CeO2-based rare earth materials are widely used as the main component in three-way catalysts for automobile exhausts emission control by converting hazardous hydrocarbons (HCs), CO, and nitrogen oxides (NOx) to H2O, CO2, and N2, which are much less harmful to the environment [1, 2, 3, 4]. It is generally believed that the excellent catalytic performance of CeO2 is related to the unique 4f orbital of Ce, which has strong localization characteristics [5]. Owing to such special electronic properties, CeO2 is considered to be an electron storage component with high oxygen storage capacity (OSC) that can also help release/store oxygen by forming/filling oxygen vacancies under reducing/oxidizing conditions [6, 7]. Accordingly, a large number of studies have been conducted to investigate the activity of surface- and bulk-oxygen in various reactions [8, 9, 10, 11, 12]. Different experimental strategies have been used to make CeO2 surfaces more active, such as metal-ion doping and preparation of CeO2 with various nanoscale morphologies [13, 14, 15, 16, 17]. Moreover, it has been shown that the addition of Zr to CeO2 can greatly increase its OSC, that the best ratio of Zr/Ce is 1:1 and that the Ce0.5Zr0.5O2 solid solution in the specific κ-phase can give the highest OSC [7, 18].
The three common crystal facets exposed on nanocrystalline CeO2 are (100), (110), and (111), and the shape of the CeO2 nanoparticles, which can determine the type and relative amount of exposed facets, will then influence their reactivities in applications [11, 13, 19]. Single-crystal CeO2 nanorods with well-defined crystal planes have been designed and prepared, and are predominantly covered with (110) facets. These nanorod materials exhibit higher activity for CO oxidation than CeO2 particles, which mainly contain (111) facets [13, 20]. These results suggest that the (110) surface, which is the second most stable surface among the low-index ceria surfaces, may exhibit higher chemical activities than the other surfaces. In several recent studies it has been reported that once oxygen vacancies formed on the CeO2(110) surface, the surface would be more active and able to interact with reductive species such as NO and NO2, which would fill the O vacancies of the surface with their O atoms. On the other hand, oxidative species such as CO can be oxidized by surface O atoms to generate CO2 and surface O vacancies. Such a surface redox cycle suggests that CeO2 not only acts as supporting material for catalytic reactions, but is also involved in surface reactions [21, 22, 23, 24].
Au catalysis has been a hot topic in heterogeneous catalysis over the last few decades, and highly dispersed Au particles supported on different metal oxides have been shown to be extremely active for CO oxidation [25, 26, 27, 28], the water-gas-shift reaction (WGS) [29, 30, 31], hydrocarbon oxidation [32], and NO reduction at low temperatures [33, 34]. Among the metal oxide supports tested for Au nanoparticles, CeO2 has been found to be one of the most active in many catalytic applications [35, 36]. Liu et al. [37] carefully investigated the activity of a single Au atom supported on the CeO2(111) surface in the WGS reaction. From their density functional theory (DFT) calculations, electron transfer to the localized nonbonding 4f orbital of CeO2(111) was found to induce the oxidation of Au, and the positively charged Auδ+ species was determined to be the site for strong adsorption of CO and subsequent reactions. In the case of CO oxidation, Wang et al. [26] suggested that gold with a wire-like or several other specific configurations could efficiently adsorb CO and O2 and promote CO oxidation in both supported and unsupported Au systems. In other studies, the interfaces between Au nanoparticles and oxide supports, as well as the structural flexibility, have been suggested to be important for low-temperature CO oxidation [27, 38, 39].
In recent years, Au-based catalysts have also attracted great interest in the field of NOx decomposition. Wang et al. [40] studied N2O formation on the Au(111) surface and found that the coupling of two NO molecules to form N2O2 was preferred and its subsequent dissociation was more likely to occur than single NO molecule dissociation. In addition, Zhang et al. [41] concluded that NO could react with pre-covered oxygen on the Au(111) surface to form NO2 under different O coverages, and, because of the very weak adsorption of atomic oxygen on the surface, such NO oxidation by pre-covered oxygen proceeds via the Eley-Rideal mechanism.
Rare-earth oxides and Au nanoparticles have extensive applications, and have attracted very wide research interest. In particular, supported Au/CeO2 catalysts have been shown to be able to simultaneously remove CO and NOx, which are both important air pollutants, in vehicle emission control applications [42, 43, 44]. Compared with the most common process of CO oxidation through reaction with O2, the CO + NOx reaction may simultaneously eliminate both the pollutants, which could bring more environmental benefits than the CO + O2 reaction. In this study, we investigated the related chemical processes and reactions catalyzed by a six-atom Au nanocluster (Au6) supported on the CeO2(110) surface with DFT calculations, with the aim of elucidating the possible reaction pathways between CO and NOx in the corresponding systems. Detailed steps of the reactions, including adsorption and all the possible surface pathways of reactants, were considered. In addition, the role of the unique electronic structures of the CeO2 supports in the various catalytic reaction steps were carefully studied and are discussed.
The DFT calculations were performed with a Hubbard-U term (DFT + U) to correct for the description of the electronic properties of ceria [45, 46], where U = 5 eV was applied to the Ce 4f states [47, 48, 49]. The Vienna Ab initio Simulation Package was used. The valence electronic states of Ce (4f, 5s, 5p, 6d, 6s), Au (5d, 6s, 6p), O (2s, 2p), C (2s, 2p), and N (2s, 2p) were described by plane-wave basis sets with a cut-off energy of 350 eV, and core-valence interactions were treated with the projector augmented wave method [50, 51]. The PW91 electron exchange-correlation functional of the generalized gradient approximation was used [52, 53]. The transition states (TSs) in the reactions were determined by the constrained minimization method [54, 55] and electron charge analyses were performed by Bader decomposition of the charge density [56].
