Payam
Kaghazchi
a,
Felice C.
Simeone
b,
Khaled A.
Soliman
b,
Ludwig A.
Kibler
b and
Timo
Jacob
*ab
aFritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, D-14195, Berlin, Germany. E-mail: jacob@fhi-berlin.mpg.de; Fax: +49-(0)30-8413-4701; Tel: +49-(0)30-8413-4816
bInstitut für Elektrochemie, Universität Ulm, Ulm, D-89081, Germany
First published on 7th August 2008
Using density functional theory calculations and the extended ab initio atomistic thermodynamics approach, we studied the adsorption of oxygen on the different surface faces, which are involved in the faceting of Ir(210). Constructing the (p,T)-surface phase diagrams of the corresponding surfaces in contact with an oxygen atmosphere, we find that at high temperatures the planar surfaces are stable, while lowering the temperature stabilizes those nano-facets found experimentally. Afterwards, we constructed the (a,T,Δϕ)-phase diagram for Ir(210) in contact with an aqueous electrolyte and found that the same nano-facets should be stable under electrochemical conditions. Motivated by this prediction from theory, experiments were performed using cyclic voltammetry and in-situ scanning tunneling microscopy. The presence of nanofacets for Ir(210) gives rise to a characteristic current-peak in the hydrogen adsorption region for sulfuric acid solution. Furthermore, first results on the electrocatalytic behavior of nano-faceted Ir(210) are presented.
One way out of this dilemma is the formation of well-defined nanostructures or facets on single-crystal surfaces, which provide a reproducible basis and model systems for studying structural sensitivity in (electro-)catalytic reactions. Surface faceting can be understood as a morphology change from a flat bulk-truncated surface to a hill-and-valley structure. While clean surfaces rarely facet, adsorbate-induced faceting of surfaces, driven by the anisotropy of surface free energy, is a general phenomenon observed in many systems.1–3 Usually the facets have more close-packed surface structures than the original surface, resulting in a minimized surface free energy although the total surface area may be increased. Therefore, in order to actively select and control a desired surface morphology, it is necessary to deepen our understanding of adsorbate-induced faceting. Furthermore, this would provide model systems to study structural sensitivity in catalytic reactions4–6 and may be used as templates to grow nanostructures.7,8
So far experimental studies of adsorbate-induced faceting of metal surfaces focused mainly on body-centered cubic or face-centered cubic metals, such as W(111),1,2Mo(111),9,10Ni(210),11,12Pt(210),13Ir(210),14Rh(553),15 and vicinal Cu surfaces.16–19 Although the enhancement of the anisotropy in surface free energy is the thermodynamic driving force for facet formation, in most cases this process is hindered by kinetic limitations. Therefore, not only is a critical adsorbate coverage required but also a minimum annealing temperature, allowing the system to overcome all kinetic barriers in the process of facet formation.
Recently the group of Madey found that on particular rough surfaces certain adsorbates are able to induce the formation of well-defined nanostructures after annealing the system to elevated temperatures.14,20 Using scanning tunnelling microscopy (STM) and low-energy electron diffraction (LEED) under ultra-high vacuum (UHV) they could demonstrate that an initially planar Ir(210) surface becomes faceted when being covered with more than 0.5 ML oxygen and annealed to temperatures above 600 K. The facets that form were characterized as an array of three-sided pyramidal nanostructures having Ir(311), Ir(31−1) and Ir(110) faces. Furthermore, higher resolution STM images showed that while the (311) and (31−1) faces are always unreconstructed, some (110) faces are partially reconstructed. This superstructure was proposed to be a “stepped double-missing-row”-(110) surface.21
After facet formation, oxygen that still remains on the surface can be removed by reaction with H2 at T < 400 K. During this reaction the nanopyramidal surface structure is not affected, since the kinetic barrier of facet destruction is not reached at these low temperatures. The clean nanofacets remain stable up to ∼600 K, and for higher temperatures the initial planar Ir(210) surface becomes stable again.
Similar behavior could also be observed in the case of Re(11−21), where by adsorption of oxygen pyramid-like facets having each two (01−11) and (10−11) faces could be generated. However, changing the adsorbate to ammonia and annealing to 900 K, led to the formation of two-sided ridges with (13−42) and (31−42) faces.22
It has also been demonstrated that planar and nano-faceted Ir(210) surfaces can be prepared outside an UHV chamber by inductive heating in a N2 + H2 mixture and a nitrogen atmosphere, respectively.23 Cooling the sample in a reducing gas atmosphere yields a planar surface according to the preparation of unreconstructed low-index planes of iridium.24–27 The presence of trace amounts of oxygen in nitrogen gas was found to be crucial for facet formation on Ir(210).23 Nano-faceted Ir(210) in contact with aqueous sulfuric acid is easily characterised by a sharp voltammetric current peak around −0.2 V vs.SCE.23 The similarity of the nano-pyramids obtained outside an UHV chamber with those reported by Madey et al.14,21 has been verified by in-situscanning tunnelling microscopy (STM).23
By the combination of theory and experiments we will demonstrate that choosing appropriate adsorbate and potential conditions it should be possible to electrochemically generate a reproducible and well-defined basis for studying catalytic reactions on unsupported monometallic nanostructures with controllable size and shape.
