When metal single atoms are anchored onto specific carbon-based materials, the electronic structure of the metal center can be significantly optimized through charge delocalization effects mediated by the support. This not only enhances metal stability but also promotes the adsorption and activation of reaction intermediates, thereby improving CO oxidation activity. For example, when Ni atoms are anchored on the h_4,4,4-gy/G heterostructure, the high Fermi level (E_F) and high conductivity of the graphene substrate provide additional electrons to the h_4,4,4-gy sheet, which in turn influences the d-states of the metal atoms. The adsorption energy of Ni-h_4,4,4-gy/G increases from 3.98 eV to 4.50 eV, indicating that the graphene substrate significantly enhances the stability of Ni^[15-16]. Moreover, Tang et al.^[17]found that the N₃-graphene substrate enhances the activity of single-metal Pd. The Pd-4d orbitals near the Fermi level strongly hybridize with the density of states (DOS) of graphene, bringing the electrons in Pd's d-shell closer to the graphene Fermi level. This electronic modulation effect substantially reduces the activation energy barrier for the CO oxidation reaction, with energy barriers of only 0.24 and 0.15 eV for the Langmuir-Hinshelwood (L-H) mechanism and the three-molecule Eley-Rideal (TER) mechanism, respectively, indicating that CO₂is more easily formed.

Carbon nitride (CN) enhances MSI by anchoring single atoms via N-metal coordination, while leveraging electron transfer to optimize the catalytic performance of active sites^[18].Rao et al.^[19]investigated the catalytic performance of Pd/Pt-embedded planar CN for CO oxidation. The strong hybridization between the d-orbitals of Pd/Pt and the N-2p orbitals results in high binding energies and diffusion barriers, enhancing metal stability. Electron transfer from the metal to O₂ leads to elongation of the O—O bond, thereby promoting O₂ activation. On Pt@CN, the rate-limiting step barrier for the L-H mechanism is 0.68 eV, and the barrier for the second step involving CO₂ formation is only 0.5 eV; in contrast, Pd@CN favors the TER mechanism, with a barrier as low as 0.48 eV, indicating that Pt/Pd@CN can efficiently catalyze CO oxidation. Secondly, Pd at N vacancies acts as an electron donor, transferring a large amount of electrons to neighboring C atoms. The hybridization between its 4d orbitals and the C 2p orbitals forms a high diffusion barrier of 3.8 eV, ensuring the stability of the Pd-C₃N species. Charge transfer effectively activates gas molecules; the energy barriers for the first step of the Eley-Rideal (E-R) mechanism and the L-H mechanism in the oxidation process are 0.64 eV and 0.72 eV, respectively, while the second step’s barrier is significantly reduced to 0.24 eV, indicating that this reaction can proceed under ambient conditions^[20].

In addition, the sp-hybridized carbon framework of graphdiyne (GDY) can form strong interactions with Ni atoms, resulting in a high diffusion energy barrier of 1.74 eV that effectively inhibits the migration and aggregation of Ni. Meanwhile, the synergistic effect of MSI enhances the adsorption of CO with a single O atom, whereas CO₂is only weakly adsorbed (-0.08 eV). The results indicate that the Ni-GDY system is more likely to complete the initial CO oxidation via the L-H mechanism (energy barrier: 0.81 eV), followed by secondary oxidation via the E-R mechanism (energy barrier: 0.25 eV). This sequential reaction pathway exhibits excellent catalytic cycling properties^[21]. In addition, Figure 1shows the volcano curve relationship between the metal-support binding energy and the CO adsorption energy for Pt/Pd@CN and other partial catalysts; higher or lower binding energy may enhance CO adsorption.

In summary, the superior electronic structure and high specific surface area of carbon-based supports facilitate the stabilization and dispersion of metals. MSIs are crucial for stabilizing single-atom metals, with electron transfer playing a vital role. Therefore, a deep understanding and precise control of MSIs are of great significance for designing and optimizing high-performance catalysts. Currently, many discoveries in carbon-based nickel-group single-atom systems are based on theoretical studies, and their practical applications still face numerous challenges, such as the precise regulation of single-atom sites and the enhancement of catalytic stability and durability.

In recent years, metal oxides such as TiO₂,CeO₂,and Al₂O₃have attracted considerable attention as support materials for SACs. Researchers have leveraged the abundant Ti and O defects on TiO₂ to anchor metals and enhance CO oxidation activity through MSI^[22]. Hoang et al.^[23]anchored single Pt atoms onto the surface of a honeycomb-like TiO₂nanowire array (NA). Pt ions are coordinated by 5–6 oxygen atoms at Ti vacancies, forming strong MSI; the T₉₀for CO is as low as 160 ℃. After 100 hours of hydrothermal aging at 700 ℃, the array structure and single-atom dispersion are still preserved, and T ₉₀increases by only 19 ℃ under sulfur-containing conditions, demonstrating excellent catalytic stability. Song et al.^[24]revealed that the chemical state of the metal is influenced by both dual dispersity and the properties of the support. On the TiO₂(101) surface, highly oxidized single Pt atoms form abundant Pt-O-Ti sites, reducing the temperature required for 100% CO conversion by 45 ℃ (T ₁₀₀is 150 ℃), with an apparent activation energy (E _a) of 55.9 kJ/mol. The results indicate that a support with a low oxygen vacancy formation energy is crucial for imparting catalytic activity, while single-atom dispersion is the key to maximizing this activity.

