Water Interface Revealed by

Article pubs.acs.org/JPCC

An Atomistic Description of the γ‑Alumina/Water Interface Revealed by Ab Initio Molecular Dynamics B. F. Ngouana-Wakou,† P. Cornette,‡ M. Corral Valero,† D. Costa,*,‡ and P. Raybaud*,† †

IFP Energies nouvelles, Direction Catalyse et Séparation, Rond-point de l’échangeur de Solaize, BP 3, 69360 Solaize, France Physico-Chimie des Surfaces, PSL Research University, CNRS, Institut de Recherche de Chimie Paris, Chimie ParisTech, 11 rue Pierre et Marie Curie, 75005 Paris, France

‡

S Supporting Information *

ABSTRACT: We report ab initio molecular dynamics (AIMD) simulations of the (100) and (110) γ-Al2O3/water interfaces at 300 K, using two sets of supercell models for each surface and two time lengths of simulation (10 and 40 ps). We first show that the effect of liquid water on the vibrational frequencies of hydroxyl groups at the interface varies according to the type of surface. This trend is explained by two key parameters affecting the interaction of both surfaces with water: the nature of the OH groups (i.e., μ1-OH, μ1-H2O, μ2-OH, and μ3-OH) and H-bond network among surface OH groups. The hydroxylated (110) surface favors the local structuration of water at the interface and the solvation of its μ1-OH and μ1-H2O groups by water similarly as in bulk liquid water. By contrast, on the (100) surface, a stronger H-bond network among μ1-OH and μ1-H2O groups reduces the water/surface interaction. We illustrate also how the interfacial interacting sites are spatially organized on the surfaces by twodimensional maps of O−H distances. On both surfaces, the interfacial water layer orientation is predominantly Hup−Hdown. For long AIMD simulation time, Grotthuss-like mechanisms are identified on the (110) surface.

1. INTRODUCTION The molecular scale understanding of the chemical phenomena that take place at the solid−liquid interface (SLI) is a common challenge present in many scientific domains such as chemistry, biology, materials, and environmental sciences.1 Among others, we can cite the study of corrosion phenomena, the formation of aerosols, micelles, colloids, cellular membranes, groundwater purification, electrochemical processes, and heterogeneous catalysis. For heterogeneous catalysis, understanding and controlling SLI chemical phenomena at the molecular level represents a challenge for the preparation of catalytic materials and for controlling their reactivity.2 Transition aluminas, such as γ-alumina (γ-Al2O3), are the most widely used oxide supports for clean fuel production from fossil and biomass resources.3 On the one hand, the synthesis steps of industrial supported catalysts are most often undertaken by aqueous phase impregnation techniques on γ-alumina supports3 where the rational understanding and design of solvated metallic precursors in interaction with the solid support at SLI is crucial. On the other hand, the industrial © 2017 American Chemical Society

production of energy from fossil and biological resources implies chemical reactions occurring often at a catalyst−liquid interface. For the Fischer−Tropsch (FT) process and for biomass conversion processes (such as the “aqueous phase reforming”, APR), water is a predominant byproduct that modifies the chemical properties of the γ-alumina supported catalyst, its stability (dissolution), and reactivity.4−8 Despite the widespread experimental knowledge of solution chemistry,9,10 there is still a crucial need to improve the molecular scale understanding and to enable rational design of SLI at the preparation steps of catalyst or in aqueous reaction conditions. In particular, it remains unclear how the acidic− basic properties of aluminas predominantly governed by the nature and concentration of hydroxyl groups evolve in aqueous medium. Received: January 4, 2017 Revised: March 24, 2017 Published: April 19, 2017 10351

DOI: 10.1021/acs.jpcc.7b00101 J. Phys. Chem. C 2017, 121, 10351−10363

Article

The Journal of Physical Chemistry C Table 1. Relevant Information about the Interfacial Models and Time Length Used for AIMD crystallographic orientation (100) (110) (100) (110)

supercell size

a, b, and c cell parameters (Å)

number of interfacial H2O molecules

length of AIMD simulations (ps)

× × × ×

8.34, 11.09, 40.00 8.03, 8.32, 36.00 16.72, 16.66, 40.00 16.06, 16.64, 36.00

43 43 180 180

40 40 10 10

(1 (1 (2 (2

2) 1) 3) 2)

The adsorption of water molecules on model surfaces of αalumina11,12 and γ-alumina13,14 has been investigated by density functional theory (DFT) calculations and revealed the various chemisorption modes (including formation of hydroxyls) and strength of water molecules as a function of coverage in the absence of surrounding liquid water. Thanks to the increase of computational performances, it has become possible to investigate through ab initio molecular dynamic (AIMD) simulations the structuration of various aluminum oxide− water interfaces such as boehmite (AlOOH)15−18 and αalumina surface.19 Beyond aluminum oxide materials (the scope of the present work), it can be noticed that a growing literature exists reporting AIMD studies on various solid−liquid interfaces such as goethite,20 silica,21 clays,22 and TiO2.23,24 Moreover, the behavior of SiO2 water interfaces has been compared to the α-alumina water interfaces.19 In particular, the effect of the H-bonding network between the OH groups on the water organization at the interface was highlighted in some of these studies. It has been shown that the α-alumina polymorph surfaces induce a significant organization of the interfacial water layer.19 On the boehmite surface, proton transfers between OH groups have been observed at monatomic steps, with the assistance of the solvent, through the Grotthuss mechanism25 involving interfacial water molecules.15 The further investigation of organic molecules such as glycine16 and polyols26 on boehmite has also allowed a rational interpretation of the effect of such surfactants in the presence of aqueous solvent on the relative stability of boehmite surfaces. Understanding the precise hydrated state of the aluminum oxide surface or interface has also helped provide an improved understanding of the surface charge effect on α-alumina.27 It has also been shown that the hydration state impacts the stability of metallic species such as Pd28,29 and Pt30,31 clusters on γ-alumina and the interaction of the metal hydroxide precursor such as Co(OH)232 or CuCl233 on γ-alumina and Ni(OH)2 on α-alumina.34 Considering γ-alumina, there is still a need to obtain atomic scale insights on the water−solid interface in order to better understand chemical mechanisms taking place at the interface during the impregnation of metallic salts or during reactions in aqueous environment and to provide more rational guides. Powerful spectroscopic techniques such as sum-frequency vibrational spectroscopy have been successfully applied to αalumina model surfaces35 in order to gain insights into the local structure of the water−solid interface including water orientation and H-bond networks. For model systems, a direct interpretation can be given with the help of AIMD simulations.19 However, at the present time, this experimental technique cannot be directly applied to more complex γalumina nanopowders where several surfaces are simultaneously exposed on the γ-alumina nanoparticles. Another spectroscopic technique, such as in situ attenuated total reflection infrared (ATR-IR) spectroscopy, may help to analyze chemical processes occurring at the solid−liquid interface1,36−38 and should be well adapted for investigating γ-alumina nano-

