Surface structural-chemical characterization of a single-site d⁰ heterogeneous arene hydrogenation catalyst having 100% active sites

Abstract

Structural characterization of the catalytically significant sites on solid catalyst surfaces is frequently tenuous because their fraction, among all sites, typically is quite low. Here we report the combined application of solid-state ¹³C-cross-polarization magic angle spinning nuclear magnetic resonance (¹³C-CPMAS-NMR) spectroscopy, density functional theory (DFT), and Zr X-ray absorption spectroscopy (XAS) to characterize the adsorption products and surface chemistry of the precatalysts (η⁵-C₅H₅)₂ZrR₂ (R = H, CH₃) and [η⁵-C₅(CH₃)₅]Zr(CH₃)₃ adsorbed on Brønsted superacidic sulfated alumina (AlS). The latter complex is exceptionally active for benzene hydrogenation, with ∼100% of the Zr sites catalytically significant as determined by kinetic poisoning experiments. The ¹³C-CPMAS-NMR, DFT, and XAS data indicate formation of organozirconium cations having a largely electrostatic [η⁵-C₅(CH₃)₅]Zr(CH₃)₂⁺···AlS⁻ interaction with greatly elongated Zr···O_AlS distances of ∼2.35(2) Å. The catalytic benzene hydrogenation cycle is stepwise understandable by DFT, and proceeds via turnover-limiting H₂ delivery to surface [η⁵-C₅(CH₃)₅]ZrH₂(benzene)⁺···AlS⁻ species, observable by solid-state NMR and XAS.

Organometallic molecule-derived heterogeneous catalysts are of increasing interest owing to their enhanced thermal stability and activity vs. their homogeneous analogs, and their atomically precise tailorable metal-ligand structures vs. other heterogeneous catalysts (1, 2). Furthermore, it is becoming increasingly evident that the inorganic support in many systems is noninnocent and can function as both a ligand and an activator, with the chemically important but poorly understood nature of the catalyst–support interaction strongly modulating catalytic activity and selectivity (3, 4). When adsorbed on Lewis acidic, dehydroxylated alumina surfaces, group 4 complexes such as Cp₂ZrR₂ (Cp = η⁵-C₅H₅; A, R = H; B, R = CH₃) and Cp*Zr(CH₃)₃ [C, Cp* = η⁵-C₅(CH₃)₅] were argued on the basis of high-resolution solid-state NMR spectroscopy to transfer an alkyl anion to unsaturated, Lewis acidic surface sites as in Fig. 1 (complexes B, C → qualitative model D) (5, 6). The resulting catalysts are extremely active for olefin hydrogenation and polymerization, and analogous ion-paired species form the basis for large-scale industrial polymerization processes (7, 8). However, kinetic poisoning experiments in which the catalytic sites are titrated in situ with H₂O or ^tBuCH₂OH indicate that ≤5% of D-type sites are catalytically significant, likely reflecting, among other factors, the established heterogeneity of alumina surfaces (5, 6, 9), hence rendering active site structural and chemical descriptions necessarily imprecise. In contrast to these results, chemisorption of such organozirconium precursors on SiO₂, Al₂O₃, and SiO₂-Al₂O₃ surfaces having appreciable coverage by weakly acidic OH groups predominantly yields covalently bound, poorly electrophilic E-type species via Zr–CH₃ protonolysis with CH₄ evolution (5, 6, 10, 11). Although the E-type sites may be characterized in some detail by high-resolution solid-state NMR and extended X-ray adsorption fine structure spectroscopy (EXAFS), they display minimal catalytic turnover in the absence of added, complicating activators [e.g., methylalumoxane or B(C₆F₅)₃], and the fraction of catalytically significant sites is unknown (12, 13). In such situations, it is experimentally impossible to unambiguously distinguish catalytically significant sites from inactive “spectator” sites, hence to fully understand the catalytic chemistry.

Fig. 1.

Proposed model structures of chemisorbed organozirconium complexes A (A), B (B), and C (C) on (D) dehydroxylated Lewis acidic metal oxides, (E) weakly Brønsted acidic hydroxylated metal oxides, and (F) highly Brønsted acidic sulfated metal oxides.