For CeO₂, researchers have found that altering the preparation method, treatment atmosphere, and metal loading can promote the activation and migration of lattice oxygen in the support. For example, in Pd/CeO₂prepared by flame spray pyrolysis (1PdFSP), highly dispersed Pd²⁺can activate lattice oxygen, facilitating the reverse spillover of oxygen at the Pd–O–Ce interface and effectively inhibiting Pd agglomeration (E _a≈ 50 kJ/mol). In contrast, the impregnation method (1PdRods) tends to lead to Pd reduction and aggregation (E _a≈ 60 kJ/mol). The mobility of oxygen at the metal–support interface has a significant impact on the stability and reactivity of SACs, which is crucial for maintaining metal isolation^[25]. Secondly, redox dynamic regulation can significantly enhance the catalytic activity of Pt/CeO₂. After reduction followed by low-temperature oxidation treatment, Pt⁰atoms and Pt²⁺cations coexist; Figure 2aillustrates different redox treatments, whose synergistic effects manifest as follows: interactions between peripheral ions and the interface promote oxygen species spillover, while internally reducible atoms provide active sites for CO adsorption. After Red₂₅₀/Ox_RTtreatment, the catalyst’s T ₂₀decreases by 219 ℃, and E _adrops sharply from 89 ± 5 kJ/mol to 30 ± 2 kJ/mol, indicating a significant improvement in reaction kinetics^[26]. In addition, the loading of SACs directly affects the distribution of active sites and catalytic performance. In high-loading Pt/CeO₂(8–20 wt%), high concentrations of Pt²⁺(surface and boundary positions) and Pt⁴⁺ions (in the cerium lattice) are generated, with some ions aggregating into PtO_xclusters, leading to distortion of the cerium lattice and promoting oxygen release and migration. Below 0 ℃, the CO conversion rate of 8Pt/CeO₂reaches 60%, whereas 20Pt/CeO₂can achieve 90%. In the temperature range from −50 ℃ to 50 ℃, both catalysts exhibit an activation energy of 9 kJ/mol, which is much lower than the 50.7 kJ/mol observed for 1Pt/CeO₂at T≥ 150 ℃^[27]. However, when the metal surface density is too high (density ≥ 2.1 Pt/nm²), agglomeration is more likely to occur, resulting in coverage of active sites and impaired oxygen migration, which significantly reduces the CO oxidation activity^[28].

Existing studies have largely focused on single variables, while Slavinskaya et al.^[29]found that the metal loading and pretreatment method can jointly regulate the oxidation performance of Pt/CeO₂: A high-loading (8 wt%) catalyst, after oxidative pretreatment, forms PtO _xclusters composed of Pt²⁺and Pt⁴⁺, which can directly adsorb CO and O, achieving a conversion rate of approximately 70% at -40 ℃. In contrast, a low-loading (1 wt%) catalyst, after oxidation, forms only monomeric Pt²⁺and becomes active only above 100 ℃; although reduction treatment can convert Pt²⁺into metallic Pt particles, their activity remains significantly lower than that of the PtO _xformed via oxidation, confirming that PtO _xclusters are the key active centers for low-temperature CO oxidation.

In addition, adjusting the calcination temperature can optimize the local coordination environment of SACs. For example, Pt/CeO₂-800 (calcined at 800 ℃) exhibits a square-planar coordination geometry with poor activation ability for O₂, whereas the edge-site Pt@CeEdge-O configuration formed in Pt/CeO₂-550 (calcined at 550 ℃) facilitates the interaction between molecules and active sites, reducing E_a to 43 kJ/mol and lowering the activation energy barrier during oxidation (0.45 eV). Consequently, it displays superior activity in CO oxidation^[30]. When metal atoms adopt low-coordination structures, their quantum size effects can significantly enhance electronic energy levels, thereby boosting reaction activity. For instance, in Pd₁/CeO₂_AT prepared by air calcination at 800 ℃, oxygen vacancies induce the formation of an unsaturated Pd₁O₄ square coordination structure. The Pd²⁺ sites in this structure exhibit higher intrinsic activity compared to fully coordinated Pd²⁺ clusters and Pd/PdO nanoparticles, reducing the T₉₀ temperature for CO oxidation by approximately 100 ℃^[31-32].

In addition, differences in the initial atomic structures also affect catalytic activity. As shown in Figure 2b,the Pt1 single atom in the Pt/CeO₂(500 ℃) system can move dynamically as it migrates with small Pt clusters, maintaining a near-zero-order reaction activity; in contrast, the square-planar Pt₁in Pt_ATCeO₂(800 ℃) is firmly anchored during oxidation, and its restricted mobility leads to lower activity. The T ₅₀of Pt/CeO₂(180 ℃) is significantly superior to that of Pt_ATCeO₂(335 ℃)^[33].

By introducing appropriate promoters, abundant OH groups can be introduced onto the SAC surface, increasing the number of surface active sites^[34-35]. Under H₂O-containing conditions, the electron-donating promoter MgO can facilitate the formation of a large number of active OH species on the Pt/Al₂O₃surface. These OH species can directly participate in the CO oxidation reaction and also promote the desorption of bicarbonate and formate species, thereby accelerating the oxidation cycle^[36-37]. Secondly, the use of a one-step alkaline stabilizer enables the generation of a large number of [Pt₁-O _x] active sites on the Pt/Al₂O₃surface, allowing Pt to maintain a positive charge over the course of the reaction. In the presence of H₂, O₂is activated to produce abundant OH species, and CO oxidation proceeds via a water–gas shift (WGS)-like pathway (CO reacts with surface —OH groups to form CO₂). The T _99.8is 110 ℃, and its activity and selectivity are not affected by support effects^[38].

Single atoms are not always more active than their corresponding clusters or nanoparticles^[39].For example, Pt/Al₂O₃-L prepared by calcination at 550 ℃ forms poorly dispersed PtO _xclusters, which can be readily reduced by CO during the reaction to yield active Pt⁰sites. These sites exhibit both strong CO adsorption and efficient O activation, with the highest reaction rate observed at 170 ℃ (39 μmol·g_cat ^-1·s^-1), significantly outperforming the highly dispersed single-atom catalyst Pt/Al₂O₃-H (only 7 μmol·g_cat ^-1·s^-1 at 170 ℃)^[40]. In addition, Morfin et al.^[41]found that the active species Pt₁O₂-CO and Pt₁-（CO₃H）-Al on single atoms are much more stable than on clusters, resulting in lower oxidative activity compared to nanocatalysts.