powders39 in aqueous environment. However, in order to offer an unambiguous interpretation of the observed vibrational bands and adsorption modes, quantum simulations are often very useful.38,39 Finally, adsorption microcalorimetry measurements have evaluated the surface energies of γ-alumina nanopowders as a function of the water pressure which directly impacts the hydration degree of the exposed surfaces.40,41 An abrupt decay of the surface energies was shown when starting from the anhydrous states to the highest water coverages, where liquid water is present. As aforementioned for the other experimental techniques, a fruitful parallel with density functional theory (DFT) calculations has been made in order to quantify the evolution of the adsorption energies of water on the surfaces exposed by the γ-alumina particles.13,14 Indeed, this rational interpretation has been possible in the regime of the very first monolayer of the chemisorbed water molecule by using DFT models of hydrated γ-alumina surfaces13,14 with well identified hydroxyl groups located on various relevant surface orientations such as (100), (110), and (111) planes. These models have been established by considering water molecules in the vapor phase, which allowed the hydration state to be described as a function of pressure and temperature conditions. Nevertheless, to the best of our knowledge, the conditions where the surrounding water molecules are considered to be in the liquid phase have not been explored for the γ-alumina surfaces so far. In the present work, we will provide a theoretical analysis of the γ-alumina/water interface by means of AIMD calculations. Two predominant surfaces exposed on the γ-alumina nanoparticles will be studied, the (110) and (100) surfaces.3,14 The aim of our contribution is to furnish an atomic scale description of the dynamic behavior of surface OH groups and interfacial water molecules through the detailed analysis of various structural and spectroscopic properties. We will first show the impact of water on the computed vibrational spectrum of surface OH groups. Then, we will analyze the atomic structural features of the interface according to various tools such as concentration profiles, radial distribution functions, twodimensional maps, and average dipole distributions of water molecules. Finally, we discuss the impact of the types of SLI models used for the AIMD simulations.

2. METHODS In this work, we considered the two main surfaces exposed by γ-Al2O3 nanocrystals, namely, the (110) and (100) orientations. The parent bulk γ-Al2O3 structure for the slab models used in this work originates from the earlier model proposed by Krokidis et al.42 and further refined by Digne et al.13,14 The unitary cell contains 40 atoms (16 Al and 24 O). Previous to our AIMD simulations, we optimized this model at 0 K with the VASP code, and the PBE+D2 functional thus including dispersion forces with Grimme corrections43 and a cutoff energy of 400 eV. The optimized cell parameters are a = 8.034, b = 8.36, and c = 5.55, which reveals a minor discrepancy (of 1 Å for (110) and z > 1.5 Å for (100)) correspond to Os and Hs belonging to μ1-OH and chemisorbed H2O molecules, while the inner peaks (z < 1 Å for (110) and z < 1.5 Å for (100)) correspond to μ2-OH or μ3-OH groups. On the (100) surface, the concentrations of Os and Hs in the outermost peaks reach maxima for z values which are both close to 2−2.2 Å. This feature suggests that some Os−Hs bonds are located in the same plane parallel to the (100) surface. On the (110) surface, this feature is not present: the outermost peak of Os is shifted closer to the surface (z ∼ 1.2−1.5 Å), where the outermost peak of Hs is at ∼2 Å. Considering the radial distribution functions (RDFs) related to all Os−Hs pairs (Figure S4), for the (100) surface, a shoulder at ∼1.5 Å is present right after the first coordination sphere of the covalent O−H bonds (located at 1 Å), whereas the second Os−Hs peak is shifted at ∼1.75 Å on the (110) surface. The latter two peaks, corresponding to the second coordination sphere of surface OH groups, are assigned to hydrogen-bonded interactions between the Os species and other surface Hs. Table 2 shows that the overall OsHs coordination number (CN) involving exclusively H-bonds is slightly higher on the (110) surface than on the (100) surface. However, the nature of these

