Highly Active and Stable CH4 Oxidation by Substitution of Ce4+ by Two Pd2+ Ions in CeO2(111).

Methane (CH4) combustion is an increasingly important reaction for environmental protection, for which Pd/CeO2 has emerged as the preferred catalyst. There is a lack of understanding of the nature of the active site in these catalysts. Here, we use density functional theory to understand the role of doping of Pd in the ceria surface for generating sites highly active toward the C-H bonds in CH4. Specifically, we demonstrate that two Pd2+ ions can substitute one Ce4+ ion, resulting in a very stable structure containing a highly coordinated unsaturated Pd cation that can strongly adsorb CH4 and dissociate the first C-H bond with a low energy barrier. An important aspect of the high activity of the stabilized isolated Pd cation is its ability to form a strong σ-complex with CH4, which leads to effective activation of CH4. We show that also other transition metals like Pt, Rh, and Ni can give rise to similar structures with high activity toward C-H bond dissociation. These insights provide us with a novel structural view of solid solutions of transition metals such as Pt, Pd, Ni, and Rh in CeO2, known to exhibit high activity in CH4 combustion.

Introduction

Methane (CH4) is a significant greenhouse gas with a global warming potential ca. 20 times higher than that of CO2.[1] Accordingly, it is desirable to develop technologies to remove residual CH4 present from the exhaust of increasingly popular natural gas engines.[2−6] The challenge is to develop highly active catalysts that can operate at the relatively low temperatures of the exhaust gas. The high Pd loading in the current generation of preferred Pd/CeO2 catalysts poses a cost challenge, requiring a deeper understanding of the nature of the active sites.[1,7−9] In automotive three-way catalysis, Pd is already extensively used to catalyze hydrocarbon oxidation.[10−13] The function of CeO2 in these catalysts is mainly to store and release oxygen, while more recently its ability to maintain a high dispersion of transition metals has been emphasized.[14−16] Although the importance of a close interaction between Pd and CeO2 in catalytic oxidation chemistry has been clearly demonstrated,[1,3,7,17] the exact structure of the active Pd species remains unclear. It has for instance been shown that isolated PdO species on the surface of CeO2 are crucial for the low-temperature CO oxidation performance in Pd/CeO2.[18]

Several studies mention the importance of bulk PdO for high CH4 combustion activity.[1,3,7,11,19] Computational studies support this by showing that the (101) surface termination of PdO contains under-coordinated Pd atoms able to form a strong σ-adsorption complex with CH4.[11,19−22] Conventionally, CH4 dissociation on metal surfaces relies on high CH4 translational energies to enhance the sticking probability. In the context of low-temperature CH4 activation, a Langmuir–Hinshelwood mechanism for CH4 combustion is preferred. A recent study of Weaver and co-workers also showed that CH4 adsorbs strongly on under-coordinated Ir atoms in the IrO2(110) surface, resulting in low-temperature C–H bond activation.[23]

However, the activity of PdO alone is not sufficient to explain the Pd-CeO2 synergy observed for CH4 activation. Cargnello et al. showed that PdO species at the Pd–ceria interface account for the high CH4 combustion activity, while PdO reduction into metallic Pd at T > 850 °C is causing deactivation.[1] Trovarelli and co-workers proposed that highly dispersed Pd, specifically Pd in a PdO4 square-planar coordination environment, is the most likely active site for CH4 oxidation.[3] Lu and co-workers showed that removal of PdO nanoparticles from a PdO/Ce1–PdO2-σ catalyst significantly improves CH4 oxidation, emphasizing the important role of a solid solution of Pd in CeO2.[7] The relevance of such solid solutions in ceria-based catalysis is increasingly recognized.[24−27] Recent literature also suggests that solid ceria solutions of Pt, Pd, Ni, and Rh may be crucial to explain CH4 combustion at low temperature.[7,17,28−30]