In addition to the above metal oxides, Fe₃O₄ and CuO have also been used in recent years as Ni-SAC supports for CO oxidation studies. Hulva et al.^[15] found that the Bader charge of Ni (+0.68 e) lies between that of the metal surface (+0.01 e) and that of the five-coordinate cation on NiO(100) (+1.19 e). As the coordination environment changes, the d-band center of Ni shifts downward, weakening the d→2π* antibonding back-donation interaction and causing CO-Ni bonding to rely primarily on electrostatic interactions. This finding indicates that optimizing the local coordination environment of Ni is crucial for maintaining its CO adsorption capability. Zhou et al.^[42] constructed two types of Ni active sites on a CuO monolayer via thermal deposition: metallic Ni_m and cationic Ni²⁺. At 200 K, CO can be adsorbed onto Ni_m, whereas at 70 K, it cannot be adsorbed onto Ni_c. However, the latter exhibits high activity in O₂ activation at room temperature, with a dissociation energy barrier of only 0.14 eV. Ni_m and Ni_c follow the Mars–van Krevelen (MvK) and Eley–Rideal (E-R) mechanisms, respectively. This study contributes to a deeper understanding of how the chemical state of metal atoms influences oxidation pathways.

In summary, the performance of SACs is highly dependent on metal dispersion, chemical state, coordination environment, and the dynamic properties of the support, all of which are influenced by the support's intrinsic structure and regulation methods (e.g., preparation methods, calcination temperature, treatment atmosphere, loading amount, etc.). Future research should focus on the rational design and optimization of MSI, exploring multiple regulation variables and delving into the dynamic interaction mechanisms at single-atom sites, with the aim of developing more active and stable low-temperature CO oxidation catalysts.

The porous structure and ultrahigh surface area of MOFs provide abundant anchoring sites for single atoms, while the metal can induce charge transfer, promoting the adsorption and activation of reaction molecules and thereby reducing the activation energy barrier^[43].Zhang et al.^[44]discovered a magnetic MOF X₂C₁₈H₁₂(X = Mn, Fe, Co, Ni) for CO oxidation. The “synergistic charge-spin catalysis” mechanism between adjacent X atoms promotes the activation of spin-triplet O₂ via changes in magnetic moment and electron transfer, significantly reducing the oxidation energy barrier. The O₂ adsorption energy E _ads, the O–substrate charge transfer value Δρ, and the O–O stretching frequency νare all linearly correlated with ΔM(the spin interaction value), while chemical activity is negatively correlated with ΔE _SEE(the spin excitation energy) (the smaller ΔE _SEEis, the larger ΔMbecomes). The synergistic catalytic efficiency decreases in the order Mn < Fe < Co < Ni. In the future, ΔE _SEEcould serve as a criterion for designing more efficient Ni-based MOF sacs.

Covalent organic frameworks (COFs), with their C=C double-bond-enhanced π-conjugated systems and post-functionalization-introduced ligand groups, enable metal atoms to be firmly anchored within the framework^[45].For example, in trzn-COF, the nucleophilicity of the conjugated π electrons from sp² C atoms, N/O heteroatoms, and aromatic rings in conjunction with the triazine ring results in strong interactions between Pd and the support, enhancing catalyst stability. During oxidation, the W₂ site follows the TER mechanism, where the rate-determining step—the formation of the C—O bond—requires overcoming an energy barrier of 1.03 eV, while the O—O bond cleavage requires only 0.13 eV, indicating that Pd1/trzn-COF exhibits high catalytic activity^[46].

For microporous zeolites, the formation of slightly larger atomic clusters facilitates low-temperature activation of oxygen. Ultra-small Pd clusters in FER zeolite exhibit a high O₂ binding energy (142 kJ/mol) and more easily activated O—O bonds, and the small clusters display better mobility in the presence of CO, making them less susceptible to poisoning. In a humid gas stream, 100% CO conversion can be achieved below 125 ℃; in contrast, at low temperatures, CO-covered nanoparticles inhibit O₂ activation, resulting in a conversion rate of less than 20% at 120 ℃^[47]. Furthermore, by adjusting the reaction temperature and gas atmosphere, the dynamic transformation of metals between single atoms and clusters can be effectively controlled (Figure 3): Under high-temperature oxidation conditions, single Pt ions reside in the six-membered rings of the zeolite, with their electrons withdrawn by the support to form a highly oxidized, electron-deficient state; under reducing conditions, Pt²⁺ forms a homogeneous Pt-dicarbonyl complex, which is subsequently converted in H₂ to a Pt ^{δ +} monocarbonyl. This process weakens the bonding between the metal and the zeolite, leading to a tendency for Pt species to aggregate. This chemical behavior is observed in several commonly used zeolites (ZSM-5, Beta, mordenite, and Y), suggesting that this phenomenon has a certain degree of generality^[48-49]. Therefore, the transformation behavior of metals can be controlled by altering environmental conditions, thereby optimizing catalyst performance and providing a direction for research on zeolite-based SACs. However, whether at the nanoscale or as single atoms, the metal-support/reactant MSI remains the key factor determining catalytic activity^[50].

In summary, although the anchoring strategies of different supports vary, their core focus lies in the dynamic synergy of electron and interface interactions between the metal, support, and reactants. For (non)metal framework materials, future research can concentrate on the following directions: (1) elucidating the microscopic coupling mechanisms of spin synergy and charge transfer; (2) exploring the stability and reconstruction patterns of metal-support interfaces under extreme conditions.

