A [Cu₂O]²⁺ core in Cu-ZSM-5, the active site in the oxidation of methane to methanol

Abstract

Driven by the depletion of crude oil, the direct oxidation of methane to methanol has been of considerable interest. Promising low-temperature activity of an oxygen-activated zeolite, Cu-ZSM-5, has recently been reported in this selective oxidation and the active site in this reaction correlates with an absorption feature at 22,700 cm⁻¹. In the present study, this absorption band is used to selectively resonance enhance Raman vibrations of this active site. ¹⁸O₂ labeling experiments allow definitive assignment of the observed vibrations and exclude all previously characterized copper-oxygen species for the active site. In combination with DFT and normal coordinate analysis calculations, the oxygen activated Cu core is uniquely defined as a bent mono-(μ-oxo)dicupric site. Spectroscopically validated electronic structure calculations show polarization of the low-lying singly-occupied molecular orbital of the [Cu₂O]²⁺ core, which is directed into the zeolite channel, upon approach of CH₄. This induces significant oxyl character into the bridging O atom leading to a low transition state energy consistent with experiment and explains why the bent mono-(μ-oxo)dicupric core is highly activated for H atom abstraction from CH₄. The oxygen intermediate of Cu-ZSM-5 is now the most well defined species active in the methane monooxygenase reaction.

Methane, the major component of natural gas, is highly abundant and its selective oxidation to a liquid form such as methanol would help harness its potential as a petroleum alternative for fuels and in the petrochemical industry. Currently, industrial methanol production from methane is accomplished by the steam reforming of methane into syngas, a mixture of H₂ and CO, followed by methanol synthesis. These processes require high temperatures and pressures and as a result, alternatives such as the selective direct oxidation of methane to methanol are of considerable interest. However, methane has the strongest C—H bond of any hydrocarbon (104 kcal/mol), thus its selective oxidation to methanol without further oxidation is extremely challenging.

In nature, the methane monooxygenase (MMO) enzymes accomplish the direct conversion of methane into methanol at ambient temperatures at iron- or copper-oxygen active sites (1–3). Researchers have attempted to mimic MMO's activity synthetically by creating oxygen-activated iron and copper catalysts (4). Recently, an O₂-activated zeolite Cu-ZSM-5 has been shown to convert methane to methanol at low temperatures (100 °C) with selectivity >98% (5). Reactivity only occurs at a small fraction of the total copper sites in the zeolite (5). Researchers have attempted to determine the identity of the active copper-oxygen core, although the challenge of distinguishing it from spectator sites has not been overcome and no consensus has been reached as to its structure (6–8). The oxygen-activated site of Cu-ZSM-5 was correlated to an absorption feature at 22,700 cm⁻¹, a spectroscopic handle unique to the reactive core (6). An absorption feature at this energy, however, is not sufficient to unambiguously define the nature of the active site. In the present study, the Raman vibrations corresponding only to the active site are selectively resonance enhanced, providing detailed experimental insight. The same active site can be generated with N₂O and is shown to be capable of selective methane oxidation. The resonance Raman (rR) and absorption data are coupled to normal coordinate (NCA) and density functional theory (DFT) calculations to define the geometric and electronic structure of the active site of oxygen-activated Cu-ZSM-5 and the nature of its activation for H atom abstraction.

Results and Discussion

Spectroscopic Characterization of the Active Core in Oxygen-Activated Cu-ZSM-5.

