Understanding the Impact of Defects on Catalytic CO Oxidation of LaFeO3-Supported Rh, Pd, and Pt Single-Atom Catalysts.

Understanding the intrinsic catalytic properties of perovskite materials can accelerate the development of highly active and abundant complex oxide catalysts. Here, we performed a first-principles density functional theory study combined with a microkinetics analysis to comprehensively investigate the influence of defects on catalytic CO oxidation of LaFeO3 catalysts containing single atoms of Rh, Pd, and Pt. La defects and subsurface O vacancies considerably affect the local electronic structure of these single atoms adsorbed at the surface or replacing Fe in the surface of the perovskite. As a consequence, not only the stability of the introduced single atoms is enhanced but also the CO and O2 adsorption energies are modified. This also affects the barriers for CO oxidation. Uniquely, we find that the presence of La defects results in a much higher CO oxidation rate for the doped perovskite surface. A linear correlation between the activation barrier for CO oxidation and the surface O vacancy formation energy for these models is identified. Additionally, the presence of subsurface O vacancies only slightly promotes CO oxidation on the LaFeO3 surface with an adsorbed Rh atom. Our findings suggest that the introduction of La defects in LaFeO3-based environmental catalysts could be a promising strategy toward improved oxidation performance. The insights revealed herein guide the design of the perovskite-based three-way catalyst through compositional variation.

Introduction

The development of new low-cost and highly active catalysts for automotive emission control is of great importance to the protection of the environment. Among practical solutions, supported platinum group metals (PGMs) are widely used to catalyze reactions to abate the emission of noxious gases from gasoline engines.[1−6] The cost of PGMs and their low-temperature performance as well as their aggregation into larger particles when operated at elevated temperatures pose significant challenges in developing better catalysts.[7,8] These aspects explain why already for a long time substantial efforts have been made to reduce the content of precious metals in these catalysts.[9−15] In the extreme case, catalysts contain these precious metals in an atomically dispersed form, which in recent years has become a topic of frontier research.[16−22]

Another solution to the problem of costly precious metals is to replace them by a cheaper class of materials. Recently, perovskite (ABO3 type) materials have been explored as a potential replacement for conventional three-way catalysts.[23−26] As in general these materials are not active enough, one strategy is to promote these base metal-based oxides with precious metals. This only makes sense when these are present in high dispersion. Early studies indicated that the noble metals can retain good dispersion on perovskite surfaces.[27,28] For instance, LaFeO3 (LFO) perovskites exhibit excellent resistance against sintering of precious metal overlayers.[29] The most prominent reaction investigated in this respect is CO oxidation.[30−32] Previous studies demonstrated that the physical and chemical and, accordingly, catalytic properties of perovskites can be tuned by partial substitution of the A and B sites with transition-metal ions.[32−36] For example, Pd doping of LFO can strongly improve the CO oxidation activity.[37] Another recent experiment highlighted that the catalytic activity of CO oxidation on LFO-supported Rh and Cu catalysts was enhanced by the presence of La defects.[38] Onn and co-workers reported about a high-surface-area LFO-supported Pd catalyst obtained by atomic layer deposition with good catalytic performance.[39]

There are very few studies that focus on the underlying chemical aspects of the surface reactivity of perovskites by computational chemistry. Some studies have dealt with the stability of transition-metal atoms and adsorption of simple molecules on these complex oxide surfaces.[40−43] This represents a lack of the fundamentals of the reaction mechanism and the relation with the electronic structure that is modified by aspects such as common defects and hinders the guided development of more active systems.

In this study, we aim for this kind of understanding by employing density functional theory (DFT) in combination with microkinetics analysis to understand the chemical and catalytic properties of single Rh, Pd, and Pt atoms dispersed on or in the stable (001) surface of LFO with a view on CO oxidation catalysis. We investigate the stability, CO and O2 adsorption energies, and CO oxidation activation barriers for stoichiometric and defective surfaces. Especially, La and O defects can improve the stability of the introduction of precious metal atoms and tune the electronic structure toward high CO oxidation activity. A general finding is that La defects can lead to highly active CO oxidation catalysts based on the three explored precious metals, while the presence of subsurface O vacancies boosts the performance of single Rh atoms adsorbed to the LFO surface.

