The prototypical σ complexes are the dihydrogen (H₂) complexes first described by Kubas, which contain an H-H unit coordinated to a metal (2, 4). In the case of alkanes, the interaction of a C-H bond with a metal is particularly weak, largely because σ-complexation is often stabilized by a backbonding component that is highly inefficient for a C-H bond when compared with a H₂ ligand. An understanding of this mode of bonding is interesting and important both from a theoretical point of view and to explain the role they play in the reactions of the C-H moiety.

The C-H activation reaction is a key step in potential functionalization of the normally inert alkane molecule, and its integration into catalytic processes is an important goal (5, 6). The C-H activation of alkanes has been shown to involve alkane complexes as intermediates in a number of elegant IR studies (5–7). After coordination, the C-H bond breaking step is usually fast and irreversible, precluding observation of the alkane complex with a relatively “slow” technique such as NMR spectroscopy. However, alkane complexes are not always unstable with respect to C-H activation. Two important examples that are formed photolytically are the long known M(CO)₅(alkane) (M = Cr, Mo, W) systems (1, 8) and Re(Cp)(CO)₂(n-heptane). This latter complex, characterized by time-resolved IR (TRIR) spectroscopy, had by far the longest lifetime measured (≈25 ms at 298 K) of any alkane complex in solution at the time (9). This crucial result paved the way for NMR studies of Re(Cp)(CO)₂(alkane) (10–12) and further IR studies (13), because at ≈183 K, the lifetime of these complexes (≈1 h) is sufficient for NMR experiments. The formation of long lived Re(Cp)(CO)₂(alkane) complexes has recently prompted the study of the photolysis of Re(Cp′)(CO)₃ (Cp′ = Cp, Cp*) in supercritical methane (scCH₄) and liquid ethane. After photolysis in scCH₄, a rapid equilibrium between Re(Cp′)(CO)₂(CH₄) and Re(Cp′)(CO)₂(CH₃)H is observed (14).

There are also many examples of what may be considered model complexes for the coordination of alkanes, namely agostic alkyl complexes (other than α-agostic species) (15, 16). In these compounds, the agostic η²-C,H interaction is stabilized by the chelate effect, because these ligands are tethered to the metal elsewhere.

Using combined IR, NMR, and theoretical studies, we have recently thoroughly characterized an organometallic xenon complex Re(ⁱPrCp)(CO)(PF₃)Xe (17, 18). It was noted that the stability of this complex was slightly higher than that of the closely related Re(ⁱPrCp)(CO)₂Xe species, formed in the same reaction mixture. Hence, the question was whether the introduction of a PF₃ ligand would stabilize alkane complexation in species of the type Re(Cp)(CO)(PF₃)(alkane) compared with Re(Cp)(CO)₂(alkane). Alternatively, would the [Re(Cp)(CO)(PF₃)] fragment display reactivity patterns more like those of [Re(Cp*)(CO)(PMe₃)] (19) or [Re(Cp)(PMe₃)₂] (20) in which the presence of the phosphorus donor is known to promote activation of the C-H bond to form alkyl hydride species?

Our studies of the interaction of the [Re(Cp)(CO)(PF₃)] fragment with three alkanes cyclopentane, cyclohexane, and pentane described here suggest behavior that is “in between” these two extremes of alkane complex vs. alkyl hydride, and this behavior is surprisingly dependent on the alkane.

NMR Studies.

Photolysis in cyclopentane. A solution of Re(Cp)(CO)₂(PF₃) (1) in 95% cyclopentane:5% pentane-d₁₂ (added for locking purposes) was cooled to 185 K. Before photolysis, there was a single peak in the ¹H NMR spectrum at δ 5.20 due to the Cp protons of 1, and one doublet in the ¹⁹F NMR spectrum at δ −2.56 (d, ¹J_FP = 1,232 Hz). The sample was irradiated with UV light for 1 min resulting in the depletion (≈15%) of the ¹⁹F resonance of 1 and the concomitant growth of two new resonances at δ −3.28 (dt, ¹J_PF = 1,211 Hz, 2 × J_FH = 7.9 Hz) labeled species A and at δ −32.4 (d, ¹J_PF = 1,398 Hz), due to free PF₃ (Fig. 1).

Fig. 1. 282.4-MHz 19F NMR spectra of 1 in cyclopentane at 185 K. (Lower) Before photolysis. (Upper) After 1 min of photolysis. (Inset) Expansion of the highlighted half of resonance A.

This observation is consistent with previous experiments (17) where photolysis of 1 led to the loss of either a CO or a PF₃ ligand to form [Re(Cp)(CO)(PF₃)] and [Re(Cp)(CO)₂]. These fragments can interact with the alkane solvent as outlined in Fig. 2. Resonance A can be assigned to either a cyclopentane (3a) or a cyclopentyl hydride (4a) complex or to a rapidly equilibrating mixture of both. The triplet splitting of resonance A, due to J_FH, suggests a coupling to two effectively equivalent hydrogens.