Tuning a laser into the characteristic absorption feature (Fig. 1Inset A) of the oxygen-activated Cu-ZSM-5 active site leads to resonance enhancement of Raman vibrations associated only with this chromophore and therefore the active site for methanol synthesis (Fig. 1). The rR spectra show multiple vibrations that profile the absorption band (Fig. 2A)*, confirming that these vibrations are in-resonance with this electronic transition. These vibrations gain intensity with increasing Cu/Al ratio (as does the 22,700 cm⁻¹ absorption feature) (Fig. 2B) and are not observed after the site reacts with CH₄ (at 200 °C) or is heated in He >350 °C (Fig. 2C) [both lead to loss of the 22,700 cm⁻¹ band (9)], confirming that the vibrations observed are from the active site. A number of these resonance enhanced vibrations are sensitive to isotope perturbation when the active site is generated with ¹⁸O₂ (Fig. 1 blue). These isotope-sensitive vibrations involve the active oxygen species, as the reaction of the ¹⁸O₂-generated active site with CH₄ results in CH₃¹⁸OH formation (5). The most intense isotope-sensitive vibration is at 456 cm⁻¹ (Δ(¹⁸O₂) = 8 cm⁻¹) and additional isotope-sensitive vibrations are at 237 cm⁻¹ (Δ(¹⁸O₂) = 3 cm⁻¹), 870 cm⁻¹ (Δ(¹⁸O₂) = 40 cm⁻¹), 1,725 cm⁻¹ (Δ(¹⁸O₂) = 83 cm⁻¹), and 1,852 cm⁻¹ (Δ(¹⁸O₂) = 52 cm⁻¹) (Fig. 1). A broad resonance enhanced feature is observed at 974 cm⁻¹ (Δ(¹⁸O₂) = 10 cm⁻¹) corresponding to the combination band of the 514 cm⁻¹ and 456 cm⁻¹ vibrations based on energy and isotope shift. The second member of a progression of the 456 cm⁻¹ vibration is observed at 908 cm⁻¹ (Δ(¹⁸O₂) = 15 cm⁻¹). An intense isotope-insensitive feature is also observed at 514 cm⁻¹ with a shoulder near 540 cm⁻¹.

Fig. 1.

rR spectra (λ_ex = 457.9 nm) of Cu-ZSM-5 + ¹⁶O₂ (red), ¹⁸O₂ (blue). Inset A: Absorption spectrum of oxygen activated Cu-ZSM-5. Inset B: “^16,18O₂” (green), and 1:1 normalized sum of ¹⁶O₂ and ¹⁸O₂ (black).

Fig. 2.

rR spectra of O₂-activated Cu-ZSM-5. (A) rR spectra of O₂-activated Cu-ZSM-5 (Cu/Al = 0.54) collected at eight λ_exs from 351 nm to 568 nm with corresponding absorption spectrum Inset. (B) rR spectra (λ_ex = 457.9 nm) of O₂-activated Cu-ZSM-5 with varying Cu/Al ratios from 0.10 to 0.54 with corresponding absorption spectra Inset. (C) rR spectra (λ_ex = 457.9 nm) of Cu-ZSM-5 (Cu/Al = 0.54) pretreated in O₂ at 450 °C, recorded before and after heating in He at 450 °C and after reaction with CH₄ at 200 °C.

These data uniquely define the geometric and electronic structure of the oxygen-activated Cu-ZSM-5 active site. In past work, the oxygen intermediate of Cu-ZSM-5 was assigned as a bis(μ-oxo)dicopper(III) core based on the observation of an absorption feature at 22,700 cm⁻¹ and a Cu—Cu distance of ≈2.9 Å from EXAFS studies of O₂ activated Cu-ZSM-5 (6). However, this activated Cu core accounts for only a small fraction of the total Cu content in Cu-ZSM-5 (≈5%) (5, 9), making it impossible to distinguish its contribution to the EXAFS data from that of the abundantly present spectator Cu sites. Thus, EXAFS cannot provide structural information for this core. A bis(μ-oxo)dicopper(III) complex is characterized by an intense isotope-sensitive stretch ≈600 cm⁻¹ in its rR spectrum (10, 11). The absence of such a feature in the rR spectrum of the oxygen intermediate of Cu-ZSM-5 excludes the assignment of the active site as a bis(μ-oxo)dicopper(III) complex. Further, the observed stretches are inconsistent with the assignment of the site structure as a μ-(η²:η²) peroxo dicopper(II) (νCu—Cu∼270 cm⁻¹, νO—O∼750 cm⁻¹) (12) or a Cu(II)-superoxo (νO—O∼1,050–1,100 cm⁻¹) species (13, 14) (Table 1). The 870 cm⁻¹ vibration does occur at a reasonable energy for an O—O stretch of an η¹-hydroperoxo copper(II) or a μ-1,2-peroxo or hydroperoxo dicopper(II) species (15–17) (Table 1). To evaluate these possibilities, the Cu-ZSM-5 oxygen intermediate was formed with mixed isotope O₂ (“^16,18O₂”), a statistical mixture of ¹⁶O₂:^16/18O₂:¹⁸O₂ (1:2:1 ratio). If the 870 cm⁻¹ vibration is an O—O stretch, vibrations would be observed at 830, ≈845 and 870 cm⁻¹ with a 1:2:1 intensity pattern, with the feature at ≈845 cm⁻¹ representing the ¹⁶O¹⁸O and ¹⁸O¹⁶O stretches (18, 19). Fig. 1 B Inset (green) shows the rR spectrum of the active species prepared with “^16,18O₂,” which overlays a 1:1 normalized sum of the pure ¹⁶O₂ (red) and ¹⁸O₂ (blue) spectra (black in Fig. 1, Inset B). The absence of an intermediate stretch ≈845 cm⁻¹ eliminates the possibility that the site is a μ-1,2-peroxo or hydroperoxo dicopper(II) species. Thus, all presently known copper-oxygen active site structures for the oxygen intermediate of Cu-ZSM-5 are excluded by the rR data (Table 1).