Computational Methods

DFT Calculations

We employed spin-polarized DFT calculations using the projector-augmented wave[44] approach as implemented in VASP 5.3.5.[45,46] To account for the effect of the exchange–correlation and on-site Coulomb interaction, the Perdew–Burke–Ernzerhof[47] functional with the Hubbard + U correction was used. We set Ueff to 4 eV for the Fe atoms. In order to accurately describe the electronic structure of LFO, we consider its G-type antiferromagnetic structure. To understand the catalytic properties, we constructed the FeO2-terminated (001) surface, which is known to be a stable and exposed facet of this type of materials in CO oxidation.[48] In order to study the influence of noble metals on the catalytic performance, we considered doping of the surface by replacing a surface Fe atom by an M atom with M being Rh, Pd, or Pt as well as the adsorption of a single M atom on the surface of the LFO slab model.

Different adsorption sites were considered for the latter purpose. Here, the constructed slab models are 2 × 2 unit cells with six atomic layers and a slab thickness of 9.93 Å. The top four layers were relaxed and the bottom two layers were frozen to the configuration of the bulk. A vacuum thickness of 12 Å was used to avoid spurious interactions of adsorbates between neighboring super cells. For the Brillouin-zone integration, a 1 × 1 × 1 Monkhorst–Pack k-point was used. The geometry optimizations were assumed converged when the Hellmann–Feynman forces acting on atoms were less than 0.05 eV/Å. Further, computational details including a discussion of the magnetic structure of the LFO models are provided in the Supporting Information.

The formation energies of M atom-doped LFO surfaces are defined asHerein, EM+surf, Esurf, EFe, and EM are the electronic energies of the M-doped LFO surface, the empty LFO surface, the bulk Fe atom, and the bulk M atom, respectively.

Here, the adsorption energies are calculated bywhere Em+surf and Em are the electronic energies of the adsorbed system and the corresponding gas-phase species, respectively.

Microkinetics Simulations

On the basis of the DFT calculations, we carried out microkinetics simulations to gain the kinetic properties for CO oxidation on LFO-supported Rh, Pd, and Pt single-atom catalysts. For instance, the reaction rate, the rate-determining step, and species distribution are investigated. For surface reactions, the calculated activation barriers are used to estimate the forward and backward rate constants using the Eyring equationwhere k represents the reaction rate constant, kb and h are Boltzmann and Planck’s constants, respectively, T is the temperature (in K), and Ea refers to the electronic activation energy (in J). fTS and f indicates the partition functions of the transition state and the ground state, respectively. We assumed that all vibrational partition functions equal unity which gives a prefactor for all surface elementary reaction steps of ∼1013 s–1.

In the case of adsorption reactions, there is an assumption that the molecule loses only one of its translational degrees of freedom with respect to the gas phase. Accordingly, the rate of molecular adsorption was calculated asHerein, P indicates the pressure of the adsorbate in the gas phase, A′ is the surface area of the adsorption site, m is the mass of the adsorbate, and S refers to dimensionless sticking coefficient.

For the CO2 desorption process, we assumed that the activated species has two translational and three rotational degrees of freedom. Therefore, the molecular desorption rate is defined as[49]where σ refers to the symmetry number, θ is the characteristic temperature for rotation, and Edes is the desorption energy.

The details for the microkinetics modeling have been described in our previous works.[50,51] To calculate the reaction rate of all surface reaction intermediates, we constructed the differential equations by using the rate constants of the elementary reaction steps. For each of the X components presented in the reaction network, a single differential equation is defined as

In the above equation, k is the elementary reaction rate constant, ν indicates the stoichiometric coefficient of component i in elementary reaction step k, and c refers to the concentration of component k on catalyst surface.

To identify the rate-determining step of the CO oxidation reaction, Campbell’s degree of rate control (DRC) analysis[52,53] was employed. For a specific elementary step i, the coefficient of DRC (χRC,) is determined byHerein, r refers to the overall reaction rate and k and K indicate the forward rate and the equilibrium constants for step i, respectively.

The first-principles-based microkinetics simulations were carried out using the MKMCXX code.[54] The overall conversion rates of CO oxidation, steady-state coverages, and products distribution were calculated as a function of temperature by integrating the ordinary differential equations with respect to time using the backward differentiation formula method.[55−57]

Results and Discussion

Geometry and Stability of LFO-Supported Rh, Pd, and Pt Single-Atom Catalysts

We first investigated the structure and stability of Rh-, Pd-, and Pt-doped LFO surfaces. Figure a,b shows the structures of the stoichiometric doped LFO surfaces and the same surfaces with VLa. The M atom-doped stoichiometric LFO and VLa-promoted LFO structures are denoted in the following as LFO-M and LFO-VLa-M, respectively. The structures of the corresponding models with VSO are denoted as LFO-VSO-M (Figure S2). We compare the stabilities of these structures by determining their formation energies. To this end, we compute the exchange energy of the stoichiometric LFO for replacing an Fe atom with an M atom compared to bulk Fe and M. From Table , it can be seen that the formation energies of the stoichiometric LFO-Rh, LFO-Pd, and LFO-Pt models are 0.09, 1.41, and 0.79 eV, respectively. Doping Rh, Pd, and Pt in an LFO surface with a La defect is relatively speaking more favorable than doping in the stoichiometric LFO surface.

