess oxygen are relatively unstable when the partial pressure of O2 is lower than the standard pressure. However, with increased chemical potential of oxygen (as in oxygen plasma), the terminations with excess oxygen may be thermodynamically stable. In the environments where the humidity is reduced, the terminations with only doubly coordinated hydroxyl groups (configuration C3) and the clean surface (configuration A) have lower surface energies. We have found their occurrence of these two configurations depends on temperature, as shown in the temperature-dependent phase diagram for the chemical potential of oxygen equal to that in ambient air (Figure 8). At room temperature, the surface has doubly coordinated hydroxyl groups in the subsurface layer, while at elevated temperatures the 23027 |J. Phys. Chem. C 2011, 115, 23023–23029

The Journal of Physical Chemistry C


ciated with the iron oxide applications (such as the overpotential of oxygen evolution at the hematite photoanodes) may be resolved via suitable surface modifications.



Supporting Information. Additional details on surface dipole of the stoiciometric termination, construction of surface hydroxylation, and surface hydroxylation configurations. This material is available free of charge via the Internet at http://pubs.

’ AUTHOR INFORMATION Corresponding Author Figure 8. The temperature-dependent phase diagram of the (100) surface. The bold blue line represents the saturated vapor pressure of water in the temperature range from 273.15 to 373.15 K. The vertical dash-dotted lines mark 298.15 K (the standard temperature) and 373.15 K (the boiling temperature of water under 1 atm). The horizontal dashdotted line marks the standard pressure (1 atm).

configuration C3 with reduced density of hydroxyl should appear, before the formation of the clean surface after complete dehydration. The transition temperatures are sensitive to the chemical potential of H2O or the humidity level. We should point out that since the surface stability varies with temperature, it is possible that varying the temperature during growth provide a means of tuning the surface chemistry and perhaps controlling the morphology of nanostructures. The various morphologies synthesized at different temperatures may well be explained by the surface hydroxylation and stability, so an insightful understanding of the temperature dependence of the surface properties and nanomorphologies is planned in our future work.

’ CONCLUSIONS Hematite (100) is the only nonpolar low-index surface, but we have shown that it exhibits a small but appreciate surface relaxation, which has not been reported before. The Fe atoms in the terminal layer relax toward the bulk, and the oxygen atoms relax outward away from the bulk, so as to create an electrostatic potential that prevents electrons from escaping from the surface. Even with the dipolar compensation, the surface energy of the stoichiometric termination of this nonpolar surface is close to other polar surfaces, indicating strong contributions other than the electrostatic interactions. We have shown that surface hydroxylation, in varying degrees, can have an important impact on the stability of this surface (with respect to the polar alternatives) but that the preferred surface hydroxylation configurations vary with temperature. Since surface hydroxylation is related to the proton affinity, surface reactivity, and adsorbance capacity,69 the temperature dependence of the surface hydroxylation configurations should be included in relevant studies. For example, thermodynamic models that use the surface hydroxylation configurations as input parameters can be refined by taking account of the temperature dependence (as we have shown in the temperature-dependent surface phase diagram), and more accurately describe the states of realistic systems. In addition to this, these results can be used to refine synthesis processes, and some of the difficulties asso-

*E-mail: [email protected].