Usually, doping of transition metals in ceria is modeled by replacing a surface cerium atom by a transition metal atom. For Pd-doped CeO2(111), we recently considered a conventional octahedral configuration as well as a novel and more stable square-planar configuration for CH4 oxidation.[27] The predicted CH4 dissociation rates on these models were lower than on PdO(101) due to weak CH4 adsorption.[24,27] Earlier, Janik and co-workers considered models in which more than one Pd atom are embedded in the ceria lattice and which can lead to facile CH4 activation in combination with PdO clusters.[26,31] We emphasize that these models including the thermodynamically stable (111)-3Pd2+ configuration also cannot explain the experimentally observed high CH4 oxidation activity of Pd-CeO2 solid solutions. Janik and co-workers proposed that the Pd2+ ↔ Pd4+ transitions contribute to the high activity of Pd-CeO2 solid solution.[26,31] However, structures containing Pd4+ (e.g., (111)-1Pd4+/2Pd2+) are thermodynamically unstable under reaction conditions. Thermodynamically stable models will only contain Pd2+,[27] in agreement with experimental XPS and XRD data on Pd-CeO2 solid solution.[3,32−34]

An important corollary of previous data for single-Pd-atom modified CeO2 surfaces is that CH4 bonding is extremely weak, i.e., only physical adsorption occurs.[24,27] Accordingly, adsorbed CH4 is not effectively activated by Pd2+ in these configurations. As a result, the computed energy barriers for CH4 dissociation are higher than 0.9 eV and the dissociative adsorption of CH4 is endothermic.[27] Thus, from a computational point of view models considered hitherto cannot explain the high catalytic activity of transition metal doped CeO2.[24−27,31,35] In the present work, we show for the first time that one Ce ion is substituted by two Pd ions, resulting in a very thermodynamically stable configuration which is not only sinter-stable but also very reactive toward the C–H bonds in CH4. One of the Pd ions is 3-fold coordinated by lattice O atoms, while the other is 4-fold coordinated by lattice O atoms. The coordinative unsaturation of this Pd atom results in strong CH4 chemisorption and activation. We will demonstrate the broader applicability of this concept by computing stability and reactivity of other transition-metal-doped CeO2 surfaces. These insights are in good agreement with available experimental data.

Computational Details

Density Functional Theory (DFT) Calculations

We carried out spin-polarized calculations within the DFT framework as implemented in the Vienna Ab initio Simulation Package (VASP).[36] The ion–electron interactions were represented by the projector-augmented wave (PAW) method[37] and the electron exchange-correlation by the generalized gradient approximation (GGA) with the Perdew–Burke–Ernzerhof (PBE) exchange-correlation functional.[38] The Kohn–Sham valence states were expanded in a plane-wave basis set with a cutoff energy of 400 eV. The Ce(5s,5p,6s,4f,5d), O(2s,2p), Pd(4d5s), and C(2s,2p) electrons were treated as valence states. We have used the DFT+U approach, in which U is a Hubbard-like term describing the on-site Coulombic interactions.[39] This approach improves the description of localized states in ceria, where standard LDA and GGA functionals fail. For Ce, a value of U = 4.5 eV was adopted, which was calculated self-consistently by Fabris et al.[40] using the linear response approach of Cococcioni and de Gironcoli[41] and which is within the 3.0–6.0 eV range that provides localization of the electrons left upon oxygen removal from ceria.[42] For the calculations of TM-doped CeO2(111) (TM = transition metal = Pd, Pt, Ni, Rh, Cu and Zn), we used a periodic slab with a (4 × 4) supercell in which one of the surface Ce atoms was substituted by one TM atom. The CeO2(111) slab model is three Ce–O–Ce layers thick and the vacuum gap was set to 15 Å. The atoms in the bottom layer were frozen to their bulk position and only the top two Ce–O–Ce layers were allowed to relax. The bulk lattice constant (5.49 Å) as previously calculated using the PBE+U (U = 4.5 eV) functional was used.[43] For the Brillouin zone integration, a Monkhorst–Pack 1 × 1 × 1 mesh was used. To examine the influence of the size of the Monkhorst–Pack grid, we computed the CH4 adsorption energy on a Pd1/Pd-dop-II model at Monkhorst–Pack 1 × 1 × 1 and 2 × 2 × 2 meshes. Both the absolute and CH4 adsorption energies are the same for both mesh sizes. Accordingly, we computed all energies in the Γ-point. The climbing image nudged-elastic band (CI-NEB) algorithm[44,45] was used to identify the transition states for the CH4 activation and dissociation on selected models.