The CeO2(110) surface was modeled as a periodic slab with seven CeO2 layers, with the bottom two layers fixed, while the other layers were allowed to relax during all the calculations. The force threshold of all the relaxed atoms was set to 0.05 eV/Å. Spin polarization was considered in all the calculations. The vacuum height between slabs was ~10 Å to eliminate the interaction between neighboring slabs. The CeO2(110) surface was built by a 2 × 3 supercell, and we removed a whole CeO2 row along the [1(_)10] direction to model the characteristic structure of reconstructed CeO2(110) determined by experimental measurement [57] (Fig. 1(a)). The supported Au nanoparticle was constructed using a six-atom cluster, which was placed and optimized only on one side of the slab (Figs. 1(b) and (c)). The Brillouin zone was sampled with a k-point meshes of 1 × 1 × 1 because of the large size of the surface cell (~11 × 11 Å2).
To determine the chemisorption energies (Ead) of adsorbed molecules to the supported catalyst, we used the following equation:
Ead = -(Emolecule + Au6/CeO2 - EAu6/CeO2 - Emolecule)
in which Emolecule + Au6/CeO2 mol is the total energy of the system involving the supported Au6/CeO2(110) catalyst with the adsorbed molecule, EAu6/CeO2 is the total energy of the supported catalyst, and Emolecule is the total energy of the molecule in the gas phase.
Automobile exhaust contains mainly CO, NOx, and HCs. Much attention has been paid to the disposal of CO through low-temperature catalytic oxidation. As another important process in emission control, NOx decomposition is still challenging and its by-products, such as nitrite and nitrate, usually exist in the processes and may reduce the overall activity of the catalysts. It is widely expected that the direct reaction between CO and NOx would be highly beneficial because it can simultaneously remove both compounds. Therefore, the current work aims to study the possible reaction pathways of the reaction of NO with CO on Au6/CeO2(110) by DFT+U calculations.
Scanning tunneling microscopy (STM) of the CeO2(110) surface structures has been conducted by Nörenberg and co-workers [57]. They found that rows of CeO2 in the STM images were separated by ~11 Å, which indicates a 2 × 1 surface reconstruction. In Fig. 1(a), we show the structures of the CeO2(110) surface with such a missing-row reconstruction mode. It exposes three-fold coordinated O (O3c) and six-fold coordinated Ce (Ce6c) atoms on the top surface layer as well as Ce7c and O3c atoms in the missing-row area, which can be actually taken as (111) micro-facets.
In our previous work, we investigated the trimeric Au nanocluster (Au3) on the unreconstructed CeO2(110) surface [58]. In that work, we showed that the supported Au3 cluster could reduce the surface, and the electron donated by Au3 could localize on one surface cation (Ce3+) and further promoted O2 adsorption and activation by forming negatively charged O2-. These results clearly suggest the possible catalytic activities of such CeO2 supported systems. In this work, we simulated the Au6 cluster supported on the reconstructed CeO2(110) surface. From the calculated structure (Fig. 1(b)), the supported Au6 cluster exhibits a 2 + 4 double layer configuration, with the bottom 4 atoms binding to the missing-row site O3c at different heights and the top two atoms sitting on top of the 4-atom bottom layer not directly interacting with the CeO2 surface. The adsorption energy of the Au6 cluster was calculated to be 4.24 eV with respect to the same 6-atom Au cluster optimized in the gas phase, which is rather high for supported Au clusters. With respect to adsorbed Au6, the average bond distance between the bottom Au atoms of the cluster and surface O3c atoms that they directly interact with was measured to be ~2.17 Å, and the adsorption has very small effect on the structural parameters of the CeO2 support.
From the calculations, we determined that Au6 adsorption induced partial reduction of the supporting CeO2(110) surface. As shown in Fig. 1(c), there are two localized 4f electrons on two neighboring Ce6c atoms of the top of the CeO2 row directly bonding with the Au6 cluster, which is consistent with our early study [58] and the work reported by several other groups regarding transition metal clusters supported on CeO2 surfaces [59, 60].
On this supported Au6 cluster, there are three main types of adsorption sites (Fig. 1(c)): (1) the top site on a single Au atom that is not interacting with the CeO2 support (T); (2) the bridge site between the two top Au atoms (B1) and between one top and one bottom Au atom (B2); and (3) the hollow site surrounded by one top Au and two bottom Au atoms (H). In this work, the adsorption energies of CO and NO were calculated for adsorption to the above Au sites as well as to interfacial Ce cations.
The adsorption of CO and NO on transition metals (including Au) has been extensively investigated in many experimental and theoretical studies [61, 62]. In our recent study of the structures and activities of Au nanoclusters on TiO2 supports [28], we found that CO preferred to adsorb on the top site of Au. In the current work, we again found that the top site (T) was the most favorable adsorption site for both CO and NO on CeO2(110)-supported Au6. In Table 1, we list the calculated adsorption energies (Ead) of these two molecules on the various adsorption sites. Adsorption of CO (Fig. 1(d)) and NO (Fig. 1(e)) on the top site of Au6 was much stronger than on the other Au sites or Ce3+ at the interface. Moreover, the calculated Ead of CO (1.20 eV) was significantly higher than that of NO (1.03 eV), which suggests that there is competitive adsorption between CO and NO on supported Au6, and that adsorption of CO is more energetically favorable than NO.
In addition to the adsorption of a single NO and CO molecule, we also considered their adsorption at higher coverages. As shown in Table 1, the co-adsorption of two CO molecules on the two top Au sites had a total adsorption energy of 2.06 eV (Fig. 1(f)). Interestingly, for the co-adsorption of two NO on the top of the Au6 cluster, the calculations showed that they could actually form a N2O2 dimer (Fig. 1(g)), and the adsorption energy estimated with respect to two gas-phase NO molecules was 1.56 eV. From the above results, we can therefore conclude that adsorption of CO is much more energetically favorable than NO on supported Au6 at both low and high coverages.
