We investigated adsorption of N2 on stoichiometric and O-rich IrO2(110) surfaces using temperature programmed desorption (TPD) experiments and density functional theory (DFT) calculations. TPD shows that N2 desorbs predominantly from the stoichiometric-IrO2(110) surface in a well-defined peak at 270 K for N2 coverages below about 0.5 ML and that a shoulder centered near 235 K develops in the N2 TPD traces as the coverage approaches saturation, indicating that adsorbed N2 molecules destabilize at high N2 coverages. Experiments of N2 adsorption onto O-rich IrO2(110) surfaces provide evidence that N2 adsorbs exclusively on the coordinatively unsaturated Ir atoms (Ircus) of the surface and that pre-adsorbed O-atoms (“on-top” oxygen) stabilize adsorbed N2 molecules, causing the main N2 TPD peak to shift toward higher temperature with increasing oxygen coverages. Consistent with prior results, our DFT calculations predict that an N2 molecule preferentially adsorbs into an upright configuration on an Ircus atom of the IrO2(110) surface and achieves a binding energy of about 100 kJ/mol. The computed binding energy agrees well with our experimental estimate of ∼90 kJ/mol for low N2 coverages on stoichiometric IrO2(110). The DFT calculations also quantitatively reproduce the observed stabilization of N2 by co-adsorption on-top O-atoms and predict the destabilization of N2 on IrO2(110) as the N2 adlayer becomes crowded at high coverages.

The stochiometric-IrO2(110) surface is highly active in promoting the dissociation of light alkanes and H2 at temperatures as low as 100 K.1–4 Prior studies demonstrate that alkanes and H2 form strongly bound σ-complexes on IrO2(110) by datively bonding with Ircus atoms present at the surface and that the adsorbed σ-complex state serves as the precursor for initial bond dissociation.1–9 Researchers have shown that the formation and facile C–H bond activation of strongly bound σ-complexes is a common feature of alkane activation on late transition-metal oxide surfaces that expose pairs of coordinatively unsaturated (cus) metal and oxygen atoms, including the PdO(101), RuO2(110), and IrO2(110) surfaces.10,11 In general, the formation of strongly bound alkane or H2 σ-complexes promotes the initial dissociation of these compounds because dative interactions strengthen the binding in the initial state and also weaken the metal-coordinated C–H and H–H bonds. The higher dehydrogenation activity of IrO2(110) compared with PdO(101) and RuO2(110) appears to arise partly from the ability of this surface to bind alkanes and H2 more strongly than other surfaces.

Prior computational studies provide insights for understanding the origins of strong adsorbate bonding on IrO2(110). Wang and co-workers originally reported that bonding on rutile (110) surfaces involves σ-donation and π* back-donation interactions between the cus-metal atoms and adsorbed molecules, including NHx species, N2, and CH4, and that these bonding interactions are enhanced on IrO2(110) because the Ircus atoms feature higher densities of empty and filled d-states near the Fermi level compared with RuO2(110) and TiO2(110).7,12,13 For example, Wang et al. reported that N2 binds in an upright geometry and achieves a binding energy that is about two times higher on IrO2(110) compared with RuO2(110) (106 kJ/mol vs 51 kJ/mol).12 Those authors showed that the binding of N2 with a cus-metal atom is analogous to that of CO and stronger on IrO2(110) because the Ircus atoms have higher densities of empty dz2 states as well as filled dxy and dxz states near the Fermi level that enhance the σ-donation and π* back-donation interactions, respectively, with the N2 molecule. More recent studies support these general ideas, showing that σ-donation makes the dominant contribution in the bonding of CH4 on rutile (110) surfaces and that, in general, the bonding strength increases as the empty dz2 band becomes more narrow and positioned close to the Fermi level.14,15

Accurate experimental measurements of adsorbate binding energies are important for confirming computational predictions of strong adsorbate bonding on IrO2(110). Our previous experimental studies show that alkane and H2 σ-complexes react with nearly unit probability on the clean IrO2(110) surface at low temperatures and that molecular desorption only becomes appreciable at high coverages when the binding in the σ-complex state is weakened by neighboring co-adsorbates, especially bridging OH groups, and thus is not representative of the strong binding predicted for the clean IrO2(110) surface.1–4,8,9,16 A consequence is that temperature programmed reaction spectroscopy (TPRS) measurements have been unable to provide a direct measure of the binding strengths of alkane and H2 σ-complexes on clean IrO2(110). In contrast, N2 should be well-suited as a probe molecule to test the predictions of strong bonding on IrO2(110),12 since it is reasonable to expect that molecularly chemisorbed N2 will desorb rather than dissociating and thus produce TPRS peaks at low N2 coverages from which N2 binding energies can be computed.

In this study, we investigate the adsorption of N2 on IrO2(110) using TPD and obtain good quantitative agreement between experimental and computational estimates of the N2 binding energy at low coverages. We further present experimental evidence that the N2 molecules adsorb exclusively on Ircus sites during adsorption at 90 K and that co-adsorbed O-atoms stabilize N2 molecules on IrO2(110), whereas neighboring N2 molecules destabilize one another at high N2 coverage.

The experiments were performed in an ultrahigh vacuum (UHV) chamber with a base pressure of 2.0 × 10−10 Torr, the details of which have been documented in Refs. 17–20. Briefly, the chamber is equipped with an inductively coupled RF plasma source (Oxford Scientific Instruments) for generating atomic oxygen beams, an electron-beam metal evaporator (McAllister Technical Services) for the deposition of metallic Ir, an ion source (SPECS), a low energy electron diffraction (LEED) optics, and a shielded quadrupole mass spectrometer (Hiden HAL 201) for temperature programmed desorption (TPD) and TPRS. The chamber is also equipped with a dual Mg/Al anode x-ray source and a hemispherical analyzer (SPECS) for x-ray photoelectron spectroscopy (XPS) and low energy ion scattering spectroscopy (LEISS) measurements.

The Ru(0001) single crystal sample (9 mm × 1 mm) is attached to two 0.016 in. tungsten wires mounted onto an LN2 cooled sample holder. A type K thermocouple was spot welded onto the back side of the sample for temperature measurements. The sample was cleaned by alternating rounds of Ar+ sputtering (2 keV) and annealing in an O2 background of 5 × 10−7 Torr, both at 1000 K, and finally annealing in UHV at 1500 K. This procedure has been shown to produce a high level of surface crystallinity and cleanliness as ascertained by LEED and CO TPRS.21–23 The CO TPRS spectra collected after exposing the clean Ru(0001) surface to O2 showed no trace of recombinative CO desorption, indicating that a negligible quantity of carbon was present on the surface prior to oxidation.

A stoichiometrically terminated IrO2(110) film [s-IrO2(110)] was prepared for the present studies using a method that closely follows steps reported for the “template-assisted growth” of IrO2(110) onto a s-RuO2(110) surface. Recently, Abb et al.24 have shown that initially generating small IrO2(110) domains (“seeds”) on RuO2(110) via post-oxidation of metallic Ir clusters is effective in promoting the growth of large and flat IrO2(110) domains during the subsequent deposition of Ir in an O2 background, with the IrO2(110) domains developing into a conformal layer at Ir coverages above ∼2 ML. Our procedure is similar to that reported previously,24 but we use atomic oxygen as the oxidant during several steps, for convenience.