In marked contrast to the above results, chemisorption of these same organozirconium molecules on highly Brønsted “superacidic” sulfated metal oxides (14–16) such as sulfated zirconia (ZrS, H₀ = −16.1), sulfated titania (TiS, H₀ = −14.6), or sulfated alumina (AlS, H₀ = −14.6) yields, via Zr–CH₃ protonolysis with methane evolution, highly electrophilic adsorbate species that tentatively have been assigned F-type structures (Fig. 1) (17, 18). Here, H₀ is the standard Hammett acidity function, determined spectroscopically, by reactivity, or by temperature-programmed desorption/reaction (14–16). Remarkably, adsorbate structure C/AlS is one of the most active arene hydrogenation catalysts yet discovered, and by kinetic poisoning experiments, 97 ± 2% and 87 ± 3% of the Zr centers are catalytically significant for benzene hydrogenation and ethylene polymerization, respectively (17, 18). This unusually high percentage of catalytically significant d⁰ sites, the relatively clean adsorption/activation chemistry, and the unusual catalytic properties present a unique opportunity to structurally characterize such electrophilic surface catalyst structures in quantitative detail, hence to understand the origin of the unusual catalytic properties. Herein we report a combined ¹³C-cross-polarization magic angle spinning nuclear magnetic resonance (¹³C-CPMAS-NMR) spectroscopic, periodic density function theory (DFT) computational, and Zr K-edge XAS structure/reactivity characterization study of these supported organozirconium cations having nearly 100% active sites, and the informative active catalyst structural/catalytic chemistry description that emerges.

Results and Discussion

Complexes B and C were prepared, purified, and chemisorbed on AlS with rigorous exclusion of oxygen and moisture, using techniques described elsewhere (Materials and Methods) (3, 17, 18). Studies of C/AlS-mediated benzene hydrogenation were carried out as described in SI Materials and Methods, under conditions minimizing mass transport effects (3, 17, 18). The turnover frequency (N_t) for benzene hydrogenation at 25.0(±1) °C/1.0 atm H₂ is measured to be 120 (mol benzene)(mol Zr)⁻¹⋅h⁻¹, in good agreement with previously reported data (17, 18). As a control, studies of benzene hydrogenation in homogenous solution using Cp*Zr(CH₃)₃ + B(C₆F₅)₃ or Cp*Zr(CH₃)₃ + Ph₃C⁺B(C₆F₅)₄ˉ catalysts yield N_t = 1–2 (mol benzene)(mol Zr)⁻¹⋅h⁻¹, highlighting the dramatic role that chemisorption on this particular support plays in the catalysis. Furthermore, negligible turnover occurs on the sulfated oxides in the absence of chemisorbed complex C. To our knowledge, the only other supported early-transition metal catalysts active for arene hydrogenation are M[μ-CSi(CH₃)₃][CH₂Si(CH₃)₃]₂/SiO₂ (M = Nb, Ta), where N_t = 800 (mol benzene)(mol Nb)⁻¹⋅h⁻¹ and 333 (mol benzene)(mol Ta)⁻¹⋅h⁻¹) under far more drastic conditions (120 °C, 80–95 atm H₂), and little is known about their structures (19).

Chemisorption of ¹³C-labeled Cp*Zr(¹³CH₃)₃ (C′) and Cp*Zr(CH₃)₃ (C) onto AlS, dosing of C/AlS with 10% (vol/vol) ¹³C-enriched benzene (SI Materials and Methods), and the structure of C/AlS after benzene hydrogenation were first investigated by solid-state ¹³C-CPMAS-NMR spectroscopy. The spectra of C′/AlS and C/AlS (Fig. 2 A and B, respectively), exhibit three major resonances at δ 122.8, 49.7, and 9.2 ppm, with those at δ 122.8 and 9.2 straightforwardly assigned to Cp* framework and Cp*-CH₃ carbon atoms (6, 17, 18), whereas the downfield shifted Zr–CH₃ signal at δ 48.9 ppm indicates formation of a “cation-like” electron-deficient organozirconium species (20, 21), and negligible amounts of alkyl anion transfer to a quadrupolar Al site (Al-¹³CH₃ˉ; as in structure D) (17, 18, 21). After exposing a pentane slurry of C′/AlS to H₂ (SI Materials and Methods, Hydrogenolysis of Cp*Zr¹³Me₃/AlS and Fig. S1), the downfield Zr–¹³CH₃ signal at δ 48.9 ppm disappears, whereas the resonances associated with the Cp* ligand are unchanged, indicating formation of a catalytically active cationic Cp*Zr–hydride species (20–23). The ¹³C chemical shift of ¹³C-enriched benzene physisorbed on AlS is assigned at δ 120.3 ppm (Fig. 2C) (24), whereas exposure of C/AlS to ¹³C-enriched benzene yields a downfield shifted resonance at δ 127.5 ppm, consistent with benzene coordination to a cation-like d⁰ species (Fig. 2D) (25–27). Subsequent treatment of this species with substoichiometric H₂ in a benzene slurry yields a signal assignable to cyclohexane at δ 28.5 ppm, presumably physisorbed on the AlS surface (Fig. 2E and SI Materials and Methods, Benzene Hydrogenation Experiment with Cp*ZrMe₃/AlS). The breadth of the Cp* resonance suggests that some benzene molecules are still coordinated to the Zr center [see the discussion on EXAFS below].