Figure 1

Top view of the different LFO surfaces used in the DFT modeling: (a) LFO-M, (b) LFO-VLa-M, (c) Mads-LFO, and (d) Mads-LFO-VLa (color code: light blue, La; lavender, Fe; O, red; blue, M = Rh, Pd, and Pt).

Table 1

Stabilities of the Different Surface Models Used for the DFT Calculationsa

	formation energy (eV)			adsorption energy (eV)
model	LFO-M	LFO-VLa-M	LFO-VSO-M	Mads-LFO	Mads-LFO-VLa	M-LFO-VSO
Rh	0.09	–1.44	–0.60	2.26	–2.69	0.71
Pd	1.41	0.30	–2.16	1.94	–2.58	0.70
Pt	0.79	–0.31	–1.39	2.40	–2.48	1.06

For the doped models, formation energies are reported; for the models consisting of a noble metal adsorbed on the LFO surface, the adsorption energies are listed.

It is worth noting that the most stable position of an O vacancy is at the surface. We found, however, that upon doping the LFO surface by one of the noble metals considered in this study, the O vacancy prefers to be located in a subsurface position, subjacent to the M dopant. The O vacancy formation energies in these locations are lower than those associated with creating an O vacancy in the surface itself (Table ).

Table 2

Calculated Oxygen Vacancy Formation Energy (eV) for the Different Surface Modelsa

model	LFO-M	LFO-VLa-M	LFO-VSO-M	Mads-LFO	Mads-LFO-VLa	Mads-LFO-VSO
Rh	2.42 (2.10)	1.00	2.72	3.02	2.82	3.07
Pd	2.28 (1.53)	0.41	2.62	2.40	2.14	2.49
Pt	2.47 (1.56)	0.73	2.77	2.31	2.65	2.42

Values in parenthesis are corresponding subsurface oxygen formation energies.

Accordingly, we used a model for LFO-VSO-M which includes the subsurface O vacancy to investigate stability and catalytic properties. Compared with LFO-M, the formation energies for LFO-VSO-Rh, LFO-VSO-Pd, and LFO-VSO-Pt are lower, that is, −0.60, −2.16, and −1.39 eV, respectively. These data confirm the expected trend that creating oxygen vacancies in doped LFO structures is more facile because the noble metal dopants are less reactive than Fe.

For the adsorption structures, we find a preference for M adsorption on the hollow site of the LFO surface (Figures c,d and S2). We denote the adsorption of M on different LFO surfaces as Mads-LFO, Mads-LFO-VLa, and Mads-LFO-VSO. The stability of the different models is compared on the basis of the adsorption energy. Table shows that VLa enhances strongly the adsorption energies of Rh (−2.69 eV), Pd (−2.58 eV), and Pt (−2.48 eV) compared with the stoichiometric LFO surface. The data also show that the presence of an O vacancy also results in lower adsorption energies of Rh (0.71 eV), Pd (0.70 eV), and Pt (1.06 eV) in comparison to the stoichiometric surface. Thus, we can conclude that La and O defects increase the stability of isolated noble metal atoms at the LFO surface.

To understand the electronic nature of these LFO surfaces, we analyzed their partial density of states (PDOS). We focus in Figure on the bonding of the different stoichiometric and defective LFO surfaces with an Rh atom. The bonding is governed by hybridization of Rh 4d and O 2p orbitals. The O 2s orbitals lie too low in energy to be involved in the bonding. In all cases, the overlap between Rh 4d and O 2p orbitals is higher for Rh-doped LFO structures than for Rh adsorbed on LFO. For the LFO-VLa surface, the calculated Bader charges of the Rh atom in LFO-VLa-Rh and Rhads-LFO-VLa are +1.85|e| and +1.31|e|, respectively, consistent with the lower occupancy of the 4d states. Thus, we can say that Rh transfers effectively electrons to O 2p orbitals. For LFO-VSO-Rh and Rhads-LFO-VSO, the Rh atom and the nearest O atoms form a stable square-planar configuration. Figures S3 and S4 confirm that similar trends occur for Pd and Pt doping in LFO-VLa. We note however that the overlap for the d orbitals of Pd and Pt is smaller than for Rh, explaining the generally observed higher stability of Rh-doped structures.

