Unraveling Catalytic Mechanisms for CO Oxidation on Boron-Doped Fullerene: A Computational Study.

By means of spin-polarized density functional theory (DFT) computations, we unravel the reaction mechanisms of catalytic CO oxidation on B-doped fullerene. It is shown that O2 species favors to be chemically adsorbed via side-on configuration at the hex-C-B site with an adsorption energy of -1.07 eV. Two traditional pathways, Eley-Rideal (ER) and Langmuir-Hinshelwood (LH) mechanisms, are considered for the CO oxidation starting from O2 adsorption. CO species is able to bind at the B-top site of the B-doped fullerene with an adsorption energy of -0.78 eV. Therefore, CO oxidation that occurs starting from CO adsorption is also taken into account. Second reaction of CO oxidation occurs by the reaction of CO + O → CO2 with a very high energy barrier of 1.56 eV. A trimolecular Eley-Rideal (TER) pathway is proposed to avoid leaving the O atom on the B-doped fullerene after the first CO oxidation. These predictions manifest that boron-doped fullerene is a potential metal-free catalyst for CO oxidation.

Introduction

Catalytic CO oxidation reaction has attracted considerable attention due to its significance in purifying environmental pollution and eliminating electrode poisoning in fuel cells.[1−5] CO oxidation is a simple and prototypical reaction and is suitable for unraveling the intrinsic reaction mechanism of heterogeneous catalysis.[6,7] How to effectively detect and remove CO has been one of the most challenging subjects nowadays. Conversion of CO into nontoxic substances, CO2, by catalytic oxidation process has been extensively studied using noble metals (Au,[8−12] Ag,[13] Pd,[14−16] Ru, Rh, Os, Ir,[14] and Au–Pd[17,18]) and metal oxides (TiO2-,[19] CeO2-,[20−23] and ZrO2-based materials[24]) as excellent catalysts toward CO oxidation. However, their high cost and scarcity have limited their applications. As a result, development of nonmetal or metal-free catalysts to reduce or replace the use of those metals in a large-scale process has become urgent.

Carbon nanomaterials, such as fullerenes, carbon nanotubes, graphenes, etc., have been used widely as either catalyst supports or catalysts for heterogeneous catalytic reaction. There have been several studies on single-atom catalysts (SACs) supported on carbon nanomaterials, for example, Pt-,[25−27] Cu-,[28] Fe-,[29] and Au-graphene,[30] and Fe-BN-fullerene,[31] which efficiently catalyze oxidation of CO. The activation energy for the rate-determining step (RDS) of CO oxidation on the above catalysts is predicted to be within 0.51–0.58 eV, which is comparable to those of metal catalysts. Doping heteroatoms (N, B, P, S, etc.) into carbon nanomaterials could result in electron modulation to modify the electronic properties and chemical activities that make heteroatom-doped carbon materials potential catalysts for CO oxidation.[32−38] Previous investigations have shown that N-doped fullerene,[39] N-doped carbon nanotube,[40,41] N-doped penta-graphene,[42−44] graphdiyne,[45] B-doped and fullerene-like BN cage,[46] B-doped carbon nanotube,[47] and so on can serve as metal-free catalysts and exhibit superior catalytic activity toward CO oxidation. Unraveling the detailed reaction mechanisms of CO oxidation is essential to design novel metal-free catalysts. Boron-doped carbon nanotubes have been shown to exhibit high catalytic activity both experimentally and theoretically.[47,48] Previous studies demonstrated that the energy barrier of RDS for CO oxidation on B-doped fullerene using the Eley–Rideal (ER) mechanism is predicted to be 0.48 eV.[49,50] They also showed that second CO oxidation occurs by the reaction of CO with the remaining O atoms, which have an energy barrier of 1.47 eV, which limits its occurrence. To solve this issue, a trimolecular Eley–Rideal (TER) pathway is proposed to avoid leaving the O atom on the B-doped fullerene after the first CO oxidation, and it will be discussed below. Understanding the intrinsic mechanisms of CO oxidation is important to guide researchers to design the most effective catalyst.

Herein, we carry out spin-polarized first-principles calculations to unravel the catalytic mechanisms of CO oxidation on B-doped fullerene. In this study, diverse mechanisms of CO oxidation reaction on B-doped fullerene are revealed. The best reaction pathway of CO oxidation is verified by discussing the energy barrier of the RDS.