Microkinetic Model

The dissociation of CH4 from a molecularly adsorbed precursor state was described by the following kinetic scheme[23,46]

Here, [O] is a lattice oxygen atom neighboring a Pd atom. To compute the reaction rate of CH4 dissociation, we follow a Langmuir–Hinshelwood type kinetic model,[47] in which we assume that the C–H bond dissociation step is slower than the adsorption/desorption steps of CH4 (i.e., adsorption of CH4 is quasi-equilibrated). CH4 dissociation is usually considered to be irreversible. The equilibrium constant K for molecular CH4 adsorption is given byin which kads and kdes are the adsorption and desorption rate constants, PCH is the methane partial pressure and θCH and θ are the coverage with CH4 and the fraction of empty sites, respectively. The site balance leads to the following expression for the coverage of molecularly adsorbed CH4

The equilibrium constant can be written aswhere ΔG(T,P), Eads, and μ(T,P) are the Gibbs free energy change due to molecular CH4 adsorption, the DFT-computed adsorption energy of molecularly adsorbed CH4 and the chemical potential of gaseous CH4 at the temperature T and pressure P, respectively. The chemical potential of methane was given as belowwhere Pθ is the standard atmospheric pressure, and the enthalpy H and entropy S of methane were obtained from standard thermodynamic tables.[48]

The rate for CH4 dissociation can then be written asin which the rate constant k is computed from the computed activation barrier for C–H bond dissociation in adsorbed CH4 according to

Results and Discussion

Focusing first on Pd, its substitution for Ce in the CeO2 surface is energetically favorable and lowers the energy to remove an oxygen surface atom, which creates an oxygen vacancy (VO).[26,27] While most theoretical studies assume an octahedral Pd coordination arising from the Pd for Ce substitution (retaining the initial octahedral coordination of Ce4+ in the surface), we demonstrated that a stable structure exists with nearly the same energy in which Pd adopts a square-planar configuration.[27] This structure is much more active in CO oxidation than the octahedral structure. The conventional octahedral (Pd-dop-I) and the alternative square-planar (Pd-dop-II) models for Pd-doped CeO2 are shown in Figure a. CH4 adsorption is however weak on these Pd-doped surfaces, resulting in a low rate of C–H bond dissociation,[27] orders of magnitude lower than experimentally observed rates.[7] Given the importance of under-coordinated surface atoms in PdO, we should also consider atomically dispersed Pd atoms on the CeO2(111) surface (Pd1/CeO2(111) in Figure a) as candidate active sites. However, a previous computational study showed that these isolated Pd atoms on CeO2(111) will easily agglomerate into larger clusters via Ostwald ripening.[49] Therefore, we explored an alternative structure in which we added a Pd atom to the single Pd-doped CeO2.

Figure 1

(a) Optimized configurations of Pd atoms dispersed on CeO2(111). (color code: white = Ce; gray = O; red = O neighboring Pd; cyan = Pd); (b) chemical potential μ of Pd atoms dispersed on CeO2(111) (dashed lines) and Pd NPs on CeO2(111) as a function of the nanoparticle radius R (full line). The chemical potential of bulk Pd is set to zero: μbulk = 0.