Rodriguez et al. [63] found that Au supported on metal oxides can act as bi-functional catalysts for the WGS reaction, i.e., the adsorption and dissociation of water take place on the metal oxides while CO adsorbs to Au, and the subsequent reactions occur at Au/metal oxide interfaces. In this work, we also tested if there were other favorable adsorption configurations for NO when the supported Au6 was covered with CO.
Our calculated results reported in Section 3.2 have already shown that a single NO molecule only weakly adsorbs to interfacial Ce3+ sites (0.44 eV, Table 1). In contrast, when two NO molecules adsorbed to two neighboring Ce3+ atoms, they were found to be able to readily couple to form the N2O2 dimer with almost no energy barrier. Four possible adsorption structures of the N2O2 dimer were determined at the interface, which are shown in Fig. 2. Moreover, the Ead calculated with respect to the energies of the gas-phase dimer and the surface with the CO pre-covered Au6 cluster are listed in Table 2.
From the calculated adsorption structures and energies, the N2O2 dimer only binds to the interfacial Ce3+ atom, and it prefers an O/O-down configuration, in which the two O atoms of the dimer bind with two Ce3+ atoms (Fig. 2(b)). The adsorption energy was calculated to be 1.24 eV, which is significantly higher than that of the dimer in N/N-down (0.73 eV, Fig. 2(a)) or N/O-down (0.92 eV, Fig. 2(c)) configurations. Interestingly, we also located the adsorption state in which the dimer sits across the interface with one O atom bound to a Ce3+ atom and one N atom bound to an interfacial Au atom, and the corresponding Ead was calculated to be 0.98 eV (Fig. 2(d)).
Au nanoparticles exhibit catalytic activities in low- temperature CO oxidation by O2, and the detailed mechanisms have been extensively studied both experimentally and theoretically [64, 65]. However, the study of CO oxidation by NO is still limited. In this work, we systematically calculated the reaction pathways for the reaction of CO and NO on the Au6/CeO2 supported catalytic system.
According to our calculations regarding the adsorption of CO and NO in this system, NO is unable to compete with CO for the adsorption sites on the supported Au6 cluster under low and high coverages. In contrast, in the system with the Au cluster already covered with CO, two NO molecules can still form a strongly adsorbed N2O2 dimer at the interface. Accordingly, we studied the redox reaction between CO and N2O2 taking the co-adsorbed system with one CO on top of a Au atom and N2O2 in the bridging configuration at the Au6/CeO2(110) interface with the terminal O atom pointing toward CO as the initial state (IS1, Fig. 3, Fig. 2(d)). As shown in the calculated energy profiles of the reaction (Fig. 3), the co-adsorption energy of CO and N2O2 in IS1 was calculated to be 2.18 eV. When the reaction occurs, CO moves toward N2O2 to combine with the terminal O atom for CO2 formation. In the corresponding transition state (TS1, Fig. 3) located through the calculations, the distance of OC-ONNO was determined to be 1.785 Å, the C-O bond in CO increased to 1.340 Å, and the distance of the terminal O-N bond in N2O2 increased to 1.341 Å.
By comparing the structures of TS1 and IS1, the Au atom occupied by the CO molecule actually moves with the CO molecule in the reaction, which may facilitate its approach toward the N2O2 molecule. Accordingly, from their calculated energies, the reaction barrier of this process was determined to be only 0.39 eV, suggesting that this combination would easily occur under mild conditions. This result is consistent with the work of Wang et al. [40], who found that formation of N2O from the N2O2 dimer could easily occur on the Au(111) surface.
After TS1, a relatively stable intermediate complex OCONNO (IM1, Fig. 3) was determined to occur at the interface. The bond between the C atom and the terminal O atom of N2O2 decreased to 1.417 Å, while the terminal O-N bond of N2O2 further increased to 1.437 Å. Compared with TS1, this intermediate state was calculated to be 0.19 eV more stable. From Fig. 3, it then needs to go through another transition state (TS2) to fully break the terminal N-O bond to form CO2. The energy barrier corresponding to this second TS (TS2) was 0.80 eV, and the distances between the terminal O atom and OC and NNO were determined to be 1.318 and 1.858 Å, respectively. After TS2, N2O and CO2 form and naturally desorb because of their very weak adsorption to the surface (~0.2 eV for N2O and ~0.1 eV for CO2). Nevertheless, the final state (FS1) with gas-phase N2O and CO2 is still remarkably stable, and, as shown in Fig. 3, the overall process for CO oxidation by adsorbed N2O2 is highly exothermic on the Au6/CeO2 catalyst.
As another important hazardous nitrogen oxide species, N2O and its decomposition has been the topic of many studies [66, 67, 68]. In particular, it has been found that transition metal catalysts can promote the activation and dissociation of N2O [69, 70, 71].
In this work, we also calculated the interaction of N2O with the supported catalyst and its dissociation. We found that N2O did not have a strong interaction with Au6 or the interface of Au6/CeO2(110) (Ead = ~0.2 eV). Therefore, direct N2O dissociation by transfer of the terminal O atom to the supported Au6 cluster was studied. Fig. 4 shows the calculated structures of the important states during the dissociation and the corresponding energy profile. The initial state of dissociation (IS2) involves weakly bonded N2O with the O atom pointing toward the Au6 cluster. In the transition state (TS3), the length of the dissociating O-N bond was determined to be 1.372 Å, and the reaction barrier was 1.78 eV, suggesting that dissociation is kinetically unfavorable, which is in agreement with experimental results that N2O decomposition on Au-based catalysts require high temperatures of ~500-600 K [72, 73, 74].
After TS3, the terminal O atom of N2O would adsorb on the bridge site (B2) of the Au6 cluster and the N2 molecule would leave into the gas phase. The above N2O dissociation process for N2 formation was exothermic by 0.64 eV (Fig. 4).