Different SACs follow different reaction mechanisms, leading to varying CO oxidation activities. The widely accepted reaction mechanisms include the L-H, E-R, MvK, and TER mechanisms^[51].

In the L-H mechanism, CO and O₂first undergo co-adsorption and activation at a single active site, followed by the formation of peroxo-type intermediates^[52-54]. The electron-rich environment of the graphene substrate optimizes reactant adsorption strength and mass transfer efficiency, thereby promoting the catalytic reaction. Liu et al.^[55]studied the performance of PdGr in CO oxidation; the MSI between graphene and Pd induces substantial charge transfer, exposing sp-type boundary states of Pd at vacancy sites, which in turn facilitates the co-adsorption of O₂and CO. Under the same conditions, the overall reaction rate via the rL-H pathway (2.76 × 10^-2 s^-1) is more than 10⁶times that of ER1 (5.55 × 10^-11 s^-1) and TER (5.36 × 10^-8 s^-1). The results indicate that the reaction rate depends on the thermal stability of the species, and the superior catalytic performance of PdGr is attributed to its specific electronic structure.

Strain can modulate the effective adsorption energy of gas molecules, thereby enhancing molecular selectivity. Jiang et al.^[56]investigated the effects of strain on the molecular adsorption behavior on Ni-SG surfaces and on CO oxidation. Compressive strain enhances the adsorption of SO₂;tensile strain has the opposite effect. Tensile strain increases the Hirshfeld charges and orbital energy levels of O₂ and CO, leading to a narrowing of the orbital energy gaps for 5σ CO/1π CO and 5σ O₂/1π O₂, which reduces the reaction barrier for OOCO. The three-step reaction energy barriers decrease by 24% to 48% as the tensile strain ε increases from -2% to 10%, ultimately endowing Ni-SG with excellent resistance to water and sulfur as well as superior gas selectivity.

In addition, the dual active sites of multi-component single-atom catalysts can significantly enhance catalytic performance through synergistic effects^[57].Wang et al.^[58]reported the structure–activity relationship of Fe-Ni diatomic atoms anchored on N-doped graphene (Fe1Ni1@NGr): Fe carries a greater positive charge (1.09 |e|) and serves as an electrophilic adsorption center for O₂, while Ni (0.51 |e|) acts as a nucleophilic adsorption center for CO. This dual-site synergy effectively reduces the risk of CO poisoning; furthermore, the L-H mechanism energy barrier of the heteronuclear catalyst (0.47 eV) is also significantly lower than that of the homonuclear Ni₂@NGr (0.80 eV) and Fe₂@NGr (0.64 eV).

In addition, heterogeneous structured supports can significantly promote the molecular co-adsorption behavior during CO oxidation: ZnMn₂O₄@MnO₂ enhances MSI, enabling effective embedding of Pd²⁺. In the oxygen-vacancy-rich support, Pd²⁺ single atoms form coordinatively unsaturated Pd₁O₃, which is transformed into Pd₁O₅ upon O₂ activation. On the other hand, CO reduces the distorted Pd₁O₄ to Pd₁O₃, thereby generating an effective redox pair of Pd₁O₃ and Pd₁O₅. PZMO-30 follows the “MvK-induced L-H” mechanism (Figure 4), achieving 90% CO conversion at room temperature (26 ℃), with an Eₐ (34.87 kJ/mol) that is significantly lower than that of other comparative catalysts^[59].

In summary, the differences in SACs' activity in CO oxidation stem from the synergistic regulation of reaction mechanisms and electronic/geometric structures. The core of catalytic performance lies in the matching of the electronic microenvironment at the active site—such as the d-band center, charge distribution, and dynamic adsorption behavior—while catalytic efficiency can be enhanced through regulation strategies such as mechanical strain, the construction of heterogeneous structures, or multi-element single-atom systems. In the future, further research could explore the multi-mechanism coupled conversion between dynamic L-H and MvK, and attempt multi-parameter synergistic optimization involving factors such as strain and support heterogeneity.

Metal oxide-supported lattice oxygen not only participates in the formation of intermediate reaction complexes but also significantly influences the catalytic rate^[60-61]. Lu et al.^[62]investigated the oxidation mechanism of CO on Pt/CeO₂to identify the reaction intermediates and the rate-determining step, in which Pt(O)₄and Pt(CO)(O)₃are two key intermediate complexes. The reaction initiation is limited by the reaction between CO and the lattice oxygen of Pt(O)₄, forming Pt(O)₃(with an energy barrier of 62 kJ/mol), while the lattice-oxygen-derived active oxygen species (O*) plays a crucial role in subsequent reactions: it can either directly react with Pt(O)₃to regenerate Pt(O)₄or react with Pt(CO)(O)₃to restore the catalyst to the Pt(O)₄state; however, both oxidation pathways are constrained by O*. Meanwhile, the choice of support is also critical to the performance of SACs; for instance, compared with traditional oxides, novel high-entropy oxides can significantly enhance catalytic activity through their electronic effects and multi-element synergistic effects^[63]. In high-entropy fluorite HEFO, a single Pd atom is incorporated into the sublattice via stable Pd-O-M bonds. Compared with traditional Pd@CeO₂, Pd₁@HEFO exhibits a surface lattice oxygen reduction temperature as low as 160 ℃ and a twofold increase in the proportion of active oxygen species, with a significantly reduced Eₐ value (43.40 kJ/mol), indicating a lower reaction energy barrier^[64].

In summary, existing scoring systems have limited predictive capabilities for bleeding events, and their results are inconsistent^[25,30,33].