Figure a also shows the surface obtained by adding a Pd atom to the Pd-dop-II surface model (denoted as Pd1/Pd-dop-II). Adding a Pd atom is favorable (ΔE = −1.66 eV) with respect to bulk Pd, which indicates that the surface Pd atom is very stable in this position. This is further confirmed by the much higher diffusion barrier to an adjacent CeO2 surface site of 3.37 eV (Figure S2) in comparison to the value of 0.14 eV for Pd1/CeO2.[49] When we add a Pd atom to the Pd-dop-I model, the resulting structure spontaneously relaxed to Pd1/Pd-dop-II during geometry optimization in Figure S3. Adding another Pd atom to Pd1/Pd-dop-II is unfavorable (ΔE = +1.51 eV with respect to bulk Pd). Figure b shows the chemical potential of the different models in comparison to the chemical potential of supported Pd metal clusters as a function of their size.[50] While Pd1/CeO2(111) and Pd2/Pd-dop-II are less stable compared to ceria-supported Pd clusters, Pd-dop-II and Pd1/Pd-dop-II are favored over supported Pd clusters and bulk Pd. Pd1/Pd-dop-II is also stable with respect to the O stoichiometry under typical reaction conditions. The removal of surface O atoms bound to the 3-fold and 4-fold Pd atoms from this structure costs 2.48 and 2.72 eV, respectively. Thus, the O atoms are more strongly bound to the Pd atoms than the O atoms in the stoichiometric CeO2 surface.[51] It is also important to mention that adsorption of an O atom on the coordinatively unsaturated Pd atom of Pd1/Pd-dop-II is very weak (−0.10 eV). Taken together, these results imply that Pd1/Pd-dop-II is the thermodynamically expected structure under typical oxidative reaction conditions. In essence, the structure is the result of the substitution of one Ce4+ ion by two Pd2+ ions, which is in line with the much larger radius of Ce4+ (0.97 Å) compared to Pd2+ (0.64 Å).[33,52−54] The ionic radii are shown in Table , while Figure highlights the possibility to replace a Ce4+ ion by two Pd2+ ions. Table also shows that other transition metals can lead to similar structure and this topic will be discussed below. For the Pd case, one of the Pd atoms is coordinated by four lattice O atoms, while the other one is coordinated by three O atoms. We emphasize that this structure is very different from earlier proposed doping models.[24−27,31,35] The unusual aspect of the novel model is that, unlike earlier models, it contains a stable three-coordinated Pd ion, which we expect to strongly adsorb CH4. Therefore, we explored in the following the adsorption and activation of CH4 on the Pd1/Pd-dop-II structure.

Table 1

Ionic Radii of Various Metal Ions

metal	cerium		nickel		palladium		platinum		rhodium			copper		zinc
ion	Ce3+	Ce4+	Ni2+	Ni4+	Pd2+	Pd4+	Pt2+	Pt4+	Rh2+	Rh3+	Rh4+	Cu+	Cu2+	Zn2+
R (Å)	1.04	0.97	0.69	0.48	0.64	0.63	0.80	0.62	0.72	0.67	0.615	0.46	0.62	0.74

Figure 2

Illustration of the replacement of (left) one lattice Ce4+ ion by (right) two Pd2+ ions (ionic radii of Ce4+ and Pd2+ used).

CH4 adsorption and dissociation are the crucial steps in the CH4 combustion process.[23,55] The adsorption energy of CH4 on the Pd1/Pd-dop-II model is −0.59 eV, which is slightly higher than the value of −0.47 eV computed for the PdO(101) surface. CH4 adsorbs only weakly on Pd-dop-II. The adsorption energy of −0.15 eV is nearly similar to the adsorption energy on the stoichiometric CeO2 surface.[27] From this difference, we infer that the presence of a coordinated unsaturated Pd2+ ion leads to stronger adsorption. Figure a shows the initial adsorption structure of CH4 as well as the transition and final states for its dissociation into CH3 and H, resulting in Pd-CH3 and OH fragments. The potential energy surfaces for CH4 adsorption and dissociation of CH4 on this and other models are depicted in Figure b. The activation barrier for CH4 dissociation on Pd1/Pd-dop-II is 0.60 eV. The adsorption energies of CH4 on most other models are significantly lower than for Pd1/Pd-dop-II with comparable or higher barriers (e.g., 0.64 eV for PdO(101) and 0.70 eV for Pd1/CeO2(111)). The exception is Pd-dop-II, on which CH4 can be activated with an energy barrier of only 0.27 eV. As discussed before, the low barrier is due to the high activity of the surface O radical.[27] The Pd-dop-II structure is however not stable, and the reactive O atom is spontaneously removed to form the thermodynamically much more stable Pd-dop–II-VO structure. The activation energy of CH4 dissociation on this stable surface is 0.99 eV.[27]

Figure 3

(a) Initial, transition, and final states of CH4 dissociation by the Pd1/Pd-dop-II structure; (b) energy profiles of CH4 adsorption and dissociation by various Pd-containing models; (c) computed CH4 dissociation rates (T = 300–1000 K, PCH = 0.1 atm).

In order to compare C–H bond dissociation for the different surface models, we considered a Langmuir–Hinshelwood kinetic scheme in which gaseous CH4 is quasi-equilibrated with adsorbed CH4, followed by the slow C–H bond dissociation step, resulting in Pd-CH3 (CH3*) and O–H (H*) species.[23] This kinetic model together with the computed DFT-energetics predicts that CH4 dissociation proceeds with the highest rate on Pd1/Pd-dop-II and then decreases in the order Pd-dop-II > PdO(101) > Pd(211) > Pd-dop-II-VO (Figure c). Although the activation barrier for the C–H bond dissociation on Pd-dop-II is the lowest among the models investigated, the weak adsorption of CH4 results in a very low CH4 coverage and, henceforth, a low overall reaction rate. The thermodynamically preferred structure Pd-dop-II-VO has the lowest overall reaction rate.