Considering that CO may prefer to adsorb to the Au cluster because of its strong adsorption, we also investigated another pathway for N2O reduction by the direct reaction of N2O with a pre-adsorbed CO molecule. Fig. 5 shows the calculated structures of the transition state (TS4) of the reaction between gas-phase N2O and CO adsorbed on the top site of Au6 together with the energy profile for the complete process. In TS4, the distance between the terminal O atom of N2O and the C atom of CO was determined to be 1.775 Å, and the O-NN bond distance increased from ~1.200 Å in gas-phase N2O to 1.509 Å. After the transition state, N2 and CO2 form and desorb as the desired products. From the energy profile, this process has an energy barrier of 1.86 eV and is highly exothermic by 2.25 eV, suggesting that it is favorable, although high temperatures are needed.
Au nanoparticles supported on CeO2 substrates have been extensively studied for WGS and CO oxidation reactions. In our recent work [58], we investigated O2 adsorption on CeO2(111), (110), and (100) surfaces with a supported Au trimer cluster (Au3). It was found that the Au3 cluster could reduce the supports by transferring one electron to be localized in the 4f orbital of one surface Ce cation, turning it into Ce3+. Moreover, the O2 molecule can only adsorb to a Ce cation of the CeO2(110) support, which exhibits a flat open morphology without protruding lattice O atoms, by accepting the localized electron into its anti-bonding 2π state. In other words, the O2 molecule actually becomes an activated O2- species, and the relatively strong electrostatic attraction between O2- and the Ce4+ cation largely contributes to its favorable adsorption.
In this work, NO adsorption was found to be significantly weaker than CO adsorption on supported Au6 at both low and high coverages. However, NO adsorption exhibits unique properties at the Au6/CeO2 interface. Similar to CO, a single NO molecule has relatively low adsorption energy at the interface. Interestingly, two NO molecules can easily combine to form the N2O2 dimer at the interface with a calculated adsorption energy of ~1 eV. In addition, such favorable adsorption of N2O2 is not affected by the co-adsorption of CO on the Au6 cluster, suggesting that there is not competitive co-adsorption of the two reactants in these two areas.
By performing electronic analyses, we found that these results regarding the adsorption of N2O2 could be explained by the characteristic electronic properties of the CeO2 support. As shown in Fig. 1(c), the adsorption of Au6 causes the reduction of CeO2(110) by transferring two electrons to the support, which are localized in 4f orbitals of two surface Ce atoms beside the Au6 cluster. Interestingly, the occurrence of these Ce3+ cations is not affected by the adsorption of CO on the Au6 cluster (Fig. 6(a)). The localized electrons can then be transferred to N2O2 to increase its stability (see below) and its electrostatic attraction with the support Ce4+ cations. For instance, as shown in Fig. 6(b), N2O2 adsorbed at the interface to one Ce atom and one Au atom accepts one localized electron while the other electron is still on the nearby free Ce3+ cation. These results clearly show the unique catalytic role of the localized 4f electron of Ce that, as well as directly transferring to molecules like O2, can promote the formation of intermediate species, such as negatively charged N2O2, and strong electrostatic attraction to the surface. It also needs to be emphasized that the formation of negative N2O2 as the result of electron transfer from surface Ce3+ can actually strengthen the N-N bond within the dimer [75]. This is particularly desirable for the treatment of NO-based hazardous molecules because it would greatly increase the chance of the occurrence of N2.
From our calculations of the detailed processes of N2O2 reaction with CO, we have shown the important catalytic properties of Au nanoparticles, especially from a geometric point of view. As we have mentioned, in the CO + N2O2 → N2O + CO2 reaction, the high mobility of the Au atom that binds CO greatly facilitates the movement of CO toward N2O2, so that the corresponding barrier of their reaction is only ~0.4 eV. Moreover, in the case of the CeO2(110)-supported Au6 cluster, the above results indicate that this unique mobility will not be affected by oxidation of the cluster.
We performed DFT + U calculations to investigate CO + NOx redox reactions on a Au6 cluster supported on the reconstructed CeO2(110) surface. The results showed that the Au6 cluster could reduce the surface through donation of two localized 4f electrons to two neighboring surface Ce cations at the anchoring site. CO adsorbs significantly stronger than NO to the supported Au6 catalyst under different coverages, and it may cover all available sites on the cluster. Interestingly, adsorbed CO has no effect on the electron localization of the supporting CeO2(110) surface, and the localized electron can be transferred to NO to facilitate the formation and adsorption of negatively charged N2O2. For the CO + N2O2 reaction, we found that CO adsorbed on the top site of the Au6 cluster could react with a N2O2 dimer adsorbed at the Au-CeO2 interface with an overall barrier of ~1.0 eV to produce CO2 and N2O through a Langmuir-Hinshelwood mode. For the subsequent N2O decomposition process, we determined that N2O direct dissociation could occur by an O atom of N2O transferring to the Au cluster or a CO molecule pre-adsorbed on the Au cluster, which both have an energy barrier of ~1.8 eV, through an Eley-Rideal mode because N2O has a rather weak interaction with Au and CeO2. This work shows that the electron transferred from Ce3+ to the N2O2 dimer can strengthen the N-N bond to facilitate N2 formation, suggesting that the unique electronic properties of CeO2-based rare earth materials may make it useful for the elimination of NO and its derivatives.
稀土材料CeO2作为三效催化剂(TWC)的主要成分被广泛地用于汽车尾气排放控制. 其目的是将有毒有害的碳氢化合物(HCs), CO和氮氧化合物(NOx)转化成对环境无害的H2O, CO2和N2[1, 2, 3, 4]. 通常认为, CeO2出色的催化性能与其独特的Ce原子4f轨道强局域特征有关[5]. 正是由于这种特殊的电子结构, CeO2能够作为电子存储器在还原/氧化的条件下形成/填补氧空穴从而释放/储存O2, 因此CeO2具有很高的储放氧性能(OSC)[6, 7]. 因此, 大量研究探讨了CeO< sub>2在不同反应中表面氧和体相氧的活性[8, 9, 10, 11, 12]. 例如, 利用金属离子掺杂的方法或者制备出不同纳米尺寸形貌的CeO2都有助于提高催化活性[13, 14, 15, 16, 17]. 通过将Zr元素添加入CeO2材料中能明显提高其OSC性能, 最佳的Zr/Ce比例为1:1, 而且以κ相结构的Ce0.5Zr0.5O2固溶体表现出了最优的储放氧性能[7, 18].