’ ACKNOWLEDGMENT This project has been partially supported by the Australian Research Council under Grant Number DP0986752. The authors acknowledge the supercomputer usage and support from the NCI National Facility in Australia, under Merit Allocation Scheme grant p00. H.G. thanks Professor Huifang Xu for suggestions and fruitful discussions on the morphology of hematite nanoparticles. ’ REFERENCES (1) Cornell, R. M.; Schwertmann, U. The iron oxides, 2nd ed.; Wiley-VCH: Weinheim, Germany, 2003. (2) Al-Abadleh, H. A.; Grassian, V. H. Surf. Sci. Rep. 2003, 52, 63–161. (3) Eggleston, C. M.; Shankle, A. J. A.; Moyer, A. J.; Cesar, I.; Gr€atzel, M. Aquat. Sci. 2009, 71, 151–159. (4) Mason, S. E.; Iceman, C. R.; Tanwar, K. S.; Trainor, T. P.; Chaka, A. M. J. Phys. Chem. C 2009, 113, 2159–2170. (5) Li, Y.; Zhang, J. Z. Laser Photonics Rev. 2010, 4, 517–528. (6) Waychunas, G. A.; Kim, C. S.; Banfield, J. F. J. Nanopart. Res. 2005, 7, 409–433. (7) Wigginton, N. S.; Haus, K. L.; Hochella, M. F., Jr. J. Environ. Monit. 2007, 9, 1285–1432. (8) Chernyshova, I. V.; Hochella, M. F., Jr.; Madden, A. S. Phys. Chem. Chem. Phys. 2007, 9, 1736–1750. (9) Li, P.; Miser, D. E.; Rabiei, S.; Yadav, R. T.; Hajaligol, M. R. Appl. Catal., B 2003, 43, 151–162. (10) Larcher, D.; Masquelier, C.; Bonnin, D.; Chabre, Y.; Masson, V.; Leriche, J. B.; Tarascon, J. M. J. Electronchem. Soc. 2003, 150, A133– A139. (11) Madden, A. S.; Hochella, M. F., Jr. Geochim. Cosmochim. Acta 2005, 69, 389–398. (12) Wu, J.-J.; Lee, Y.-L.; Chiang, H.-H.; Wong, D. K.-P. J. Phys. Chem. B 2006, 110, 18108–18111. (13) Wu, C.; Yin, P.; Zhu, X.; OuYang, C.; Xie, Y. J. Phys. Chem. B 2006, 110, 17806–17812. (14) Lindgren, T.; Wang, H.; Beermann, N.; Vayssieres, L.; Hagfeldt, A.; Lindquist, S.-E. Sol. Energy Mater. Sol. Cells 2002, 71, 231–243. (15) Xia, C.; Hu, C.; Xiong, Y.; Wang, N. J. Alloys Compd. 2009, 480, 970–973. (16) Wang, W.; Howe, J. Y.; Gu, B. J. Phys. Chem. C 2008, 112, 9203–9208. (17) Jia, C.-J.; Sun, L.-D.; Yan, Z.-G.; You, L.-P.; Luo, F.; Han, X.-D.; Pang, Y.-C.; Zhang, Z.; Yan, C.-H. Angew. Chem., Int. Ed. 2005, 44, 4328– 4333. 23028 |J. Phys. Chem. C 2011, 115, 23023–23029

The Journal of Physical Chemistry C (18) Yamamoto, S.; Kendelewicz, T.; Newberg, J. T.; Ketteler, G.; Starr, D. E.; Mysak, E. R.; Andersson, K. J.; Ogasawara, H.; Bluhm, H.; Slameron, M.; Brown, G. E., Jr.; Nilsson, A. J. Phys. Chem. C 2010, 114, 2256–2266. (19) Goniakowski, J.; Finocchi, F.; Noguera, C. Rep. Prog. Phys. 2008, 71, 016501. (20) Eggleston, C. M.; Stack, A. G.; Rosso, K. M.; Higgins, S. R.; Bice, A. M.; Boese, S. W.; Pribyl, R. D.; Nichols, J. J. Geochim. Cosmochim. Acta 2003, 67, 985–1000. (21) Trainor, T. P.; Chaka, A. M.; Eng, P. J.; Newville, M.; Waychunas, G. A.; Catalano, J. G.; Brown, G. E., Jr. Surf. Sci. 2004, 573, 204–224. (22) Kim, C.-Y.; Escuadro, A. A.; Bedzyk, M. J.; Liu, L.; Stair, P. C. Surf. Sci. 2004, 572, 239–246. (23) Kurtz, R. L.; Henrich, V. E. Surf. Sci. 1983, 129, 345–354. (24) Wang, X.-G.; Weiss, W.; Shaikhutdinov, S. K.; Ritter, M.; Petersen, M.; Wagner, F.; Schl€ogl, R.; Scheffler, M. Phys. Rev. Lett. 1998, 81, 1038. (25) Bergermayer, W.; Schweiger, H.; Wimmer, E. Phys. Rev. B 2004, 69, 195409. (26) Weiss, W.; Ranke, W. Prog. Surf. Sci. 2002, 70, 1–151. (27) Shaikhutdinov, S. K.; Weiss, W. Surf. Sci. 1999, 432, L627– L634. (28) Becker, U.; Hochella, M. F., Jr.; Apra, E. Am. Mineral. 1996, 81, 1301–1314. (29) Junta-Rosso, J. L.; Hochella, M. F., Jr. Geochim. Cosmochim. Acta 1996, 60, 305–314. (30) Guo, H.; Barnard, A. S. J. Mater. Chem., 2011, DOI: 10.1039/ c1jm13362d. (31) Tanwar, K. S.; Lo, C. S.; Eng, P. J.; Catalano, J. G.; Walko, D. A.; Brown, J., G. E.; Waychunas, G. A.; Chaka, A. M.; Trainor, T. P. Surf. Sci. 2007, 601, 460–474. (32) Waychunas, G. A. Powder Diffr. 2009, 24, 89–93. Author is in Earth Sciences Division, Lawrence Berkeley National Lab. (33) Henderson, M. A. Surf. Sci. 2010, 604, 1197–121. (34) Spagnoli, D.; Gilbert, B.; Waychunas, G. A.; Banfield, J. F. Geochem. Cosmochim. Acta 2009, 73, 4023–4033. (35) Henderson, M. A. Surf. Sci. 2002, 515, 253–262. (36) Jones, F.; Rohl, A. L.; Farrow, J. B.; van Bronswijk, W. Phys. Chem. Chem. Phys. 2000, 2, 3209–3216. (37) Henderson, M. A.; Joyce, S. A.; Rustad, J. R. Surf. Sci. 1998, 417, 66–81. (38) Wasserman, E.; Rustad, J. R.; Felmy, A. R.; Hay, B. P.; Halley, J. W. Surf. Sci. 1997, 385, 217–239. (39) Wasserman, E.; Rustad, J. R.; Felmy, A. R. Surf. Sci. 1999, 424, 19–27. (40) Rustad, J. R.; Wasserman, E.; Felmy, A. R. Surf. Sci. 1999, 424, 28–35. (41) Gurlo, A.; Lauterbach, S.; Miehe, G.; Kleebe, H.-J.; Riedel, R. J. Phys. Chem. C 2008, 112, 9209–9213. (42) Guo, H. B.; Barnard, A. S. J. Mater. Chem. 2011, 21, 11566–11577. (43) Tasker, P. W. J. Phys. C: Solid State Phys. 1979, 12, 4977–4984. (44) Noguera, C. J. Phys.: Condens. Matter 2000, 12, R367. (45) Pentcheva, R.; Wendler, F.; Meyerheim, H. L.; Mortiz, W.; Jedrecy, N.; Scheffler, M. Phys. Rev. Lett. 2005, 94, 126101. (46) Barron, V.; Torrent, J. J. Colloid Interface Sci. 1996, 177, 407–410. (47) Catalano, J. G.; Zhang, Z.; Fenter, P.; Bedzyk, M. J. J. Colloid Interface Sci. 2006, 297, 665–671. (48) Hass, C., K; Schneider, W. F.; Curioni, A.; Andreoni, W. Science 1998, 282, 165. (49) yodziana, Z.; Nørskov, J. K.; Stoltze, P. J. Chem. Phys. 2003, 118, 11179. (50) Kresse, G.; Furthm€uller, J. Phys. Rev. B 1996, 54, 11169. (51) Kresse, G.; Furthm€uller, J. Comput. Mater. Sci. 1996, 6, 15–50. (52) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865.


