The rational design of different catalysts for heterogeneous catalysis is a popular subject among scientists, especially in environmental catalysis [1]. It is well known that aromatic chlorinated volatile organic compounds, such as polychlorinated dibenzofurans (PCDFs) and polychlorinated biphenyls (PCBs), which are also known as dioxins, are considered the most toxic pollutants [2]. It can be particularly challenging to treat such complex molecules. Chlorobenzene (PhCl) is often used as a model molecule to study the degradation of hazardous pollutants due to its similarity with dioxin. Many efforts have been made to degrade it [3], and catalytic oxidation is one of the most commonly used strategies [4]. However, most studies of catalytic reactions tend to focus only on the overall redox and acid abilities of catalysts while neglecting the enhancement of the activation of PhCl molecule by rational design of active sites.

Ceria-based catalysts are effective for PhCl oxidation reactions due to their excellent redox ability and highly localized 4f electrons [5], and phosphorizing the surface of Ce-based catalysts is a common strategy to improve their catalytic activity further [6-9]. The introduction of phosphate can generate a variety of active oxygen species and enhance the redox capability during the oxidation and chlorine removal stages of PhCl catalytic oxidation. This process mainly involves the cleavage of the C-Cl bond, the opening of the benzene ring, and the deep oxidation of PhCl. However, the results of previous studies revealed that the adsorption of PhCl during the reaction was underestimated due to its physical adsorption properties [6, 10].

In this work, we systematically calculated the different phosphate groups (H_xPO₄) on the CeO₂ surface and confirmed a covalent interaction between H₂PO₄ and PhCl by spin-polarized density functional theory (DFT) calculation. The results show that the PhCl is physically adsorbed on the CeO₂ system, which helped us design active sites to activate the PhCl molecules. We also find that the Ru₃/CeO₂ surface can effectively enhance the adsorption activation of PhCl, which is mainly attributed to the fact that the loading planar sites with a suitable orbital of the system can distort and break the p-π conjugation of PhCl. This work provides theoretical insights into the rational design of effective Ce-based catalysts for complex molecule conversion, which will also give some valuable explanations for experimental results.

We constructed a series of phosphorylated CeO₂ (111) models (P/CeO₂) and analyzed their stability (Fig. 1). The calculated results show that the P/CeO₂-a system is less stable due to the presence of an unstable P═O bond (Figs. 1a and b). In addition, with the dissociation process of H atoms in H₃PO₄ (Figs. 1c and d), the co-adsorption energies of H and H_xPO₄ species on the CeO₂(111) surface increased and then decreased (Table S1 in Supporting information). Moreover, we also noticed that the geometries and electronic structures of Figs. 1e–h are similar, so they can be treated as the same surface model. Thus, the P/CeO₂-a, P/CeO₂-b, and P/CeO₂-c surfaces were considered models for PhCl adsorption (Table S2 in Supporting information).

Next, we also analyze the electronic properties of different P/CeO₂ surface models. The calculated results show that the position of the H_xPO₄ orbital (-4.053 eV, Fig. 1d) is also much higher than that of other orbitals (-0.442 eV ~ -0.430 eV) and that the occupancy states of all the orbitals (Figs. 1b, f and h) are slightly lower than the Fermi energy (E_f), resulting in their internal orbitals being in the lower states with lower activity. We also find that the orbital energy level of P/CeO₂-b is closest to -4 eV, which is more likely to interact with the PhCl's p-π orbitals than the other orbitals with similar energies. As shown in Figs. S1 and S2 (Supporting information), the orbital in P/CeO₂-b and PhCl's π orbitals have the same symmetry. Thus, there might be a possibility that an orbital interaction could appear when PhCl approaches this group.

We next performed systematic calculations on the adsorption of PhCl on the P/CeO₂ surface, which showed that the parallel adsorption of PhCl on the P/CeO₂ surfaces are much more stable than the perpendicular adsorption structure and the corresponding adsorption energies are -0.663 eV and -0.293 eV, respectively (Figs. S3 and S4, Table S2 in Supporting information). A careful analysis of the above adsorption configurations also reveals that PhCl will adsorb between a vertical and parallel configuration due to the presence of tetrahedral H_xPO₄. We also performed an in-depth analysis of the geometrical and electronic structures of the most stable adsorption structures of PhCl on the P/CeO₂-a, P/CeO₂-b, and P/CeO₂-c surfaces.

