Conversion of Methane with Oxygen to Produce Hydrogen Catalyzed by Triatomic Rh3 - Cluster Anion.

Metal catalysts, especially noble metals, have frequently been prepared upon downsizing from nanoparticles to subnanoclusters to catalyze the important reaction of partial oxidation of methane (POM) in order to optimize the catalytic performance and conserve metal resources. Here, benefiting from mass spectrometric experiments in conjunction with photoelectron spectroscopy and quantum chemical calculations, we successfully determine that metal cluster anions composed of only three Rh atoms (Rh3 -) can catalyze the POM reaction with O2 to produce 2H2 + CO2 under thermal collision conditions (∼300 K). The interdependence between CH4 and O2 to protect Rh3 - from collapse and to promote conversion of CH4 → 2H2 has been clarified. This study not only provides a promising metal cluster displaying good catalytic behavior in POM reaction under mild conditions but also reveals a strictly molecular-level mechanism of direct partial oxidation for the production of hydrogen, a promising renewable energy source in the 21st century.

Introduction

Hydrogen, a clean and renewable energy source, is of great practical importance for fuel cells as well as the synthesis of chemicals such as ammonia, methanol, and Fischer–Tropsch products.[1−4] Nowadays, the feedstock of natural abundant methane, with a high H/C ratio, accounts for 48% of industrial hydrogen production through steam reforming of methane (SRM),[5] the reaction of which is highly endothermic and energy-intensive. Contrary to the SRM reaction, the partial oxidation of methane (POM) that follows the reaction of CH4 + O2 → 2H2 + CO2 enables the generation of highly exothermic H2 (ΔH298 = −3.31 eV) and it is considered as a promising strategy to yield H2 from an economical perspective.[6,7] However, limited by the inertness of CH4 and O2 molecules, the spontaneous POM reaction could occur only at high temperatures (T > 700 °C) in the absence of a catalyst.[8] Since the first experimental detection of free H2 from POM reaction catalyzed by Ni/Al2O3 in 1929,[9] various heterogeneous catalysts supported with base and precious metals have been engineered for CH4 conversion at relatively low temperatures.[10−14] The dominant size-dependent catalytic behavior of supported metals motivates elaborate controls of metal species spanning from nanoparticles to subnanoclusters, aiming for optimization of the catalytic performance toward the POM reaction and the conservation of metal resources particularly for precious metals (e.g., supported rhodium catalysts with sizes of 30.5 nm → 2.5 nm → 1.3 nm → 0.6 nm have been tailored).[15−18] However, it remains challenging experimentally to achieve catalytic hydrogen production under mild conditions. The genuine mechanisms of H2 production from POM reaction have also not been completely clarified at a strictly molecular level.

Mass spectrometry coupled with the mass selection technique provides a unique approach to distinguish the atomically precise metal clusters in the gas phase and to explore the intrinsic reactivity property of size-specific metal clusters under isolated conditions.[19−26] Available experiments demonstrated that base metal clusters are commonly unreactive with CH4 under thermal collision conditions.[27−29] Although most of the unary base metal clusters (M) could be facilely oxidized by O2, the generated oxide clusters (MO) possessing active sites such as oxygen-centered radicals[19,22] and Lewis acid–base pairs Mδ+–Oδ−[22] are capable of bringing about thermal methane activation, the process of which generally results in MOH product that are difficult to reform back to M. Although the noble metal clusters (NM) are able to dehydrogenate methane under thermal collision conditions,[30−32] the catalytic cycle for the reaction of CH4 with O2 mediated by metal clusters has not been experimentally identified. So far, only the positively charged atomic Pt+ as well as diatomic Au2+ and PtO+ were reported to catalyze the conversion of CH4 with O2 at room temperature.[33,34] Among these catalytic reactions, only the Pt+ system can convert one CH4 molecule to generate one H2 molecule, whereas both of Au2+ and PtO+ convert CH4 and O2 to produce formaldehyde and water. Note that three CH4 molecules were required to complete the catalytic cycle for the Au2+ system. Herein, we report the first experimental identification that the thermal reaction of one CH4 molecule with O2 to produce two H2 molecules can be achieved by using a Rh3– cluster as a catalyst.