(53) Kresse, G.; Joubert, D. Phys. Rev. B 1999, 59, 1758. (54) Monkhorst, H. J.; Pack, J. D. Phys. Rev. B 1976, 13, 5188–5192. (55) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Hmphreys, C. J.; Sutton, A. P. Phys. Rev. B 1998, 57, 1505–1509. (56) Guo, H. B.; Barnard, A. S. Phys. Rev. B 2011, 83, 094112. (57) Otte, K.; Pentcheva, R.; Schmahl, W. W.; Rustad, J. R. Phys. Rev. B 2009, 80, 205116. (58) Thevuthasan, S.; Kim, Y. J.; Yi, S. I.; Chambers, S. A.; Morais, J.; Denecke, R.; Fadley, C. S.; Liu, P.; Kendelewicz, T.; Brown, G. E., Jr. Surf. Sci. 1999, 425, 276–286. (59) Eggleston, C. M.; Hochella, M. F., Jr. Am. Mineral. 1992, 77, 911–922. (60) Rohrbach, A.; Hafner, J.; Kresse, G. Phys. Rev. B 2004, 70, 125426. (61) Kobayashi, K. Jpn. J. Appl. Phys. 2000, 39, 4311–4314. (62) Tagawa, M.; Kawasaki, T.; Oshima, C.; Otani, S.; Edamoto, K.; Nagashima, A. Surf. Sci. 2002, 517, 59–64. (63) Price, D. L.; Wills, J. M.; Cooper, B. R. Phys. Rev. Lett. 1996, 77, 3375. (64) Rustad, J. R.; Felmy, A. R.; Hay, B. P. Geochim. Cosmochim. Acta 1996, 60, 1553–1562. (65) Schimka, L.; Harl, J.; Stroppa, A.; Gr€uneis, A.; Marsman, M.; Mittendorfer, F.; Kresse, G. Nat. Mater. 2010, 9, 741–744. (66) Bulgakov, N. N.; Sadykov, V. A. React. Kinet. Catal. Lett. 1996, 58, 397–402. (67) Reeves, N. J.; Mann, S. J. Chem. Soc., Faraday Trans. 1991, 87, 3875–3880. (68) Pliego, J. R., Jr. Chem. Phys. Lett. 2003, 381, 246–247. (69) Venema, P.; Hiemstra, T.; Geidler, P. G.; van Riemsdijk, W. H. J. Colloid Interface Sci. 1998, 198, 282–295.

23029 |J. Phys. Chem. C 2011, 115, 23023–23029