In the first step, a stoichiometrically terminated RuO2(110) surface is generated by exposing clean Ru(0001) to an O-atom beam at 750 K. Similar to previous results in our laboratory,21–23,25 a total exposure of ∼76 MLRu(0001) of O atoms is sufficient to grow an s-RuO2(110) film of 4.2 nm thickness that contains ∼17 MLRu(0001) of oxygen atoms, where 1 MLRu(0001) is defined as the surface density of Ru(0001), i.e., 1.57 × 1015 atoms/cm2. In the next step, ∼1 MLRu(0001) of metallic Ir (Goodfellow, 99.9%) was deposited onto the s-RuO2(110) surface at 300 K, followed by post-oxidation at 700 K with ∼13 MLRu(0001) of O-atoms to produce initial IrO2(110) domains at the surface. An additional ∼1 MLRu(0001) of metallic Ir was subsequently deposited in an O2 background of 5 × 10−7 Torr at 700 K to ensure the growth of a closed IrO2(110) overlayer. In the final preparation step, the s-IrO2(110) film was post-oxidized with an additional ∼13 MLRu(0001) O-atom exposure at 700 K and then exposed to 20 L O2 at 600 K, in an effort to further clean the surface and fill bridging oxygen (Obr) vacancies. We present surface characterization results in the Results and Discussion section and the supplementary material, which show that our preparation method produces a conformal s-IrO2(110) film on RuO2(110) of about 1.0 nm thickness that is well-suited for UHV studies of surface reactivity.

Figure 1 shows a ball and stick model of IrO2(110) partially covered with Oot atoms. The IrO2(110) surface unit cell is rectangular, with bulk-terminated dimensions of a = 3.16 Å and b = 6.36 Å, and the surface consists of alternating rows of Ircus and Obr atoms along the [001] direction. The lattice constants of rutile RuO2(110) are a = 3.11 Å and b = 6.38 Å, suggesting that an epitaxial IrO2(110) layer on RuO2(110) experiences a strain of +1.6% and −0.3% in the a and b directions, respectively. The Ircus and Obr atoms each has a single dangling bond due to the decrease in bond coordination relative to bulk IrO2; the Ircus atoms have five fold coordination, whereas bulk Ir atoms have sixfold coordination, and Obr atoms have two fold coordination, whereas bulk O-atoms have three fold coordination. On the basis of the IrO2(110) unit cell, the areal density of Ircus atoms and Obr atoms is equal to 37% of the Ir(100) surface atom density of 1.36 × 1015 cm2. Since Ircus atoms are active adsorption sites, we define 1 ML as equal to the density of Ircus atoms on the IrO2(110) surface. On-top oxygen atoms (Oot) bond directly on Ircus atoms and also expose a dangling bond perpendicular to the surface (Fig. 1). TPD measurements show that Oot atoms on IrO2(110) are less stable than Obr atoms with Oot atoms desorbing between 400 K and 650 K and Obr atoms desorbing in a TPD peak near 950 K (Fig. S2 of the supplementary material).1,8,9

FIG. 1.

Model representation of top and side views of the IrO2(110) structure with an Oot atom. The Ircus, Ir6f, Obr, Oot, and O3f atoms are indicated, where Ir6f and O3f correspond to sixfold and threefold coordination, respectively. Red and blue atoms represent O and Ir atoms, respectively, and the Oot atom is shown in yellow.

FIG. 1.

Model representation of top and side views of the IrO2(110) structure with an Oot atom. The Ircus, Ir6f, Obr, Oot, and O3f atoms are indicated, where Ir6f and O3f correspond to sixfold and threefold coordination, respectively. Red and blue atoms represent O and Ir atoms, respectively, and the Oot atom is shown in yellow.

Close modal

We investigated the adsorption of N2 (Airgas, 99.999%) on the s-IrO2(110) surface at 90 K using TPD. All of the N2 adsorption experiments were performed using the same ∼1 nm IrO2(110) film. After N2 exposure, the sample was positioned in front of a shielded mass spectrometer at a distance of ∼5 mm and heated at a constant rate of 1 K/s until the sample temperature reached 600 K. We note that a small amount of H2 (<∼ 0.1 ML) from the vacuum background adsorbs on the IrO2(110) surface during cooling to 90 K prior to N2 adsorption and desorbs as H2O during the TPD experiments. After each TPD experiment, the sample was exposed to 3 L of O2 at 300 K and subsequently heated to 600 K to remove residual hydrogen and restore oxygen vacancies that may have been created during the TPD experiment. Reproducibility in our TPD results provides evidence that IrO2(110) films with nominally the same surface structure and composition can be repeatedly generated. Adsorbate coverages and desorption yields were quantified from the TPRS data using established procedures.1 XPS measurements were performed using Mg Kα x-rays (hν = 1253.6 eV) and a hemispherical analyzer operating in retarding mode at a pass energy of 27 eV. The data are presented after averaging 20 scans and translating the spectra so that the zero point is set to the minimum counts observed from the low binding energy side of the spectra. The thickness of the IrO2 film is estimated by assuming that the Ru 3d5/2 peak intensity decreases exponentially with the IrO2(110) film thickness. Using the TPP-2M equation, we compute an inelastic mean free path (IMFP) of 13.5 Å for 973 eV photoelectrons passing through an IrO2 film26 and estimate an average thickness of 10.2 Å for the IrO2 film that was investigated. This thickness is equivalent to 3.2 layers of IrO2(110), where a “layer” is defined as the separation between Ir-containing planes along the [110] direction of rutile IrO2.

We also investigated N2 adsorption on O-rich IrO2(110) surfaces with different, initial on-top oxygen coverages. The O-rich surfaces were prepared by exposing the s-IrO2(110) film to varying amounts of O2 at 90 K and subsequently exposing each surface to an amount of N2 (0.5 L) that is sufficient to saturate the s-IrO2(110) surface with molecularly adsorbed N2 at 90 K. In Fig. S2 of the supplementary material, we report O2 TPD spectra obtained as a function of the O2 exposure at 90 K to the IrO2(110) film on RuO2(110), and note that the spectra agree well with our previous reports of the desorption of on-top oxygen from an IrO2(110) layer grown by oxidizing Ir(100) at elevated O2 pressure and temperature.4,8,9 Briefly, on-top oxygen species desorb from the IrO2(110) film in TPD features between ∼100–225 K and 350–600 K (Fig. S2a of the supplementary material) that arise from molecular and atomic oxygen species adsorbed on Ircus atoms, referred to hereafter as O2,ot and Oot, respectively.8,9 The total on-top oxygen coverage saturates at ∼0.8 ML for adsorption at 90 K, and the atomic state (Oot) populates preferentially at low coverages, while the desorption yield of the O2,ot increases more sharply above a total oxygen coverage of ∼0.6 ML.