In addition, surface OH generated from the decomposition of an appropriate amount of H₂O can effectively promote the CO oxidation reaction. Wang et al.^[67] investigated the effect of Pt particle size on catalytic behavior under dry and H₂O-containing conditions. Under H₂O-containing conditions, the single-atom catalyst 0.2Pt/Cr_1.3Fe_0.7O₃, due to its weaker CO adsorption strength compared to the nanoparticle 2Pt/Cr_1.3Fe_0.7O₃, reacts more readily with surface OH, exhibiting a stronger H₂O-promoting effect. At 60 ℃, the TOF increased from 0.177 s^-1 to 0.396 s^-1, and at 80 ℃, the CO conversion rate rapidly rose from 15% (dry) to 80% (H₂O-containing). The addition of H₂O not only increased the reaction rate constant but also promoted the decomposition of surface carbonates and accelerated the process by which CO reacts with OH to form formate.

Recent studies have also revealed that the Pt_SA/a-TiO₂system exhibits a complex dynamic reaction mechanism in CO oxidation. As shown in Figure 5, the reaction initiates from structure (I), where oxygen vacancies are filled through the adsorption of CO and O₂. The E-R mechanism then leads to the formation of the key intermediate Pt(CO)(IV). The steady-state reaction cycle involves the dissociative adsorption of O₂ to form Pt(O)(O)(CO)(VI), followed by the formation of CO₂ via competitive L-H and E-R mechanisms. At 160 ℃, the system exhibits stable CO oxidation activity (TOF = 3.8 × 10^-3 s^-1), with an E _aof 69 kJ/mol. The fractional reaction order arises from multiple competing rate-controlling steps, and further experiments are needed to validate these findings^[68].

In summary, the dynamic participation of lattice oxygen in metal oxide supports and the coordinated evolution of metal oxidation states are central to performance regulation. The dynamic equilibrium between single atoms and clusters highlights the synergistic necessity of highly dispersed active sites and dynamic responsiveness. Moreover, environmental factors such as the introduction of H₂O enhance the anti-poisoning capability of single-atom sites. Furthermore, the validity of these conclusions regarding complex reaction pathways must be verified through real-time dynamics obtained from characterization and multi-scale simulations. Future research could focus on: (1) tracking the migration of lattice oxygen and the evolution of metal oxidation states; (2) exploring the development of novel supports such as high-entropy composite oxides; and (3) synergistically regulating catalyst activity and stability by introducing environmental factors like H₂O to maximize efficiency.

In exhaust gases, H₂O poisons active sites and competes with CO for adsorption, leading to catalyst deactivation and reduced stability^[69]. Graphene double carbon vacancies can significantly enhance the surface reactivity of the support through local structural reconstruction and π-electron density modulation. The vacancy-induced σ–π orbital rehybridization optimizes charge distribution^[70], thereby regulating H₂O adsorption behavior and enhancing the hydrophilic resistance of SACs. Jiang et al.^[71]compared the adsorption behaviors of Pt-loaded graphene catalysts with single carbon vacancies (Pt-SG) and double carbon vacancies (Pt-DG) and found that on Pt-SG, the adsorption energies of SO₂and H₂O are comparable to that of CO, leading to competition for active sites and hindering CO oxidation; on Pt-DG, H₂O is physically adsorbed (-0.30 eV), and its Pt d orbitals exhibit weak hybridization with the 1b1 and 3a1 orbitals, without forming chemical bonds. At the same time, CO can displace pre-adsorbed H₂O with an energy barrier of 0.37 eV, causing H₂O to switch to physical adsorption. Thanks to its selective adsorption properties, Pt-DG still exhibits excellent CO oxidation performance in aqueous environments, with a TER pathway energy barrier of only 0.49 eV and a room-temperature reaction rate as high as 7.16 × 10⁷ s^-1. Similarly, the adsorption energy of CO (-1.21 eV) on Ni-DG is significantly higher than that of H₂O (-0.43 eV), and CO can displace pre-adsorbed H₂O with a low energy barrier of 0.12 eV. More importantly, the reaction energy barrier between H₂O and two pre-adsorbed CO molecules is as high as 1.16 eV (well above the critical barrier of 0.75 eV), indicating that H₂O does not block the active sites. This ensures that Ni-DG can still efficiently catalyze CO oxidation in humid environments via a novel TER mechanism, with an energy barrier of only 0.34 eV for the rate-determining step^[72].

In addition to creating double carbon vacancies in the graphene structure, applying tensile strain can also effectively regulate the adsorption performance of Ni-SG for gas molecules. As the strain increases from -2% to 10%, the amount of charge transferred from H₂O molecules to Ni-SG remains stable within the range of 0.169 e to 0.184 e, and the degree of hybridization between its 1b1/3a1 orbitals and the Ni 3d orbitals is almost unaffected by strain. The adsorption energy of H₂O decreases from -0.98 eV to -0.87 eV. This selective regulation results in Ni-SG under 10% strain exhibiting a CO adsorption energy (-1.11 eV) that is superior to that of H₂O^[56].

In summary, double carbon vacancies in graphene can enhance the hydrophobicity of SACs and effectively reduce the desorption energy barrier of H₂O, thereby providing better resistance to water molecules. Therefore, the selective adsorption of gas molecules can be enhanced by introducing double carbon vacancies into the graphene support or by applying tensile strain. Currently, there is a lack of detailed mechanistic studies on the hydrophobic performance of Ni-SACs, which is essential for gaining a deeper understanding of the H₂O inhibition mechanism and for developing catalysts for CO oxidation in humid air.

N-doped graphene can effectively modulate the electronic structure of SACs, enhancing MSI by elevating the metal’s Fermi level (EF). While optimizing electronic performance, it retains the inherent advantage of the support’s high specific surface area^[73-74].Chen et al.^[75]studied tricoordinate Ni SACs supported on N-doped graphene. As the number of N coordination sites increases, the atomic binding energy gradually weakens, the Ni d-orbitals shift toward the Fermi level, and more valence electrons interact with the adsorbate, thereby enhancing the adsorption energies of *CO and *O₂.The reaction energies and activation barriers for each step in the L-H, E-R, and TER mechanisms exhibit a linear relationship with the sum of the adsorption energies. Among these, Ni-N₁C₂ exhibits the highest overall reaction rate (1.72 × 10^-2 s^-1), whereas the highly N-doped Ni-N₃C₀ displays the lowest activity (5.59 × 10^-6 s^-1). This indicates that appropriate N doping helps synergistically promote CO oxidation under different reaction mechanisms.