We compare these data to the work of Janik’s group on Pd/CeO2 for CH4 activation, who studied doping of ceria with multiple Pd atoms.[26,31] Considering only Pd-doping of ceria, a higher activity than PdO(101) is only predicted for their (111)-1Pd4+/2Pd2+ structure.[26] However, according to their ab initio phase diagram this structure is not stable under relevant reaction conditions, while the most stable (111)-3Pd2+ structure exhibits a more than 2 orders of magnitude lower activity than PdO(101). We also note that the C–H bond distance in the transition states for CH4 dissociation presented by Janik and co-workers is longer than 2 Å. This distance is too long to represent a C–H bond as one would expect to exist in the transition state for C–H bond dissociation of CH4. For instance, the C–H distance in the transition states of CH4 dissociation shown in Figure a is around 1.4 Å, close to values reported in the literature.[22,23,27,30] Detailed configurations and the potential energy surface for CH4 dissociation on Pd1/Pd-dop-II are shown in Figure S6. Janik and co-workers also investigated PdO clusters supported on Pd-doped ceria.[31] The computed reactivity of these models for CH4 dissociation are much lower than that of PdO(101).[31] Thus, we can firmly conclude that our proposed model composed of two Pd2+ ions in the CeO2(111) surface as a model for a solid Pd-CeO2 solution can provide a good explanation for the experimentally observed high CH4 dissociation activity after removal of PdO from a catalyst containing PdO clusters on a Pd-CeO2 solid solution.[7] In essence, the coordinatively unsaturated Pd cation in Pd1/Pd-dop-II is highly reactive toward methane’s C–H bonds due to the formation of a strong σ-complex similar to the complex proposed for CH4 adsorption on IrO2(110).[23]

We compared in more detail the (electronic) structure of CH4 adsorbed on Pd1/Pd-dop-II and PdO(101). In both surfaces, the Pd cation is 3-fold coordinated (Figure ). Charge analysis shows that CH4 adsorption on Pd1/Pd-dop-II leads to an increase of the Pd cation charge, while the reverse holds for the Pd cation in PdO(101) (Table S3). The latter is in line with an earlier computational study.[11] Electron density difference plots before and after CH4 adsorption on the two surfaces in Figure show that there is a redistribution of electron density between CH4 and the surface. Moreover, there is clear evidence for the formation of a σ-complex in Pd1/Pd-dop-II. A density-of-state analysis indicates that 4d-electrons from the Pd surface cation in Pd1/Pd-dop-II effectively overlap with the 1t2 frontier molecular orbitals of CH4 and henceforth strengthen the adsorption of CH4 (Figure S8). This type of interaction is absent in the CH4 adsorption complex with PdO(101). Figure also shows that the local coordination environment around CH4 adsorbed on Pd1/Pd-dop-II is more favorable for the formation of an O–H bond during C–H bond cleavage. Notably, the C–H bond that will finally dissociate is elongated more in the CH4 adsorption complex with Pd1/Pd-dop-II (1.13 Å) than in the corresponding complex with PdO(101) (1.09 Å). This result is important as it shows that a Pd cation at the interface with Pd-doped CeO2 can strongly adsorb and activate CH4.

Figure 4

(a,b) Adsorption and transition states of CH4 by Pd1/Pd-dop-II. (c,d) Adsorption and transition states of CH4 by PdO(101). (e,f) Electron density difference contours of CH4 adsorption on Pd1/Pd-dop-II and PdO(101). (Color code: cyan = Pd; red = O; off-white = Ce; gray = C; white = H).