Fig. 2.

Solid-state ¹³C-NMR spectra of organozirconium species on sulfated aluminum oxide. ¹³C-CPMAS-NMR spectra (100-MHz, 16K scans; repetition time, 5 s; contact time, 2 ms; spinning speed, 5 kHz) of (A) Cp*Zr(¹³CH₃)₃, C′/AlS; (B) Cp*Zr(CH₃)₃, C/AlS; (C) ¹³C-enriched benzene physisorbed on AlS; (D) C/AlS/H₂ exposed to ¹³C-enriched benzene; and (E) C/AlS after ¹³C-enriched benzene hydrogenation. *Rotational sidebands; #, impurity.

Molecular catalyst–surface interactions were modeled next using the DFT periodic formalism (28) (SI Materials and Methods), beginning with the AlS surface. First, it is found that the AlS surfaces expose two predominant sulfate species, sites S_A and S_B (Fig. S2), as well as (Al)_nOH hydroxyl groups, in which the OH is coordinated either to three Al (n = 3, O3 in Fig. S2) or to two Al ions (n = 2, O2 in Fig. S2). The formation of S_A sites involves a double-exchange/condensation reaction with the surface and does not afford an acidic proton, whereas S_B arises from a single exchange/condensation reaction with the surface, preserving one acidic proton that is then transferred to an Al-O surface site via an acid/base exchange process. Surface Brønsted acid properties and computed ν(S = O) vibrational modes, both before and after organozirconium complex chemisorption, are in good agreement with the experiment (17, 18). In principal, the precursor Zr–CH₃ protonolysis/chemisorption may occur at either a surface (Al)₂O(2)H or an (Al)₃O(3)H site. The DFT results indicate that Zr–CH₃ protonolysis is favored at the (Al)₃O(3)H site owing to the greater Brønsted acidity, which may be quantified by the relative computed stabilities of the anionic surfaces formed on deprotonation. The chemisorbed organozirconium surface species were next modeled for complexes A–C by placing the cationic species on the anionic/deprotonated AlS surface and seeking energy-minimized structures, as shown in Fig. 3. In all cases, two structures with similar energetic stabilizations are located, arising from the interaction between the Zr cation and sulfate species S_A and S_B. Any attempt to simulate the interaction of a Zr cation with a surface (Al)_nO⁻ species fails because of the sterically encumbered approach of the Zr metal-ligand complex to the surface. For all adsorbate complexes investigated, the interaction with S_A forms structures with rather long computed Zr···O distances of 2.20 Å, 2.22 Å, and 2.24 Å for complexes A, B, and C, respectively. Alternatively, when the interaction involves surface S_B units, the cationic Zr complex lies between two S = O groups with computed Zr···O mean distances of 2.37 Å, 2.42 Å, and 2.36 Å for complexes A, B, and C, respectively (Fig. 3). The different computed Zr···O distances are ascribable to the differing electronic and steric characteristics of the A–C ligands. As is discussed below, these results are in good agreement with the EXAFS data, and represent significantly longer distances than in typical molecular covalent Zr(IV)-OR bonds (1.94–2.01 Å),† suggesting very weak ion pairing. Moreover, the computational models of C/AlS yield Cp*Zr(CH₃)₂⁺···O_AlS⁻ structures in excellent agreement with the present solid-state NMR spectroscopic data and NMR data for metallocenium electrophiles in solution (20, 26, 29).