In recent years, numerous scholars have investigated the effects of doping metal oxides with the metal elements Zr, Pr, Co, and Cu. When dopant atoms substitute Ce cations in the CeO₂lattice, the resulting lattice distortions can modulate atomic arrangements, and changes in the electronic states lead to an increase in oxygen vacancies both in the bulk and at the surface, thereby enhancing the redox properties of the catalyst^[76]. Tan et al.^[77]constructed a support rich in surface defects (CZO) by doping CeO₂with Zr. Due to the smaller ionic radius of Zr⁴⁺, lattice strain is induced, promoting the transformation of Ce⁴⁺to Ce³⁺, which significantly enhances the dispersion and stability of Pt single atoms. After activation, Pt/CZO-a forms smaller Pt clusters, exposing more active Pt⁰sites, and the T ₉₀for CO oxidation drops to as low as 120 ℃. Moreover, after high-temperature aging, the activity of Pt/CZO-800A-a decreases only slightly below 100 ℃, while the T ₉₀of Pt/CZO-1000A-a remains at 212 ℃, confirming the significant enhancement of thermal stability conferred by Zr doping. Similarly, Deng et al.^[78]explored the impact of Pr doping in CeO₂nanorods supported on Pd on catalytic performance and its underlying mechanism. Moderate Pr doping can increase the concentration of surface-active oxygen species O_βand the content of Ce³⁺while reducing the binding energy of Pd. As the doping level increases from 2.5% to 10%, the catalytic activity exhibits a volcano-shaped trend. Among them, Pd/Pr-CeO₂-5% has the highest Pr³⁺/Pr⁴⁺ratio and oxygen vacancy concentration. This catalyst displays the lowest E _a(41.15 kJ/mol) and an increased reaction order with respect to O₂, achieving a TOF of 32 × 10^-3s^-1at 130 ℃—four times that of pure Pd/CeO₂. In addition, as the Co doping level increases to 10%, more Ce and Co in Pt₁/CeO₂form coordination structures, generating a large number of highly active Pt—O—Co/Ce sites. At 100 ℃, the CO conversion rate of 0.5% Pt/10% Co-CeO₂reaches 36.6%, which is 3.6 times that of 0.5% Pt/CeO₂(10%) and 4.9 times that of 10% Co-CeO₂(7.4%). However, at low doping levels, there are more inert sites, and at high doping levels, CoO _xspecies block some sites. Furthermore, when the Pt loading exceeds 1%, clusters or nanoparticles begin to form, leading to a decline in performance^[79]. Therefore, optimizing the single-atom loading and Co doping level is crucial for achieving efficient CO oxidation at low temperatures.

Cu doping can synergistically enhance CO catalytic performance through interfacial effects and Pt size regulation: Cu²⁺substitutes for CeO₂lattice sites, inducing lattice distortion. The resulting Cu-O-Ce sites are more easily reduced than Ce-O-Ce sites, significantly promoting oxygen migration and activation. However, Cu doping provides only limited enhancement of the activity of Pt₁/Ce (the T ₅₀for Pt₁/CeCu is approximately 116 ℃). In contrast, sub-nanometer clusters (Pt _n) exhibit a general advantage due to their greater number of active sites. The introduction of Cu not only weakens the CO adsorption strength but also accelerates CO₂desorption, resulting in much higher low-temperature activity for Ptₙ/CeCu (T ₅₀is only 34 ℃, and the TOF reaches 0.26 s^-1 at 80 ℃), far surpassing that of Pt₁/CeCu^[80].

Through various modulation methods to prepare specialized interfaces, one effective approach for modifying multiphase catalysts has emerged, with recent research focusing predominantly on the modification of metal oxides. Among these methods, acid etching can significantly enhance MSI by adjusting the electronic states of the substrate, effectively activating oxygen and thereby reducing the oxidation energy barrier. Wang et al.^[59]used HNO₃to etch ZnMn₂O₄, forming a heterogeneous structure that strengthens the interaction between the substrate and Pd²⁺. The unique coordination configuration of atoms leads to elongation of the Mn—O bonds, further enhancing the activation effect of Olatt. In PZMO-30, the activation of Oads and Olatt by Pd²⁺is most pronounced; compared to PZMO, the former's distorted Pd₁O₄structure and the superior redox couple of Pd₁O₃/Pd₁O₅enable it to effectively catalyze CO at room temperature.

Second, specific multi-component interfaces not only provide strong binding sites for single atoms, but the resulting synergistic effects can also activate lattice oxygen, significantly enhancing catalytic performance. For example, during the formation of the Pt-SA/CeO _x-TiO₂interface, Ce³⁺is enriched at the interface, facilitating electron transfer between Pt and the interface, enhancing metal stability and activating lattice oxygen to participate in the reaction. This unique interfacial structure results in nPCT exhibiting a specific mass activity (MA) at 140 ℃ that is 15.1 times that of nPT with the same loading; in particular, 0.25PCT can still efficiently oxidize CO above 135 ℃ under O₂-free conditions (relying solely on lattice oxygen), whereas 0.25PT is inactive. This indicates that the stable distribution of high-density Pt-SA at the interface and its resistance to CO poisoning are key to its superior catalytic performance^[81].