representation of the overall spectra, considering all atoms in the simulation box (as in Figure S3). The stretching frequencies of the different surface OH groups obtained in vacuum from AIMD simulations are in qualitative agreement with those previously reported by Digne et al.13,14 Interestingly, the (110) surface contains OH groups of μ1-OH and μ2-OH types which exhibit higher stretching frequencies than the same groups on the (100) surface. However, due to the higher hydroxyl coverages used in the present study, values reported from static calculations13,14 and those resulting from our AIMD simulations cannot be directly compared (also according to the use of different exchange functionals). Parts a and b of Figure 2 clearly show that interfacial water broadens the spectra significantly, especially in the stretching (3000−4000 cm−1) and bending regions (1500−1700 cm−1). This is particularly true for the (110) surface where, in addition to this broadening effect, the stretching frequencies are more significantly red-shifted. Overall, the “center of mass” of the OH stretching region is shifted toward lower values when water is present and this shift is more pronounced on the (110) surface than on the (100), which suggests that the interaction of water with surface OH groups is stronger on the (110) surface. This trend will be explained further by analyzing the radial distribution function (RDF) which will reveal that the Hbond network of the μ1-OH group is denser on the (100) surface and more reluctant to form new H-bonds with liquid water. By contrast, on the (110), the μ1-OH and H2O groups are more easily solvated by the liquid water molecules, and as a consequence, their stretching bands are more significantly redshifted. The comparison between parts a and b of Figure 2 also reveals that μ2- or μ3-OH groups of the (110) surface either in vacuum or in water present vibrational bands in the bending region. This intriguing feature is the consequence of a hopping proton between two adjacent μ2-OH and μ1-OH groups leading thus to a resonance of the μ2-OH stretching mode with the μ1OH2 bending mode. This feature is not present in the spectra obtained on the (100) surface. Compared to the (100) surface, the (110) presents a high degree of rugosity13,14 which induces stronger H-bonding interactions between μ2-OH and μ1-OH groups when some thermal motion is allowed, as is the case of AIMD simulation. This explanation will also be analyzed by using the RDF analysis presented below. Hence, the nature of the interactions between water and alumina is obviously surface dependent and this could induce 10355


Article

The Journal of Physical Chemistry C

3a). On the (100) surface, this hydrogen bond network involving μ2-OH and μ3-OH groups does not exist. The impact of these distinct hydrogen bond networks on the water solvent structuration at the interface is described in what follows. 3.3. Analysis of the Liquid Water−Surface Interactions. 3.3.1. Atomic Concentration Profiles and Radial Distribution Profiles. On the concentration profiles, we consider water molecules located at z < 4 Å to be involved in the interface. For them, a striking difference between the two surfaces is revealed in Figure 3a and b. On the (110) surface, a first peak of Hw atoms is present at z ∼ 2.5 Å, while the first peak of Ow atoms is shifted further from the surface at z ∼ 3.2 Å. By contrast, on the (100) surface, the first Ow peak is closer to the surface at z ∼ 2.5 Å. This result reveals two key distinct features of the two interfaces which are further analyzed through RDF. First of all, Figure S5a reveals clearly that the RDFs of (Ow, Os) pairs differ from one surface to another. For Os atoms belonging to μ1-OH and H2O groups, Figure S5a shows that the first peak is significantly more pronounced on the (110) surface (similarly as in bulk water) than on the (100). This means that the structuration of Os atoms of μ1-OH and H2O groups and Ow atoms of liquid water is similar to that of (Ow, Ow) pairs in bulk water (Figure S2). This result explains why the OH stretching region of the (110) surface is strongly impacted by the presence of liquid water as previously illustrated by Figure 2b. By contrast, on the (100) surface,

Table 2. Coordination Number (CN) of the Hydrogen Bonded or Second Coordination Sphere of Surface Os Atoms with Hs Atoms Located on Surface OH and Adsorbed H2O Groups (so CN Calculation Excludes the Covalent OH Bond as Given by the First Coordination Sphere at 1 Å) surface groups

(110)

(100)

all groups μ1-OH/H2O with μ1-OH/H2O μ2/μ3-O(H) with μ2/μ3-O(H) μ2/μ3-O(H) with μ1-OH/H2O

0.5 0.3 0.3 0.25

0.25 0.7 0.0 0.0

bonds differs significantly from one surface to another. On the (100) surface, Figure 4a reveals that the shorter hydrogen bonds between Os and Hs atoms originate mainly from μ1-OH and adsorbed H2O groups. Thus, this means that, on this surface, μ1-OH and chemisorbed H2O molecules interact through a denser hydrogen bond network, also corresponding to the narrow region located at z ∼ 2−2.2 Å on the atomic concentration profile. The CN involving these μ1-OH and adsorbed H2O groups is thus higher (0.7 according to Table 2) than on the (110) surface. By contrast, on the (110) surface, the RDF of Os−Hs pairs reveals that H-bonds at ∼1.75 Å involve μ2-OH and μ3-OH groups interacting either among themselves (Figure 4b) or with μ1-OH (Figure 4c). As also shown in Table 2, the CN involving those μ2-OH and μ3-OH groups are higher on the (110) surface where a denser H bond network involves μ2-O(H) and μ3-O(H) groups corresponding to the region of z ∼ 1.5 Å on the concentration profile (Figure

Figure 4. Radial distribution functions between Os and Hs atoms: (a) among μ1-OH groups and H2O; (b) among μ2-OH and μ3-OH groups; (c) Os belonging to μ1-OH groups or H2O and Hs belonging to μ2-OH and μ3-OH groups; (d) Os belonging to μ2-OH or μ3-OH groups and Hs belonging to μ1-OH groups or H2O. 10356


Article


Figure 5. Radial distribution functions between (Ow, Hs) and (Os, Hw) pairs: a) Hs from μ1-OH groups; b) Hs from μ2-OH and μ3-OH groups; c) Os from μ1-OH, surface H2O, μ2-OH and μ3-OH groups; d) Os belonging to μ2-OH and μ3-OH groups.