Table 1.

Spectroscopically characterized mononuclear and binuclear Cu/O₂ species

In fact, the most intense isotope sensitive feature is the 456 cm⁻¹ (Δ(¹⁸O₂) = 8 cm⁻¹) stretch; the 870 cm⁻¹ (Δ(¹⁸O₂) = 40 cm⁻¹) vibration has relatively low intensity but its second quantum (based on isotope shift) is observed at 1,725 cm⁻¹ (Δ(¹⁸O₂) = 83 cm⁻¹) with 6 times higher intensity than the fundamental. This intensity pattern has been observed in μ-oxo-diferric complexes (20). The observed high intensity of the 456 cm⁻¹ isotope sensitive vibration in Fig. 1 supports its assignment as the symmetric stretch of an oxo group bridging two Cu centers (21). The weak 870 cm⁻¹ vibration is then assigned as an antisymmetric metal-oxo stretch (ν_as), which should not be enhanced in the rR spectrum, while its second quantum (2ν_as) is symmetric and consequently would have higher rR intensity than the fundamental, as observed for the 1,725 cm⁻¹ vibration.^† The energy of the second quantum of the ν_as for ¹⁶O₂-activated Cu-ZSM-5 is expected to be at twice the energy of the fundamental (1,740 cm⁻¹) but is instead observed at 1,725 cm⁻¹. Further, both the energy splitting and the intensity of the 1,725 cm⁻¹ peak relative to the nearby 1,852 cm⁻¹ peak change upon ¹⁸O₂-isotopic labeling. This is consistent with a Fermi resonance between two ¹⁸O-isotope sensitive features i.e., the fourth quantum of ν_s (4ν_s, predicted at 1,824 cm⁻¹) and 2ν_as (Fig. S2). The presence of both strong symmetric and weak antisymmetric stretches leads to the assignment of the Cu-ZSM-5 active site as a bent Cu—O—Cu core.

The presence of such a core in Cu-ZSM-5 has been proposed as an intermediate during dehydration based on EXAFS data (22), but as no absorption feature is observed under these reaction conditions (9) formation of the bent Cu—O—Cu core defined here can be excluded for this process. Computational studies on Cu-ZSM-5 have surveyed the possible stabilization of several binuclear Cu sites in Cu-ZSM-5, including Cu—O₂—Cu and Cu—O—Cu cores (7, 23), but without spectroscopic data to discriminate among them, no assignment of the active site structure could be made. The presence of a bent Cu—O—Cu core in synthetic model complexes has been proposed (24–26), but to date, no such complex has been unambiguously defined. As a result, the data reported in the present study are the first to definitively characterize a mono-oxygen bridged binuclear Cu site in any system. Activation of Cu-ZSM-5 by N₂O instead of O₂ also results in the formation of an absorption feature at 22,700 cm⁻¹. Starting from a dehydrated Cu-ZSM-5 (450 °C in He), this absorption feature is observed at lower temperatures (100 °C) than for the O₂ activation reaction (175 °C) (Fig. 3, Inset). rR spectra were collected on the N₂O activated core of Cu-ZSM-5 and compared with the rR vibrations observed for the O₂ activated core in Fig. 3. Identical rR features are observed after both treatments, confirming that activation of Cu-ZSM-5 by either O₂ or N₂O results in formation of the same active site (Fig. 3). This N₂O-generated active core is also capable of low-temperature selective oxidation of CH₄ to CH₃OH to an extent comparable to the O₂ activated Cu-ZSM-5 site (14 ± 2 μmol per gram of Cu-ZSM-5).