Based on morphology control, interface structure optimization can be further achieved. Zhao et al.^[82]synthesized Co _xO _ynanoarrays with different morphologies and structures via facet-induced synthesis to regulate the metal–support interface. Among these, the interlaced nanoneedle morphology of Co _xO _y-I exhibits abundant Co³⁺ species and the highest oxygen vacancy concentration (16.57%). On one hand, the oxygen vacancies enriched on the crystal facets can anchor Pd single atoms, forming stable Pd–O–Co bonds; on the other hand, CO molecules preferentially interact with surface Co³⁺ and Pd species, facilitating CO adsorption and reaction. The strong CO adsorption energy (-7.55 kcal/mol) combined with the weaker CO₂ desorption energy (-0.92 kcal/mol) enables complete CO oxidation at 90 ℃, with an activation energy E _aof 51 kJ/mol.

In addition, efficiency can also be enhanced through the regulation of metal coordination structures. For example, the square-planar coordination structure of PdO₄differs from the octahedral (111) structure of Pd/CeO₂. In the former, when the metal cation is reduced, electrons are transferred from lattice oxygen, leading to the breaking of metal–O bonds and thereby facilitating the removal of surface oxygen^[83]. Pt₁/CeO₂_TS prepared by the thermal shock method exhibits an asymmetric square-planar Pt₁O₄configuration, which can dynamically transform into a Pt₁O_{4- x}coordination under CO oxidation conditions, generating partially reduced, active Pt₁ ^{δ +}. XANES analysis (see Figure 6) shows that the white-line intensity of Pt₁/CeO₂_TS is lower than that of Pt₁/CeO₂_AT; combined with the rising-edge features of the first-derivative spectrum (inset), this confirms the reduced coordination symmetry and the formation of a partially reduced state. This structural feature enhances the adsorption and activation of reactants: at a space velocity of 200 L/g-h, T ₅₀(150 ℃) is approximately 140 ℃ lower than that achieved by the AT atom-trapping method (287 ℃), and even after oxidation treatment at 500 ℃, T ₅₀increases by only 2–10 ℃^[84].

In addition to interface optimization of metal oxide supports, an appropriate amount of water can promote CO oxidation on transition metal surfaces. When a water layer is deposited on the surface of a single-atom alloy (SAA), the formation of hydrogen bonds and charge transfer effectively enhance the adsorption (from -1.07 eV to -1.52 eV) and activation of O₂at the H₂O/Ni@Au(100) interface, increasing the coverage of O₂. At the same time, the water layer reduces the CO adsorption energy from -1.85 eV to -1.66 eV. This dual regulation lowers the energy barrier for the rate-determining step of CO oxidation on Ni@Au(100) from 0.66 eV to 0.38 eV, increasing the reaction rate by approximately 10¹¹-fold. Studies have shown that this strategy is applicable to SAA systems involving Ni, Pd, Pt, and other metals^[85-86].

Different types of intrinsic defects, such as carbon defects, oxygen defects, and cationic defects, are often engineered to stabilize metal atoms^[87]. For example, the interactions between pristine and vacancy-containing graphene and TMs (transition metals) give rise to significant differences in binding energy, migration barriers, and reaction energy barriers (Fig. 7). Due to perturbations in their local electronic structure, defective graphene exhibits enhanced electron-transfer capabilities and can also mitigate metal-clustering issues^[88-89]. If the number of missing atoms is even, the carbon atoms undergo complete reconstruction without leaving any dangling bonds; if the number of missing atoms is odd, the dangling bonds render the graphene more reactive^[90]. This results in a higher migration barrier for double carbon vacancies compared to single carbon vacancies. For instance, the diffusion barrier for Ni in double-vacancy graphene (Ni-DG) reaches as high as 6.23 eV, demonstrating exceptional stability. The binding energy of Ni-DG (-7.93 eV) exceeds that of the single-vacancy Ni-SG system (-7.43 eV), indicating that double vacancies can anchor Ni atoms more firmly. Furthermore, the adsorption energy of CO on Ni-DG (-1.21 eV) is significantly higher than that of O₂(-0.41 eV). This selective adsorption property enables Ni-DG to preferentially capture CO molecules in mixed gas environments. Coupled with the low energy barrier of 0.34 eV associated with the TER mechanism, Ni-DG exhibits superior low-temperature CO oxidation performance compared to Ni-SG (LH mechanism: 0.59 eV), with a CO oxidation rate as high as 1.06 × 10⁷ s^{-1 [72]}. In addition, double carbon vacancies not only have higher migration barriers but also exhibit better sulfur and water resistance: For Pt-SG, the atomic diffusion barrier is 3.61 eV, whereas the diffusion barrier of Pt-DG (5.80 eV) indicates that double carbon vacancies can more effectively suppress atomic aggregation. In Pt-SG, SO₂ and H₂O compete with CO for adsorption sites, making it difficult for O₂ and CO to co-adsorb. In contrast, on Pt-DG, the adsorption energies of SO₂ and H₂O are significantly reduced, greatly diminishing their interference with CO oxidation. The interaction between O—O’s 1π orbital and CO’s 5σ and 1π orbitals is weakened, leading to the breaking of the O—O bond and ultimately promoting the formation of CO₂ ^[71].