this structuration is less pronounced: the intensity of the first peak is reduced. Regarding μ2-OH and μ3-OH groups, the (Os, Ow) pairs are far more distant than the (Os, Ow) pairs involving μ1-OH and physisorbed H2O, whatever the surface (Figure S5b). Moreover, the first Os−Ow distance is shorter (∼2.8 Å) on the (100) surface than on the (110) surface (∼3.8 Å). It indicates that Os−Ow proximity involving Os of μ2-OH and μ3-OH is more numerous on the (100) surface than on the (110) surface. This feature is consistent with the previous analysis of the concentration profiles. Considering the RDF of the (Os, Hw) and (Ow, Hs) pairs, the (110) surface exhibits short Ow−Hs distances involving Hs belonging to chemisorbed water or to μ1-OH groups in a greater extent than the (100) surface (Figure 5a). The first peak (Ow−Hs ∼ 1 Å) corresponding to transient protonation of liquid water and the second peak at ∼1.7 Å corresponding to Ow−Hs hydrogen bonding are both more intense on the (110) surface than on the (100) surface. Table 3 also shows that the CN of Ow−Hs involving μ1-OH groups is slightly higher on the (110) than on the (100). As for the atomic concentration profile, this trend is inverted when considering the μ2-OH and μ3-OH surface groups: the Ow peaks at ∼1.75 and ∼2.5 Å are more intense on the (100) than on the (110). This trend may be induced by the different nature of the internal hydrogen bond networks of the two surfaces. On the (100) surface, the H bond network among μ1OH and chemisorbed H2O prevents these groups from becoming H-donors and from interacting with Ow from solvent water molecules. This explains why the vibrational stretching modes of OH groups in Figure 2a are less affected by the

Table 3. Coordination Number of Os−Hw and of Hs−Ow Calculated from the RDF Analyses (CN Calculation Excludes the Covalent OH Bond as Given by the First Coordination Sphere at 1 Å) surface and water atoms

(110)

(100)

Hs of the surface and Ow of liquid water at the interface Hs of μ1-OH/H2O and Ow of liquid water at the interface Hs of μ2-OH and μ3-OH surface groups and Ow of liquid water Os of the surface and Hw of liquid water at the interface Os of μ1-OH and H2O surface groups and Hw of liquid water Os of μ2-O(H) and μ3-O(H) surface groups and Hw of liquid water

0.5 0.5 0.0

0.5 0.4 0.2

0.25 0.6 0

0.25 0.6 0

surrounding bulk water. On the (110) surface, the H bond network involving μ2-OH and μ3-OH groups plays this role. The CN of Ow−Hs involving μ2- and μ3-OH groups is thus slightly higher on the (100) than on the (110). This observation is consistent with the atomic concentration profiles, showing that some Ow atoms (those interacting with μ2-OH and μ3-OH) are found much closer (z ∼ 2.5 Å) to the (100) surface than to the (110) surface (z ∼ 3.2 Å) where they only interact with μ1-OH. For both surfaces, the H-bond networks of surface OH groups are not strongly perturbed by the presence of water. In fact, the H-bond donors of surface OH groups to water are those which do not exhibit a strong H bond with other OH groups. Concerning the (Os, Hw) pairs (Os acceptor), parts c and d of Figure 5 show that they are almost insensitive to the surface. This is also true for the CN of Os−Hw. This means that the H 10357


Article


Figure 6. Surface solvent average interatomic distances from the AIMD trajectories at 300 K for the large slab models. From left to right: Os−Hw distances where Os atoms belong to μ1-OH groups and chemisorbed H2O; Os atoms belong to μ2- and μ3-OH groups; surface molecular model. Top, (110) surface represented by the (2 × 2) model (17.9 OH/nm2); bottom, (100)(2 × 3) surface (17.2 OH/nm2). The color legend for the length scale (in Å) is given on the left. We applied a cutoff of 3.5 Å (all distances beyond this value are considered to be at this cutoff value).

Figure 7. Surface solvent average interatomic distances from the AIMD trajectories at 300 K for the large slab models. From left to right: Ow−Hs distances where Hs atoms belong to μ1-OH groups and chemisorbed H2O; Hs atoms belong to μ2- and μ3-OH groups; surface molecular model. Top, (110) surface represented by the (2 × 2) model (17.9 OH/nm2); bottom, (100)(2 × 3) surface (17.2 OH/nm2). The color legend for the length scale (in Å) is given on the left. We applied a cutoff of 3.5 Å (all distances beyond this value are considered to be at this cutoff value).

10358


Article


the previous RDF analysis and simultaneously furnishes a spatial insight into the sites involved in H-bonding. 3.4. Orientation of the Water Molecules at the Interface. Figure 8 presents the average fraction for the

bonding acceptor character of OH surface groups with respect to water does not depend on the surface but rather on the type of OH group. μ1-OH and chemisorbed H2O groups exhibit the stronger H-bond as revealed by transient protonation of surface OH group (first peak at 1 Å), and Os−Hw bond at a distance of 1.7 Å (second peak in Figure 5c). The CN of the Os atoms due to these H-bonds with liquid water is about 0.6 in both cases (Figure 5c). By contrast, the μ2-OH and μ3-OH groups are not H-bond acceptors (H−O distances greater than 2.2 Å). The reasons for this result are the following ones: the lower accessibility of their Os atoms being located in a plane below the μ1-OH groups (Figure 3), their predominant H-bond donor character on the (100) surface as previously discussed, the close to saturation of their coordination sphere (each oxygen being bonded to more than three atoms), and the reduced availability of the lone pairs. 3.3.2. Two-Dimensional Mapping of the Water−Surface Interactions. To offer a complementary spatial analysis of the H-bond interactions between the surface OH groups and interfacial water, Figures 6 and 7 illustrate the two-dimensional (2D) mapping of the average bond lengths for Hw−Os (when water is a H-bond donor to the surface) and Ow−Hs (when water is a H-bond acceptor from the surface) interactions, respectively. Blue spots correspond to the shortest distances. If we compare the dynamical behavior of the Hw atoms (Figure 6) and of the Ow atoms (Figure 7) interacting with the Os and Hs atoms, respectively, it appears clearly that the interaction domains of Ow are more extended than those of Hw which are far more localized on specific spots. Thus, this implies that Ow atoms interacting with the surface through H bonds remain more mobile than Hw atoms, whose position is strongly determined by the H-bond formed with the surface. If we consider the (Os, Hw) pairs, Figure 6 shows that the interacting pairs are more preferentially localized on the μ1-OH and H2O groups than on the μ2- and μ3-OH groups. As already shown by the RDF analysis, μ2- and μ3-OH groups are very weak H-bond acceptors. On the (110) surface, more numerous blue spots are revealed for μ1-OH and H2O groups which also confirms that this surface exhibits more H-bond acceptor sites than the (100) surface. Regarding the (Hs, Ow) pairs (Figure 7), the domains (brown and blue) are spread over larger areas which illustrates a nonnegligible mobility of Ow atoms as previously said. However, it must be added that the domains identified for the (Hs, Ow) pairs are strongly correlated to those found for the (Hw, Os) pairs. The brown domains correspond to larger Hs−Ow distances where Ow atoms belong to water molecules interacting through one Hw atom with the Os surface reported in Figure 6. This is particularly true for the (100) surface where most of the brown domains are located in a position very close to the domains of short Hw−Os distances involving μ1-OH and H2O groups (Figure 6). The dark blue domains corresponding to the stronger Hbonds involving Hs atoms are more numerous and more extended on the (110) surface than on the (100) surface. They are predominantly located in close vicinity to μ1-OH and H2O groups for the (110) surface. For the (100) surface, only one region (with three close dark blue areas) exhibits short Hs−Ow distances involving μ1-OH and H2O groups. Finally, the (100) surface exhibits some dark blue spots corresponding to short Hs−Ow distances involving μ2- and μ3-OH groups identified in Figure 7, whereas the (110) does not. This trend thus confirms