All plane wave DFT calculations were performed using the projector augmented wave pseudopotentials27 provided in the Vienna ab initio simulation package (VASP).28,29 The Perdew–Burke–Ernzerhof (PBE) exchange-correlation functional30 was used with a plane wave expansion cutoff of 400 eV. We used four layers to model the IrO2(110) film, which has an ∼12 Å thick slab. The PBE bulk lattice constant of IrO2 (a = 4.54 Å and c = 3.19 Å) is used to fix the lateral dimensions of the slab. The bottom two layers are fixed, but all other lattice atoms are allowed to relax during the calculations until the forces are less than 0.05 eV/Å. A vacuum spacing of ∼25 Å was included, which is sufficient to reduce the periodic interaction in the surface normal direction. In terms of system size, a 2 × 4 unit cell with a corresponding 2 × 2 × 1 Monkhorst-Pack k-point mesh is employed, but studying the coverage effects of adsorbed N2 uses a different system size of 1 × 8 unit cell with a corresponding 1 × 4 × 1 Monkhorst-Pack k-point mesh. Unless otherwise noted, our DFT calculations were performed for a single N2 molecule adsorbed within the 2 × 4 surface model of IrO2(110), which corresponds to a N2 coverage equal to 12.5% of the total density of Ircus atoms and 25% of the Ircus density within one Ircus row. In this study, the binding energy, Eb, of an adsorbed N2 molecule on the surface is defined using the expression

Eb=EN2+EsurfEN2+surf,

where EN2+surf is the energy of N2 on the bare surface, Esurf is the energy of the bare surface, and EN2 is the energy of an isolated N2 molecule in the gas phase. All reported binding energies are corrected for zero-point vibrational energy. From the equation above, a large positive value for the binding energy indicates a high stability of the adsorbed N2 molecule under consideration.

To understand N2 adsorption on IrO2(110), we examined the partial density of states (pDOS) of the d orbitals of Ircus atoms. In addition, the crystal orbital Hamilton population (COHP) measure was used to characterize and quantify N2–Ircus orbital interactions. The COHP provides a measure of the overlap between specific atomic orbitals and therefore can provide a relative quantification of the bonding.31–33 The LOBSTER software is used to obtain the COHP from the VASP output.34 

We characterized the s-IrO2(110) film grown on RuO2(110) using LEED, XPS, LEISS, and O2 TPD to determine the oxidation state and structure of the IrO2 layer and confirm that the IrO2(110) layer is conformal such that a negligible amount of RuO2(110) is exposed at the vacuum–solid interface. Figure 2(a) shows Ir 4f spectra obtained after depositing 1.8 ML of metallic Ir onto RuO2(110) in UHV and after oxidizing the Ir by exposure to an O-atom beam at 700 K. The Ir 4f spectrum obtained after oxidation exhibits a dominant Ir 4f7/2 peak near 61.9 eV that agrees well with that reported for IrO235 and appears at ∼1 eV higher binding energy compared with the Ir 4f7/2 peak obtained from metallic Ir deposited onto RuO2(110). Ru 3d spectra obtained before and after growing an IrO2(110) film on an ∼13 layer RuO2(110) film demonstrate that the IrO2 film strongly attenuates the Ru 3d peaks [Fig. 2(b)]. Based on the Ru 3d5/2 peak attenuation, the IrO2(110) film is estimated to be 3.2 layers thick. LEISS measurements performed with Ne+ ions exhibit only a single peak that is consistent with Ir atoms at the surface; a peak for Ne+ scattering from Ru atoms is unobservable in the LEISS spectrum obtained from the IrO2 film (Fig. S1 of the supplementary material). Furthermore, O2 TPD spectra obtained after adsorbing on-top oxygen on the IrO2 film at 300 K closely resemble those obtained from a thick IrO2(110) layer grown on Ir(100)4,8,9 and lack distinguishing features that arise from the recombination of Oot-atoms on RuO2(110)36,37 (Fig. S2b of the supplementary material). These findings provide strong evidence that the growth method produces a fully oxidized IrO2 layer and that this layer completely covers the underlying RuO2(110) substrate.

FIG. 2.

(a) Ir 4f XPS spectra obtained before and after exposing Ir-covered RuO2(110) to atomic oxygen at 700 K. (b) Ru 3d XPS spectra obtained from the clean RuO2(110) film and after generating a 3.2 layer IrO2 film on the RuO2(110) film. LEED patterns obtained at 80 eV primary energy from (c) an initial 13-layer RuO2(110) film grown on Ru(0001) and (d) after preparing a 3.2 layer IrO2 film on the RuO2(110) surface using the procedure discussed in the text. The blue circle in (a) denotes the location of a primary LEED spot for Ru(0001). The absence of a LEED spot at this position demonstrates that the RuO2(110) film fully covers the Ru(0001) substrate. The lack of a similarly positioned spot from the IrO2(110) film confirms the absence of metallic Ir(111) domains.

FIG. 2.

(a) Ir 4f XPS spectra obtained before and after exposing Ir-covered RuO2(110) to atomic oxygen at 700 K. (b) Ru 3d XPS spectra obtained from the clean RuO2(110) film and after generating a 3.2 layer IrO2 film on the RuO2(110) film. LEED patterns obtained at 80 eV primary energy from (c) an initial 13-layer RuO2(110) film grown on Ru(0001) and (d) after preparing a 3.2 layer IrO2 film on the RuO2(110) surface using the procedure discussed in the text. The blue circle in (a) denotes the location of a primary LEED spot for Ru(0001). The absence of a LEED spot at this position demonstrates that the RuO2(110) film fully covers the Ru(0001) substrate. The lack of a similarly positioned spot from the IrO2(110) film confirms the absence of metallic Ir(111) domains.

Close modal

Figures 2(c) and 2(d) show LEED patterns obtained from the ∼13 layer, clean RuO2(110) surface and after growth of the ∼1 nm IrO2 layer, respectively. As discussed previously, clean RuO2(110) grows in three equivalent orientations on Ru(0001) in which the shorter a lattice vector of the (110) unit cell aligns with the close-packed directions of the substrate.37 The reciprocal lattice of RuO2(110) has dimensions of (0.862, 0.420) relative to the lattice constant of the Ru(0001) substrate, and the three rotational domains give rise to characteristic triplets of LEED spots located near the primary substrate spots in addition to a hexagonal arrangement of single spots, at the (0, 0.42) position, that are closer to the (0,0) spot [Fig. 2(c)]. The LEED pattern obtained from the ∼13 layer RuO2(110) surface exhibits negligible intensity at the locations of the primary spots from the Ru(0001) substrate [blue circle, Fig. 2(c)], demonstrating that the initial RuO2(110) film completely covers the Ru(0001) surface.

The LEED pattern obtained from the IrO2(110) layer grown on RuO2(110) [Fig. 2(d)] exhibits spots that are consistent with three rotational domains of IrO2(110) with the spots at nearly the same positions as observed for the clean RuO2(110) substrate. The LEED spots originating from an IrO2(110) layer should lie at similar positions as those from RuO2(110) since the a and b lattice constants of IrO2(110) and RuO2(110) are close in value (+1.6%; −0.3%). Our LEED observations thus provide strong evidence that the IrO2 layer forms exclusively in the (110) orientation on RuO2(110), in agreement with the results of Abb et al.24 We emphasize that our LEISS and O2 TPD measurements (Fig. S2 of the supplementary material) demonstrate that the IrO2 layer completely covers the RuO2(110) surface and assert that the observed LEED pattern arises entirely from the IrO2 surface because electrons diffracted from the underlying RuO2(110) substrate would be too strongly attenuated by the ∼1 nm IrO2 layer to produce measurable LEED spots. The LEED spots from the IrO2(110) film are relatively sharp but exhibit streaking mainly along the reciprocal space direction that corresponds to the [1¯10] direction (b direction) of the real-space unit cell (Fig. 1). This characteristic suggests that IrO2(110) tends to form domains on RuO2(110) that are longer in the [100] direction and shorter along the [1¯10] direction. Overall, our experiments confirm that the template-assisted growth procedure is effective in generating conformal IrO2(110) thin films on RuO2(110).24 

Our TPD results show that N2 binds strongly on the s-IrO2(110) surface, in good agreement with prior DFT results,12 and that the adsorbed N2 molecules become destabilized at high coverages. Figure 3(a) shows N2 TPD spectra obtained after generating different N2 coverages on s-IrO2(110) at 90 K. The N2 TPD trace initially exhibits a well-defined peak at about 270 K as well as a wide leading edge and a small feature near 100 K. We observe negligible evolution of NOx species, indicating that N2 adsorbs reversibly on IrO2(110) without reacting during the TPD measurements. The main TPD peak intensifies and downshifts to 265 K as the N2 coverage increases to nearly 0.5 ML, while the features below about 200 K remain largely unchanged. A desorption temperature of 270 K is significantly higher than that for physisorbed N238 and is thus indicative of a chemical bonding interaction between N2 and the IrO2(110) surface, in agreement with prior DFT results.12 Below, we discuss N2 binding energies on IrO2(110) determined from the TPD data.