Fig. 3.

rR spectra (λ_ex = 457.9 nm) of Cu-ZSM-5 (Cu/Al = 0.54) activated in O₂ (Upper) or N₂O (Lower). Inset Diffuse reflectance UV-vis spectra recorded at room temperature after treatment of Cu-ZSM-5 with O₂ (Upper) or N₂O (Lower) in the temperature range of 100 °C-200 °C and 50–125 °C respectively. The arrows on Inset represent the increasing temperature of the O₂ and N₂O treatment (temperature difference between two spectra is 25 °C).

Geometric and Electronic Structure of the Cu₂O Core in Cu-ZSM-5.

Normal coordinate analysis (NCA) was used to correlate the observed rR symmetric (ν_s) and antisymmetric (ν_as) vibrations and their isotope shifts to the bending angle for a Cu₂O core, ∠CuOCu. The symmetric (ν_s) and antisymmetric (ν_as) stretch energies correlate with ∠CuOCu, and by fitting the four observables (ν_s and ν_as with ¹⁶O₂ and ¹⁸O₂) to NCA calculations, the observed stretches are consistent with a ∠CuOCu of 140° (Table S1). We further used these NCA calculations to evaluate the dependence of ν_s and ν_as of a Cu₂O site on ∠CuOCu. As this angle decreases from 140° to 100°, the latter angle consistent with a bis-μ-oxo dicopper structure, the predicted ν_s increases to 600 cm⁻¹ (Table S1). Thus, the observed ν_s (≈600 cm⁻¹) in bis-μ-oxo dicopper complexes is dominantly the result of the tight Cu—O—Cu angle associated with the presence of a second single atom bridge. The observed ν_s and ν_as for the active site of Cu-ZSM-5 require that this site has a wide ∠CuOCu (≈140°), and exclude the possibility of a second oxygen atom bridge.

Using the NCA-calculated ∠CuOCu of 140° as a starting point, these studies were expanded with DFT calculations to evaluate the capacity of the ZSM-5 lattice to host a Cu₂O site. To form this active site, 4 electrons would have to be transferred to O₂ from Cu^I ZSM-5, forming the Cu₂O core and an additional oxo group. This can result in an active core containing either two Cu^IIIs or two Cu^IIs with an accompanying 2-electron transfer from nearby Cu^I ions to reduce the additional O atom.^‡ In ZSM-5, the 10-membered rings of the lattice are the only sites with pore-sizes appropriate for access of O₂, N₂O, and CH_4, and within these rings, a Cu₂O core could only bind two lattice Os to each Cu (28). We therefore initially modeled the site with a tetrahedral Al(OH)₄⁻ ligand (referred to as an Al T-site) bound bidentate to each Cu atom of the Cu₂O core (Fig. 4A, small Al T-site model “T—Cu₂O” where the Hs cap the oxides in the Si lattice positions). This simplified ligation has been found to model the contribution of the lattice reasonably well (7). The wavefunction of T—Cu^III₂O (singlet) is electronically unstable and relaxes to a T—Cu^II₂O core with two additional holes delocalized to the T-site oxygens. A T—Cu^III₂O can only be stabilized by adding a fourth ligand (OH⁻) to each Cu^III. However, no feature associated with a Cu-OH stretch [computationally predicted at 655 cm⁻¹ (Δ(¹⁸O₂) = 16 cm⁻¹)] is observed in the rR spectrum (Fig. 1). Thus, the experimental evidence argues against the presence of a fourth O-ligand that would be necessary to stabilize a Cu^III₂O site. The calculations do, however, support a stable Cu^II₂O electronic structure for the active site^§. We evaluated the effect of the ∠CuOCu of the T—Cu^II₂O model on the DFT calculated ν_s and ν_as and found that, in agreement with the NCA calculations, an ∠CuOCu of 138° results in the best correlation with experimental data (Table 2, T—Cu^II₂O).

Fig. 4.

Structural models of ZSM-5 and the Cu₂O intermediate used for DFT calculations. (A) Small model, T—Cu₂O. (B) 10-membered ring of ZSM-5 containing two Al-sites separated by two Si-sites C. Large model constructed from part of a 10-membered ring (boxed atoms, B.), L-Cu₂O.

Table 2.