Figure 8. Fraction of the different orientation of the dipole of the first interfacial water molecules with respect to the normal of the surface (z). Light gray: dipole-z angle 140°, Hdown−Hdown orientation (two Hw directed toward the surface).

different orientations of the dipole moment in interfacial water molecules (only the closest water molecules to the surface are considered). The orientation of the water dipole moment has been computed from the angle between the dipole moment of the water molecules and the surface normal vector. When this angle is close to 0°, water molecules are in a perfect Hup−Hup orientation with Ow atoms pointing toward the surface. When this angle is close to 180°, water molecules are in a Hdown− Hdown orientation with Hw pointing toward the surface. For all surfaces, the dipole is between 80 and 110° which may correspond to molecules either oriented in a Hup−Hdown position or lying parallel to the surface. We subsequently analyze the orientation of each OH group of the water molecule, revealing that, on average, the first OH group is pointing toward the surface, whereas the second is mostly oriented at 130° from the z axis. This clearly shows that the water dipole, although being flat, corresponds to a Hup−Hdown orientation which is typical from water being a H-bond donor and H-bond acceptor with respect to the surface sites, thus suggesting the amphoteric character of the surfaces. According to Figure 8, this parameter does not depend strongly on the type of surfaces. 3.5. Comparison of the Different Models Used: Effect of Hydroxyl Coverages. The main trends of the previous analysis remain valid also for the AIMD simulation undertaken on the smaller supercells and with a longer simulation time (40 ps). The concentration profiles and RDF of the (110) surface (figures in Supporting Information S5 and S6) exhibit very similar features for the two sets of AIMD simulations. 10359


Article


Figure 9. Proton transfer between two surface μ1-OH groups on the small cell of the (110) surface/water interface. Evolution of the four OH distances involved (d1, d2, d3, and d4) during one part of the trajectory. Right: molecular view of the proton exchange between one μ1-H2O group and one μ1-OH group through a bridging liquid water molecule as in a Grotthus mechanism.

Short Os−Hw and Ow−Hs covalent bonds at 1 Å are observed for Os and Hs atoms belonging to the μ1-OH and H2O groups located on the two (110) surface models. To a lesser extent, this is also observed on the larger model of the (100) surface, whereas it is not on the smaller model. This implies that some water molecules may transiently exchange protons with the surface OH groups, particularly on the (110). A more detailed analysis of the AIMD simulations on the smaller model of the (110) surface enables identification of a Grotthuss-like mechanism25 which involves the proton transfer from a surface H2O group to a μ1-OH group, through the transient formation of hydronium ion from a bridging liquid water molecule, as illustrated in Figure 9. It is shown that the lifetime of the hydronium ion is very short, since the elongation (shortening) of d1 (d2) and d3 (d4) distances are simultaneous events. Note that this Grotthus-like mechanism was already observed on the boehmite/water interface.15 As was shown by the previous RDF and 2D-plot analysis, this observation may reveal that liquid water interacts more strongly with the (110) than with the (100) where its internal hydrogen bond network refrains the solvation effect of liquid water.

Nevertheless, one slight discrepancy can be noticed between the two models used for the (100) surface. As explained in the Methods section, the two (100) models exhibit different OH coverages. Regarding the surface network of OH groups, the (2 × 3) supercell of the (100) surface with the higher OH coverage exhibits a double peak of Hs (z ∼ 0.5 and 1 Å) close to the surface (Figure 3b), while a single peak close to the surface (z ∼ 0.5 Å) is present on the (100) surface with lower OH coverage (Figure S6b). These distinct features are induced by the presence of one additional H2O molecule, and the dissociation of one existing H2O into two μ1-OH and μ2-OH groups on the model with higher coverage (Figure 1a and b). As a consequence, on the (100) surface with lower OH coverage (Figure S7c), the first peak in the RDF of OsHs pairs where Os belong to μ1-OH or surface H2O and Hs to μ2-OH and μ3-OH groups is obviously shifted at larger distance (∼2.8 Å). Regarding the water−alumina interface, on the (100) model with lower water coverage, the presence of a Hw peak at z < 2 Å (Figure S6b) reveals that some liquid water molecules may interact through their H atoms with the surface hydroxyls. This is confirmed by the RDF reported in Figure S8d), showing a small bump at ∼1.8 Å corresponding to (Os, Hw) interacting pairs involving μ2-O or μ3-O groups as H-bond acceptors. This small bump was not present in the model with higher water coverages (Figure 5b). Simultaneously, the RDF of (Ow, Hs) pairs (Figure S8b) shows that Ow atoms are interacting more weakly with μ2-OH or μ3-OH groups (Ow−Hs distances at ∼1.9 Å instead of 1.8 Å) for the (100) model with higher coverages (Figure 5b). Nevertheless, the Ow atoms of liquid water are still closer to μ2-OH or μ3-OH groups than on the (110) surface which confirms the general trend previously reported for the larger systems.