We investigated the feasibility of complete CH4 oxidation on the Pd1/Pd-dop-II model. Figure shows the potential energy diagram for the complete catalytic cycle. After CH4 dissociation, the resulting CH3* species can be further dehydrogenated to Pd-CH2 (CH2*) species and another O–H species (Int2 → Int3). This process is endothermic by 0.55 eV and requires overcoming an activation barrier of 1.05 eV. The CH2 species on the 3-fold Pd atom will migrate to a neighboring lattice O atom to form a CH2O species (ΔE = −1.79 eV) with a barrier of only 0.48 eV. Next, an O2 molecule will adsorb on the 3-fold Pd atom (Int4 → Int5, Eads = −0.86 eV). The reaction then further proceeds via a sequence of facile H-transfer steps on the surface forming OOH species that are involved in C–H bond dissociation and H2O formation. The third C–H bond dissociation step in CH2O is highly exothermic (ΔE = −3.09 eV). After H2O desorption (Int7 → Int8), rotation of OCHO occurs, enabling further O2 adsorption (Int9 → Int10), C–H bond cleavage and formation of CO2. CO2 desorption costs 0.52 eV, leaving one oxygen vacancy (Int11 → Int12). Dissociation of OOH heals this vacancy (Int12 → Int13), while the other OH fragment reacts with the remaining H atom to form another H2O molecule (Int13 → Int14). H2O desorption (ΔE = +1.55 eV) completes the catalytic cycle.

Figure 5

Energy diagram for the reaction CH4(g) + 2 O2(g) → CO2(g) + 2 H2O(g) on Pd1/Pd-dop-II. The structure of selected intermediates (Int) are depicted (all structures in the Supporting Information). The active site * is the 3-fold Pd atom of Pd1/Pd-dop-II; [O] represents a surface lattice O atom involved in the reaction; VO represents the oxygen vacancy. (color code: white = Ce; gray = O; red = lattice O neighboring Pd or O of molecular O2, H2O or CO2; cyan = Pd; blue = H).

Overall, the reaction pathway for complete CH4 oxidation for the novel Pd1/Pd-dop-II model appears to be feasible. Weaver and co-workers reported that the formation of the CH2 intermediate is the most difficult step (Ea = 1.44 eV) for CH4 oxidation on PdO(101).[21] The highest barrier for our model is significantly lower (Ea = 1.05 eV). Taking into account entropy, we computed that the rate of this C–H bond activation step at 623 K is still 2 orders of magnitude higher than the overall process of adsorption and dissociation of CH4. The reaction energy diagram also emphasizes the influence of competitive adsorption of O2 and H2O. This will shift the operating window of this catalyst to a temperature regime where vacant Pd sites are available. We can compare the O2 and H2O adsorption energies of, respectively, −0.82 and −1.55 eV to those for PdO(101), i.e. −1.58 and −1.01 eV, respectively. Thus, while O2 inhibition is alleviated in our model with respect to PdO(101), competition with H2O is more prominent. Based on the gas phase entropy of H2O, we can predict that the free energy of H2O desorption is lower than the highest barrier at around 500 K. The occurrence of these competitive effects has been experimentally demonstrated by Weaver and co-workers in a TPD study for PdO(101).[56] The systematic study for PdO(101) by Bossche and Gronbeck showed that CH4 oxidation is inhibited by molecular H2O adsorbed on under-coordinated Pd sites at low temperature.[19] Summarizing, Pd1/Pd-dop-II provides a model on which CH4 can be activated in a facile manner and a complete reaction cycle leading to CO2 and H2O is possible.

Encouraged by these insights, we also investigated the effect of modification of CeO2 with Ni, Pt, Rh, Cu, and Zn, because the former three give rise to active CH4 combustion catalysts in solid solutions,[17,29,30] while the latter two are expected to be low active catalysts. The substitution of all of these transition metals in the CeO2(111) surface is favorable against the bulk of the corresponding transition metal (Table ). Except for Pt, these transition metals adopt the same square-planar configuration as Pd (Figure S11).[27] The most stable configuration for Pt is the octahedral one, which is likely related to the larger size of Pt in comparison with the other substituents (Table S2). Placement of another like transition metal atom on the doped site is also exothermic, showing the generality of the stabilization of transition metal atoms on doped CeO2. Next, we investigated CH4 adsorption on these structures. The adsorption energy is highest on Pt1/Pt-dop and decreases in the order Pt1/Pt-dop > Pd1/Pd-dop > Rh1/Rh-dop > Ni1/Ni-dop ≈ Cu1/Cu-dop > Zn1/Zn-dop. The electron density difference plots for CH4 adsorbed on Pt1/Pt-dop and Zn1/Zn-dop in Figure S13 show an effective σ-complex formation for the Pt case similar to Pd, while it is absent for Zn. The activation barrier for C–H bond dissociation follows roughly the reverse trend with Zn1/Zn-dop having the highest activation barrier and Pt1/Pt-dop the lowest.