CeO2纳米晶体中最常见的三个晶面是(100), (110)和(111)面, 而其形貌决定了暴露晶面的类型和数量, 影响着它在反应中的活性[11, 13, 19]. 通过设计和制备出的单晶纳米棒状的CeO2可以暴露出大量(110)晶面, CO催化氧化反应表明, 这种纳米棒状材料比纳米颗粒形貌的CeO2表现出了更高的催化活性, 而后者主要是由稳定性较高的(111)晶面构成的[13, 20]. 这表明在低指数晶面中, 第二稳定的CeO2(110)面表现出更高的活性. 最近的研究发现, CeO2(110)一旦形成氧空穴之后, 表面将更加活泼并与可还原性的气体NO, NO2相互作用, 这些气体分子中的O原子能填补CeO2表面的氧空穴. 相反地, 可氧化性物质CO能够被CeO2表面的晶格氧氧化成CO2而再次产生表面的氧空穴. 因此, 表面氧化还原意味着CeO2不仅作为催化反应中的载体材料, 而且自身能够参与到表面反应当中[21, 22, 23, 24].
与此同时, Au催化剂在过去数十年间在多相催化领域中成为一个研究的热点. 氧化物载体负载的高分散Au颗粒在CO氧化[25, 26, 27, 28], 水汽变化(WGS)[29, 30, 31], 烃类氧化[32]和NO还原[33, 34]等反应中都表现出极高的催化活性. 在Au纳米颗粒负载的氧化物载体中, CeO2是最具化学活性的载体之一[35, 36]. Liu等[37]详细研究了单个Au原子负载CeO2(111)的WGS反应. DFT计算结果表明, Au原子上的部分电子能够转移到CeO2(111)中高度局域的非键Ce-4f轨道从而引起Au颗粒氧化, 带正电荷的Auδ+物种提供了CO分子的强吸附位并促进了后续的反应. 在CO氧化的研究中, Wang等[26]提出, 线性结构和其他一些特殊构型的Au纳米簇能有效地吸附CO和O2分子, 提高CO氧化的能力. 研究还发现, Au与氧化物载体的界面以及Au团簇的结构流动性同样对CO低温氧化起到非常重要的作用[27, 38, 39].
在最近几年, Au基催化剂在NOx降解的应用领域也引起了研究者们的广泛兴趣. Wang等[40]研究了Au(111)表面N2O的形成机理. 他们发现, 相比单个NO分子的解离, 两个NO分子聚集成N2O2之后更容易解离产生N2O. 除此之外, Zhang等[41]认为, 在Au(111)表面的不同O原子覆盖度的情况下, NO将与O反应生成NO2. 这是因为O原子在表面的吸附作用很弱, 与NO的氧化反应是通过Eley-Rideal机理发生的.
综上所述, 稀土氧化物和Au纳米颗粒都有着广泛的应用和研究价值. 尤其是负载型Au/CeO2催化剂在汽车尾气排放控制这一领域的应用. 这类催化剂能够同时消除CO和NOx这两类主要空气污染物[42, 43, 44]. 相比最为主流的采用O2氧化来消除CO, CO + NOx的相互反应能够同时消除这两种污染物, 显然更具环保效益. 然而, 该类型反应的研究却是比较有限的. 为了阐明CO与NOx之间相互作用的可能的反应路径, 本文通过建立CeO2(110)负载的6个Au原子簇(Au6)模型, 用Hubbard-U校正的密度泛函理论(D FT + U)研究了相关的表面化学过程和催化反应. 详细考察了各种反应步骤, 包括吸附和所有可能的表面反应路径. 此外, 我们还仔细探讨了CeO2在不同催化反应步骤中, 其独特的电子结构所体现的作用.
本文采用了DFT + U计算方法以准确描述Ce元素的电子结构[45, 46], 其中U = 5 eV作用到Ce的4f轨道[47, 48, 49]. 采用的是Vienna Ab-initio Simulation Package程序包, Ce (4f, 5s, 5p, 6d, 6s), Au (5d, 6s, 6p), O (2s, 2p), C (2s, 2p)和N (2s, 2p)作为价电子组态, 用投影缀加平面波的方法描述核-价电子的相互作用, 平面波基组展开的截断能取350 eV[50, 51]. 同时, 采用广义梯度近似框架下的PW91作为交换相关泛函[52, 53]. 计算的反应过渡态(TS)采用了约束最小化方法[54, 55], 并用Bader电荷密度分解来分析反应中的电子结构变化[56].
我们构建了7层的CeO2(110)板层模型, 最底部的两层原子被固定而其他原子做结构弛豫. 弛豫原子力的收敛标准设定为0.05 eV/Å. 在计算中同时考虑了自旋极化效应. 在两个板层模型之间的真空层高度选取了~10 Å以消除相邻板层之间的相互作用. CeO2(110)表面由2 × 3的超晶胞模型构建, 并根据实验[57]沿着[1(_)10]方向去除一条CeO2来模拟出重构(110)表面的特征结构(见图1(a)). 构建了6个Au原子模型来模拟Au纳米团簇, 将其负载于CeO2一侧的板层模型上(见图1(b)和(c)). 鉴于整个超晶胞表面大小为~11 × 11 Å2, 因此布里渊区的k点取值为1 × 1 × 1.