Figure 2

PDOS of Rh 4d, O 2p, and O 2s orbitals (the O atom adjacent to Rh) in (a) LFO-Rh, (b) Rhads-LFO, (c) LFO-VLa-Rh, (d) Rhads-LFO-VLa, (e) LFO-VSO-Rh, and (f) Rhads-LFO-VSO.

Adsorption of CO and O2

Before we explore the reaction mechanism of CO oxidation, we first investigate the adsorption properties of CO and O2 on these surfaces. All relevant data are collected in Tables S2 and S3. Our calculations indicate that the stoichiometric LFO surface cannot bind CO or O2. Therefore, we predict that the stoichiometric surface should present a low activity. This is in line with the early experimental result.[50] Notably, the surface O vacancy formation energy for the LFO surface is as high as 2.18 eV, which implies that spontaneous formation of surface O vacancies will only take place at relatively high temperature. Accordingly, we did not consider CO adsorption on such vacancies. For LFO-Rh, the CO adsorption energy is −2.14 eV, which is stronger than for CO adsorption on LFO-Pd (−0.53 eV) and LFO-Pt (−1.38 eV). The stronger CO adsorption on Rh can be correlated to the enhanced donation of CO electron density into the more unoccupied Rh d-orbitals. In the catalytic mechanism to be discussed below, the adsorbed CO molecule reacts with a surface O atom to form CO2. This constitutes an M–vK mechanism. Molecular oxygen will then adsorb on the created O vacancy. Its adsorption energy on the defect sites of LFO-Rh, LFO-Pd, and LFO-Pt are −2.09, −1.53, and −2.17 eV, respectively. We find that the strength of O2 adsorption correlates strongly to the surface O vacancy formation energy as follows from the computed surface O vacancy formation energies for LFO-Rh (2.42 eV), LFO-Pd (2.28 eV), and LFO-Pt (2.47 eV).

We also considered the effect of VLa on the CO and O2 adsorption energies for the LFO-M surfaces. The presence of a La defect results in a substantial decrease of the CO adsorption strength on the Rh-doped surface and a decrease of the O2 adsorption strength on the O defect obtained after one catalytic CO2 formation cycle. On the other hand, the CO adsorption strengths on LFO-VLa-Pd and LFO-VLa-Pt are enhanced with respect to the corresponding stoichiometric surfaces. The O2 adsorption strengths are however decreased, similar to the Rh case. The different influence of a La defect on the CO adsorption strength for Rh on the one hand and Pt and Pd on the other has to do with the degree of hybridization between the CO and Rh 4d orbitals. For Rh, this leads to a lower hybridization, lowering the CO adsorption strength. For Pd and Pt, the shift of the relevant d orbitals due to the La defect causes a stronger hybridization with the CO orbitals. Finally, we determined that the presence of a subsurface O vacancy upon Rh doping in LFO-VSO decreases both the CO and O2 adsorption energies. Neither Pd- nor Pt-doped LFO-VSO can bind CO or O2. The reason is that the square-planar coordination of Pt and Pd with O atoms precludes CO adsorption, which is not the case for Rh. We also investigated CO and O2 adsorption on Rhads-LFO. CO adsorbs weakly on Rhads-LFO (−0.99 eV) than on Rhads-LFO-VLa (−1.30 eV), while O2 adsorption is slightly weakened for Rhads-LFO-VLa compared to its stoichiometric counterpart. CO does not adsorb on the Pdads-LFO and Ptads-LFO surfaces. The main reason for this is that the Pd and Pt atoms are coordinatively saturated in their square-planar geometry.

In summary, the above results show that the presence of La and subsurface O vacancies has a profound effect on the CO and O2 adsorption energies. We expect that the CO oxidation rate should benefit from intermediate adsorption strengths of the reactants (Sabatier principle). To investigate this in detail, we determined the complete potential energy diagrams for the CO oxidation on the various LFO surfaces. To this end, we explored all relevant stable and transition states. We also included CO oxidation involving lattice O atoms, as experimental results indicate that the structural O atoms of transition metal-substituted LFO participate in CO2 formation.[58] Moreover, the modeling of adsorption complexes showed that co-adsorption of CO and O2 is unfavorable on the modified LFO surfaces explored in this work. Accordingly, a Langmuir–Hinshelwood reaction mechanism is less likely.