Experimentally observed rR features, DFT vibrational predictions (¹⁸O-isotopic shifts in parentheses) and DFT-optimized geometric parameters for the small Al T-site (T-Cu^II₂O) and large lattice (L-Cu^II₂O) models.

	rR Features				Geometric Parameters
	Cu-O-Cu bend (Δ18O2)/cm−1	νs, (Δ18O2)/cm−1	νas, (Δ18O2)/cm−1	Al-T-site, (Δ18O2)/cm−1	∠Cu-O-Cu	Cu-O, Å	Cu-Cu, Å
Experiment	237 (−3)	456 (−8)	870 (−40)	514/540 (0)
T-CuII2O (calcd)	300 (−3)	498 (−6)	890 (−42)	571 (0)	138°	1.74	3.29
L-CuII2O (calcd)	253 (−2)	456 (−5)	852 (−37)	556/568 (0)	139°	1.75/1.76	3.29

Within the 10-membered ring of ZSM-5, the experimentally calibrated NCA and DFT calculated structural parameters most closely fit an active site bound to two Al T-sites connected by two intermediate Si T-sites (Fig. 4B). This placement is also consistent with the crystal structure of Cu^I-ZSM-5, which resolves two Cu^I atoms in the 10-membered ring separated by two intermediate T-sites (28). Thus, a larger model was formed with each Al T-site ligand extended with covalently bonded Si T-sites (Fig. 4C, Large Lattice Model, “L-Cu₂O”). The L-Cu^III₂O also electronically relaxes to a Cu^II₂O core upon optimization, confirming that the ZSM-5 lattice can only stabilize a Cu^II₂O core^¶. The calculated ∠CuOCu is 139° with no constraints imposed on this core. The calculated ν_s (456 cm⁻¹) andν_as (852 cm⁻¹) and ¹⁸O-isotopic shifts (5 and 37 cm⁻¹ respectively) for L-Cu^II₂O agree well with the experimentally observed values for these vibrations [ν_s = 456 cm⁻¹ (Δ(¹⁸O₂) = 8 cm⁻¹) and ν_as = 870 cm⁻¹ (Δ(¹⁸O₂) = 40 cm⁻¹)] (Table 2). Further, L-Cu^II₂O has calculated Al-T-site vibrations between 556 and 568 cm⁻¹, consistent with the observed ¹⁸O-isotope insensitive vibrations at 514 cm⁻¹ and 540 cm⁻¹ in Fig. 1. The Cu—O—Cu bending mode of L-Cu^II₂O is calculated at 253 cm⁻¹ (Δ(¹⁸O₂) = 2 cm⁻¹), in agreement with the observed weak vibration at 237 cm⁻¹ (Δ(¹⁸O₂) = 3 cm⁻¹) in Fig. 1.^‖ To investigate the origin of the absorption feature at 22,700 cm⁻¹, time dependent DFT calculations (TD-DFT) were performed on L-Cu^II₂O. These predict a single dominant feature at 23,400 cm⁻¹ originating from the bridging oxo ligand to Cu^II charge transfer (Fig. S4), providing an assignment for the 22,700 cm⁻¹ band in Fig. 1 Inset A. The DFT calculations on L-Cu^II₂O thus accurately reproduce both the observed rR vibrations and the dominant absorption feature of oxygen-activated Cu-ZSM-5, a further confirmation of this geometric and electronic structural assignment.

Reaction with CH₄.

The disappearance of the 22,700 cm⁻¹ absorption band as a function of temperature was used to evaluate the activation energy (E_a) for the reaction of oxygen-activated Cu-ZSM-5 with CH₄ (Fig. 5A). From the Arrhenius plots of the reaction at temperatures between 110 and 200 °C, the activation energy is 15.7 ± 0.5 kcal/mol (Fig. 5C, open square). This reaction has a kinetic isotope effect (KIE) (3.1 at 175 °C) resulting in an increase in the E_a of the reaction by 3.1 ± 0.5 kcal/mol, obtained from an Arrhenius plot of the reaction with C²H₄ (Fig. 5 B and C, filled triangle). This indicates that C—H bond breaking is involved in the rate limiting step of the oxidation of CH₄. Product analysis (by ¹H-NMR, Fig. S5A) of the reaction of oxygen-activated Cu-ZSM-5 with CH₂²H₂ confirms this KIE. Analysis of the KIE as a function of reaction temperature shows a similar increase in E_a (2.1 ± 0.3 kcal/mol) for reaction with a C—²H as compared with a C—H bond (Fig. S5B). Because comparable increases in E_a are observed by monitoring the decay of the 22,700 cm⁻¹ band and by product analysis, the absorption feature can be directly correlated to the reactive site.