Next, we calculated the adsorption PhCl molecules on the P/CeO₂-a, P/CeO₂-b, and P/CeO₂-c surfaces (Figs. 2a-c), which are called PhCl-P/CeO₂-a, PhCl-P/CeO₂-b and PhCl-P/CeO₂-c, respectively. The results show that the PhCl molecule is physically adsorbed at these surfaces. The corresponding adsorption energies are exothermic, about 0.963, 0.701, and 0.824 eV, respectively. One can note that the PhCl adsorbs most weakly on the P/CeO₂-b surface, which is attributed to the fact that the PhCl-P/CeO₂-b system has two small new states. In addition, we also find that the PhCl's wavefunction images (WF) of two new states around seventh highest occupied molecule orbital (HOMO7) are broken HOMO7 shapes, while the middle one is a whole HOMO7 shape (Figs. 2d–l). These three fully occupied states can be characterized as bonding, anti-bonding, and non-bonding orbitals from low to high energy levels according to Hückel method [11-13]. PhCl's HOMO7 and H₂PO₄ orbitals can be treated as two individual states, forming these new orbitals through coupling and overlapping, i.e., covalent or orbital interactions. Therefore, the fully occupied anti-bonding makes PhCl-P/CeO₂-b unstable.

Moreover, the dipole-dipole or Coulomb interactions on the adsorption strength of the reactant molecules must be addressed. Using the charge density difference calculations, we show a significant charge transfer between PhCl and P/CeO₂ surfaces (Figs. 3a–c) and that species with high electron density can attract species with low electron density. Aside from the attraction between O and H (Figs. 3a and c), and Cl and H (Fig. 3b), there is also an accumulation of electron density in the ortho H of Cl on the PhCl-P/CeO₂-b system (Fig. 3b). Combined with the anti-bonding image in Fig. 2, this phenomenon again confirms the existence of strong orbital interactions between PhCl and PO₄²⁻ on the PhCl-P/CeO₂-b. Though the adsorption energy exceeds the usual physical adsorption energy, the charge density difference shows no electron transfer between the adsorption site and PhCl. And the distances between PhCl and H_xPO₄ are all larger than 2 Å with PhCl no reformation. This evidence indicates that PhCl still physically adsorbs on these P/CeO₂ slabs.

From the above discussion, we can find that by constructing planar catalytic material structures, we can fully ensure that the active site orbitals match the symmetry of the PhCl π-frontier orbitals. Therefore, a new type of catalytic site consisting of three single-atom clusters (M₃ clusters) has been proposed, and these models are called single-atom cluster catalysts (SACCs), which can interact with the PhCl, e.g., the CeO₂ (111) surface anchors M₃ clusters differ from conventional transition metal close-packed surface models (Fig. S5 in Supporting information) in that M₃ can remain in a narrow state such as atomic orbitals to interact with molecular orbitals, thereby enhancing covalent interactions with a more precise energy match, whereas Fe₃ and Ru₃ are preferentially considered.

As a classical elemental, Fe was chosen due to its ability to donate electrons to N₂, which is also a similar π orbital system [14], and since the ground state of Fe₃ is ferromagnetic (Table S3 in Supporting information) with high magnetic moment values, and all calculations in this work are based on this. Considering the two adsorption configurations according to the symmetry of Fe₃ (Fig. 4), we can find that PhCl adsorption is slightly distorted: the bond angle of Cl-C-C_para (C_para represents the para-C of Cl) is 175.688° in PhCl-Fe₃/CeO₂-a (Fig. 4a) and 166.848° in PhCl-Fe₃/CeO₂-b (Fig. 4b) systems. The bond length of C-Cl is 1.753 Å in the PhCl-Fe₃/CeO₂-a and 1.740 Å in the PhCl-Fe₃/CeO₂-b systems. In addition, Bader charge calculations (Table S4 in Supporting information) show that Fe₃ transfers 0.648 |e| of electrons to PhCl on the PhCl-Fe₃/CeO₂-a and 0.666 |e| to PhCl-Fe₃/CeO₂-b, while the remaining electrons of Fe₃ are transferred to the surface. In parallel, we have calculated an in-depth analysis of the electronic structures of the more stable PhCl-Fe₃/CeO₂-a. The electron localization function (ELF) of PhCl-Fe₃/CeO₂-a (Fig. 4c) shows the weak electron localization between Fe and C atoms, and the LDOS (Fig. 4d) shows an occupied lowest unoccupied molecule orbital (LUMO), respectively. This calculation suggests that the bonding is more ionic and that Fe₃ activates PhCl by donating electrons [15].