CO Oxidation

Figures and 4 show the potential energy diagrams for the Rh-, Pd-, and Pt-containing LFO surfaces. For the stoichiometric doped LFO surfaces, the activation barriers for the reaction of adsorbed CO with a lattice O to CO2 are 1.38 eV (Rh), 0.48 eV (Pd), and 0.52 eV (Pt). CO2 desorption energies for these three cases are 0.70, 0.05, and 0.78 eV, respectively. After O2 adsorption on the O vacancy (see above), another CO molecule will adsorb strongly on LFO-Rh (−1.97 eV) and LFO-Pt (−1.62 eV), but relatively weakly on LFO-Pd (−0.52 eV). The adsorbed CO molecule moves toward adsorbed O2, which leads to CO2 formation; the barriers for this process are 2.33 and 2.22 eV for LFO-Rh and LFO-Pt, respectively, and are moderate for LFO-Pd (0.58 eV).

Figure 3

Potential energy diagrams for CO oxidation on (a) LFO-Rh, LFO-VLa-Rh, and LFO-VSO-Rh, (b) Rhads-LFO and Rhads-LFO-VLa, and (c) corresponding geometries for CO oxidation on the different models. The structures of stable and transition states are indicated by corresponding numbers in the different panels.

Figure 4

Potential energy diagrams for CO oxidation on (a) LFO-Pd and LFO-VLa-Pd and (b) LFO-Pt and LFO-VLa-Pt. (c) Corresponding geometries for CO oxidation on the different models. The structures of stable and transition states are indicated by the corresponding numbers in the different panels.

As CO and O2 adsorption energies change due to the presence of La and subsurface O defects, we also investigated their influence on the potential energy diagrams. The overall barrier for CO oxidation on LFO-VLa-Rh is significantly lower (0.84 eV) compared to the value 1.38 eV for the stoichiometric LFO-Rh surface. Similar differences are found for LFO-VLa-Pd and LFO-VLa-Pt (Figure ). In particular, CO oxidation on LFO-VLa-Pd involved a very low overall activation barrier of 0.42 eV. Importantly, we identified a linear correlation between the activation barrier for CO oxidation to CO2 and the surface O vacancy formation energy for the LFO surfaces with a La defect (Figure ). This strongly indicates that the surface O vacancy formation energy is a good descriptor for CO oxidation on transition metal-doped LFO-VLa catalysts. An experimental indication of the benefit of La defects can be found in the work on Rh- and Cu-doped LFO.[39] The trend observed in Figure , therefore, might be useful to predict new nonstoichiometric LFO perovskite-based catalysts. On the other hand, we admit that such correlations were not observed for the stoichiometric and oxygen-defective models.

Figure 5

Activation barriers of CO oxidation on LFO-VLa-M (M = Rh, Pd, and Pt) are linearly related to the corresponding surface O vacancy formation energies. (Blue dot: first half, CO oxidation; red square: second half, CO oxidation.)

We also investigated CO oxidation on the LFO-VSO-M models that contain a subsurface O vacancy. The activation barrier for the first CO2 formation step for M = Rh is 0.54 eV, which is substantially lower than that for LFO-Rh. CO2 desorption from LFO-VSO-Rh costs 1.10 eV, the high value being ascribed to the high energy for surface O vacancy formation on this surface. The second CO2 formation step for LFO-VSO-Rh is 2.11 eV. Even though this barrier is lower than the barrier of 2.54 eV for LFO-Rh, it is too high to be relevant for catalytic CO oxidation. For LFO-VSO-Pd and LFO-VSO-Pt, CO oxidation is less likely to proceed on these surfaces because of the very weak adsorption of CO.

The CO oxidation potential energy diagrams for the single Rh, Pd, and Pt atoms adsorbed on the LFO surfaces are quite different. CO adsorbs on Rh of Rhads-LFO relatively weakly compared to LFO-Rh, which facilitates the migration of CO during the CO2 formation step. Indeed, the barrier for this step is only 0.93 eV, substantially lower than the value of 1.38 eV for LFO-Rh. The relatively high CO2 desorption energy (0.64 eV) is consistent with the high surface O vacancy formation (3.02 eV) for stoichiometric Rhads-LFO. After adsorption of O2 and CO, the following CO2 formation step needs to overcome a barrier of 1.32 eV. In contrast to LFO-Rh, Rhads-LFO exhibits a lower overall activation barrier and CO2 desorption energy. It is noteworthy that the square-planar local structures formed on single Pd and Pt atoms adsorbed on the LFO surfaces preclude the simultaneous binding of CO and O2.