Fig. 3.

Energy-minimized computed chemisorbed catalyst structures for (A) Cp₂ZrH⁺ (complex A), Cp₂ZrMe⁺ (complex B), and Cp*Zr(CH₃)₂⁺ (complex C) coordination to the S = O groups of the sulfated alumina surface at S_A sites, and (B) Cp₂ZrH⁺ (A), Cp₂ZrMe⁺ (B), and Cp*Zr(CH₃)₂⁺ (C) coordination to the S = O groups of the sulfated alumina surface at S_B sites. Distances in angströms (Å). Al, yellow-green; Avg, average distance; C, olive; H, blue; O, red; S, orange; Zr, purple.

Finally, for all chemisorbed complexes, the DFT-derived internal organometallic catalyst metrical parameters were investigated and found to be very similar to those in the respective precatalyst molecules. Only minor distortions of the neutral molecule bond lengths and bond angles are observed, and these accommodate closer cation approach to the surface. Structural parameters for A/AlS, B/AlS, and C/AlS are summarized in Table 1. Similar patterns are observed in comparing the single-crystal diffraction-derived molecule structures of analogous ion pairs vs. their neutrally charged molecular precursors (†, 30).

Table 1.

DFT-computed geometrical parameters (angströms and degrees) for catalysts Cp₂ZrH₂, Cp₂ZrMe₂, and Cp*ZrMe₃ chemisorbed on sulfated alumina

	Cp2ZrH2		Cp2ZrMe2		Cp*ZrMe3
	SA	SB	SA	SB	SA	SB
Zr–H	1.82	1.82
Zr–CH3			2.26	2.29	2.24*	2.24*
Zr–Cpcenter	2.19*	2.20*	2.22*	2.22*	2.20	2.21
Zr–C(Cp)	2.51	2.51	2.53	2.53	2.51	2.51
Zr–O = S	2.20	2.37*	2.22	2.42*	2.24	2.36*
∠Cp1-Zr-Cp2	133.4	133.1	129.6	131.2
∠Me1-Zr-Me2					100.3	99.3
∠Cp1-Zr-Cp2-CH3			60.4	65.6
∠Cp1-Zr-Cp2-H	73.4	77.4

*Average values reported. ∠, angle.

Zr K-edge EXAFS and X-ray absorption near-edge structure (XANES) data for the neat and supported complexes were collected using strictly anhydrous/anaerobic methodology (31). Attempts to fit the C/AlS EXAFS spectrum with models involving four scattering shells, Zr–C(Cp*), Zr–CH₃(Cp*), Zr–CH₃, and Zr–O, were complicated by the overlap of the scatterers arising from the insignificant differences in both bond distances and scattering characteristics (phase and amplitude). Therefore, to better define catalyst structural details, the EXAFS spectra were analyzed by fitting the difference spectra of the organozirconium complexes before and after chemisorption on AlS, after exposure to benzene (Fig. S3), and after benzene hydrogenation (Fig. S4), thus retaining only the new scattering contributions associated with the structural changes while eliminating invariant contributions (Materials and Methods) (31, 32). In this way, the Cp₂ZrH₂ spectrum was used to isolate the scattering contributions of the Cp₂Zr fragment. To validate the EXAFS accuracy of the FEFF-generated phase shifts and backscattering amplitudes (33) for Zr–C scattering, compound A (Cp₂ZrH₂) with a single Zr–C scattering shell (Cp) first was examined. The spectrum of neat A and the fitted spectrum are shown in Fig. 4A, and relevant data are compiled in Table 2. The derived coordination number of 10.1 (two η⁵-C₅H₅ rings) is within the 10% uncertainty associated with EXAFS techniques (31, 32), whereas the derived average Zr–C distance of 2.52(±2) Å falls within the 2.39–2.58-Å Zr–C(Cp) bond distance range for numerous Cp₂ZrX₂ complexes (20, 34, 35). Compound B [Cp₂Zr(CH₃)₂] with two distinct Zr–C scattering shells (Cp and CH₃) similarly was characterized, and the spectrum can be fit to a model with 2.1 and 10.3 C atoms at distances from Zr of 2.27(±2) Å and 2.53(±2) Å, respectively (Table 2). These parameters are in excellent agreement with single-crystal diffraction data for B, where Zr–CH₃ = 2.276(5) Å and Zr–C(Cp) = 2.525(±12) Å (20, 35), and lend confidence in the present data analysis procedure. Fitting the difference EXAFS spectrum of B/AlS – B (Fig. 4B) as above reveals that the Zr–O scattering contribution in supported catalyst B/AlS involves 2.1 O atoms at an average Zr–O distance of 2.37(±2) Å. There is no evidence of a close Zr···Zr contact as might be expected for a dimeric species. These results are in excellent agreement with the aforementioned DFT models of B/S_B (Fig. 3B), which yield an average Zr···O bond distance of 2.37 Å, significantly longer than in typical molecular covalent Zr–O bonds, as noted above.