Results and Discussion

Cluster Reactivity

The reactions of laser ablation generated Rh3– cluster anions with CH4 and O2 were studied by double ion trap experiments in which methane and molecular oxygen were spatially separated by using two ion trap reactors. After thermalization of the mass-selected Rh3– anions by collision with He atoms in the first reactor (Figure a), CH4 was injected and reacted for about 3.6 ms. It can be seen from Figure b that although the Rh3– anions depleted significantly at the CH4 pressure of 83 mPa, a new strong peak assigned as Rh3CH2– appeared, suggesting that the dehydrogenation of methane with the loss of one H2 molecule took place (reaction ):

Figure 1

Mass spectra for the reactions of mass-selected Rh3– with (a) He and (b) 83 mPa CH4; Rh3CH2– with (c) He, (d) 1.33 mPa 16O2, and (e) 1.05 mPa 18O2; and Rh3O– with (f) He and (g) 0.7 mPa CO under thermal collision conditions. The reaction times are (b) 3.6 ms, (d, e) 1.9 ms, and (g) 3.6 ms. The peaks marked with asterisks in panels d and e can be assigned to Rh2O3– originating from the oxidation of Rh3O– by O2 (Figure S1).

The isotopic labeling experiment with CD4 confirmed the dehydrogenation channel (Figure S1). The rate constant (k1) of a pseudo-first-order reaction between Rh3– and CH4 was determined to be 1.9 × 10–11 cm3 molecule–1 s–1,[35,36] corresponding to a reaction efficiency of 1.9%. The kinetic isotopic effect (k1,CH4/k1,CD4) was estimated to be 7.9.

After Rh3– reacted with CH4 in the first reactor, the resulting product ions Rh3CH2– were then mass-selected (Figure c) and thermalized in the second reactor for reaction with O2. As shown in Figure d, upon the interaction of Rh3CH2– with 1.33 mPa 16O2 for about 1.9 ms, a strong product peak with the mass larger than that of Rh3CH2– by 2 amu emerged and it can be assigned as Rh316O–, which was further confirmed by the isotopic labeling experiment with 18O2 (Figure e). In addition, a weak product signal corresponding to Rh3– was clearly identified. Two possible channels could be considered for the generation of Rh3O– or Rh3– in the reaction of Rh3CH2– with O2. The Rh3O– could be produced by either the successive desorption of H2 and CO molecules (reaction ) or the direct evaporation of a CH2O molecule (reaction ). Likewise, the successive liberation of H2 and CO2 molecules results in the formation of Rh3– (reaction ). The Rh3– may also originate from the evaporation of CH2O2 (reaction ).

Further theoretical calculations (see below) support that the Rh3O– and Rh3– are more likely to be generated through the channels involving liberation of H2 and CO/CO2 molecules (reactions and 3a). The measured k1(Rh3CH2– + O2) is 7.2 × 10–10 cm3 molecule–1 s–1 that is larger than k1(Rh3– + CH4) by more than 1 order of magnitude.

Further experiments demonstrated that the produced Rh3O– (Figure f) in reaction could be very efficiently reduced to Rh3– by CO (Figure g and reaction ) with k1 of 1.6 × 10–9 cm3 molecule–1 s–1. As a resulton the basis of reactions , 2a, 3a, and 4, it can be concluded that the reaction of CH4 with O2 can produce two H2 molecules with concomitant formation of CO2 under thermal collision conditions catalyzed by the Rh3– cluster. Furthermore, two pathways (i and ii) to complete the catalytic cycle are identified at a strictly molecular level, as shown in Figure .

Figure 2

Proposed catalytic cycles for the reaction of CH4 + O2 → 2H2 + CO2 mediated by Rh3– cluster.

Structures of Reaction Intermediates and Reaction Mechanisms

The ground-state structure of Rh3– cluster was previously characterized to have an isosceles triangle in the quintet state.[35,37] To explore the mechanistic details for the reaction of CH4 with O2 catalyzed by Rh3–, we conducted the photoelectron spectroscopy (PES) experiments and density functional theory (DFT) calculations with PBE functional[38] in an attempt to determine the structures of the important reaction intermediates Rh3CH2– and Rh3O– involved in the catalytic cycles. The comparison between experimental PES spectra and simulated density of states (DOS) spectra of isomeric structures (Figure , Figures S3 and S4) shows that (i) the structure of Rh3CH2– can be in the triplet HRh3(CH)− by capping a C–H unit on the top of the Rh3 plane and suspending the second H atom by two Rh atoms; (ii) the Rh3O– is also in a triplet state that possesses an O atom terminally attached to one Rh atom of the Rh3 moiety.