吸附分子在催化剂表面的化学吸附能的计算公式为:
Ead = -(Emolecule +Au6/CeO2-EAu6/CeO2-Emolecule)
其中, Emolecule + Au6/CeO2为Au6/CeO2(110)与吸附分子相互作用的体系总能量, EAu6/CeO2为负载型催化剂的体系总能量, Emolecule为气相分子的能量.
汽车尾气中主要含有CO, NOx和HCs三种主要成分, 人们多关注于CO低温催化氧化的处理上. 作为汽车尾气控制另外一个非常重要的过程, NOx的降解处理仍然是一个挑战. 另外, 反应的副产物诸如硝酸盐和亚硝酸盐也会降低整个催化剂的活性. 与此同时, 通过CO与NOx的直接反应能同时消除这两类空气污染物, 所以这类方法是非常高效的. 因此本文通过DFT + U计算研究了NO与CO在Au6/CeO2(110)表面上反应的可行路径.
Nörenberg等[57]通过扫描隧道显微镜(STM)观察了CeO2(110)的表面结构. 结果发现, 每条CeO2晶列之间的间隔为~11 Å, 表明该表面经历了2 × 1扩展的重构过程. CeO2(110)表面失去一条CeO2结构的重构模型如图1(a)所示. 最表层的CeO2暴露了3配位的O原子(O3c)和6配位的Ce原子(Ce6c), 而在缺失一排CeO2的局部表面处, 暴露了7配位的Ce原子(Ce7c)和3配位的 O原子(O3c), 此处也能看作成(111)晶面的微观结构.
在前期工作中, 我们计算了3个Au原子组成的团簇(Au3)在CeO2(110)单晶表面的吸附[58]. 结果发现, Au3纳米团簇能够还原CeO2表面, 并且Au3上的电子可以转移并局域到表面邻近的Ce4+上, 将其还原到Ce3+, 同时进一步促进O2的吸附并将O2活化成带负电荷的O2-. 这充分说明了CeO2负载型催化剂的催化活性. 本文模拟了Au6纳米团簇负载于重构CeO2(110)表面结构. 优化后的结构如图1(b)所示, 负载的Au6纳米团簇呈现了2 + 4的双层原子构型. 底层的4个Au原子与失去一排CeO2晶列处的不饱和O3c成键, 其高度位置略有差异. 而顶端的2个Au原子位于底层4个Au原子的上方, 其配位数较少, 与CeO2表面没有任何直接的相互作用. 经计算得到Au6团簇的吸附能高达4.24 eV (参考气相中该Au6原子团簇优化后的能量), 与CeO2载体有相互作用的底层4个Au原子与表面直接相连的O3c原子的平均距离只有2.17 Å. 因此, 对于负载的金属来说, Au6与载体有很强的相互作用. 从结构上来看, Au6纳米颗粒的负载对于CeO2(110)载体的结构形变影响也很小.
通过计算, 我们再次确定了Au6纳米团簇的吸附引起了CeO2(110)载体表面部分还原. 正如图1(c)所示, 在与Au6团簇成键的那条表面CeO2晶列上, 两个邻近的表层Ce6c上出现了两个4f局域电子, 与本课题组之前的计算结果[58]和其他课题组计算的过渡金属团簇负载于CeO2表面的结果高度一致[59, 60].
在负载的Au6团簇颗粒上, 主要存在着3种可吸附位点(见图1(c)), 分别是(1)与CeO2载体无任何相互作用的独立Au原子的顶位吸附位点(T); (2)处于两个顶层Au原子之间的桥位吸附位点(B1)和处于一个顶层Au原子和一个底层Au原子之间的桥位吸附位点(B2); (3)处于两个底层Au原子和一个顶层Au原子之间的三重空位吸附位点(H). 本文计算了CO和NO分子在上述Au团簇吸附位以及界面Ce阳离子上的吸附能.
事实上, 许多实验和理论计算都研究了CO, NO分子在过渡金属包括Au催化剂上的吸附情况[61, 62]. 本课题组近期研究了Au纳米颗粒负载于TiO2载体的结构和催化活性[28], 我们发现CO的最佳吸附位点是在Au纳米颗粒的顶位. 本文再次发现, CO和NO分子在Au6/CeO2催化剂上最佳吸附位置依然在Au的顶位处(T). 表1列出了CO和NO分子在Au6/CeO2催化剂上不同吸附位上的吸附能(Ead). 通过比较, CO (图1(d))和NO (图1(e))分子在Au 团簇顶位处的吸附都明显强于各自在Au团簇其他的吸附位或是载体Ce3+上的吸附位. 而且CO的吸附能(Ead = 1.2 eV)明显高于NO (Ead = 1.03 eV), 说明CO和NO在负载的Au6纳米团簇上还存在竞争吸附的关系, 而前者与Au原子的相互作用更强, 更有利于CO在Au6上的吸附.
除了单个CO和NO分子在Au6/CeO2(110)表面上吸附, 我们同样考虑了两者在高覆盖度下的吸附情况. 如表1所示, 2个CO分子在2个Au顶位处的共吸附能为2.06 eV (图1(f)). 有趣的是, 当2个NO分子共吸附在2个Au原子顶位处的时候, NO能自发地结合成N2O2二聚体的结构(图1(g)), 算得其吸附能为1.56 eV (参考2个NO气相分子的能量). 由此可见, 无论是在低覆盖或者高覆盖度的情况下, CO都比NO更容易占据Au6团簇上的吸附位.
针对WGS反应, Rodriguez等[63]发现Au负载于氧化物载体是一类双功能催化剂, 即CO分子在Au上吸附而H2O的吸附和解离则发生在氧化物表面, 后续的反应发生在金属-载体的界面. 因此本文也考虑了当Au6团簇上被CO覆盖之后, NO能否与表面其他吸附位有强的相互作用.