Fig. 4.

k²-weighted Fourier transforms (FT) of organozirconium complex Zr K-edge EXAFS. (A) k²-weighted Fourier transforms of neat Cp₂ZrH₂ (complex A). Red, Δk = 2.5–10.0 Å⁻¹ data; blue, fit (N_Zr-Cp = 10.1 at 2.52 Å) for ΔR = 1.4–2.6 Å; solid line, FT magnitude; dotted line, FT imaginary part. (B) k²-weighted Fourier transform of the Zr K-edge EXAFS of B/AlS − B. Red, Δk = 2.5–10.2 Å⁻¹ data; blue, fit (N_Zr-O = 2.1 at 2.37 Å) for ΔR = 1.5–2.6 Å; solid line, FT magnitude; dotted line, FT imaginary part. (C) Zr K-edge XANES from 17.96 to 18.04 keV. Red, C/AlS + benzene (E_o = 18.0013 keV); blue, C/AlS (E_o = 18.0005 keV). (D) k²-weighted Fourier transform of the Zr K-edge difference EXAFS of C·C₆H₆/AlS – C/AlS. Red, Δk = 2.7–10.6 Å⁻¹ data; blue, fit for ΔR = 1.6–2.3 Å (N_Zr-Bz = 3.0 at 2.35 Å); solid line, FT magnitude; dotted line, FT imaginary part.

Table 2.

EXAFS data for neat and supported organozirconium complexes

Entry	Complex	Scatterer	CN (±10%)	R (±0.02 Å)	Δσ2 (Å2 × 103)	E0 (eV)
1	A	Zr–Cp	10.1	2.52	0.0	0.3
2	B	Zr–CH3	2.1	2.27	−3.0	2.0
		Zr–Cp	10.3	2.53	0.0	0.4
3	B/AlS – B	Zr–Osupport	2.1	2.37	−3.0	10.8
4	C·C6H6/AlS – C/AlS	Zr–Cbenzene	3.0	2.35	2.0	11.6
5	C·C6H6/AlS – C·C6H6/AlS/H2	Zr–Cbenzene	1.6	2.36	2.0	9.7

XAS-monitored catalytic experiments next were carried out in which C/AlS was exposed to benzene, then H₂ at 25 °C. Fig. 4C compares the XANES data for C·C₆H₆/AlS with those of C/AlS. Note that C·C₆H₆/AlS exhibits a shift in the edge to higher energy of about 0.8 eV (18.0013 keV with chemisorbed benzene vs. 18.0005 keV without benzene) broadening of the XANES beyond the edge vs. untreated C/AlS, consistent with benzene coordination. Although Zr K-edge XANES (corresponding to a 1s to 5p orbital electronic transition) does not probe the d-band structure directly, and is thus less sensitive to changes in the coordination environment, it is reasonable that benzene π* orbital–Zr 5p orbital mixing would provide additional transition probability for the Zr 1s core electrons, leading to the observed spectral changes (36). More interestingly, as shown in Fig. 4D (and summarized in Table 2), fitting of the C·C₆H₆/AlS – C/AlS difference EXAFS spectrum reveals that each Zr center has obtained ∼3.0 additional carbon neighbors. In principle, this coordination might correspond to possible scenarios: (i) ∼50% of the organozirconium centers are coordinated to benzene in an η⁶-coordination fashion; (ii) ∼100% of the active sites are coordinated to benzene in an η³ mode; or (iii) metal coordination induces significant benzene ring deformation from planarity (37). The last hypothesis, however, is not supported by the EXAFS metrical data. Furthermore, the average EXAFS-derived Zr–C_benzene distance is 2.35(±2) Å, in good agreement with single-crystal diffraction-characterized benzene coordination to electrophilic d⁰ centers (27, 38). Although η³ coordination is relatively uncommon for π-complexed arenes (39, 40), the present results are consistent with the solid-state NMR results and DFT calculations, considering the overlap of four scattering shells.