FIG. 3.

(a) N2 TPD spectra as a function of N2 coverage obtained after adsorbing N2 on s-IrO2(110) at 90 K. (b) N2 coverage on s-IrO2(110) as a function of the N2 exposure at 90 K.

FIG. 3.

(a) N2 TPD spectra as a function of N2 coverage obtained after adsorbing N2 on s-IrO2(110) at 90 K. (b) N2 coverage on s-IrO2(110) as a function of the N2 exposure at 90 K.

Close modal

Increasing the N2 coverage to above ∼0.6 ML causes the intensity of the N2 TPD peak at 265 K to diminish slightly, while an intense shoulder near 235 K emerges and the desorption rate below 200 K increases. The main peak and shoulder as well as TPD features below ∼150 K further intensify as the N2 coverage approaches saturation. The appearance of a shoulder on the leading edge of the main N2 TPD peak suggests that N2 molecules adsorbed on the Ircus rows are destabilized as the N2 adlayer becomes crowded at high N2 coverages. Furthermore, we suggest that a fraction of the adsorbing N2 molecules become kinetically trapped in meta-stable configurations at 90 K and that such species are mainly responsible for the TPD features observed below ∼200 K. Overall, our TPD results show that N2 binds strongly on IrO2(110) and mainly populates a well-defined state in which the N2 molecules begin to repel one another as the layer becomes crowded at high N2 coverages.

Figure 3(b) shows the N2 uptake curve on IrO2(110) at 90 K as determined from our TPD results. We find that the N2 coverage initially increases sharply with N2 exposure until reaching a plateau at ∼0.8 ML, corresponding to the effective saturation coverage of N2 on s-IrO2(110) at 90 K. The uptake agrees well with Langmuirian behavior [dashed line, Fig. 3(b)], and the high N2 saturation coverage is consistent with the binding geometry predicted by DFT in which each N2 molecule binds on top of a single Ircus site and adopts an upright geometry analogous to adsorbed CO.12 

We applied inversion analysis of our TPD data to estimate the activation energy for N2 desorption from IrO2(110) as a function of N2 coverage. The inversion analysis, reported by Tait et al.,39–41 describes the desorption rate (Rd) as a function of the temperature (T) and adsorbate coverage (θ) using the Polanyi–Wigner equation

Rd=dθdt=νθnexpEd(θ)RT,
(1)

where ν represents the pre-factor for desorption, Ed(θ) is the coverage-dependent activation energy for desorption (“desorption energy”), and n is the reaction order. Rearranging the Polanyi–Wigner equation gives the following expression:

Edθ=RTlnRdνθn.
(2)

To apply this equation, the desorption rate (Rd) measured during TPD is first expressed in units of ML/s [e.g., Fig. 3(a)] and the TPD spectrum is numerically integrated to determine the N2 coverage (θ) at each temperature of the measurement. We then set the desorption pre-factor at a fixed value for all coverages and temperatures, set n = 1 (i.e., first order desorption), and solve Eq. (2) to obtain Ed(θ) for a TPD trace obtained for a given initial N2 coverage.

We applied the inversion analysis using two values of the desorption pre-factor that were computed using the transition-state theory formula and the so-called 3N and 3N − 2 models.42,43 The 3N model treats all of the adsorbate motions as harmonic vibrations, while the 3N − 2 model treats two of the frustrated motions as free and the remaining 3N − 2 motions as harmonic vibrations. Vibrational frequencies determined from DFT (Table S1 of the supplementary material) were used to calculate the pre-factors, and the motions that yield the lowest two vibrational frequencies were treated as a free translation and a free rotation in the 3N − 2 model. Desorption pre-factors of ν3N = 3.1 × 1015 s−1 and ν3N−2 = 2.7 × 1014 s−1 were used in the inversion analysis, where these values are equal to the 3N and 3N − 2 pre-factors computed at 270 K. Prior studies provide evidence that the 3N desorption pre-factors are more accurate for describing adsorbate desorption from IrO2(110) as well as RuO2(110)1–3,44 because adsorbed molecules have low mobility on these surfaces. The large surface corrugation perpendicular to the cus-metal rows and the formation of strong, localized bonds between the adsorbate and cus-metal site give rise to high barriers for adsorbate diffusion on these (110) oxide surfaces.4,8,9,11,37

Figures 4(a) and 4(b) show coverage-dependent desorption energies of N2 from s-IrO2(110) determined from TPD spectra obtained at initial N2 coverages of 0.46 ML and 0.75 ML and computed using the 3N and 3N − 2 desorption pre-factors. The sharp decrease in Ed with increasing N2 coverage to about 0.01–0.02 ML is an artifact of the analysis. Thereafter, inversion analysis predicts that Ed decreases linearly [red dashed lines, Figs. 4(a) and 4(b)] with increasing N2 coverage to ∼0.3 ML and 0.5 ML for the TPD spectra obtained at [N2]o = 0.46 ML and 0.75 ML, respectively. The desorption energies decrease more sharply as the N2 coverage increases to values that coincide with the leading-edge regions of the TPD spectra.

FIG. 4.

N2 desorption energy (E) as a function of the N2 coverage [N2] on s-IrO2(110) determined from inversion analysis of TPD data obtained at initial N2 coverages of 0.46 ML and 0.75 ML. (a) shows the E vs [N2] curves computed for the two initial N2 coverages using a desorption pre-factor computed using the 3N model (ν3N = 3 × 1015 s−1), and (b) shows the E vs [N2] curves computed for two initial N2 coverages using a desorption pre-factor computed using the 3N − 2 model (ν3N−2 = 2 × 1014 s−1). The linear fits mentioned in the text are shown as red dashed lines, and the corresponding equations are given in each graph.

FIG. 4.

N2 desorption energy (E) as a function of the N2 coverage [N2] on s-IrO2(110) determined from inversion analysis of TPD data obtained at initial N2 coverages of 0.46 ML and 0.75 ML. (a) shows the E vs [N2] curves computed for the two initial N2 coverages using a desorption pre-factor computed using the 3N model (ν3N = 3 × 1015 s−1), and (b) shows the E vs [N2] curves computed for two initial N2 coverages using a desorption pre-factor computed using the 3N − 2 model (ν3N−2 = 2 × 1014 s−1). The linear fits mentioned in the text are shown as red dashed lines, and the corresponding equations are given in each graph.