In addition, Ru is also a commonly used catalytic material in PhCl degradation and then we also carried out a systematic study of the properties of the Ru₃ cluster (Fig. 5) [16, 17]. As similar to Fe₃, the magnetic arrangement was also tested, and the results showed that the Ru species was in a ferromagnetic state (Table S3). We first calculated the adsorption structures of PhCl on the Ru₃/CeO₂ surfaces (Figs. 5a and b). The results show that the PhCl adsorption was stronger (corresponding adsorption energies are exothermic by 2.796 and 2.829 eV), and there also have similar structures at Ru₃ (Figs. 5a and b): The bond angle of Cl-C-C_para is 160.395° in PhCl-Ru₃/CeO₂-a and 169.246° in PhCl-Ru₃/CeO₂-b, the bond length of C-Cl is 1.761 Å in PhCl-Ru₃/CeO₂-a and 1.746 Å in PhCl-Fe₃/CeO₂-b. In addition, we also find that the PhCl species get around 0.455 |e| electrons from Ru₃ in PhCl-Fe₃/CeO₂-a and 0.434 |e| on the PhCl-Fe₃/CeO₂-b surfaces (Table S5 in Supporting information), and these results indicate that the Ru₃ transfers more electrons than Fe₃. Moreover, the ELF results (Fig. 5c) show the same weak electron sharing between C and Ru sites. And the local density of states (LDOS) of PhCl and enlarged around E_f with WF indicate its electron has been spin-polarized after the PhCl adsorption (Fig. 5d and Fig. S6 in Supporting information). There are two group orbitals near E_f as WF images: The two orbitals have a p character in the Cl atom (HOMO1, Fig. S1, purple), and they are similar to the shape of LUMO1 (blue). The results also show that the PhCl accepts electrons and loses electrons simultaneously in the PhCl/CeO₂ systems, suggesting the bidirectional electron transfer mechanism of the M₃ cluster [18]. In Fig. 5d, the states and wavefunctions noted by purple and brown dots are both HOMO1-like shapes in Fig. S1 with 2p_z at the Cl atom and π shape orbitals at the aromatic ring. The purple one is below Fermi energy, and the brown one is above that, which means that HOMO1 loses electrons. On the opposite, the blue and black noted states are both LUMO1 shapes in Fig. S1 without 2p_z at Cl and π orbitals at the aromatic ring. The blue state is below Fermi energy, which indicates that LUMO1 gets electrons. Compared with the Fe₃ case, the Bader charge accumulation is much smaller in the Ru₃ case. Thus, we believe there is a bidirectional electron transfer mechanism. From the above discussion, we can learn that the PhCl can be activated at the Ru₃ site similarly to Fe₃, but the mechanism is more complex.

Interestingly, to achieve the bonding between M and C, we have also calculated a particular main group element: Boron (B), mainly because B can form new species through multi-central bonds like B-C-N compounds [19-21]. In addition, the relevant literature reports that B is often used as an additive for CeO₂ catalysts to improve the catalytic performance [22]. Therefore, we systematically calculated the PhCl adsorbed at the B₃ site on the B₃/CeO₂ surfaces (Figs. 6a–d). The results indicate that the C-B bond length is shorter, approximately 1.6 Å, compared to Fe₃ and Ru₃ clusters, typically within the range of a covalent bond. The Bader charge shows that the adsorbed PhCl molecules on the B₃/CeO₂ systems accept more electrons (Table S6 in Supporting information) than other systems (such as Fe₃/CeO₂ or Ru₃/CeO₂).

In addition, the calculated results show that the B-O bond length ranges from 1.2 Å to 1.3 Å, indicating that the loaded B₃ forms a B₃O₃ compound on the CeO₂ (111) surface, which is coordinated with Ce. PhCl can accept electrons from B₃ sites and the CeO₂ surface. We also find that the electron localized at the boron and carbon is much stronger (Fig. 6e), indicating the presence of covalent bonding between B and C species [23]. The calculated LDOS and WF of states below E_f also show two obvious C-B orbital overlaps and twisted frontier orbital shapes (Fig. 6f and Fig. S1); these results also suggest that the molecule accepts electrons through the covalent bonding. We can also find that the heavy deformation energy consumption may be why the adsorption energy is lower than that of Ru₃ and Fe₃ systems. All total and local DOS of three PhCl-M₃/CeO₂ are shown in Fig. S7 (Supporting information).