Fig. 2. Possible photoproducts obtained from photolysis of 1 in alkanes.

¹H NMR spectra showed a corresponding decrease in the signals of 1 upon irradiation. A quintet grows in at δ −2.29 in the ¹H spectrum that is due to the bound CH₂ unit of the cyclopentane ligand in the previously characterized Re(Cp)(CO)₂(cyclopentane) (2a) (10), and a singlet at δ 4.92 is due to the Cp ligand of 2a. The alkane in complexes 2a–2c is known not to undergo oxidative cleavage to form alkyl hydrides under these conditions (10–12).

In addition to the resonances due to 2a in the ¹H spectrum, another peak due to a Cp ligand of a species with a higher concentration was also observed to grow in at δ 5.21. Intensities suggested that this peak was likely from the same species that produces the ¹⁹F resonance A, i.e., 3a/4a. Further photolysis (1 min) increased the concentration of all of the newly observed ¹⁹F and ¹H signals. Significantly, there was no evidence for any shielded ¹H resonance (δ < 0) expected for 3a/4a. As a result, the temperature of the NMR probe was raised to 215 K in 10-K steps in the absence of photolysis. As the temperature was increased, the ¹H NMR spectra showed the emergence of a broad resonance A′ at δ −2.92, which sharpened and increased in intensity at higher temperature. A′ was assigned to the bound CH₂ protons of cyclopentane in Re(Cp)(CO)(PF₃)(cyclopentane) (3a), and resonance A is also assigned to 3a (Fig. 3). The intensity of signal A in the ¹⁹F NMR spectra was not temperature-dependent.

Fig. 3. Variable-temperature, 300-MHz 1H NMR spectra after photolysis of 1 in cyclopentane.

The apparent increase in concentration of 3a at higher temperature is an artifact of the selective excitation scheme used to suppress the resonances of the free alkane solvent. The selective excitation can cause broad peaks to be attenuated or even “lost” with the broadest peaks suffering the most severe attenuation, as was observed at 185 K. Temperature cycling between 215 and 185 K confirms that this is occurring as A′ disappears again at lower temperature. The cause of the broadening of the A′ resonance is the onset of a decoalescence phenomenon that is starting to occur, leading to increasing linewidths at lower temperatures. For the two hydrogens in a bound CH₂ unit of 3a to be equivalent, it is necessary for them to average their environments through a combination of both (i) an exchange of complexed and uncomplexed hydrogens as shown in Fig. 4 and (ii) rotation of the cyclopentane moiety around the metal-ligand bond axis. A slowing of either the rate of exchange or rotation could lead to the observed decoalescence.

Fig. 4. Exchange of complexed C-H bonds in a bound CH2 unit.

The resonances associated with 2a started to disappear as the temperature increased above 185 K, due to the documented thermal decomposition of 2a (10). At 215 K, signals due to 3a (including A, A′) also decreased with time, due to the thermal decomposition of 3a. Clearly 3a is thermally more stable than 2a. Resonance A′ at 215 K shows a splitting attributed to a J_PH coupling, confirmed by a ³¹P decoupling experiment. The ¹⁹F and ¹H resonances of 3a were connected to the same species by using a pair of shift correlation experiments, 2D ¹H-³¹P and ¹⁹F-³¹P. Both resonances A (¹⁹F) and A′ (¹H) were shown to correlate to the same ³¹P frequency at δ 124.2, confirming that they are from the same species. The ¹⁹F-³¹P correlation experiment shows the ¹H couplings in both the ¹⁹F and ³¹P dimensions seen earlier and reveals that J_PH ≈ 26 Hz [see the supporting information (SI)].

Key indicators of the presence of an alkane complex from an NMR perspective are a bound CH₂ or CH₃ unit in which δ ¹H is moderately shielded and, most importantly, a ¹J_CH value that is slightly reduced from that observed in a free alkane. For example, the protons of the bound CH₂ group in 2a appear at δ −2.3 with ¹J_CH = 113 Hz (cf. cyclopentane δ 1.50; ¹J_CH 129.4 Hz) (10). In all of the alkane complexes observed in NMR studies to date, the exchange of the complexed and uncomplexed C-H bond has always been fast on the NMR time scale (Fig. 4). This means that the observed δ ¹H and ¹J_CH values of the protons on the bound carbon are an average of the values for complexed and uncomplexed hydrogens. Density functional theory calculations suggest some deshielding of the ¹H resonance and an increase in ¹J_CH of the uncomplexed hydrogen(s) in Re(Cp)(CO)₂(cyclohexane) (12) compared with the free alkane. Therefore, the true shielding and reduction of ¹J_CH for the complexed hydrogen may be just over twice the observed changes in the averaged ¹H NMR shift and ¹J_CH for a CH₂ group compared with the free alkane. The trends observed in both agostic alkyl and dihydrogen complexes suggest that the more stretched the complexed C-H moiety becomes, the smaller ¹J_CH will become. Typically the ¹J_CH of a complexed C-H bond may be reduced by up to 50% relative to the uncomplexed unit (free, ≈125 Hz; complexed, 60–120 Hz) based on data from agostic species [although cases of ¹J_CH < 40 Hz are known (15)]. In an exchange averaged situation, one expects ¹J_CH in a bound CH₂ of an alkane complex to be ≈90–128 Hz.