Close modal

Similar linear relations between Ed and [N2] are predicted from the TPD spectra obtained at different initial N2 coverages and a fixed pre-factor. Specifically, a zero-coverage N2 binding energy of ∼90 kJ/mol and a slope of about −30 kJ/mol per ML are estimated for the inversion analysis performed with the 3N pre-factor. The intercept and slope are about 84 kJ/mol and −28 kJ/mol per ML, respectively, for the linear region of Ed([N2]) determined with the 3N − 2 pre-factor. Comparison shows that the N2 desorption energies differ by less than ∼10% for the 3N vs 3N − 2 pre-factors and are thus only weakly sensitive to the pre-factor values determined from these models. As mentioned above, we expect that the zero-coverage desorption energy predicted using the 3N pre-factor is more accurate. Our analysis quantitatively demonstrates that N2 binds strongly to the IrO2(110) surface, achieving a binding energy (90 kJ/mol) that is indicative of a chemical bonding interaction with the Ircus atoms in agreement with DFT calculations by Wang et al.12 The N2 binding energy on IrO2(110) is also significantly larger than the values reported for N2 adsorbed molecularly on Fe(111) and Ru(0001) surfaces,45,46 which are metals used in ammonia synthesis. Furthermore, the linear decrease in the N2 binding energy with increasing N2 coverage initiates at the lowest coverage studied. This behavior suggests that N2 molecules populate Ircus adsorption sites randomly and thus have limited ability to diffuse along the Ircus rows at 90 K to avoid destabilizing interactions with neighboring molecules.

Figure 5(a) shows a series of N2 TPD spectra obtained after exposing different O-rich IrO2(110) surfaces to a saturation dose of N2 at 90 K. As discussed in the Experimental Details section, the O-rich IrO2(110) surfaces were prepared by exposing s-IrO2(110) to varying quantities of O2 at a surface temperature of 90 K prior to N2 adsorption. Our results show that the main N2 TPD peak in addition to the TPD features at lower temperature diminishes with increasing oxygen pre-coverage. In fact, the N2 coverage decreases approximately linearly with increasing oxygen pre-coverage to 0.6 ML [Fig. 5(b)], and [N2] decreases to ∼0.04 ML as the initial on-top oxygen reaches saturation at a coverage of ∼0.8 ML. This behavior demonstrates that on-top oxygen species suppress N2 adsorption on IrO2(110) and thus provides strong evidence that N2 adsorbs nearly exclusively on Ircus sites at 90 K.

FIG. 5.

(a) N2 TPD spectra as a function of the on-top oxygen coverage adsorbed prior to a saturation N2 exposure (0.5 L) at 90 K, and (b) N2 coverage at 90 K as a function of the initial on-top oxygen coverage. The dashed line shown in (b) represents a linear fit to the data up to an O-coverage of ∼0.6 ML.

FIG. 5.

(a) N2 TPD spectra as a function of the on-top oxygen coverage adsorbed prior to a saturation N2 exposure (0.5 L) at 90 K, and (b) N2 coverage at 90 K as a function of the initial on-top oxygen coverage. The dashed line shown in (b) represents a linear fit to the data up to an O-coverage of ∼0.6 ML.

Close modal

The main N2 TPD peak also remains broad and shifts toward higher temperature as the oxygen pre-coverage increases. The rate maximum shifts from ∼265 K to 290 K for N2 adsorbed on the saturated oxygen layer, and the trailing edges of the spectra lie at higher temperature for the oxygen pre-covered vs clean IrO2(110) surfaces. The upshift in the N2 desorption temperature indicates that on-top oxygen species (Oot, O2,ot) stabilize N2 molecules on the IrO2(110) surface. Inversion analysis using the 3N pre-factor predicts a binding energy of ∼98 kJ/mol for N2 in the limit of low coverage on O-rich IrO2(110) with 0.6 ML of oxygen. We suggest that the N2 TPD peaks are broad for the oxygen-covered IrO2(110) surfaces, irrespective of the N2 coverage, because the total coverage (N2 + Oot, O2,ot) was high (∼0.7 ML to 0.8 ML) in each of our TPD experiments with O-rich IrO2(110). Intermolecular interactions would influence the binding of a larger fraction of the N2 molecules at these high total coverages and thus broaden the N2 TPD peak.

We performed PBE calculations to investigate how co-adsorbed N2 and Oot-atoms influence the adsorption of N2 on the IrO2(110) surface for comparison with the experimental results. Wang et al. have previously reported that N2 binds strongly on an Ircus atom of s-IrO2(110) with a binding energy of 106 kJ/mol and preferentially adsorbs into an upright configuration that is analogous to the bonding geometry of CO adsorbed on metal and metal-oxide surfaces.12 Those authors presented an analysis of the partial density of states (p-DOS) projected onto the cus-metal atoms of IrO2(110), RuO2(110), and TiO2(110), which reveals that the Ircus atom possesses higher densities of empty 5dz2 states as well as filled 5dxz and 5dyz states near the Fermi level (EF) compared with Rucus and Ticus atoms. The high densities of these filled and empty d-states near EF should enhance σ-donation and π* back-donation interactions, respectively, between N2 and an Ircus atom, resulting in stronger bonding of N2 on IrO2(110) compared with RuO2(110) and TiO2(110). The IpCOHP analysis presented below supports the interpretation reported by Wang et al.12 

Figure 6 shows the favored configurations and binding energies of N2 on IrO2(110) with 0, 1, and 2 adjacent Oot atoms that we computed using DFT-PBE. Our DFT calculations of N2 adsorption on IrO2(110) agree well with the results of Wang et al.,12 as we find that N2 strongly prefers to adsorb in an upright geometry and achieves a binding energy of 101 kJ/mol. The computed binding energy lies close to our experimental estimate of ∼90 kJ/mol for N2 adsorbed on s-IrO2(110) at low coverage. Test calculations incorporating dispersion within the D3 method47 further overestimate the N2 binding energy on IrO2(110) by ∼30 kJ/mol. The calculations predict that the N2 stretch frequency redshifts from 2426 cm−1 to 2286 cm−1 after N2 adsorbs on the s-IrO2(110) surface and that the N–N bond elongates by about 0.01 Å (Table S2 of the supplementary material). The N–N stretch frequency is predicted to be slightly larger for N2 adsorbed on s-IrO2(110) vs s-RuO2(110) (2286 vs 2280 cm−1), even though the computed binding energy is about twice as high for N2 on s-IrO2(110). As discussed below, we have reported similar behavior for CO adsorbed on single vs multiple layer PdO(101) surfaces on Pd(100).48 Our calculations also reveal that Oot atoms strengthen the N2 binding energy on IrO2(110) by ∼5 kJ/mol per neighboring Oot atom [Figs. 6(b) and 6(c)], in close agreement with our analysis of TPD data. Prior studies report that Oot atoms also enhance the binding of CH4, H2, and NHx species on IrO2(110).4,7,8,12 Although H-bonding with Oot atoms contributes to the stabilization of NHx species, Wang et al. presented evidence that a bonding interaction between the Oot atom and co-adsorbed species also plays an important role.12 

FIG. 6.

Molecular configurations of the preferred upright configuration of an N2 molecule (green) on IrO2(110) with (a) zero Oot atoms, (b) one Oot atom, and (c) two Oot atoms as determined using DFT-PBE calculations.

FIG. 6.

Molecular configurations of the preferred upright configuration of an N2 molecule (green) on IrO2(110) with (a) zero Oot atoms, (b) one Oot atom, and (c) two Oot atoms as determined using DFT-PBE calculations.