In the following, we will first describe the theoretical and experimental methods that were used to investigate the faceting of Ir(210). Afterwards, a brief description on calculations for oxygen-induced facet-formation on Ir(210) is given, which provides the basis for generating the electrochemical phase diagram. The theoretical prediction that facet formation should also be possible electrochemically was then studied experimentally. In this context, cyclic voltammetry and in-situSTM studies on the electrochemical behaviour of Ir(210) in contact with perchloric acid solution are described, followed by electrocatalytic investigations for different simple reactions.
ΔGform = ΔGsurface + ΔGedge + ΔGkink + ΔGstrain + … | (1) |
As long as the facets are large enough, such that contributions from step-edges, kinks, and strain are negligible compared to surface contributions, the overall formation energy can be approximated by the surface contribution only. This condition, usually referred as Herring-condition, is comparable to the so-called Wulff-construction. On the basis of this condition, facet formation should occur when
![]() | (2) |
![]() | (3) |
Here S311 and S110 specify the partial contributions of the different faces to each pyramidal-shaped facet, while θ311 and θ110 are the tilt angles of the faces with respect to the initial substrate, T is the temperature, a the water activity, and Δϕ the electrode potential. Experimentally and geometrically obtained values for Sf and θf are summarized in Table 1. It should be noted that since Ir(311) and Ir(31−1) show the same surface morphology, both have been combined.
The interfacial free energies γ, which are relevant for eqn (3), give the stability of the corresponding electrode/electrolyte-interfaces. As described in ref. 28 and 29, an exact evaluation of the interfacial free energies is in principle possible, but requires a self-consistent modeling of the entire interfacial region, which might range up to several 100 Å. Since this is currently beyond capabilities of ab initio approaches, we reduce our model to the electrode and the adlayer only and assume a constant influence of the electrolyte, allowing us to neglect its presence when studying relative stabilities only. Consequently, the interfacial free energy reduces to
![]() | (4) |
![]() | (5) |
With eqn (4) we now can evaluate (approximate) the interfacial free energies of the different surface faces of the facets, since all relevant quantities can be deduced from first principles, here density functional theory calculations. These can then be used together with eqn (3) to finally obtain the electrochemical phase diagram.
In order to calculate the total energies of different surface structures, which are required for eqn (4), we performed DFT slab calculations using the CASTEP code31 with Vanderbilt-type ultrasoft pseudopotentials32 and the generalized gradient approximation (GGA) exchange–correlation functional proposed by Perdew, Burke and Ernzerhof (PBE).33 Layer-converged supercells consisting of 16-layer slabs for Ir(210), 11-layer slabs for Ir(311), 12-layer slabs for Ir(110), and 7-layer slabs for Ir(110)-superstructure were used to model oxygen adsorption with different coverages and adlayer structures. To decouple the interactions between neighboring slabs in the supercell geometry, repeated slabs were separated by a ∼12 Å vacuum. For Ir(210), Ir(311) and Ir(110)-superstructure, the bottom three layers, and for Ir(110) the bottom four layers, were fixed at the calculated bulk structure, while the geometry of the remaining layers plus adsorbates were fully optimized (to <0.03 eV Å−1). The Brillouin zones of the (1 × 1)-surface unitcells of Ir(210), Ir(311), Ir(110), and the superstructure were sampled with 10 × 8, 14 × 8, 14 × 10, and 4 × 4 Monkhorst-Pack k-point meshes, respectively. Finally, a plane-wave basis set with an energy cutoff 340 eV was used.
Investigating the error sources related to slab thickness, vacuum size, plane-wave cutoff and k-point mesh, we found the maximum overall error bar in the surface free energy to be <5 meV Å−2, when using optimized values for each parameter.
Throughout this section oxygen binding energies are with respect to half an gas-phase oxygen molecule.
![]() | ||
Fig. 1 Hard-sphere models of Ir(210), Ir(311), Ir(110), and Ir(110)-superstructure, which are the surfaces relevant for faceting of Ir(210). |
Besides the external parameters temperature and pressure, it is well-known that under electrochemical conditions even the electrode potential is able to cause surface oxidation. Therefore, in the following we will combine both concepts, the oxygen-induced surface faceting and the potential-induced electrooxidation. We will show that under electrochemical conditions, potential-induced surface faceting should be possible.