Figure 3

Experimental (Expt.) photoelectron spectra and simulated (Sim.) density of states (DOS) spectra for isomers of Rh3CH2– (left) and Rh3O– (right). The DOS spectrum for Rh3O– is red-shifted by 0.08 eV. Both the DFT calculated isomers of Rh3CH2– and Rh3O– are in the triplet state.

The most possible pathways computed with DFT method for the elementary reactions of Rh3– with CH4 (R1) and Rh3CH2– with O2 (R2) are shown in Figures and S5–S8. The quintet Rh3– traps CH4 by one Rh atom (Rhα) to form encounter complex I1 with a binding energy of 0.20 eV. The oxidative addition of the first C–H bond onto Rhα then takes place with a small energy barrier of 0.12 eV (5I1 → 5TS1), resulting in formation of the CH3–Rhα–H–Rhβ moiety. After the bridging H atom (Hb) moves away from Rhα (5I2 → 5I3, Figure S5), the CH3 moiety that has been bonded with Rhα also connects to the Rhγ atom (5I3 → 5I4) so that the second and the third C–H bonds can be successively activated by Rhα (I4 → TS4 → I5) and Rhγ (I5 → TS5 → I6), respectively. Note that a spin inversion from quintet state to triplet state occurs during activation of the second C–H bond (5I4 → 3TS4, Figure S6). The facile cleavage of three C–H bonds of CH4 leads to the formation of a more stable intermediate I6, in which each Rh atom is coordinated with a terminal H atom (Ht). Subsequently, the residual C–H bond in I6 moves toward the Rhβ atom and the most stable intermediate I7 is formed by capping a C–H unit on the top of Rh3 plane. Meanwhile, two of the three Ht atoms in I6 become Hb atoms in I7. The Ht in I7 then combines with the adjacent Hb to form an H2 molecule and then desorb from Rhγ (I7 → I8 → P1), leading to the generation of triplet Rh3CH2– ion.

Figure 4

DFT calculated potential energy profiles for reactions of 5Rh3– + CH4 (R1) and 3Rh3CH2– + O2 (R2). The relative energies (ΔH0) are given in eV. The structures of R, I1, I2, I4–I7, I9, I10, I12, I13, I15, I18, I19, I22, P1 (3Rh3CH2– + H2), P2 (3Rh3O– + H2 + CO), and P3 (5Rh3– + H2 + CO2) are plotted, whereas those of I3, I8, I11, I14, I16, I17, I20, I21, I23, and TS1–TS20 can be found in the Supporting Information. The superscript denotes the spin multiplicity.

In the reaction of 3Rh3CH2– with O2, the spare Rhγ adsorbs the O2 molecule to form intermediate 3I9 and 2.55 eV of energy is released. Thus, the 3I9 has enough energy to overcome the barrier (1.05 eV, 3I9 → 3TS8) for cleavage of the O–O bond by cooperation between Rhγ and Rhβ, resulting in the more stable intermediate 3I10 with an Ot and an Ob attached to Rh atoms. The 3I10 then undergoes a series of structural rearrangements involving Ht transfer from Rhα to Rhβ (I10 → I11, Figure S7) and rupture of the HC–Rhγ bond (I11 → I12). Subsequently, the HC–Ob coupling (3I12 → 3I13) occurs, followed by activation of the fourth C–H bond by Rhα (3I13 → 3I14 → 3I15) and formation of 3I15 that has two Ht atoms coordinated with Rhα and Rhβ. After the Ht originally bonded with Rhα transfers to Rhβ (3I15 → 3I16 → 3I17 → 3I18), the two Ht atoms on Rhβ in 3I18 could make an H2 unit and are finally evaporated to form triplet Rh3CO2– (3I19), the structure of which contains an isolated CO unit and an Ot atom.