Close modal

We also used DFT to investigate the binding of N2 on s-IrO2(110) as a function of the N2 coverage to make comparisons with our experimental finding that N2 molecules become destabilized on IrO2(110) at high N2 coverages. Figures 7(a) and 7(b) show the computed energies required to sequentially remove N2 molecules from the ends [Fig. 7(a)] vs the middle [Fig. 7(b)] of –N2–N2– rows on IrO2(110), where the initial N2 coverage is 1 ML and the removal of one N2 molecule is equivalent to decreasing the N2 coverage by 0.125 ML. As expected, our results predict that N2 desorption from the middle of a –N2–N2– row is more favorable (smaller energy requirement) than N2 desorption from the end of a row, demonstrating larger destabilization of N2 molecules adsorbed next to two vs one N2 molecules. The destabilization also becomes less pronounced as the length of a –N2–N2– chain decreases, indicating that relaxation at the chain ends relieves the repulsive interactions experienced by the interior N2 molecules. For example, the energies required to remove an N2 molecule from the middle of an infinitely long vs three-molecule long chain are 86 kJ/mol and 92 kJ/mol, respectively. Our DFT results capture the destabilization of N2 adsorbed at high coverages on IrO2(110) but underestimate the decrease in N2 binding energy compared with our experimental estimates (∼15 kJ/mol vs ∼30 kJ/mol per ML of N2). We speculate that the slightly overestimated binding strength (∼10 kJ/mol) of N2 on an Ircus atom offsets the destabilization caused by neighboring N2 molecules. Overall, our simulations correctly predict destabilization of the N2 adlayer with increasing N2 coverage and reveal that the N2 binding energy is sensitive to the local N2 environment. The latter finding provides a viable explanation for the broadening of the N2 TPD peak that we observe at high N2 coverages.

FIG. 7.

Energies required to sequentially remove an N2 molecule from N2-covered IrO2(110) initially with 1 ML of N2 molecules adsorbed on Ircus atoms computed using DFT-PBE, where each step lowers the N2 coverage by 0.125 ML. Energy changes are shown for two scenarios: (a) N2 desorption from the end of an –N2–N2– chain and (b) N2 desorption from the middle of –N2–N2– chains. A more positive value of the energy change corresponds to a higher barrier for the N2 molecule to desorb from the surface.

FIG. 7.

Energies required to sequentially remove an N2 molecule from N2-covered IrO2(110) initially with 1 ML of N2 molecules adsorbed on Ircus atoms computed using DFT-PBE, where each step lowers the N2 coverage by 0.125 ML. Energy changes are shown for two scenarios: (a) N2 desorption from the end of an –N2–N2– chain and (b) N2 desorption from the middle of –N2–N2– chains. A more positive value of the energy change corresponds to a higher barrier for the N2 molecule to desorb from the surface.

Close modal

To understand the origins of strong binding of N2 on IrO2(110) (101 kJ/mol), we evaluated pCOHP between the N2 molecule and the d-orbitals of an Ircus atom. For comparison, we also performed the same analysis for N2 on s-RuO2(110) for which we predict a binding energy of 55 kJ/mol at a low coverage. The IpCOHP, obtained by integrating the pCOHP up to the Fermi level, provides a quantitative approach for evaluating the strength of specific orbital interactions. Table I lists the bonding and antibonding contributions to the σ- and π-bonding of adsorbed N2 on IrO2(110) and RuO2(110). The pCOHP analysis predicts strong bonding and antibonding interactions that lead to more negative, net interactions of σ- and π-bonding for N2 on IrO2(110) compared to RuO2(110). The σ bonding involves donation from the N2σ orbital into empty 5dz2 states and can be strengthened by an increase in the density of empty 5dz2 states near the Fermi level. Consistent with the larger σ-donation predicted for N2 on IrO2(110), Fig. 8(a) shows that the IrO2(110) surface has a higher density of unoccupied dz2 states near the Fermi level compared with RuO2(110). Wang et al. also argued that the lower energy level of the unoccupied dz2 state on IrO2(110) vs RuO2(110) triggers strong σ donations from N2 to an Ircus atom and enhances the adsorption strength.12 Our analysis quantitatively supports this interpretation. The pCOHP analysis reveals that π-bonding interactions are also stronger for N2 on IrO2(110) vs RuO2(110). Since π bonding interactions involve back-donation from 5dxz + 5dyz states into the N2 π* orbital, an increase in the density of occupied 5dxz + 5dyz states near the Fermi level should enhance π back-donation. Figure 8(b) shows that an Ircus atom has a higher density of occupied dxz + dyz states near the Fermi level compared with a Rucus atom of RuO2(110).

TABLE I.

IpCOHP values of upright N2 on IrO2(110) and RuO2(110) with their respective Eb (kJ/mol) values. The σ interaction involves bonding between Ir (Ru) 5dz2 (4dz2) states and N 2pz and 2s orbitals and the π interaction involves bonding between Ir (Ru) 5dxz (4dxz) states and the N 2px orbital as well as bonding between Ir (Ru) 5dyz (4dyz) states and the N 2py orbital. A more negative (positive) value for the IpCOHP is indicative of stronger bonding (antibonding) interactions.

σ interactionπ interaction
SurfacesEbTotalBondingAntibondingTotalBondingAntibonding
IrO2(110) 101.3 −2.12 −2.97 0.84 −1.12 −1.61 0.49 
RuO2(110) 55.3 −1.69 −2.34 0.65 −0.84 −1.16 0.33 
σ interactionπ interaction
SurfacesEbTotalBondingAntibondingTotalBondingAntibonding
IrO2(110) 101.3 −2.12 −2.97 0.84 −1.12 −1.61 0.49 
RuO2(110) 55.3 −1.69 −2.34 0.65 −0.84 −1.16 0.33 
FIG. 8.

Projected density of states of (a) 5dz2 and 4dz2 and (b) 5dxz + 5dyz and 4dxz + 4dyz for IrO2(110) (blue) and RuO2(110) (red) with respect to the Fermi level (EF). The Fermi level is located at 0 eV and marked by the dashed line.

FIG. 8.

Projected density of states of (a) 5dz2 and 4dz2 and (b) 5dxz + 5dyz and 4dxz + 4dyz for IrO2(110) (blue) and RuO2(110) (red) with respect to the Fermi level (EF). The Fermi level is located at 0 eV and marked by the dashed line.

Close modal

The COHP analysis shows that the bonding of N2 on IrO2(110) involves both σ-donation and π* back-donation interactions, with the σ-donation interaction making a lager contribution to the bonding. Each type of interaction is stronger on IrO2(110) compared with RuO2(110) because the former has higher densities of unoccupied dz2 and occupied dxz + dyz states near the Fermi level (Fig. 8). We have recently reported an analogous bonding mechanism of CO adsorbed on PdO(101) surfaces and showed that CO binds more strongly on multiple vs single-layer PdO(101) due to both enhanced σ-donation and π* back-donation interactions.48 These interactions have offsetting effects on the C–O bond order, resulting in similar C–O stretch frequencies but different CO binding energies on the PdO(101) surfaces. The similar N2 stretch frequency but a larger binding energy of N2 on IrO2(110) vs RuO2(110) is analogous to the behavior reported for CO on PdO(101) surfaces, further demonstrating the similarities in the bonding of N2 and CO on these late transition-metal oxide surfaces.