4. CONCLUSIONS The (110) and (100) γ-alumina surfaces have been studied by AIMD simulations at room temperature in the presence of liquid water. Four sets of AIMD simulations have been undertaken to check the effect of simulation time, supercell size, and hydroxylation state on the atomistic description of the γ-alumina/water interfaces. It has been shown that the main trends reported here are rather independent of the simulation time (10 ps vs 40 ps) and supercell sizes. Some effects of the hydroxylation coverages have been highlighted on the (100) 10360



■

surface, even though they do not change the main conclusions of this study for the two similar hydroxylation states considered here. We cannot exclude however that a more significant difference in the choice of initial hydroxylation coverage would induce different effects. The comparison of the two surfaces allowed us to extract some critical spectroscopic and structural features for depicting the interaction of the alumina surfaces with liquid water. First, the vibrational analysis of the stretching and bending modes has revealed that the hydroxyl bands of the (110) surface are much more affected (larger red shift) by the presence of water than those of the (100) surface. To explain this observed trend, it has been shown that the hydrogen bond network of the OH groups located on the γalumina surfaces plays a noninnocent role on the behavior of the liquid water molecules close to the interface. On the (100) surface, the μ1-OH and μ1-OH2 groups predominantly interact through a H-bond network exhibiting shorter O−H distances and higher CN than on the (110) surface. This H-bond network reduces the interaction with liquid water molecules. On the (110) surface, the H-bond network among μ1-OH and μ1-OH2 groups weaker than the network among μ2-OH and μ3OH groups. These μ2-/μ3-OH also act as a H-bond donor toward μ1-OH. As a consequence, liquid water interacts differently at the two interfaces. On the (110) surface, water interacts with μ1-OH and H2O groups through H-bonds as donor and acceptor which allows a structuration of water at the interface similarly as in bulk water. Qualitatively, μ1-OH and H2O groups are solvated by liquid water as if they were in bulk water. On the (100) surface, the stronger H-bond network of μ1-OH and H2O groups prevents such a solvation process by liquid water which interacts predominantly with μ2 and μ3-OH groups acting as H-bond donor. On the (110) surface, μ2 and μ3-OH groups interacting through an H-bond network are less prone to interact with liquid water. These are the key distinct features of the two interfaces revealed by the concentration profiles, RDF analyses, and 2D mapping which also allowed identification of the spatial distribution of relevant sites on the surfaces. These key features are at the origin of the distinct spectroscopic trends revealed by our simulations. In general, the water molecules at the interface follow the preferential Hup−Hdown orientation whatever the surface. The infrared spectra revealed that the presence of water induces a more pronounced redshift of the bands located in the OH stretching region for the (110) interface than for the (100). This can also be interpreted as an impact of the different Hbond networks of the two surfaces. Protonation and deprotonation of water molecules occur mainly with μ1-OH and H2O groups. A possible Grotthuss-like mechanism has been observed on the (110) surface model for the long AIMD simulation time (40 ps) and not on the (100) which confirms the stronger interaction of liquid water with the (110) surface. Finally, this atomistic description of the γ-alumina/water interfaces revealed the distinct behaviors of hydroxyl groups located on the two main surfaces constitutive of the γ-alumina nanoplatelets. Hopefully, the results obtained here could help for the interpretation of future experimental in situ spectroscopic analyses. They may also profitably be used to better rationalize morphology effects closely connected to acidic−basic properties of γ-alumina supports in water during the impregnation step of heterogeneous catalysts.

Article

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.7b00101. (S1) Simulation cells used for the two larger models, (S2) radial distribution functions of liquid water, (S3) raw infrared spectra of the SLI before smoothing, (S4) complementary radial distribution functions calculated on the larger supercells, (S5) atomic relative concentrations at SLI for the smaller supercells, and (S6) radial distribution functions for the smaller supercells (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

D. Costa: 0000-0002-3781-9867 P. Raybaud: 0000-0003-4506-5062 Author Contributions

B.F.N.-W. and P.C. contributed equally. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors would like to thank M.-L. Bonnet (Chimie ParisTech), C. Michel (ENS Lyon), R. Réocreux (ENS Lyon), and M. Rivallan (IFPEN) for fruitful scientific exchanges. This work was supported by the French National Research Agency within the framework of the ANR-14-CE08-0019 SLIMCAT project. This work was performed using HPC resources from GENCI-CINES (Grant 2016-c2016087386).

■

REFERENCES

(1) Zaera, F. Probing Liquid/Solid Interfaces at the Molecular Level. Chem. Rev. 2012, 112, 2920−2986. (2) Bourikas, K.; Kordulis, C.; Lycourghiotis, A. The Role of the Liquid-Solid Interface in the Preparation of Supported Catalysts. Catal. Rev.: Sci. Eng. 2006, 48, 363−444. (3) Euzen, P.; Raybaud, P.; Krokidis, X.; Toulhoat, H.; Le Loarer, J. L.; Jolivet, J. P.; Froidefond, C. Alumina. In Handbook of Porous Solids; Wiley-VCH Verlag GmbH: Weinheim, Germany, 2008; pp 1591− 1677. (4) Lefèvre, G.; Duc, M.; Lepeut, P.; Caplain, R.; Fédoroff, M. Hydration of γ-Alumina in Water and Its Effects on Surface Reactivity. Langmuir 2002, 18, 7530−7537. (5) Carrier, X.; Marceau, E. C.; Lambert, J. F.; Che, M. Transformations of γ-alumina in aqueous suspensions 1. Alumina chemical weathering studied as a function of pH. J. Colloid Interface Sci. 2007, 308, 429−437. (6) Hahn, M. W.; Copeland, J. R.; van Pelt, A. H.; Sievers, C. Stability of Amorphous Silica-Alumina in Hot Liquid Water. ChemSusChem 2013, 6, 2304−2315. (7) Larmier, K.; Nicolle, A.; Chizallet, C.; Cadran, N.; Maury, S.; Lamic-Humblot, A. F.; Marceau, E.; Lauron-Pernot, H. Influence of Coadsorbed Water and Alcohol Molecules on Isopropyl Alcohol Dehydration on gamma-Alumina: Multiscale Modeling of Experimental Kinetic Profiles. ACS Catal. 2016, 6, 1905−1920. (8) Abi Aad, J.; Casale, S.; Michau, M.; Courty, P.; Diehl, F.; Marceau, E.; Carrier, X. Chemical weathering of alumina in aqueous