Typically, defect sites with unsaturated coordination can act as “traps” to capture metal precursors, enabling atomic-level dispersion and stabilization; on the other hand, they expose more active sites and facilitate electron transfer, thereby influencing catalytic performance^[91-92]. For example, in Pt/CA-T, small CeO₂particles stabilize and anchor Pt single atoms at step-edge defect sites; after H₂activation, these sites can transform into highly active small Pt clusters. Meanwhile, the Pt–CeO₂interface synergistically promotes oxygen activation and CO adsorption, with the following hierarchy of active sites: step sites ≈ terrace sites > corner sites > Pt single atoms anchored on CeO₂. This design enables the catalyst to achieve 50% CO conversion at 106 ℃; after aging treatment, the T ₅₀value of Pt/CA-T-800A-a (118 ℃) remains significantly superior to that of Pt/CA-800A-a (178 ℃), and its TOF (0.065 s^-1) is 6.5 times that of the latter^[93]. When metals are loaded onto supports containing oxygen vacancies and cationic vacancies, changes in the metal’s electron density can significantly alter the adsorption strength of CO molecules^[94]. Xie et al.^[95]employed a surface-defect-enrichment strategy to precisely control the local coordination environment of the metal, using Pt₁/CA and Pt₁/CA-e (rich in oxygen defects) as precursors to prepare surface-adsorbed Pt_ASL/CA and lattice-embedded Pt_ASL/CA-e. Because the latter exhibits a lower Pt oxidation state and a shorter Pt–O bond length, the average CO adsorption energy is moderate (-1.38 eV), resulting in a significant reduction in E _a(43 kJ/mol). At 125 ℃, the TOF value (0.84 s^-1) is 3.5 times that of the former (0.238 s^-1), and the high reactivity of lattice oxygen significantly lowers the rate-determining energy barrier (0.82 eV) compared to the adsorption-dominated structure (1.78 eV), demonstrating outstanding low-temperature activity and interfacial synergy.

For defect sites within MOFs, the unique interplay of electronic and spatial effects can significantly reduce the activation energy barrier required for reactions^[96].Guo et al.^[97] uniformly dispersed Pt in a Ce-MOF with tailored defects. The Pt²⁺ species form strong electronic coupling with Ce³⁺/Ce⁴⁺, and the coordination sites of Pt and Ce, along with the hybridization of Pt 4f and 5d orbitals with neighboring O atoms, significantly lower the reaction activation energy to 50.5 kJ/mol. The results show that at 150 ℃, only 0.12 wt% Pt is sufficient to achieve 100% CO conversion, and the catalyst maintains near-perfect activity even after 96 hours of continuous operation and 10 cycles of high-temperature treatment at 300 ℃.

In confined spaces, functional confinement constrains the behavior of guest atoms through interactions between functional groups or atoms on the confining substrate and the guest atoms; geometric confinement, on the other hand, creates isolated microenvironments using rigid channels or interlayer gaps, which can prevent the aggregation of single atoms while shortening diffusion paths and enriching reactants. The synergistic effect of these two mechanisms can optimize reaction energy barriers and charge transfer, significantly enhancing catalytic performance^[98].

The electronic energy level structure between the layers of two-dimensional materials can confine the electrons of guest materials within the interlayer space, forming energy well structures^[99],thereby enhancing catalyst stability and intrinsic activity. In the graphene/amorphous ZrO₂interlayer structure, the rotational effect of graphene not only shifts part of the Pd-4d states toward the valence band, effectively inhibiting atomic migration, but also accelerates the movement of reactant orbitals near the Fermi level. Meanwhile, Zr-4d enhances the carrier mobility at the interface (with an effective mass as low as 0.424 m ₀). In the L-H mechanism of CO catalytic oxidation, the energy barrier for the first step—formation of the OOCO intermediate—ranges from 0.075 to 1.085 eV; in the TS2 process, only Model 2 (Pd@0° graphene/ZrO₂) overcomes an energy barrier of 1.762 eV to form the final CO₂, whereas the energy barriers for other models are all below 0.2 eV^[100]. Second, the lattice constraints imposed by metal oxide supports both stabilize single atoms and reduce the energy barrier for extracting lattice oxygen^[101]. Liao et al.^[102]prepared Pd/MnO₂catalysts with different loading levels: low loading resulted in lattice-confined single Pd atoms, while high loading led to the formation of Pd clusters. Among these, the 1.7Pd-MnO₂catalyst exhibited the lowest Mn oxidation state and the highest oxygen vacancy content, with a bulk lattice oxygen desorption temperature 100 ℃ lower than that of Pd cluster catalysts. The lattice-confined single Pd atoms showed an extremely low oxygen vacancy formation energy (−0.58 eV), indicating their ability to efficiently activate lattice oxygen. This catalyst achieved a TOF as high as 0.203 s⁻¹at 50 ℃ and enabled complete CO conversion at 70 ℃, with an activation energy E _aof only 9.5 kJ/mol. For MOFs, the confinement effect within the pore space is one of the three primary strategies for stabilizing metals^[103]. Xue et al.^[104]investigated the stability and CO oxidation performance of Ni²⁺, Pd²⁺, and Pt²⁺encapsulated in MOF-808. The metals encapsulated in the pores are reduced due to electron transfer from the framework, leading to significant electron accumulation on the surface. Among them, the substitution energy of Pt²⁺ions is −480 kJ/mol, exhibiting the best stability. In oxidation reactions, MOF-808-Pt^IIdemonstrates superior catalytic performance via the L-H mechanism, with an activation energy of only 44 kJ/mol for the first step of CO oxidation. Its relative TOF (1.00) is also significantly higher than that of MOF-808-Ni^II(2.09 × 10⁻¹⁴) and other metal catalysts. Therefore, confining Pt within the MOF pore structure facilitates the redistribution and enrichment of charge.

In summary, the electronic behavior of metals is significantly influenced by support defects, frontier orbital characteristics, and the overlap of electron clouds between the metal and the support^[105].The above four strategies—precisely regulating the local electronic structure, the dynamic behavior of oxygen species, the coordination environment, and the chemical and spatial environment of the metal—have emerged as key approaches for optimizing SAC performance. Future research can focus on: (1) the dynamic balance between doping concentration and loading amount; (2) exploring multi-element co-doping strategies to overcome the performance limitations of single-element doping; (3) the continued challenge of precisely controlling the distribution of defect types; and (4) the limited versatility of confinement strategies in Ni-SAC systems, which may be constrained by factors such as support selection and synthesis methods.

Finally, this article also summarizes the CO oxidation performance of some Ni-SACs prepared in recent years, their preparation methods (Table 1),and theoretical studies (Table 2).