We have investigated the adsorption on N2 on an IrO2(110) film using TPD experiments and DFT calculations. Our results show that N2 desorbs predominantly from s-IrO2(110) in a well-defined TPD peak at ∼270 K at N2 coverages below about 0.5 ML and that a new feature centered near 235 K develops as the N2 layer becomes crowded at higher N2 coverages. Inversion analysis of the TPD data predicts a binding energy of ∼90 kJ/mol for N2 adsorbed at a low coverage on s-IrO2(110) and shows that the N2 binding energy decreases with an increase in N2 coverage by about 30 kJ/mol per ML of N2. Our TPD results also demonstrate that on-top oxygen strongly suppresses N2 adsorption on IrO2(110), thus providing experimental evidence that N2 binds exclusively on Ircus atoms and that Oot atoms stabilize the binding of N2 by 5–10 kJ/mol at a high Oot coverage. The binding energy of an isolated N2 molecule on s-IrO2(110) computed using DFT-PBE agrees well with our experimental estimate from TPD data (101 kJ/mol vs 90 kJ/mol). Furthermore, the DFT calculations accurately reproduce the trends in N2 binding energy with increasing N2 and Oot coverage on IrO2(110). Our results demonstrate that DFT-PBE predictions of N2 binding energies on s-IrO2(110) agree well with experimental estimates and provide insights for understanding how co-adsorbed species influence the binding on IrO2(110).

See the supplementary material for Ne+ LEISS spectra from the IrO2(110)/RuO2(110) film, O2 TPD spectra obtained as a function of on-top oxygen coverage on the 3.2 layer IrO2(110) film on RuO2(110), O2 TPD spectra from saturated Oot layers on pure RuO2(110) and an IrO2(110) film on RuO2(110), normal mode frequencies of N2 on IrO2(110) computed using DFT, and computed properties of N2 adsorbed on IrO2(110) and RuO2(110) surfaces.

R.M. and M.K. contributed equally to this work.

We acknowledge the Ohio Supercomputing Center for providing computational resources. We also acknowledge financial support from the Department of Energy, Office of Basic Energy Sciences, Catalysis Science Division through Grant No. DE-FG02-03ER15478 and ExxonMobil Research and Engineering.