Results and Discussion

Interaction and Adsorption Behavior of O2 and CO on B-doped Fullerene

To unravel the reaction mechanisms of CO oxidation on B-doped fullerene, we first investigate the adsorption behavior of O2 and CO on B-doped fullerene. A polarized C––B+ bond, which becomes an active site for molecular adsorption, is created by doping the boron atom into fullerene (see Figure ). Gao and Chen have demonstrated that B-doped fullerene can resist temperatures up to 1000 K, suggesting its structural stability.[50] Bader charge calculation shows that the carbons bound to boron gain negative charges of 0.88–1.16 |e| from the substitutive boron. The O2 and CO species are placed above those C–B sites containing B-top, hex-C–B, and pent-C–B sites of the B-doped fullerene at a proper distance during the geometric optimizations.

Figure 1

Optimized model structures, possible active sites, and Bader charge of B-doped fullerene. The gray and pink spheres represent C and B atoms, respectively. The charges are given in |e|.

The optimized configurations of O2 and CO adsorption on the B-doped fullerene as well as the related adsorption energies are displayed in Figure and Table . It is found that O2 species chemisorbs on the C–B site of the B-doped fullerene via two types of configurations: side-on and end-on configurations. Calculations demonstrate that the O2 species prefers to strongly chemisorb at the hex-C–B site via the side-on configuration. The adsorption energy is computed to be −1.07 eV, while those are −0.61 and −0.63 eV for the O2 side-on configuration at the pent-C–B site and the O2 end-on configuration at the B-top site, respectively. The related distance of O–O bond is calculated to be elongated from 1.222 Å of free O2 molecule to 1.462, 1.461, and 1.318 Å, respectively, as depicted in Figure . Bader charge analysis (see Table ) shows that the adsorbed O2 species gain −1.42, −1.33, and −1.28 |e| from the B-doped fullerene, indicating that the O2 species is chemisorbed on the B-doped fullerene. It is comparable to those on the catalysts of metals and SACs, Pt surfaces (−0.58 to −0.72 eV for (111),[51] −1.48 eV for (110),[52] and −1.30 eV for (001)[53]), Pt clusters (−0.72 eV for Pt2 and −1.08 eV for Pt3),[54,55] Au clusters (−0.24 eV for Au4,[55] ca. −0.50 eV for Au7 and Au9,[30] −0.60 eV for Au29,[10,56,57] −0.99 eV[17,58] for Au38), Pt-graphene (−1.34 eV),[25] Cu-graphene (−2.67 eV),[28] Fe-graphene (−2.09 eV),[29] Au-graphene (−1.34 eV).[30]

Figure 2

Optimized structures of O2 and CO adsorption on B-doped fullerene. The gray, pink, and red spheres represent C, B, and O, atoms, respectively. The distances are given in angstroms.

Table 1

Computed Adsorption Energies (eV) and Bader Charges (|e|; in Parentheses) of O2 and CO Species on the B-doped Fullerene

adsorbate/site	hex-C–B	penta-C–B	B-top
O2	–1.07 (−1.42)	–0.61 (−1.33)	–0.63 (−1.28)
CO			–0.78 (−0.49)

For the CO adsorption, CO is placed at the mentioned active sites with a 1.5–2.0 Å distance away from the B-doped fullerene during structural optimization. One should note that CO species is able to adsorb at the B-top site of the B-doped fullerene with an adsorption energy of −0.78 eV. Compared to the cases of B-doped carbon nanotube[47] and N-doped fullerene[59] (−0.05 and −0.20 eV for B-doped carbon nanotube and N-doped fullerene, respectively), CO is chemisorbed on the B-doped fullerene. More charge transfer (−0.49 |e|) from the B-doped fullerene occurs, resulting in stronger interaction between the O2 species and the B-doped fullerene. This result indicates that CO is likely to bind to B-doped fullerene. The O adsorption, which is depicted in Figure S1, takes place by the first oxidation reaction of CO + O2 → CO2 + O. It is found that the atomic O prefers to bind at the C–B site with the computed adsorption energies of −5.10 and −5.24 eV at the penta-C–B and hex-C–B sites, respectively.