It is noteworthy that the production of H2 in the reaction of 3Rh3CH2– + O2 is highly exothermic (by 4.59 eV) so the resulting 3Rh3CO2– possesses sufficient internal energy to enable the CO desorption and generation of triplet Rh3O– ion (I19 → Rh3O– + CO, P2). Alternatively, the CO moiety in 3Rh3CO2– could directly couple with the Ot atom to form a CO2 unit, which is further evaporated to reform quintet Rh3– through a spin conversion (3I19 → 3I20 → 3I21 → 5I22 → 5I23 → 5Rh3– + CO2, P3, Figures S7 and S8). Despite that the channel of CO2 elimination (ΔH0 = −2.52 eV) is thermodynamically more favorable than that of CO desorption (ΔH0 = −1.66 eV), the kinetic analysis on the basis of Rice−Ramsperger−Kassel−Marcus (RRKM) theory[39] demonstrates that the rate (6.1 × 1010 s–1) of CO desorption from Rh3CO2– is about six times the rate of internal conversion (1.0 × 1010 s–1) of I19 → TS19 involved in CO2 formation. Thus, the reaction of Rh3CH2– with O2 should generate Rh3O– + CO (rather than Rh3– + CO2) as the major product, in consistent with the experimental observation (Figures d, e). The pathway for direct elimination of CH2O to produce Rh3O– was also considered for the reaction of Rh3CH2– + O2 (Figure S9); however, it is kinetically less favorable than the path of H2 + CO production. The 3Rh3O– generated after H2 + CO desorption can further barrierlessly adsorb and oxidize a CO molecule to CO2, and 3Rh3O– is reduced to 5Rh3– following the same path as I20 → P3. As a result, the computational results support the idea that the reaction of CH4 with O2 to produce 2H2 + CO2 can be achieved under thermal collision conditions using a Rh3– catalyst.

Insights into the Gas-Phase Catalysis and Related Condensed-Phase Catalysis

The transformation of important molecules with molecular oxygen mediated by atomic clusters has received considerable attention in gas-phase studies, but only a very limited number of bare metal clusters including Pt3–6–,[40] Au6–,[41] and Ag7,9,11–[42] have been reported to exhibit catalytic behavior. The elaborate measurement on elementary reactions demonstrated that the completion of catalytic cycles requires preferential adsorption of O2 onto M– clusters and the resulting MO– then oxidizes small molecules (e.g., CO) to reform M–. In this study, we identify the first example for catalytic conversion of CH4 with O2 mediated by a metal cluster (Rh3–), moreover, a distinct mechanism to complete the catalytic cycle has been definitely determined. The Rh3– should react with CH4 at first to produce Rh3CH2– + H2, after which O2 is dissociatively adsorbed onto the Rh3CH2– intermediate, enabling formation of the second H2 molecule and CO2 as well as regeneration of Rh3– cluster. The preferential oxidation of Rh3– by O2 has also been experimentally tested. Unfortunately, the Rh3– rapidly disintegrates into Rh2O2– + Rh and RhO2– + Rh2 with k1 of 7.1 × 10–10 cm3 molecule–1 s–1 (Figures S1 and S2). The stepwise reaction of Rh3– with CH4 and then with O2 to close the catalytic cycle for production of 2H2 + CO2 (Figure ) thus implies an interdependence between CH4 and O2 that the preferential reaction with CH4 could protect Rh3– cluster from collapse, whereas the O2 adsorption could promote transformation of CH4 to two H2 molecules afterward.