前文的计算结果表明, 单个NO分子与界面Ce3+的作用依然非常弱. (Ead = 0.44 eV, 表1). 然而, 当2个NO在2个邻近的Ce3+离子上共吸附时, 它们能非常容易地结合, 从而形成N2O2二聚体的结构, 该过程几乎没有能垒.
于是, 我们又计算了N2O2分子在CeO2载体上4种不同的吸附构型, 见图2. 其吸附能(Ead)的计算以气相N2O2分子与Au6预吸附CO的表面为参考, 如表2所示.
由计算得到的结构和吸附能, 我们发现当N2O2二聚体分子仅与界面处Ce3+相互作用时, 最好的吸附结构是O/O原子端朝下与表面的2个Ce3+成键(图2(b)), 该结构的吸附能为1.24 eV, 明显强于N/N原子端朝下吸附的构型(0.73 eV, 图2(a))或者N/O原子端朝下吸附的构型(0.92 eV, 图2(c)). 有趣的是, 我们还计算了一种N2O2横跨Au-CeO2界面的吸附构型, 末端一个O原子与载体表面Ce3+成键, 而另一头的N原子与一个处于界面的Au原子成键, 其吸附能的大小为0.98 eV (图2(d)).
大量的实验和理论计算都表明, Au纳米颗粒在低温CO+O2氧化反应中具有很好的催化活性, 其相关的反应机理也被深入研究[64, 65]. 然而, CO + NO氧化还原反应的研究则相对较少. 因此, 本文系统地研究了CO和NO在Au6/CeO2负载型催化体系中的化学反应过程.
通过上述CO和NO在Au6/CeO2上的吸附计算可知, 在低覆盖度和高覆盖度条件下, NO在负载的Au颗粒团簇上的吸附能力都不如CO分子. 有趣的是, 即使CO分子已经吸附在Au团簇的顶位, NO仍然可以在Au-CeO2界面处形成强吸附的N2O2二聚体. 于是, 我们研究了CO和N2O2之间的氧化还原反应, 并将斜跨在Au6/CeO2(110)界面处的N2O2末端的O原子指向着吸附在Au顶位处的CO分子这样的共吸附结构作为反应的初始态(IS1, 图3和图2(d)). 由图3可见, CO和N2O2分子在Au6/CeO2(110)表面的共吸附能达到了2.18 eV(IS1). 当CO和N2O2开始相互反应时, CO分子逐渐地向N2O2中的末端O原子靠拢, 目的是与之结合生成CO2. 此时达到第一个过渡态TS1, OC-ONNO的距离为1.785 Å, CO分子内的键长被拉长到1.340 Å, 而N2O2分子中的末端O-N的键长也增加到了1.341 Å.
通过比较IS1和TS1这两个状态, 我们发现吸附CO的Au原子在反应过程中, 随着CO一起移动, 从而促进CO向N2O2分子靠拢并与之反应. 从计算的能量来看, 该步反应的活化能垒只有0.39 eV, 说明在较温和的条件下就能发生. 这与Wang等[40]的研究结果十分一致. 他们通过计算得出, 在Au(111)表面上, N2O2二聚体分子很容易生成N2O.
经过了TS1过渡态之后, 界面处生成了OCONNO这一中间物种(IM1). C原子同N2O2中末端O原子相互结合成键, 缩短到了1.417 Å. 而末端的O-N键则增加到1.437 Å. 与TS1过渡态能量相比, IM1这一中间态能量仅稳定了0.19 eV. 由图3可见, OCONNO分子需要经历第二个过渡态TS2以完全断裂N-O键来生成CO2, TS2的活化能垒为0.80 eV. 在TS2结构中, ONNO末端O原子同C原子和N原子的距离分别被优化为1.318和1.858 Å. 在越过TS2之后, N2O和CO2分子由于同Au6/CeO2表面都是弱吸附作用(N2O约为0.2 eV; CO2约为0.1 eV), 易于从表面脱附进入气相中. 从图3反应能量曲线上看, 反应末态FS1十分稳定, 在整个CO + N2O2反应过程中放出大量热能.
作为汽车尾气排放中另一类有毒有害污染物, N2O的降解也已成为研究热点[66, 67, 68]. 特别是研究发现, 过渡金属催化剂能够有效地活化和解离N2O分子[69, 70, 71].
本文也详细地计算了N2O与Au6/CeO2(110)催化剂表面的相互作用和相关的解离反应. 结果发现, N2O物种与负载的Au6纳米团簇或Au-CeO2载体界面处并无较强的相互作用(Ead = ~0.2 eV). 因此, 我们研究了N2O通过断裂分子中的末端N-O键使O原子留在Au纳米团簇上方法来直接解离N2O分子. 图4给出了N2O单分子解离的反应能量曲线图以及相关结构. 可以看到, 在解离初态(IS2)时, 与表面作用非常弱的N2O分子的O端指向了Au6纳米团簇的B2吸附位. 在TS3中, O-N键的距离为1.372 Å, N2O的直接解离能垒为1.78 eV, 说明这步解离反应还是较难发生的, 与之前的研究结果一致, 即N2O分子在Au基催化剂上的解离通常需要~500-600 K的高温[72, 73, 74].
经过TS3之后, N2O中末端的O原子最终吸附在Au6团簇的B2桥位上, 而N2从表面脱离进入气相. 整个反应过程的能量变化如图4所示, 尽管N2O分子需要一个较高的解离能垒, 但是总体反应仍然放热了0.64 eV.
考虑到CO分子原本就在Au6纳米团簇的顶位上具有很强的吸附能, 本文同样计算了N2O与预吸附CO直接反应来还原. 图5阐明了气相N2O的末端O原子直接与CO中的C原子碰撞的反应过程, 并给出了整个反应的能量曲线图. 在过渡态TS4结构中, N2O末端的O原子与CO中C原子的距离计算得为1.775 Å, 而O-NN键从气相的~1.20增至1.509 Å. 经历过渡态之后, 目标反应产物N2和CO2最终从表面脱附. 由整个能量曲线图可见, 该步反应的能垒为1.86 eV并最终放热2.25 eV, 说明尽管该反应路径需要在高温环境下实现, 但最终的热力学状态是非常稳定的.