The minimum-energy paths of O2 and CO adsorption at B-doped fullerene have been illustrated by CINEB calculation, as depicted in Figure S2. The calculation shows that no energy barrier is required to surmount for the CO adsorption process, but it needs a barrier of 0.2 eV to surmount for the O2 adsorption process, indicating that the CO adsorption process is kinetically more favored than the O2 adsorption process, which is thermodynamically favored. Therefore, CO oxidation process occurs starting from both O2 and/or CO adsorption, as discussed the following section.

CO Oxidation Reaction Starting from O2 Adsorption and/or CO Adsorption on B-doped Fullerene

Here, we illustrate the mechanisms of CO oxidation beginning from O2 adsorption and/or CO adsorption on B-doped fullerene. Two traditional pathways, Eley–Rideal (ER) and Langmuir–Hinshelwood (LH) mechanisms, are considered for the CO oxidation.

CO Oxidation Starting from the O2 Side-On Adsorption

The potential energy surface (PES) of the CO oxidation reaction via the LH mechanism at the hex-C–B site of B-doped fullerene is mapped by the CINEB method, as described in Figure . The first step occurs by attachment of the gas-phase CO molecule to fullerene to achieve O2 and CO co-adsorption (intermediate C) on the fullerene. The step is found to be exothermic by 4.24 eV with an energy barrier of 0.51 eV. The O–O bond distance in the O2 and CO co-adsorption (intermediate C) is found to increase to 2.154 from 1.461 Å. Then, CO2 desorption takes place from the intermediate C with a small energy barrier of 0.08 eV and an exothermicity of 0.58 eV. The overall reaction of O2 + CO → CO2 + O on the B-doped fullerene is exothermic by ∼5.00 eV and requires to surmount a 0.51 eV barrier of the rate-determining step. The second CO oxidation takes place by the remaining O atom on fullerene, which will be studied below.

Figure 3

Energy diagrams for CO oxidation starting from the O2 side-on adsorption via the LH mechanism at the hex-C–B site of B-doped fullerene.

Furthermore, the ER mechanism is also discussed beginning from the O2 side-on adsorption based on the PES depicted in Figure . The process begins by the attachment of a gas-phase CO molecule to insert into the O–O bond of the O2 end-on adsorption to form a carbonate-like CO3 (intermediate C) with a large exothermicity of 4.57 eV. In TS B, the O–O bond is slightly elongated to 1.472 from 1.461 Å with 0.34 eV higher in energy than the reactant A. Subsequently, CO2 desorption can occur by cleavage of Bfullerene–O and Cfullerene–O bonds with very high barriers of 2.36 and 1.95 eV, respectively. The high barrier of this process suggests that it is kinetically unfavored. To solve this issue, a trimolecular Eley–Rideal (TER) pathway, reaction of the second CO species with the CO3 intermediate, is proposed and will be discussed below.

Figure 4

Energy diagrams for CO oxidation starting from the O2 side-on adsorption via the ER mechanism at the hex-C–B site of B-doped fullerene.

CO Oxidation Starting from the O2 End-On Adsorption

First, the LH mechanism is discussed beginning from the O2 end-on adsorption, as depicted in Figure . The CO species initially approaches the B-doped fullerene and interacts with the O2 end-on adsorption to form a peroxo-type (OOCO) intermediate with an exothermicity of 2.03 eV, as seen in Figure . This step only requires 0.24 eV to surmount. The O–O bond length is predicted to be elongated to 1.475 from 1.314 Å in the OOCO intermediate. Then, the cleavage of the O–O bond in the OOCO intermediate takes place to desorb CO2 via a transition structure (TS D), with a barrier of 0.89 eV and endothermicity of 1.64 eV. In TS D, the cleavage of the O–O and C–C distances is predicted to be 1.831 and 1.673 Å, respectively. The overall reaction beginning from the O2 end-on adsorption is exothermic by 3.67 eV and needs to overcome a 0.89 eV barrier of the rate-determining step.

Figure 5

Energy diagrams for CO oxidation starting from the O2 end-on adsorption via the LH mechanism at the B-top site of B-doped fullerene.