10361


Article


bar02) surfaces: implication on surface charge. Phys. Chem. Chem. Phys. 2011, 13, 6531−6543. (28) Valero, M. C.; Raybaud, P.; Sautet, P. Nucleation of Pd-n (n = 1−5) clusters and wetting of Pd particles on gamma-Al2O3 surfaces: A density functional theory study. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 75, 045427. (29) Corral Valero, M.; Raybaud, P.; Sautet, P. Influence of the hydroxylation of gamma-Al2O3 surfaces on the stability and diffusion of single Pd atoms: A DFT study. J. Phys. Chem. B 2006, 110, 1759− 1767. (30) Hu, C. H.; Chizallet, C.; Mager-Maury, C.; Corral-Valero, M.; Sautet, P.; Toulhoat, H.; Raybaud, P. Modulation of catalyst particle structure upon support hydroxylation: Ab initio insights into Pd13 and Pt13/gamma-Al2O3. J. Catal. 2010, 274, 99−110. (31) Mager-Maury, C.; Chizallet, C.; Sautet, P.; Raybaud, P. Platinum Nanoclusters Stabilized on gamma-Alumina by Chlorine Used As a Capping Surface Ligand: A Density Functional Theory Study. ACS Catal. 2012, 2, 1346−1357. (32) Larmier, K.; Chizallet, C.; Raybaud, P. Tuning the MetalSupport Interaction by Structural Recognition of Cobalt-Based Catalyst Precursors. Angew. Chem., Int. Ed. 2015, 54, 6824−6827. (33) Louwerse, M. J.; Rothenberg, G. Modeling Catalyst Preparation: The Structure of Impregnated-Dried Copper Chloride on gammaAlumina at Low Loadings. ACS Catal. 2013, 3, 1545−1554. (34) Tougerti, A.; Llorens, I.; D’Acapito, F.; Fonda, E.; Hazemann, J. L.; Joly, Y.; Thiaudière, D.; Che, M.; Carrier, X. Surface Science Approach to the Solid−Liquid Interface: Surface-Dependent Precipitation of Ni(OH)2 on α-Al2O3 Surfaces. Angew. Chem., Int. Ed. 2012, 51, 7697−7701. (35) Sung, J. H.; Zhang, L. N.; Tian, C. S.; Shen, Y. R.; Waychunas, G. A. Effect of pH on the Water/alpha-Al2O3 (11̅02) Interface Structure Studied by Sum-Frequency Vibrational Spectroscopy. J. Phys. Chem. C 2011, 115, 13887−13893. (36) Andanson, J.-M.; Baiker, A. Exploring catalytic solid/liquid interfaces by in situ attenuated total reflection infrared spectroscopy. Chem. Soc. Rev. 2010, 39, 4571−4584. (37) Lefèvre, G.; Preočanin, T.; Lützenkirchen, J. Attenuated Total Reflection - Infrared Spectroscopy Applied to the Study of Mineral Aqueous Electrolyte Solution Interfaces: A General Overview and a Case Study. In Infrared Spectroscopy - Materials Science, Engineering and Technology; Theophanides, T., Ed.; InTech: Rijeka, Croatia, 2012; Vol. 1, pp 97−122. (38) Davantès, A.; Costa, D.; Lefèvre, G. Molybdenum(VI) Adsorption onto Lepidocrocite (γ-FeOOH): In Situ Vibrational Spectroscopy and DFT+U Theoretical Study. J. Phys. Chem. C 2016, 120, 11871−11881. (39) Copeland, J. R.; Shi, X. R.; Sholl, D. S.; Sievers, C. Surface Interactions of C2 and C3 Polyols with γ-Al2O3 and the Role of Coadsorbed Water. Langmuir 2013, 29, 581−593. (40) Castro, R. H. R.; Quach, D. V. Analysis of Anhydrous and Hydrated Surface Energies of gamma-Al2O3 by Water Adsorption Microcalorimetry. J. Phys. Chem. C 2012, 116, 24726−24733. (41) Drazin, J. W.; Castro, R. H. R. Water Adsorption Microcalorimetry Model: Deciphering Surface Energies and Water Chemical Potentials of Nanocrystalline Oxides. J. Phys. Chem. C 2014, 118, 10131−10142. (42) Krokidis, X.; Raybaud, P.; Gobichon, A. E.; Rebours, B.; Euzen, P.; Toulhoat, H. Theoretical study of the dehydration process of boehmite to gamma-alumina. J. Phys. Chem. B 2001, 105, 5121−5130. (43) Grimme, S. Semiempirical GGA-type density functional constructed with a long-range dispersion correction. J. Comput. Chem. 2006, 27, 1787. (44) CP2K package, www.cp2k.org/. (45) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 1996, 77, 3865−3868. (46) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized gradient approximation made simple (vol 77, pg 3865, 1996). Phys. Rev. Lett. 1997, 78, 1396−1396.