负载于CeO2的Au纳米颗粒广泛应用于WGS和CO氧化反应. 本课题组曾系统地计算了O2在3个原子尺寸的小颗粒金纳米团簇(Au3)负载于CeO2(111), (110), (100)表面的吸附情况[58]. 我们发现, Au3团簇能够转移一个电子到Ce阳离子的4f局域轨道上, 从而使载体表面还原, 将Ce4+还原为Ce3+. 我们发现, 只有CeO2(110)面具有非常平坦宽广的形貌结构, 不存在任何突出的晶格O原子. 因此, O2分子才能在(110)表面的Ce3+< /sup>上吸附, 并使4f轨道上的局域电子进入其反键2π*轨道, 从而被活化成超氧分子. 换句话说, 正是O2分子同表面Ce阳离子之间强静电吸引作用造成了O2分子在CeO2(110)表面上良好的吸附能力.
本文发现, 在低覆盖度和高覆盖度两种情况下, NO在Au6纳米颗粒上的吸附能总是小于CO. 但是, NO却在Au6/CeO2界面处表现出了非常独特的性质. 与CO类似, 低覆盖度下NO在界面处的吸附能依然很低. 但有趣的是, 当两个NO分子同时在CeO2载体上相互靠近共吸附时, 它们很容易相互结合, 生成N2O2二聚体的结构, 并且在Au-CeO2界面处的吸附能达到~1 eV. 不仅如此, N2O2的优势吸附构型并不受在Au纳米团簇上CO分子吸附的影响, 这意味着CO和N2O2这两个反应物在各自吸附的两个区域内(Au团簇和Au-CeO2界面)并不存在竞争吸附的问题.
通过进一步的电子结构分析发现, N2O2分子的强吸附依然可以通过载体CeO2特殊的电子结构来解释. 正如图1(c)中的电子局域结构所示, Au6纳米颗粒的吸附引起了CeO2(110)面的还原. Au纳米团簇上的2个电子被转移并局域到2个与Au簇邻近的表面Ce阳离子4f轨道上. 有趣的是, 2个Ce3+阳离子并不受CO吸附于Au6团簇的影响(见图6(a)). 表面局域电子能进一步地转移到N2O2中, 增加其与表面的稳定性以及与载体Ce4+的静电吸引, 从而促进N2O2分子吸附. 由图6(b)中可见, 斜跨吸附在Ce原子和Au原子上的N2O2分子夺走了表面一个4f局域电子, 而另一个4f局域电子依然保留在表面自由的Ce3+阳离子上. 这充分说明Ce元素具有非常独特的催化性能, 其4f局域电子不仅能转移到类似O2这样的小分子中, 而且能促进带有负电性的N2O2-这一类较大分子的形成, 使其与表面存在着强烈的静电吸引作用, 在反应中成为稳定的中间产物. 需要强调的是, 带负电N2O2二聚体分子的形成正是Ce3+上局域电子灵活转移的结果. Ce3+上的局域电子能够转移到整个分子中, 197;并大大增强了N-N键的键能[75]. 因此, CeO2电子转移这类特性将会提高N2的产生机会, 对于处理NO基这类有毒、有害气体是非常可行的.
本文详细计算了CO与N2O2在Au6/CeO2(110)表面的反应过程, 尤其从Au纳米团簇型貌变化的几何角度揭示了重要的催化特征. 正如前文所述, 在CO + N2O2 → N2O + CO2反应中, 具有高度流动性的Au原子有助于CO分子准确地指向并靠近N2O2进行氧化还原反应, 所以这个过程的能垒只有~0.4 eV. 而且, 针对Au/CeO2(110)体系, 上述的计算结果进一步表明, Au原子独特的流动性并不受自身氧化的影响.
采用DFT + U的计算方法, 研究了Au6纳米颗粒负载于CeO2(110)重构表面用于CO + NOx氧化还原反应. 结果表明, Au6纳米颗粒通过给出两个4f局域电子到邻近的2个表面Ce阳离子上从而引起CeO2(110)表面部分还原. 对于Au6团簇上的吸附情况, 无论是高覆盖度还是低覆盖度, CO的吸附能都要明显强于NO. 因此, CO能够完全覆盖Au簇上的吸附位点. 有趣的是, CO的吸附并不影响CeO2(110)载体上局域电子分布的形态; 而且局域电子还可以转移并协助负电性的N2O2二聚体分子的形成和吸附. 对于CO+N2O2反应而言, 吸附在Au团簇顶位的CO能够与吸附在Au-CeO2界面处的N2O2二聚体分子通过Langmuir-Hinsherwood双分子机理相互反应, 生成CO2和N2O产物. 这步反应的总能垒约为1 eV. 在随后的N2O降解反应中, N2O分子既可以通过直接解离的方式, 在Au纳米团簇上留下一个O原子, 也可以通过与Au上预吸附的CO分子相互反应来消除. 在这步反应中, 由于N2O分子本身在Au团簇或者CeO2载体上的吸附作用都很弱, 因此反应能垒都会达到~1.8 eV, 并且遵循Eley-Rideal分子碰撞机理. 通过进一步的电子分析可知, Ce3+上的局域电子在反应过程中被转移到N2O2分子中, 从而增强了N-N键能并促成N2的形成. 结果表明, CeO2稀土材料这一独特的电子特性对于NO及其衍生物的消除将会起到一定的帮助.
致谢 感谢山东济南国家超算中心提供的计算机时.
2014, Vol. 35 Issue (8): 1305-1317
PDF (1060 KB)
2014, Vol. 35 Issue (8): 1305-1317
PDF (1060 KB)