The ER mechanism is further described beginning from the O2 end-on adsorption, as depicted in the PES (see Figure ). CO species in reactant A achieves end-on O2 adsorption and extracts one O atom directly by surmounting a TS B with an energy barrier of 0.64 eV to produce the final state C (product). In TS B, the O–O bond increases from 1.318 to 1.330 Å, and the distance between the C atom of CO and the O atom of O2 decreases from 2.705 to 2.333 Å. The process is predicted to be highly exothermic by −5.01 eV. The energy barrier (0.64 eV) of this rate-determining step is slightly higher than that of the LH mechanism (0.51 eV).

Figure 6

Energy diagrams for CO oxidation starting from the O2 end-on adsorption via the ER mechanism at the B-top site of B-doped fullerene.

CO Oxidation Starting from the CO Adsorption

As mentioned above, there is no energy barrier in the minimum-energy path for CO adsorption at the B-top site of the B-doped fullerene with an adsorption energy of −0.78 eV. Therefore, CO oxidation on the B-doped fullerene starting from the CO adsorption can be achieved more easily because this process is both kinetically and thermodynamically favored. As seen in Figure , the gas-phase O2 species approaches the B-doped fullerene and directly undergoes CO adsorption to form a peroxo-type (OOCO) intermediate with an exothermicity of 0.99 eV. In addition, only a low energy barrier of 0.05 eV is required for this step. The O–O bond is found to increase to 1.490 from 1.238 Å in the OOCO intermediate. Next, the cleavage of the O–O bond in the OOCO intermediate occurs to desorb CO2 via a transition structure (TS D), with a barrier of 0.39 eV and endothermicity of 3.40 eV. In TS D, the cleavage O–O and C–B distances are computed to be 1.707 and 1.608 Å, respectively. The overall reaction beginning from the CO adsorption is exothermic by 4.39 eV and needs to overcome a 0.39 eV barrier of the rate-determining step. Compared to the proposed reaction mechanisms beginning from the O2 adsorption as alluded above, the energy barrier of the rate-determining step is the lowest one (0.39 eV), indicating that CO oxidation prefers to occur via the LH mechanism starting from the preadsorption of CO on the B-doped fullerene.

Figure 7

Energy diagrams for CO oxidation starting from the CO adsorption via the LH mechanism at the B-top site of B-doped fullerene.

Second CO Oxidation Occurring with the Remaining O Atom

As depicted in Figures S3 and S4, the gas-phase CO molecule approaches the adsorbed O species and undergoes O adsorption to produce a second CO2 species. The process via the ER mechanism (see Figure S3) is exothermic by 1.03 eV but needs to pass a high energy barrier of 1.56 eV, while the process via the LH mechanism (see Figure S4) needs to overcome a high energy barrier of 1.82 eV. The desorption energy of CO2 is predicted to be 0.03 eV, suggesting that CO2 species is easily released from the B-doped fullerene. However, the B-doped fullerene is readily poisoned by the remaining O species due to the very high barrier for removing the O atom. To solve this issue, a trimolecular Eley–Rideal (TER) pathway is proposed to avoid leaving the O atom on the B-doped fullerene after the first CO oxidation.

Trimolecular Eley–Rideal (TER) Mechanism

In this section, we consider the new TER mechanism of CO oxidation, which is called the self-promotion mechanism proposed by Lyalin et al.,[60] to avoid leaving the O atom on the B-doped fullerene. The TER mechanism can occur starting from the carbonate-like (CO3) and the peroxo-type (OOCO) intermediates. As shown in Figure , the second CO molecule can approach the CO3 intermediate and extract one oxygen in the CO3 to directly form two CO2 species by passing barriers 1.37 eV (extracting the O atom bound to the B atom) and 1.79 eV (extracting the O atom bound to C atom), which are 0.99 and 0.18 eV lower than 2.36 and 1.95 eV, respectively. The process is more exothermic by 1.01 eV. The other TER mechanism can take place through the OOCO intermediate depicted in Figure ; the second CO molecule can attach to the OOCO intermediate and extract the O bound to B and/or C to produce two CO2 species. The predicted energy barriers are 0.68 eV (extracting the O atom bound to the B atom) and/or 0.33 eV (extracting the O atom bound to the C atom), which are 0.21 and 0.06 eV lower than 0.89 and 0.39 eV for the traditional LH mechanism, respectively. Besides, the process is more exothermic by 2.33 and 1.37 eV, respectively. The calculation results demonstrate that the TER mechanism is found to be more kinetical and thermodynamical compared with the traditional LH and ER mechanisms.