Next we study the influence of VLa on the catalytic properties of a Rh atom adsorbed LFO. As shown in Figure b, the first CO2 formation involves a barrier of 0.35 eV. The second CO2 formation step on this surface also exhibits a relatively low activation barrier (0.98 eV) compared to the corresponding barrier of 1.32 eV for the stoichiometric Rhads-LFO. As to the Pdads-LFO-VLa and Ptads-LFO-VLa, we find that also these surfaces cannot catalyze CO oxidation due to the very weak adsorption of CO and O2.

Microkinetics Modeling

To systematically evaluate the catalytic activities of CO oxidation on the above LFO-supported single-atom catalysts, we carried out microkinetics simulations based on the potential energy diagrams. Figure shows that predicted CO2 formation rates on the defective LFO-supported Rh, Pd, and Pt models are substantially higher than those on their stoichiometric counterparts. This is in line with an experimental indication that La defects improve the catalytic activity of CO oxidation on Rh-doped LFO.[38] Our modeling reveals that the increased activity on such La-deficient LFO models is due to intermediate CO and O2 adsorption energies and low overall activation barriers. Notably, our data predict that LFO-VLa-Pd is the preferred composition, indicating that replacement of Rh by Pd should be promising.

Figure 6

Microkinetics simulations for CO oxidation on LFO-supported Rh, Pd, and Pt catalysts. CO2 formation rates r (in mol s–1) as a function of temperature on different models (P = 1 atm, CO/O2 ratio = 1): (a) LFO-Rh, LFO-VLa-Rh, LFO-VSO-Rh, Rhads-LFO, and Rhads-LFO-VLa; (b) LFO-Pd, LFO-VLa-Pd, LFO-Pt, and LFO-VLa-Pt.

We further analyzed the kinetics in more detail. Inspection of the steady-state surface coverages (Figure ) shows that both the Rh sites and surface O vacancy sites of LFO-Rh are occupied by CO and O2 species, respectively, up to relatively high temperature. For LFO-VLa-Rh, adsorbed CO can already react at low temperature with adsorbed O2 species because the associated barrier is much lower for the surface with a La defect in comparison to the stoichiometric LFO-Rh surface. A DRC analysis supports the coverage differences between LFO-Rh and LFO-VLa-Rh in the sense that the reaction of adsorbed CO with adsorbed O2 is the rate-determining step. Similar kinetic trends are computed for the Pd- and Pt-containing LFO surfaces (Figure S5).

Figure 7

Calculated steady-state coverages (a–e) and DRC analysis (f–j) for CO oxidation on LFO-Rh, LFO-VLa-Rh, LFO-VSO-Rh, Rhads-LFO-VLa, and Rhads-LFO. Herein, * and # represent the surface sites on the Rh atom and LFO support, respectively. (P = 1 atm, CO/O2 ratio = 1.)

Conclusions

In summary, we investigated CO oxidation on LFO-supported Rh, Pd, and Pt single-atom models by first-principles DFT calculations and microkinetics simulations. Next to the stoichiometric LFO surface, we also explored the influence of La defects and subsurface O vacancies on the catalytic rates. Models containing such defects enhance the stability of these single atoms adsorbed at the LFO surface or doped into the surface replacing Fe. These changes also affect the CO and O2 adsorption energies and, in this way, affect the barrier for CO oxidation surface reactions. An important finding is that LFO surfaces with a La defect result in optimum CO and O2 binding energies with respect to CO oxidation barriers. A strong correlation between these activation barriers and the oxygen vacancy formation energy is found. Microkinetics simulations emphasize the higher CO oxidation rates for the Rh-, Pd-, and Pt-doped LFO surfaces with a La defect compared to the cases of the stoichiometric surface. Among all the considered models, Pd-doped LFO with the La defect exhibits the highest activity for the low-temperature CO oxidation. Notably, the presence of subsurface O vacancies only facilitates CO oxidation on the Rh-adsorbed LFO surface. Thus, our results demonstrate that the introduction of La defects in LFO-based environmental catalysts is a promising strategy toward improved oxidation performance.