The DFT models of the C·C₆H₆/AlS structure indicate that the benzene fragment has inserted between the Zr···O contacts at either site S_A or site S_B and engages in ηⁿ benzene coordination (Fig. 5A), with three Zr–C(benzene) distances (Zr–C1, Zr–C2, and Zr–C6) slightly shorter than the others by ∼0.05 Å. Thus, the computed Zr–C(benzene) mean distance is found to be 2.71 Å, somewhat overestimated vs. the EXAFS-derived distance [2.35(±2) Å] and the average distance for cationic Zr(IV)–η⁶-benzene complexes [2.62(±2) Å] in the Cambridge Crystallographic Database (34). This is not completely unexpected, and in other combined DFT + EXAFS studies, DFT calculations sometimes have overestimated bond distances (41–43). As a consequence of the benzene coordination, the Zr centers of chemisorbed species C/S_A and C/S_B are displaced substantially from the anionic surface coordination sites S_A and S_B, respectively, and the C∙C₆H₆/AlS structure converges to a unique conformation in which the cationic complex lies between the S_A and S_B vicinal sulfate groups (Fig. 5A). A long contact between the Zr center and the S_B surface site is observed with a Zr···O(S_B) distance of 4.17 Å. With the benzene inserted between the cationic Zr center and the S_A anion, the Zr···O(S_A) distance is elongated even further to 5.21 Å. This benzene insertion/activation process is calculated to be exothermic by ~14 kcal/mol with respect to C/S_B, and ~4 kcal/mol with respect to C/S_A. Interestingly, the arene intrusion between the cationic Zr-alkyl center and the weakly coordinating anionic surface is reminiscent of the ethylene coordination/activation mode at homogeneous cationic single-site catalyst centers and fluoroarylborate counteranion displacement that precedes monomer enchainment and polymerization (Fig. 5B) (44–46).

Fig. 5.

Energy-minimized computed structures of surface-bound and solution-phase catalyst–substrate complexes. (A) Cp*Zr(CH₃)₂⁺ cation coordination to/activation of a benzene molecule and an S = O fragment of an anionic surface sulfate group at two different AlS sites (C·C₆H₆/AlS). (B) Cp*Zr(CH₃)₂⁺ cation coordination to/activation of an ethylene molecule, which has displaced an H₃CB(C₆F₅)₃⁻ counteranion in homogeneous solution. Distance in angströms (Å). Al, yellow; B, pink; C, olive; F, green; H, blue; O, red; S, orange, Zr, purple.