Fig. 5.

E_a and KIE measured from the decay of the 22,700 cm⁻¹ absorption band of O₂-activated Cu-ZSM-5 (Cu/Al = 0.54). (A and B) Decay of the 22,700 cm⁻¹ absorption band of an O₂ activated Cu-ZSM-5 (Cu/Al = 0.54) measured during reaction at 175 °C with CH₄ (A) and C²H₄ (B). The arrow on the figures shows the evolution in time. Time interval between two spectra is 15 seconds. (C) Arrhenius' plots for CH₄ (open squares) and C²H₄ (filled triangle).

DFT calculations were used to obtain a transition state structure and evaluate the reactivity of the spectroscopically-validated L-Cu^II₂O model with CH₄ (Fig. 6A). The H atom abstraction reaction is calculated to have a zero-point corrected activation energy of 18.5 kcal/mol and an increase E_a upon reaction with C²H₄ of 1.3 kcal/mol, both in reasonable agreement with the experimentally observed E_a of 15.7 kcal/mol and ΔE_a of 3.1 kcal/mol. The calculated H atom abstraction is endothermic by only 13.8 kcal/mol (ΔE), reflecting the difference in bond dissociation energy of H-CH₃ compared with that of the [Cu—OH—Cu]²⁺ intermediate that would be generated. The strong O—H bond of the [Cu—OH—Cu]²⁺ species (calculated bond dissociation energy of 90 kcal/mol) helps drive the reaction. The [Cu—OH—Cu]²⁺ intermediate is best described as a delocalized-radical species, with Mulliken atomic spin densities of 0.26 and 0.44 on the Cu atoms and 0.17 on the bridging O. In the subsequent step, rebound of the hydroxyl radical (leaving 2 Cu^I) to couple with the methyl radical completes the reaction. Thus, the L-Cu^II₂O model of oxygen-activated Cu-ZSM-5 can abstract an H atom from CH₄ through a low activation barrier consistent with experiment (Fig. 6A).

Fig. 6.

DFT-calculated reactivity of L-Cu^II₂O with CH₄. (A) Reaction coordinate of H atom abstraction from CH₄ by L-Cu^II₂O. (B) SOMOs at the TS shown with the line of CH₄ approach in the plane (Left) and below the plane of the figure (Right).

The electronic structure of L-Cu^II₂O provides insight into the activation of [Cu₂O]²⁺ for this reaction. There are two low-lying singly-occupied orbitals (SOMOs) that change along the reaction coordinate (Fig. 6B). Approaching the transition state, one of these orbitals, perpendicular to the Cu—O—Cu plane i.e., directed into the zeolite channel, localizes along the O—H—C vector and gains significant oxygen p-character (24% O with dominant delocalization into H^·CH₃ in the transition state, Fig. 6B, Upper). The second singly-occupied low-lying molecular orbital is covalently delocalized between one of the coppers (CuA in Fig. 6B, Lower) (46%) and the oxygen atom (20%) with some delocalization into the second copper (CuB in Fig. 6B, Lower) (22%). This is essentially a (CuB)^II-O^˙̄, cupric-oxyl species with some delocalization of the oxidized metal into the second Cu (CuA). This is further supported by changes in the Mayer bond order (29) (MBO) of the CuA—O and CuB—O bonds along the reaction coordinate. In L-Cu^II₂O, the MBO of the CuA—O bond is 1.24 and that of CuB—O is 1.02. The difference in the two MBOs is the direct result of constraints imposed by the zeolite lattice: The lattice oxygens form stronger bonds with CuB (MBOs of 0.54 and 0.44) than with CuA (MBOs of 0.46 and 0.37). The increased donation of the lattice Os to CuB leads to a decrease in the strength of the CuB—Oxo bond. At the transition state the CuA—O bond is significantly weaker than in the reactant L-Cu^II₂O (MBO of 0.72, a decrease of 0.52). While the CuB—O bond also decreases in strength at the transition state (MBO of 0.86, a decrease of 0.19), this change is less than half that of the CuA—O bond. A cupric-oxyl species has been proposed as a highly activated intermediate in copper-oxygen chemistry, however, it has not been observed in any bioinorganic or model system (30). The significant oxygen p character in the low-lying half occupied orbital renders it highly activated as an acceptor orbital in H atom abstraction (Fig. 6B Upper).