In order to generate the electrochemical phase diagram shown in Fig. 2, we made use of the assumption that oxygen binding energies as well as the electrolyte structure and properties should be potential-independent, which in turn allowed us to directly use the DFT-energies that were obtained when investigating the gas-phase system.34 This is certainly a strong assumption and different theoretical studies have been performed on the role of surrounding water on binding energies and reaction barriers.35–40 But in previous studies on Pt-oxide formation we were able to reproduce the experimental CV-curve on the basis of this approximation. However, so far it is not clear how the surrounding water might influence the nanostructured surface that shows a variety of lower coordinated sites (e.g. step-edges or kinks).
![]() | ||
Fig. 2 (a,T,Δϕ)-phase diagram for the electrochemical faceting of Ir(210) in an aqueous electrolyte. The left figure shows the interfacial free energy γ as function of the water chemical potential and electrode potential (referenced to RHE), while the right figure shows the view to the bottom. In addition, the temperature scale, which corresponds to a = 1, is given on the right side of the phase diagram. The structure-labeling corresponds to the models shown in Fig. 3. |
![]() | ||
Fig. 3 Models of the different surface structures that are present on Ir(210) at specific electrode potentials (see Table 2). For all surfaces 1 ML is defined as one oxygen atom per surface unit cell. |
Phase | Potential range/V |
---|---|
a | Δϕ < 0.10 |
b | 0.10 < Δϕ < 0.18 |
c | 0.18 < Δϕ < 0.26 |
d | 0.26 < Δϕ < 0.57 |
e | 0.57 < Δϕ < 0.65 |
f | 0.65 < Δϕ < 0.85 |
oxide | 0.85 < Δϕ |
On the basis of this approach, we distinguish between clean and oxygen-covered surfaces of:
• planar Ir(210),
• nanopyramids with (311), (31−1) and (110)-regular faces,
• and nanopyramids with (311), (31−1) and (110)-superstructure faces.
The surface free energies of the faceted surfaces were calculated using the left side of eqn (3) with the parameters for the partial surface areas and facet tilt angles summarized in Table 1. Each summand of this equation was evaluated by eqn (4), where the main temperature- and pressure-dependence is assumed to be dominated by the water chemical potential eqn (5). As already mentioned above, this approach is based on the Herring-condition, in which contributions from step-edges, kinks and surface stress or strain are considered to be small.
Although experimentally the coexistence of (110)-regular and (110)-superstructure was observed on most nanopyramids, we only consider the extremes in which the entire (110)-faces of all pyramids have one of both structures at the same time. However, in the following we will also discuss the consequences of having a mixture (coexistence).
Since all binding energies were calculated for oxygen adsorption on pure Ir, the following discussion is restricted to the part of the phase diagram where the IrO2 bulk-oxide is not stable yet. Equivalent to eqn (4), this can be translated to the following stability condition, which restricts the electrode potential range of interest
+1.29 eV − 2eΔϕ < μH2O < 0 eV. | (6) |
Fig. 2 shows the final phase diagram, where on the left side γ is plotted against the water chemical potential and the electrode potential referenced to half an oxygen molecule
![]() | (7) |
Above 0.1 V adsorption of oxygen takes place, causing the formation or stabilization of the nanofaceted surface. While at lower electrode potentials (0.1 V > Δϕ > 0.26 V) nanopyramids of the type (311)/(31−1)/(110)-superstructure are most stable, at more positive potentials (0.26 V > Δϕ > 0.85 V) the (110)-superstructure face is replaced by regular (110)-(1 × 1). In this range, increasing the potential does not change the structure, but causes the coverage of oxygen on the facets to increase. Finally the IrO2 bulk-oxide appears as stable phase for electrode potentials above 0.85 V. Overall, we find the surfaces phases summarized in Table 2. Interestingly, the overall phase transition behavior is qualitatively comparable to the temperature decrease in the gas-phase UHV system.
Finally, regarding the phase diagram, two aspects should be mentioned. While clean planar Ir(210) is most probably the stable phase at low electrode potentials (phase a), oxygen coverages of around or above 0.5 ML are required in order to stabilize the nanofacets. If one would only concentrate on coverages below this value, oxygen-covered planar Ir(210) would become stable at potentials above 0.13 V. This coverage-dependence is in agreement with experimental observations.14,21
Moreover, the presence of the (110)-superstructure face at potentials of 0.1 V > Δϕ > 0.26 V is rather remarkable. On a planar Ir(110) surface this superstructure is always less favorable than regular (110), but this is different for the faceted Ir(210) surface. There, the superstructure forms on the (110)-side of the nanopyramids at lower potential, which is a consequence of the nonlinear dependency of the surface free energy on the tilt angle [see prefactors on the left side of eqn (3)] and the fact that the (110)-faces of the nanopyramids are already tilted with respect to the (210)-substrate. Again, this behavior is also confirmed experimentally.21
In order to evaluate the influences coming from choosing the PBE exchange–correlation functional, we recalculated the most relevant surface structures with the LDA functional and generated the equivalent surface phase diagram. Comparison shows that with the LDA functional all phase transitions are shifted toward lower potentials, without causing any changes in the ordering of the stable phases. Furthermore, the stability ranges, respectively chemical potential ranges, of the different phases are almost the same with both xc-functionals. Therefore, it is reasonable to assume that the conclusions drawn above are qualitatively independent of the xc-functional.