Figure 8

Energy diagrams for CO oxidation starting from the carbonate-like (CO3) intermediate via the TER mechanism.

Figure 9

Energy diagrams for CO oxidation starting from the peroxo-type (OOCO) intermediate via the TER mechanism (a) starting from O2 end-on adsorption and (b) starting from CO adsorption.

The catalytic activity (energy barriers of the rate-determining step: 0.33–0.68 eV) of the B-doped fullerene via the TER mechanism is comparable to those on the most studied catalysts including metal surface system: Au(111) (1.97 eV)[13] and Au(211) (0.59 and 0.65 eV);[9] metal nanoparticle system: Au29 (0.69 eV),[10] Au38 (0.37 eV),[17] Pd38 (0.49 eV),[17] and core–shell Au–Pd (0.10 and 0.59 eV for Au32/Pd6 and Pd32/Au6 clusters, respectively);[17] SAC system: Pt- (0.33 eV),[26,27] Cu- (0.54 eV),[28] Fe- (0.58 eV),[29] and Au-embedded graphene (0.31 eV),[30] Fe-graphene oxide (0.61 eV);[61] and metal-free catalysts: N-doped fullerene (0.20 eV),[59] N-doped (0.45–0.58 eV),[62] and B-doped CNTs (0.34–0.42 eV).[47] Consequently, the calculations suggest that B-doped fullerene could be an effective metal-free catalyst for CO oxidation.

Conclusions

We have investigated the detailed mechanisms for catalytic CO oxidation catalyzed by the boron-doped fullerene by performing spin-polarized first-principles calculations. The calculations demonstrate that O2 is chemisorbed via side-on configuration with adsorption energies of −1.07 eV at the hex-C–B site of the boron-doped fullerene. It is found that CO species can also bind at the B-top site with an adsorption energy of −0.78 eV. There is no energy barrier required to surmount for the CO adsorption process, but it needs a barrier of 0.2 eV to surmount for the O2 adsorption process, indicating that the CO adsorption process is kinetically more favored than the O2 adsorption process. Therefore, two traditional pathways, Eley–Rideal (ER) and Langmuir–Hinshelwood (LH) mechanisms, are considered for the CO oxidation starting from both O2 and/or CO adsorption. The results show that CO oxidation takes place by the LH mechanism starting from the preadsorption CO with the lowest energy barrier of 0.39 eV for the rate-determining step. The remaining atomic O species continues to proceed second reaction of CO oxidation with a very high energy barrier of 1.56 eV, which hinders the catalytic activity of the B-doped fullerene. A TER pathway is proposed to avoid leaving the O atom on the B-doped fullerene after the first CO oxidation. The computations demonstrate that the TER mechanism is found to be more kinetical and thermodynamical compared with the traditional LH and ER mechanisms.

Computational Methods

All geometries are optimized using the periodic first-principles computations based on spin-polarized density functional theory (DFT) implemented in the Vienna Ab initio Simulation Package (VASP) code.[63,64] The exchange–correlation functional of Perdew–Wang correlation (PW91) with the generalized gradient approximation (GGA) is described.[65] The plane-wave basis set is set to be 400 eV of cutoff energy. The Brillouin zones are selected as Γ k-point mesh and 5 × 5 × 5 k-point mesh for geometric structure optimizations and electronic property calculations, respectively.[66] A 25 × 25 × 25 Å3 periodic supercell is selected for the calculations, which is large enough to avoid the interaction of periodic images.[59] The adsorption energy of adsorbates (O2 and CO) on B-doped fullerene is predicted according to the following equationwhere Etot, EB-fullerene, and Eadsorbate represent the computed energies of adsorbed species on the B-doped fullerene, the isolated B-doped fullerene, and the gas-phase molecule, respectively. It is noted that the predicted O–O bond lengths of gas-phase O2, C–O bond length of gas-phase CO, and C–O bond length of gas-phase CO2 are 1.222, 1.137, and 1.169 Å, respectively, which agree with the experimental data of 1.207,[67] 1.128,[68] and 1.193 Å.[69] The minimum-energy paths (MEP) including transition states are constructed by the climbing image nudged elastic band (CINEB) method.[70,71] The electron distribution and transfer are described by the Bader charge calculations.[72−74]