Out of the three M₃ clusters, PhCl adsorbs most strongly (-2.829 eV) on the Ru₃ site. In electron structure, according to the Bader charge analysis, the order of electrons acquisition in PhCl is B₃ (1.112 |e|) > Fe₃ (0.648 |e|) > Ru₃ (0.434 |e|) site. The smallest electron transfer can be attributed to the different bidirectional electron transfer mechanisms. As for B₃, PhCl extracts more electrons through B₃O₃ from the surface. Besides the electron structure, the adsorption energy is also determined by the molecular reformation. The first step of PhCl oxidation is believed to be the cleavage of C-Cl and the formation of phenol species, which further develop into other organic intermediates and, finally, CO₂ + H₂O [24]. The C-Cl bond length indicates the possibility of C-Cl dissociation. The order is Fe₃ (1.753 Å) > Ru₃ (1.746 Å) > B₃ (1.733 Å). This trend reveals that PhCl on the Fe₃ site dissociates Cl most easily. But another factor influencing oxidation reaction activity is the C atom bond with Cl, which drives the formation of phenol intermediate. The Bader charge of three C can reveal the activity of this (Tables S4-S6 in Supporting information): Fe₃ (C4: 0.114 e⁻), Ru₃ (C4: 0.005 e⁻), and B₃ (C4: 0.096 e⁻). PhCl on Ru₃ has the most positive charged C4. Thus, PhCl on Ru₃ can be the most active reactant for the next oxidation step.

Compared to PhCl on H_xPO₄/CeO₂, PhCl performs much differently on M₃ sites. First, the PhCl is deformed on Fe₃ or Ru₃ sites, which is different from PhCl on P/CeO₂. Second, The LDOS of PhCl in Fe₃ or Ru₃ cases shows an electron transfer, which is not observed in P/CeO₂ surfaces. Third, the adsorption ability of Fe₃ and Ru₃ sites is much stronger than P/CeO₂ surfaces. These characters illustrate again that Fe₃ and Ru₃ sites chemically bond with PhCl and P/CeO₂ adsorb PhCl physically.

So, from the above discussion, we can see that the PhCl adsorbed at the Fe₃ and Ru₃ sites donates electrons to PhCl and binds to it through stronger ionic interactions, whereas PhCl acts mainly with covalent bonding at the B₃ site. Secondly, we can also find that the C atom of the deformed PhCl can interact with boron and break its stable p-π conjugation. The discrepancy may be attributed to the overlap integral (S) and coupling matrix elemental (V) related to the orbitals of M₃ and the frontier orbitals of PhCl. In particular, the values of the wider sp orbitals of B₃/CeO₂ are all larger, enhancing the covalent character of the B-C bond according to the Newns-Anderson model [25].

In this work, the physical adsorption interaction between PhCl and different phosphorylated CeO₂ (111) was studied deeply. A covalent interaction in one case guided us to design a class of tri-cluster M₃ adsorption sites to activate PhCl in theory. The analysis involves three clusters: Fe₃, Ru₃, and B₃. Interestingly, in addition to traditional transition metal elements that transfer electrons and form more ionic bonds, boron atoms share electrons with carbon atoms and form more covalent bonds. The broken of p-π conjugation shall activate PhCl and make molecular easier to dissociate C-Cl bond. This will avoid chlorinated organic species at first reaction step. Our research does not only guide a catalyst design for the PhCl degradation but also provides a template or rational descriptor (S or V) for designing catalyst sites for complex molecules according to fundamental theory.

Declaration of competing interest

All authors declare that there are no conflicts of interest, financial or otherwise in this work; and there are no other relationships or activities that can appear to have influenced the submitted work.

CRediT authorship contribution statement

Jin Li: Writing – original draft, Software, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Xin Chen: Methodology, Investigation. Aling Chen: Writing – review & editing, Writing – original draft, Supervision, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Zhi-Qiang Wang: Writing – review & editing, Methodology, Investigation, Formal analysis, Data curation. Dengsong Zhang: Supervision, Data curation, Conceptualization.

Acknowledgments

The authors acknowledge the support from the National Key Research and Development Program of China (No. 2023YFA1508500) and the National Natural Science Foundation of China (No. 22276120).

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2024.110527.