The fundamental mechanism of the POM reaction with molecular oxygen over bulk Rh-based catalysts has been extensively investigated. Some researchers studying the catalysts of supported rhodium nanoparticles or nanoclusters (with the sizes of 0.6–3.3 nm) proposed that the H2 (and CO) production follows a combustion-reforming mechanism,[15−18] among which CH4 and O2 first undergo combustion to give CO2 and H2O and subsequently the steaming (H2O) or dry (CO2) reforming of unreacted methane and water–gas shift reactions occur. Alternatively, the direct partial oxidation mechanism with H2 (and H2O) as primary products was emphasized on the Rh(111) single crystal surface.[43] However, there remains a lack of experimental techniques to characterize the elementary steps in condensed-phase systems and to prove the proposed mechanisms. Herein, benefiting from the well-controlled gas phase cluster reactions, we identify that the rhodium cluster composed of only three Rh atoms (Rh3–) is active enough to catalyze the POM reaction under thermal collision conditions. Furthermore, the mechanism of direct H2 production has also been certified at a strictly molecular level by following each elementary step of the catalytic cycle (Figure ). It is noteworthy that the product H2 has the possibility of being oxidized into H2O by the Rh3O– ion, the important intermediate generated during the catalytic reaction. Our further mass spectrometric experiments (Figure S1 and S2) demonstrate that the reaction rate (5.3 × 10–10 cm3 molecule–1 s–1) of Rh3O– + H2 is slower than that of Rh3O– + CO by a factor of 3, which is advantageous for improving the product yield of H2. The employment of a hydrogen-permeable membrane reactor in real-life catalysis may further allow the selective separation of H2.[44]

Conclusion

In summary, the catalytic conversion of CH4 with O2 mediated by the atomic cluster Rh3– has been experimentally achieved in the gas phase. Although the Rh3– cluster is prone to disintegrate upon interaction with O2, the preferential thermal reaction of Rh3– with CH4 could generate a H2 molecule and a Rh3CH2– ion. The resulting Rh3CH2– subsequently reacts with O2 to produce the second H2 molecule and CO; meanwhile, the intact [Rh3] moiety is oxidized to Rh3O– that can be further reduced by CO to reform Rh3–. Thus, the cycle for partial oxidation of CH4 with O2 to 2H2 + CO2 is closed by using the Rh3– cluster catalyst. This study not only makes a significant step toward catalytic conversion of methane with O2 mediated by atomic clusters, but also provides molecular evidence for the direct partial oxidation mechanism for H2 production from methane under mild conditions.

Methods

Experimental Methods

The negatively charged rhodium clusters (Rh–) were generated by laser ablation of a rotating and translating Rh disk in the presence of a 6 atm He carrier gas. The reactivity experiments were carried out in a double ion trap system, which includes two quadrupole mass filters (QMFs) and two linear ion traps (LITs). Among many generated Rh– clusters, the triatomic Rh3– ions of interest were mass-selected by using the first QMF[45] and entered into the first LIT reactor,[46] where they were confined and thermalized by collisions with a pulse of buffer gas He and then reacted with 12CH4 or CD4. The second QMF mass-selected the product ions (e.g., Rh3CH2–) resulting from the first LIT to inject into the second LIT for further reaction with 16O2 or 18O2 molecule. The Rh3O– cluster was also generated by laser ablation of the Rh disk in the presence of 0.02% 16O2/He for reaction with reactant molecules of CO and H2. The temperature of cooling gas (He), reactant gases, and LIT reactor was around 300 K. A reflectron time-of-flight mass spectrometer[47] was used to detect the cluster ions ejected from the LIT reactor. The photoelectron imaging spectroscopy[48] was employed to characterize the structures of reaction intermediates Rh3CH2– and Rh3O–. Details on experimental methods can be found in the Supporting Information.

Theoretical Methods

Density functional theory (DFT) calculations using the Gaussian 09 program[49] were carried out to investigate the structures of reaction intermediates Rh3X– (X = CH2, CO, O) as well as the reaction pathways of Rh3– + CH4, Rh3CH2– + O2, and Rh3O– + CO. The PBE functional[38] has been proved to perform well for bare rhodium clusters[35] so the results by PBE method are given throughout this work. The TZVP basis sets[50] for C, H, and O atoms and the D95V basis set combined with the Stuttgart/Dresden relativistic effective core potentials (denoted as SDD in the Gaussian software)[51] for Rh atom were used. The reaction pathway calculations involved geometry optimization of reaction intermediates (IMs) and transition states (TSs) through which the IMs transfer to each other. The zero-point vibration corrected energies (ΔH0) in units of eV are reported in this work. The Rice–Ramsperger–Kassel–Marcus (RRKM) based theory[39] was used to predict the rates of internal conversion of reaction intermediates. The density of states (DOS) simulations based on the generalized Koopmans’ theorem[52] were performed to assign the structures of reaction intermediates through comparison with the photoelectron spectra in the experiment. Details on theoretical methods can be found in the Supporting Information.