Conclusions

The oxygen-activated core of Cu-ZSM-5, reactive in the selective oxidation of methane into methanol, correlates to an absorption feature at 22,700 cm⁻¹. Raman vibrations of this active core are resonantly enhanced with λ_ex within this absorption feature. ¹⁸O isotope sensitive rR features profile this absorption feature, uniquely defining the active site structure. ¹⁸O₂ and ^18,16O₂ labeling experiments exclude all previously defined oxygen bridged mononuclear and binuclear Cu sites for this active core. The energies, intensities, and isotope shifts of the observed vibrations lead to the assignment of this core as a mono-oxygen bridged binuclear Cu site. NCA and DFT calculations define the geometric and electronic structure of this core and provide insight into how the constrained lattice of ZSM-5 restricts the coordination environment of the bound Cu atoms and their spatial orientation resulting in the formation of a bent mono-μ-oxo dicupric core. These calculations accurately reproduce the observed rR vibrations and isotope shifts and the dominant absorption feature.

The observation of a KIE (4.0 to 2.0 at 115–200 °C) shows C—H bond breaking is involved in the rate limiting step. Because similar increases in E_a in C—H vs. C—²H bond breaking were observed by monitoring the decay of the 22,700 cm⁻¹ band and by product analysis, the absorption feature can be directly correlated to the reactive site. DFT calculations show the formation of a [Cu—OH—Cu]²⁺ intermediate with a calculated energy barrier in good agreement with experimental data. The strong O—H bond of the delocalized radical [Cu—OH—Cu]²⁺ species helps drive the reaction. The electronic structure of L-Cu^II₂O provides insight in the reaction mechanism and shows that upon approach of CH₄, one of the SOMOs of the Cu₂O core gains significant oxygen p-character oriented into the zeolite channel and toward the H-CH₃ bond. At the transition state this results in the formation of a cupric-oxyl intermediate that is highly activated for H atom abstraction and explains the reactivity of the bent mono-μ-oxo dicupric core in Cu-ZSM-5 in the selective oxidation of methane into methanol. While the nature of the active site in the enzyme particulate MMO is still a matter of discussion, our studies on Cu-ZSM-5 definitively determine that a binuclear cupric core can in fact perform the critical reaction of methane oxygenation.

Materials and Methods

Synthesis of Cu-ZSM-5 Samples.

A series of Cu-ZSM-5 samples with increasing Cu loading was prepared starting from Na-ZSM-5 (Si/Al = 12, ALSI-PENTA). Aqueous exchange of Na-ZSM-5 with Cu(CH₃CO₂)_2. H₂O solutions of various Cu²⁺ concentrations resulted in Cu-ZSM-5 with Cu/Al ratios of 0.10, 0.22, 0.31, 0.42 and 0.54, using procedures reported in ref. 5. The Cu and Al contents were determined by Inductively Coupled Plasma Atomic Absorption (ICP-AA).

UV-Vis Spectra for Cu-ZSM-5 Samples After O₂, N₂O, He, ¹⁸O₂ Treatment.

In situ UV-vis-NIR spectra in the diffuse reflectance spectroscopy mode (DRS) of the Cu-ZSM-5 series were recorded on a Varian Cary 5 UV-vis-NIR spectrophotometer. All spectra were recorded at room temperature and cooling of the sample was done in an atmosphere of the corresponding treatment gas. O₂ and He treatments at 450 °C were done following procedures reported in ref. 5. The UV-vis spectra after interaction of a dehydrated Cu-ZSM-5 (Cu/Al = 0.54, dehydrated at 450 °C in He) with N₂O and O₂ at temperatures <250 °C were recorded after the sample was flushed at room temperature with O₂ or N₂O (5 min, 100vol.% O₂ or N₂O flow of 50 mL·min⁻¹) and heated under this closed atmosphere (5 °C min⁻¹ to the indicated temperature).