suspension at ambient pressure: a mechanistic study. ChemCatChem 2017, DOI: 10.1002/cctc.201700145. (9) Che, M. Interfacial coordination chemistry - concepts and relevance to catalysis phenomena. Stud. Surf. Sci. Catal. 1993, 75, 31− 68. (10) Brown, G. E., Jr.; Henrich, V. E.; Casey, W. H.; Clark, D. L.; Eggleston, C.; Felmy, A.; Goodman, D. W.; Grätzel, M.; Maciel, G.; McCarthy, M. I.; et al. Metal oxide surfaces and their interactions with aqueous solutions and microbial organisms. Chem. Rev. 1999, 99, 77− 174. (11) Hass, K. C.; Schneider, W. F.; Curioni, A.; Andreoni, W. Firstprinciples molecular dynamics simulations of H2O on alpha-Al2O3 (0001). J. Phys. Chem. B 2000, 104, 5527−5540. (12) Hass, K. C.; Schneider, W. F.; Curioni, A.; Andreoni, W. The chemistry of water on alumina surfaces: Reaction dynamics from first principles. Science 1998, 282, 265−268. (13) Digne, M.; Sautet, P.; Raybaud, P.; Euzen, P.; Toulhoat, H. Hydroxyl groups on gamma-alumina surfaces: A DFT study. J. Catal. 2002, 211, 1−5. (14) Digne, M.; Sautet, P.; Raybaud, P.; Euzen, P.; Toulhoat, H. Use of DFT to achieve a rational understanding of acid-basic properties of gamma-alumina surfaces. J. Catal. 2004, 226, 54−68. (15) Motta, A.; Gaigeot, M. P.; Costa, D. Ab Initio Molecular Dynamics Study of the AlOOH Boehmite/Water Interface: Role of Steps in Interfacial Grotthus Proton Transfers. J. Phys. Chem. C 2012, 116, 12514−12524. (16) Motta, A.; Gaigeot, M. P.; Costa, D. AIMD Evidence of Inner Sphere Adsorption of Glycine on a Stepped (101) Boehmite AlOOH Surface. J. Phys. Chem. C 2012, 116, 23418−23427. (17) Ribeiro, T.; Motta, A.; Marcus, P.; Gaigeot, M. P.; Lopez, X.; Costa, D. Formation of the OOH° radical at steps of the boehmite surface and its inhibition by gallic acid: a theoretical study including DFT- based dynamics. J. Inorg. Biochem. 2013, 128, 164−173. (18) Raybaud, P.; Digne, M.; Iftimie, R.; Wellens, W.; Euzen, P.; Toulhoat, H. Morphology and Surface Properties of Boehmite (γAlOOH): A Density Functional Theory Study. J. Catal. 2001, 201, 236−246. (19) Gaigeot, M. P.; Sprik, M.; Sulpizi, M. Oxide/water interfaces: how the surface chemistry modifies interfacial water properties. J. Phys.: Condens. Matter 2012, 24, 124106. (20) Kubicki, J. D.; Paul, K. W.; Kabalan, L.; Zhu, Q.; Mrozik, M. K.; Aryanpour, M.; Pierre-Louis, A. M.; Strongin, D. R. ATR-FTIR and Density Functional Theory Study of the Structures, Energetics, and Vibrational Spectra of Phosphate Adsorbed onto Goethite. Langmuir 2012, 28, 14573−14587. (21) Musso, F.; Mignon, P.; Ugliengo, P.; Sodupe, M. Cooperative effects at water-crystalline silica interfaces strengthen surface silanol hydrogen bonding. An ab initio molecular dynamics study. Phys. Chem. Chem. Phys. 2012, 14, 10507−10514. (22) Suter, J. L.; Sprik, M.; Boek, E. S. Free energies of absorption of alkali ions onto beidellite and montmorillonite surfaces from constrained molecular dynamics simulations. Geochim. Cosmochim. Acta 2012, 91, 109−119. (23) Mattioli, G.; Filippone, F.; Caminiti, R.; Bonapasta, A. A. Short hydrogen bonds at the water/TiO2 (anatase) interface. J. Phys. Chem. C 2008, 112, 13579−13586. (24) Sebbari, K.; Roques, J.; Simoni, E.; Domain, C.; Perron, H.; Catalette, H. First-principles molecular dynamics simulations of uranyl ion interaction at the water/rutile TiO2(110) interface. Surf. Sci. 2012, 606, 1135−1141. (25) de Grotthuss, C. J. T. Sur la décomposition de l’eau et des corps qu’elle tient en dissolution à l’aide de l’électricité galvanique. Ann. Chim. 1806, 58, 54−73. (26) Chiche, D.; Chizallet, C.; Durupthy, O.; Channeac, C.; Revel, R.; Raybaud, P.; Jolivet, J. P. Growth of boehmite particles in the presence of xylitol: morphology oriented by the nest effect of hydrogen bonding. Phys. Chem. Chem. Phys. 2009, 11, 11310−11323. (27) Tougerti, A.; Methivier, C.; Cristol, S.; Tielens, F.; Che, M.; Carrier, X. Structure of clean and hydrated alpha-Al2O3 (1(1)over10362


Article

The Journal of Physical Chemistry C (47) Goedecker, S.; Teter, M.; Hutter, J. Separable dual-space Gaussian pseudopotentials. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 1703−1710. (48) Grimme, S. Semiempirical GGA-type density functional constructed with a long-range dispersion correction. J. Comput. Chem. 2006, 27, 1787−1799. (49) Humphrey, W.; Dalke, A.; Schulten, K. VMD - Visual Molecular Dynamics. J. Mol. Graphics 1996, 14, 33−38. (50) VMD website: http://www.ks.uiuc.edu/Research/vmd/. (51) Max, J. J.; Chapados, C. Isotope effects in liquid water by infrared spectroscopy. III. H2O and D2O spectra from 6000 to 0 cm−1. J. Chem. Phys. 2009, 131, 184505.

10363


Water Interface Revealed by

Recommend Documents