Following substoichiometric benzene hydrogenation (Materials and Methods), fitting of the (C·C₆H₆/AlS + H₂) – C·C₆H₆/AlS EXAFS difference spectrum reveals that about 50% of the Zr sites retain a coordinated benzene molecule [CN = 1.6; Zr–C_benzene = 2.36(±2) Å], in good agreement with the aforementioned ¹³C-CPMAS NMR results (Fig. 2E) and kinetic data showing that benzene is irreversibly captured by C/AlS, that the established rate law is zero-order in [benzene], and that the first H₂ addition is turnover limiting (17, 18). The pathway for the present, highly unusual d⁰-mediated arene hydrogenation process next was probed by DFT. The calculations (Fig. 6) reveal that the initial C∙C₆H₆/AlS hydrogenation/activation, involving Zr–CH₃⁺ hydrogenolysis to produce the active catalyst, Cp*ZrH₂∙C₆H₆/AlS, is exothermic by ∼38 kcal/mol, in agreement with the aforementioned NMR data (Fig. 2). As a consequence of the lesser Zr–H steric hindrance vs. Zr–CH₃, the metal center is drawn toward the S_B surface sites to afford a somewhat closer Zr···O contact of 2.32 Å. The initial step of the catalytic cycle consists of formal Hˉ transfer from Zr center to the activated arene and formation of a formal Zr–C σ-bond (I). A coordination site around the Zr center remains open, and the Zr···O(S_B) distance contracts to 2.18 Å, whereas the interaction with S_A is lost. The second step in the catalytic cycle involves H₂ activation (II) with slight elongation of Zr···O(S_B) to 2.30 Å, followed by Zr–C hydrogenolysis (III) to regenerate a Zr hydride. To achieve complete benzene hydrogenation, appropriate variants of the I→II→III sequence are repeated to ultimately produce cyclohexane. The potential energy and Gibbs free-energy profiles for this cycle, depicted in Fig. 6B, indicate that thermal and entropic contributions are most important in the benzene and H₂ activation steps and in the product release. In the former, this primarily reflects entropy changes associated with the bimolecularity, whereas opposite considerations hold for the product release. Note that the initial benzene capture remains exergonic because of the stronger binding vs. H₂, whereas the free-energy gain associated with H₂ activation is 2.4, 12.5, and 5.9 kcal/mol for the first, second, and third hydrogenation subcycles, respectively. These differences reflect small displacements of the Zr center from the surface, as required for H₂ binding. To locate the turnover-limiting step, we focused on the transition-state energies associated with the most relevant steps of the catalytic cycle (SI Materials and Methods). The results reveal that benzene coordination and intramolecular hydride transfer (step I) essentially are barrierless processes, whereas H₂ activation (step II) presents a significant barrier (17.0 kcal/mol) and for the Zr–C hydrogenolysis (step III), it is necessary to overcome a 3.5-kcal/mol energy barrier to break the H–H bond. These results point to the H₂ activation (step II) as the turnover-limiting step of the overall benzene hydrogenation cycle, in accord with the aforementioned experimental observations that the rate law is zero-order in [benzene] and first-order in [H₂], and with observations on a related organothorium/alumina system (5). In that case, D₂ is delivered in pairs to both arene faces, meaning that arene “flipping over” and/or rapid hydrogenate dissociation/reassociation are possible, but neither cyclohexene nor cyclohexadiene are present in detectable quantities during turnover.

Fig. 6.

Computed catalytic pathway for Cp*ZrH₂⁺/AlS-mediated benzene hydrogenation. (A) Catalytic cycle. (B) Corresponding energetic profile. Sequences of three steps are operative following Cp*ZrH₂⁺/AlS + benzene capture: I, hydride addition; II, H₂ activation; and III, Zr–C hydrogenolysis.

Conclusions

The present results provide a unique combined ¹³C-CPMAS NMR, DFT computational, and XAS picture of how molecule-derived d⁰ organozirconium arene hydrogenation catalysts are activated and turn over on sulfated oxide surfaces. The long Zr⁺···O_AlS⁻ distances indicate loose, nondirectional ion pairing, as might be expected from the conjugate base of an extremely strong solid Brønsted acid in which the negative charge is highly dispersed (15, 16), and finds surprisingly close analogy to homogeneous ion-paired early-transition metal polymerization catalysts in which the nature of the ion pairing between the cationic catalyst and the charge-dispersed, electrostatically bound, and easily displaceable counteranion strongly modulates the barrier to olefin activation and enchainment (44–47). Indeed, this description is closely analogous to Fig. 5B, in which the initial activation of the incoming olefinic substrate by the electrophilic metal center requires geometrical loosening of the ion pairing (46–48). From a coordination chemistry perspective, this also suggests the intriguing possibility that such electron-deficient surfaces may be the long-sought, ultimate “weakly coordinating” anions.

Materials and Methods

The procedure for the chemisorption of the organometallic complex Cp*ZrMe₃ on sulfated alumina was previously reported (18). For the synthesis and characterizations of the hydrogenolysis of Cp*Zr¹³Me₃/AlS, dosing of Cp*ZrMe₃/AlS with benzene, benzene hydrogenation experiments with Cp*ZrMe₃/AlS, benzene hydrogenation experiment with Cp*ZrMe₃ + B(C₆F₅)₃, benzene hydrogenation experiment with Cp*ZrMe₃ + [Ph₃C][B(C₆F₅)₄], see SI Materials and Methods. For DFT calculations detaills, as well as EXAFS measurements and data analysis, also see SI Materials and Methods.