1.
Z.
Liang
,
T.
Li
,
M.
Kim
,
A.
Asthagiri
, and
J. F.
Weaver
, “
Low-temperature activation of methane on the IrO2(110) surface
,”
Science
356
,
298
301
(
2017
).
2.
Y. X.
Bian
,
M.
Kim
,
T.
Li
,
A.
Asthagiri
, and
J. F.
Weaver
, “
Facile dehydrogenation of ethane on the IrO2(110) surface
,”
J. Am. Chem. Soc.
140
,
2665
2672
(
2018
).
3.
R.
Martin
,
M.
Kim
,
A.
Franklin
,
Y.
Bian
,
A.
Asthagiri
, and
J. F.
Weaver
, “
Adsorption and oxidation of propane and cyclopropane on IrO2(110)
,”
Phys. Chem. Chem. Phys.
20
,
29264
29273
(
2018
).
4.
M.
Kim
,
A.
Franklin
,
R.
Martin
,
F.
Feng
,
T.
Li
,
Z.
Liang
,
A.
Asthagiri
, and
J.
Weaver
, “
Adsorption and oxidation of CH4 on oxygen-rich IrO2(110)
,”
J. Phys. Chem. C
123
,
27603
27614
(
2019
).
5.
T. L. M.
Pham
,
E. G.
Leggesse
, and
J. C.
Jiang
, “
Ethylene formation by methane dehydrogenation and C–C coupling reaction on a stoichiometric IrO2(110) surface - a density functional theory investigation
,”
Catal. Sci. Tech.
5
,
4064
4071
(
2015
).
6.
T. L. M.
Pham
,
S.
Nachimuthu
,
J. L.
Kuo
, and
J. C.
Jiang
, “
A DFT study of ethane activation on IrO2(110) surface by precursor-mediated mechanism
,”
Appl. Catal., A
541
,
8
14
(
2017
).
7.
C. C.
Wang
,
S. S.
Siao
, and
J. C.
Jiang
, “
C–H bond activation of methane via sigma-d Interaction on the IrO2(110) surface: Density functional theory study
,”
J. Phys. Chem. C
116
,
6367
6370
(
2012
).
8.
T.
Li
,
M.
Kim
,
Z.
Liang
,
A.
Asthagiri
, and
J.
Weaver
, “
Hydrogen oxidation on oxygen-rich IrO2(110)
,”
Catal., Struct. React.
4
,
1
13
(
2018
).
9.
T.
Li
,
M.
Kim
,
Z.
Liang
,
A.
Asthagiri
, and
J. F.
Weaver
, “
Dissociative chemisorption and oxidation of H2 on the stoichiometric IrO2(110) surface
,”
Top. Catal.
61
,
397
411
(
2018
).
10.
J. F.
Weaver
,
C.
Hakanoglu
,
A.
Antony
, and
A.
Asthagiri
, “
Alkane activation on crystalline metal oxide surfaces
,”
Chem. Soc. Rev.
43
,
7536
7547
(
2014
).
11.
J. F.
Weaver
, “
Surface chemistry of late transition metal oxides
,”
Chem. Rev.
113
,
4164
4215
(
2013
).
12.
C. C.
Wang
,
S. S.
Siao
, and
J. C.
Jiang
, “
Density functional theory study of NHx (x = 0–3) and N2 adsorption on IrO2(110) surfaces
,”
J. Phys. Chem. C
114
,
18588
18593
(
2010
).
13.
C. C.
Wang
,
S. S.
Siao
, and
J. C.
Jiang
, “
Density functional theory study of the oxidation of ammonia on the IrO2(110) surface
,”
Langmuir
27
,
14253
14259
(
2011
).
14.
V.
Fung
,
F.
Tao
, and
D. E.
Jiang
, “
Low-temperature activation of methane on doped single atoms: Descriptor and prediction
,”
Phys. Chem. Chem. Phys.
20
,
22909
22914
(
2018
).
15.
Y.
Tsuji
and
K.
Yoshizawa
, “
Adsorption and activation of methane on the (110) surface of rutile-type metal dioxides
,”
J. Phys. Chem. C
122
,
15359
15381
(
2018
).
16.
Y. X.
Bian
,
T.
Li
, and
J. F.
Weaver
, “
Structure and reactivity of iridium oxide layers grown on Ir(100) by oxidation at sub-ambient O2 pressures
,”
J. Phys. D: Appl. Phys.
52
,
434002
(
2019
).
17.
A. L.
Gerrard
,
J. J.
Chen
, and
J. F.
Weaver
, “
Oxidation of nitrided Si(100) by gaseous atomic and molecular oxygen
,”
J. Phys. Chem. B
109
,
8017
8028
(
2005
).
18.
R.
Rai
,
T.
Li
,
Z.
Liang
,
M.
Kim
,
A.
Asthagiri
, and
J. F.
Weaver
, “
Growth and termination of an IrO2(100) film on Ir(111)
,”
Surf. Sci.
252
,
213
221
(
2016
).
19.
H. H.
Kan
,
R. B.
Shumbera
, and
J. F.
Weaver
, “
Adsorption and abstraction of oxygen atoms on Pd(111): Characterization of the precursor to PdO formation
,”
Surf. Sci.
602
,
1337
1346
(
2008
).
20.
W.
Cartas
,
R.
Rai
,
A.
Sathe
,
A.
Schaefer
, and
J. F.
Weaver
, “
Oxidation of a Tb2O3(111) thin film on Pt(111) by gas-phase oxygen atoms
,”
J. Phys. Chem. C
118
,
20916
20926
(
2014
).
21.
T.
Li
,
M.
Kim
,
R.
Rai
,
Z.
Liang
,
A.
Asthagiri
, and
J. F.
Weaver
, “
Adsorption of alkanes on stoichiometric and oxygen-rich RuO2(110)
,”
Phys. Chem. Chem. Phys.
18
,
22647
22660
(
2016
).
22.
T.
Li
,
R.
Rai
,
Z.
Liang
,
M.
Kim
,
A.
Asthagiri
, and
J. F.
Weaver
, “
Adsorption and oxidation of n-butane on the stoichiometric RuO2(110) surface
,”
J. Phys. Chem. C
120
,
9863
9873
(
2016
).
23.
Z.
Liang
,
M.
Kim
,
T.
Li
,
R.
Rai
,
A.
Asthagiri
, and
J. F.
Weaver
, “
Adsorption and oxidation of ethylene on the stoichiometric and O-rich RuO2(110) surfaces
,”
J. Phys. Chem. C
121
,
20375
20386
(
2017
).
24.
M. J. S.
Abb
,
B.
Herd
, and
H.
Over
, “
Template-assisted growth of ultrathin single-crystalline IrO2(110) films on RuO2(110)/Ru(0001) and its thermal stability
,”
J. Phys. Chem. C
122
,
14725
14732
(
2018
).
25.
R.
Rai
and
J. F.
Weaver
, “
Methanol oxidation on stoichiometric and oxygen-rich RuO2(110)
,”
Phys. Chem. Chem. Phys.
19
,
18975
18987
(
2017
).
26.
C. J.
Powell
and
A.
Jablonski
,
NIST Electron Inelastic-Mean-Free-Path Database
(
National Institute of Standards and Technology
,
Gaithersburg, Maryland
,
2010
).
27.
P. E.
Blochl
, “
Projector augmented-wave method
,”
Phys. Rev. B
50
,
17953
17979
(
1994
).
28.
G.
Kresse
, “
Ab-initio molecular-dynamics for liquid-metals
,”
J. Non-Cryst. Solids
193
,
222
229
(
1995
).
29.
G.
Kresse
and
J.
Hafner
, “
Abinitio Hellmann-Feynman molecular-dynamics for liquid-metals
,”
J. Non-Cryst. Solids
156
,
956
960
(
1993
).
30.
J. P.
Perdew
,
K.
Burke
, and
M.
Ernzerhof
, “
Generalized gradient approximation made simple
,”
Phys. Rev. Lett.
77
,
3865
3868
(
1996
).
31.
R.
Dronskowski
and
P. E.
Blochl
, “
Crystal orbital Hamilton populations (cohp): Energy-resolved visualization of chemical bonding in solids based on density-functional Calculations
,”
J. Phys. Chem.
97
,
8617
8624
(
1993
).
32.
V. L.
Deringer
,
A. L.
Tchougreeff
, and
R.
Dronskowski
, “
Crystal orbital Hamilton population (COHP) analysis as projected from plane-wave basis sets
,”
J. Phys. Chem. A
115
,
5461
5466
(
2011
).
33.
S.
Maintz
,
V. L.
Deringer
,
A. L.
Tchougreeff
, and
R.
Dronskowski
, “
Analytic projection from plane-wave and PAW wavefunctions and application to chemical-bonding analysis in solids
,”
J. Comput. Chem.
34
,
2557
2567
(
2013
).
34.
S.
Maintz
,
V. L.
Deringer
,
A. L.
Tchougreeff
, and
R.
Dronskowski
, “
Lobster: A tool to extract chemical bonding from plane-wave based DFT
,”
J. Comput. Chem.
37
,
1030
1035
(
2016
).
35.
S. J.
Freakley
,
J.
Ruiz-Esquius
, and
D. J.
Morgan
, “
The X-ray photoelectron spectra of Ir, IrO2 and IrCl3 revisited
,”
Surf. Interface Anal.
49
,
794
799
(
2017
).
36.
Y. D.
Kim
,
A. P.
Seitsonen
,
S.
Wendt
,
J.
Wang
,
C.
Fan
,
K.
Jacobi
,
H.
Over
, and
G.
Ertl
, “
Characterization of various oxygen species on an oxide surface: RuO2(110)
,”
J. Phys. Chem. B
105
,
3752
3758
(
2001
).
37.
H.
Over
, “
Surface chemistry of ruthenium dioxide in heterogeneous catalysis and electrocatalysis: From fundamental to applied research
,”
Chem. Rev.
112
,
3356
3426
(
2012
).
38.
P.
Gardner
,
R.
Martin
,
M.
Tushaus
,
J.
Shamir
, and
A. M.
Bradshaw
, “
The structure-sensitive adsorption of N2 on the (1 × 1) and (5 × 1) Surfaces of Ir(100)
,”
Surf. Sci.
287
,
135
140
(
1993
).
39.
S. L.
Tait
,
Z.
Dohnalek
,
C. T.
Campbell
, and
B. D.
Kay
, “
n-Alkanes on MgO(100). I. Coverage-dependent desorption kinetics of n-butane
,”
J. Chem. Phys.
122
,
164707
(
2005
).
40.
S. L.
Tait
,
Z.
Dohnalek
,
C. T.
Campbell
, and
B. D.
Kay
, “
n-Alkanes on MgO(100). II. Chain length dependence of kinetic desorption parameters for small n-alkanes
,”
J. Chem. Phys.
122
,
164708
(
2005
).
41.
S. L.
Tait
,
Z.
Dohnalek
,
C. T.
Campbell
, and
B. D.
Kay
, “
n-alkanes on Pt(111) and on C(0001)/Pt(111): Chain length dependence of kinetic desorption parameters
,”
J. Chem. Phys.
125
,
234308
(
2006
).
42.
A.
Antony
,
A.
Asthagiri
, and
J. F.
Weaver
, “
Pathways for C–H bond cleavage of propane σ-complexes on PdO(101)
,”
Phys. Chem. Chem. Phys.
14
,
12202
12212
(
2012
).
43.
A.
Antony
,
A.
Asthagiri
, and
J. F.
Weaver
, “
Pathways and kinetics of methane and ethane C–H bond cleavage on PdO(101)
,”
J. Chem. Phys.
139
,
104702-1
104702-12
(
2013
).
44.
L.
Chen
,
R. S.
Smith
,
B. D.
Kay
, and
Z.
Dohnalek
, “
Adsorption of small hydrocarbons on rutile TiO2(110)
,”
Surf. Sci.
650
,
83
92
(
2016
).
45.
J. J.
Mortensen
,
L. B.
Hansen
,
B.
Hammer
, and
J. K.
Norskov
, “
Nitrogen adsorption and dissociation on Fe(111)
,”
J. Catal.
182
,
479
488
(
1999
).
46.
J. J.
Mortensen
,
Y.
Morikawa
,
B.
Hammer
, and
J. K.
Norskov
, “
A comparison of N2 and CO adsorption on Ru(001)
,”
Z. Phys. Chem.
198
,
113
122
(
1997
).
47.
S.
Grimme
,
J.
Antony
,
S.
Ehrlich
, and
H.
Krieg
, “
A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu
,”
J. Chem. Phys.
132
,
154104
(
2010
).
48.
V.
Mehar
,
M.
Kim
,
M.
Shipilin
,
M.
Van den Bossche
,
J.
Gustafson
,
L. R.
Merte
,
U.
Hejral
,
H.
Gronbeck
,
E.
Lundgren
, et al, “
Understanding the intrinsic surface reactivity of single-layer and multilayer PdO(101) on Pd(100)
,”
ACS Catal.
8
,
8553
8567
(
2018
).

Supplementary Material