For the ¹⁸O-labeled O₂ experiments (¹⁸O₂ and a statistical mixture of ¹⁶O₂:^16/18O₂:¹⁸O₂ with 1:2:1 ratio purchased from Buchem BV), a calcined Cu-ZSM-5 (Cu/Al = 0.54) was heated in He (heating rate of 1 °C min⁻¹ to 450 °C). At 450 °C a flow of 30 mL·min⁻¹ of the ¹⁸O-labeled O₂ (100 vol.%) was passed over the sample for 5 min. The sample was kept under a closed ¹⁸O₂ atmosphere at 450 °C for 30 min.

Reaction with CH₄, C²H₄ and C²H₂H₂.

After O₂ calcination, the reactions with CH₄ and C²H₄ (purchased from Aldrich) (total flow of 50 mL·min⁻¹, 10% CH₄ or C²H₄ in He at reaction temperatures between 115 and 200 °C) were monitored. The setup consists of a plug-flow reactor (i.d. = 8 mm) fitted inside a furnace, a UV−Vis light source (Top Sensor Systems DH-2000 deuterium–halogen light source) and a photodiode array detector (Ocean Optics SD 2000) connected to the catalyst bed via optical fiber technology (Top Sensor Systems FCB-UV400-ME cable and FCB-UV400G-0.1-XHT high-temperature probe) (6). Each spectrum is the result of the superposition of 300 scans in the 38,000–12,000 cm⁻¹ region, each taking 50 ms. Thus, one spectrum is obtained every 15 s.

For reaction with C²H₂H₂ (purchased from Aldrich) and the subsequent extraction of deuterated methanol, an O₂ calcined Cu-ZSM-5 was reacted with C²H₂H₂. After 2 min flow of 100% C²H₂H₂ the sample was kept under a closed C²H₂H₂ atmosphere for 20 min at the reaction temperature. The sample was then cooled to room temperature and the catalyst was poured into a screwtop vial containing 2 mL of ²H₂O. The closed vial was heavily stirred at room temperature for 20 h. The supernatant was sampled and transferred to an NMR tube for ¹H-NMR analysis. ¹H-NMR spectra were recorded on a Bruker (Fällanden, Switzerland) Avance II 500 spectrometer operating at 500.130 MHz. To quantitate the amount of methanol generated by low temperature O₂ and N₂O activated Cu-ZSM-5, a dehydrated sample was activated in O₂ or N₂O at 200 °C followed by reaction with CH₄ (10% in He, 50 mL·min⁻¹) at 200 °C. A known amount of acetonitrile was added to 1 mL of the supernatant as a calibrant and methanol concentration was determined by ¹H-NMR.

Resonance Raman Spectroscopy.

Resonance Raman spectra were recorded on a Princeton Instruments ST-135 back-illuminated CCD detector and on a Spex 1877 CP triple monochromator with 1,200, 1,800, and 2,400 grooves per millimeter of holographic spectrograph gratings. Excitation was provided by a Coherent I90C-K Kr⁺ ion laser or an Innova Sabre 25/7 Ar⁺ CW ion laser. The spectral resolution was <2 cm⁻¹. Spectra were recorded at room temperature at powers ranging from 5mW to 75mW at the sample. Sample preparation was the same as for the samples prepared for UV-vis measurements. Baseline spectra were collected using ground, activated charcoal in a quartz U-tube side arm.

Calculations.

Normal coordinate analysis (NCA).

Normal coordinate analysis (NCA) was based on a Urey-Bradley force field with the FG-matrix method of Wilson (31) and was performed on a triatomic Cu—O—Cu model (C_2v symmetry) with a modified version of Schachtschneider's GMAT and FPERT programs (32, 33).

DFT calculations.

Spin-unrestricted DFT calculations were performed with Gaussian 03 (34) and the B3LYP functional. The 6–311G* basis set was used for the geometry optimizations and frequency calculations of T—Cu₂O. A split basis set was used for geometry optimizations, transition state searches, frequency calculations, and TD-DFT calculations of L-Cu₂O with 6–311G* on the three Cu—O—Cu core atoms and the four Cu-coordinating lattice Os and 6–31G* on all other atoms. For reaction coordinate calculations, the split basis set included 6–311G* on the CH₄ atoms. Broken symmetry calculations were performed for all singlet spin states. Single point calculations were performed on optimized structures using the 6–311+G* basis set. For geometry optimizations and transition state searches with the L-Cu₂O models, eight Si atoms were constrained at their crystallographically defined coordinates (35). Mulliken population analyses and MBO analyses were performed with QMForge (36) and vibrations and molecular orbitals were visualized in GaussView (37).