![]() | ||
Fig. 4 Cyclic voltammograms for planar (dashed line) and nano-faceted (solid line) Ir(210) in 0.1 M HClO4. Scan rate: 50 mV s−1. |
It is well-known that the perchlorate anion can easily be reduced by iridium to chloride. Therefore, stable curves representing stationary behaviour show slightly smaller peaks as the curves in Fig. 4 for the 1st cycles. A variety of peaks can be discerned in the hydrogen adsorption region, i.e., at potentials between −0.3 and +0.1 V. These might be attributed to well-ordered domains of unreconstructed Ir(210) in the case of the planar surface or to the (110) and (311) faces for the faceted surface. However, since details on the electrochemical behaviour of Ir(110) and Ir(311) are not available yet, a direct assignment of the peaks in Fig. 4 is not possible. When going towards the faceted surface, three peaks are emerging between −0.2 and 0 V, while the peak at −0.25 V is slightly decreasing. Even more evidently, a relatively sharp reversible peak at 0.22 V is related to the presence of nano-facets on the surface (Fig. 4). This peak is probably related to the initial stages of OH or O adsorption from water and absent in the case of planar Ir(210). Restricting the positive potential limit to 0.4 V did not give rise to significant changes in the voltammograms. Thus, we conclude that simple potential variation does not lead to an electrochemical faceting of planar Ir(210) at room temperature.
The STM image shown in Fig. 5 was recorded for nano-faceted Ir(210) in 0.1 M HClO4 after measuring the respective voltammogram in the electrochemical cell (see Fig. 4). The electrode was contacted with the solution in the STM cell at 0.4 V. The surface consists of pyramids with facets of different orientation corresponding to slightly different size. It can be seen in Fig. 5 that the pyramids are uniformly distributed. Moreover, there are no planar zones alternating with the pyramids. The vertical distance between two consecutive pyramidal-extremes (apex/valley) never exceeds 3 nm and the average width at the base of the pyramids is 30 nm. It should be mentioned that the size of the pyramids changes slightly for various measurements, pointing out that subtle differences in the preparation procedure (e.g., different annealing times and/or temperatures) may influence the faceting process. By carefully watching the disposition of the pyramids, some order can be seen. The pyramids seem to share one edge along which they align. This could be a consequence of the faceting mechanism.
![]() | ||
Fig. 5 STM-image for nano-faceted Ir(210) in 0.1 M HClO4 at 0.4 V. |
The STM measurements give also the opportunity to estimate the tilt angle that the different facets form with the horizontal plane of the STM images. The (311) facet forms a tilt angle of 18 ± 2°, a value already obtained with different techniques by Madey and co-workers.14 For the other facet, the angle is 9 ± 2° and indicates the presence of a superstructure formed by the Ir atoms of the top most plane of a reconstructed (110) facet.21 The model proposed by Madey and co-workers,21 which identifies the geometry of the pyramids as made by two (311) and one (110), describes correctly the experimental results reported here.
Since the adsorption of oxygen was seen to provoke facet formation, an attempt was made to apply more positive potential values, where surface oxidation is expected to take place. Such experiments were also performed with 0.1 M H2SO4 to rule out any disturbance by perchlorate reduction. In addition, the differences in the voltammograms for planar and faceted Ir(210) are much more pronounced for sulfuric acid compared with perchloric acid solution. Although slight changes in the voltammetric profile could be discerned after potential excursions to 0.8 V, a straightforward electrochemical method for nanofacet formation on Ir(210) as function of electrode potential has still to be explored.
![]() | ||
Fig. 6 Current–potential curves for CO adlayer oxidation on planar (dashed line) and on nano-faceted (solid line) Ir(210) in 0.1 M H2SO4. Scan rate: 10 mV s−1. |
It should be mentioned that higher overpotentials have also been reported for CO monolayer oxidation on Pt nanoparticles supported on glassy carbon electrodes.41 Spatially confined formation of oxygen containing species at active sites and slow diffusion of CO molecules to the active sites were given as main reasons for slower kinetics compared to extended Pt surfaces.41 While size effects have been addressed for the latter Pt systems,42 it still remains to be a challenge to significantly vary the size of the nano-facets on Ir(210). However, the faceted Ir(210) surface can ideally serve as a model system to study structure sensitivity of electrocatalytic reactions.
This journal is © The Royal Society of Chemistry 2009 |