Table 2

Insertion Energy of the First TM Atom into One Lattice Ce Position Atom (Eads-TM1) and the Second TM Atom (Eads-TM2) with Respect to Corresponding TM Bulk Atomsa

TM1/TM-dop-II	Ni	Pd	Pt	Cu	Rh	Zn
Eads (TM1, eV)	–4.79	–3.62	–4.04	–4.20	–4.24	–4.79
Eads (TM2, eV)	–2.03	–1.66	–0.61	–1.22	–1.21	–3.65
Eads (CH4, eV)	–0.42	–0.59	–0.81	–0.41	–0.46	–0.17
Ea (eV)	0.62	0.60	0.49	0.75	0.52	1.10
ΔE (eV)	–0.42	–0.49	–0.77	–0.70	–0.55	0.52
r (molecules·site–1·s–1)	1.41 × 100	5.57 × 101	3.30 × 104	9.42 × 10–2	2.12 × 101	1.04 × 10–6

Adsorption energy Eads, activation energy Ea, reaction energy ΔE of CH4 dehydrogenation, and the rate of CH4 dissociation by TM1/TM-dop-II. PCH = 0.1 atm, and T = 623 K.

Using the kinetic model for CH4 adsorption and dissociation, we found that the computed rates decreases in the order Pt1/Pt-dop ≫ Pd1/Pd-dop > Rh1/Rh-dop > Ni1/Ni-dop > Cu1/Cu-dop > Zn1/Zn-dop in Figure a. At a temperature of 623 K and a CH4 pressure of 0.1 atm, the dissociation rate is higher for the Pt-doped structure than the Pd-doped one. Figure b emphasizes the strong correlation between the activation barrier and the distance between the surface and CH4 in the adsorbed state. The decreasing activity at high temperature is because the free energy for desorption is higher than the activation barrier for CH4 dissociation. The free energy change for CH4 adsorption is given by Δads – TΔads, in which Δads and TΔads represent the enthalpy and entropy contributions. The activation free energy barrier is roughly equal to the activation barrier, as the entropy change during CH4 dissociation starting from the σ-complex is negligible in comparison with the entropy change during adsorption or desorption. The predicted high activity computed for Pt1/Pt-dop-II provides a good explanation for the experimentally reported high activity of a Ce1–PtO2-σ solid solution.[28,57] In a similar manner, Ni- and Rh-promoted CeO2 catalysts have also been noted for their promising activity in CH4 combustion.[58−62]

Figure 6

(a) Computed CH4 dissociation rates for various TM1/TM-dop-II structures as a function of temperature with PCH = 0.1 atm and (b) relation between CH4 dissociation barrier and the distance between the reactive TM atom and the H atom of the activated C–H bond in CH4 in the adsorbed state.

Conclusions

We investigated a novel structure of a solid solution of Pd in CeO2 with the purpose of explaining the high activity of Pd-CeO2 solid solutions toward CH4 activation. We show that two Pd2+ ions can substitute one Ce4+ ion in the stable CeO2(111) surface, resulting in a structure that is stable under oxidative conditions. CH4 will strongly adsorb as a σ-complex on the Pd cation that is coordinatively unsaturated. The CH4 adsorption energy is higher on this novel structure than on PdO(101). Consequently, the activation barrier for dissociation of the adsorbed CH4 molecule is lower for Pd1/Pd-dop-II. Kinetic simulations show that CH4 dissociation proceeds with the highest rate on this structure. We also show that similar structures can be obtained by doping the CeO2(111) surface with Pt, Ni, Rh, Cu, and Zn. Specifically, the more reactive transition metals Pt, Ni, and Rh can lead to strong CH4 adsorption complexes, low C–H activation barriers, and a high CH4 dissociation activity. The concept of substituting two transition metal ions for one Ce4+ ion in ceria is important as it results in a very stable structure containing a highly reactive coordinatively unsaturated transition metal. We expect that this insight opens up new possibilities to rationally design active and stable catalysts of surface doped oxides.