In the case where complete oxidative cleavage occurs, the complexed C-H bond is broken, and now the comparable coupling is a ²J_CH interaction, which is expected to be small, typically <10 Hz, and likely of opposite sign to ¹J_CH. Examples of complexes that undergo a rapid exchange of alkyl and hydride protons, likely via an alkane complex intermediate, are known, e.g., [Cp*Os(dmpm)(CH₃)H]⁺ (21).

It may not be possible to differentiate complexed alkanes from alkyl hydride species (4) that undergo rapid exchange of hydride and alkyl protons based on the δ ¹H or δ ¹³C values alone, because they may be similar in these two scenarios. However, the averaging of a large ¹J_CH with a small (negative) ²J_CH in the alkyl hydride case would be expected to produce an averaged J_CH value of ≈55–70 Hz, significantly less than that observed for an alkane complex.

To gain insight into the nature of the interaction of cyclopentane in 3a, photolysis experiments in partially ¹³C-labeled cyclopentane (25% cyclopentane-¹³C₁, 15% cyclopentane-d₁₀, 60% normal cyclopentane, equivalent to ≈7% ¹³C total) were carried out. A ¹³C edited 1D ¹H-{³¹P} experiment, which selects only the resonances of hydrogens directly attached to a ¹³C nucleus with simultaneous ³¹P decoupling, was used at 210 K, revealing J_CH = 75 ± 2 Hz. A 2D ¹H-¹³C correlated experiment provided δ ¹³C = −6.80 for the bound carbon of the cyclopentane of 3a compared with δ ¹³C = −31.2 for 2a. δ ¹³C of 3a is also higher than those reported for trans-Re(Cp)(PMe₃)₂(R)H (R = n-C₆H₁₃, c-C₃H₅, CH₃), which vary from δ −14 to δ −39 (20). In lieu of more data, values of δ ¹³C of the carbon in the coordinated CH₂ unit (C_α) are not a priori able to differentiate structures in this case.

The observed value of ¹J_CH ≈75 Hz for the interacting cyclopentane in 3a is significantly less than that reported for the complexed cyclopentane in 2a (112.9 Hz). It is also larger than the maximum value predicted for an averaged coupling in a cyclopentyl hydride complex 4a (≈70 Hz). This finding suggests at least three possibilities for the mode of coordination of the alkane in 3a. (i) The cyclopentane unit is coordinating to the metal center in a σ-alkane complex fashion with a highly stretched complexed C-H bond, significantly greater than the 1.15–1.17 Å calculated for the complexed C-H distance of Re(Cp)(CO)₂(cyclohexane) (2c) (12). (ii) The complexed C-H bond of the coordinated cyclopentane has undergone oxidative cleavage to form the cyclopentyl hydride 4a, but the ¹J_CH value for the C_α-H is unusually large. (iii) There is rapid exchange between the σ-alkane and its corresponding alkyl hydride complex, and there is a sizeable amount of each component (3a and 4a) in equilibrium. In this latter case, the observed J_CH would be the weighted average from all of the contributing structures in Fig. 5; i.e., ¹J_CH of an alkyl group, ¹J_CH of an uncomplexed C-H in a σ-alkane, ¹J_CH of a complexed (and possibly stretched) C-H in a σ-alkane, and ²J_CH of a hydride. As an example, if the averaged ¹J_CH value in 3a were the same as 2a, 113 Hz, and if a best estimate averaged value of ¹J_CH for 4a, ≈61 Hz, were present, an equilibrium with a ratio of 3a:4a of 27:73 would be required to produce the observed value of 75 Hz. If more stretched C-H bonds in the alkane complex component are present, the relative amount of 3a would be increased. A similar approach may be applied to the use of J_PH as well. The averaged value of J_PH in the case of a cis-alkyl hydride 4a is expected to be much larger (estimate ≈33 Hz) than for the alkane complex 3a, which is estimated to be <10 Hz but increasing with C-H bond elongation. The observed value of 26 Hz suggests that the products have character that is closer to the alkyl hydride/highly stretched end of the range of possibilities.

Fig. 5. Exchange process for alkyl and hydride protons via alkane complexes.

Based on NMR studies alone, it is not possible to differentiate these three possibilities while there are exchange processes present at the lowest useable temperatures. Other methods must therefore be applied. In the case of IR spectroscopy, separate rather than averaged bands can often be observed for equilibrating mixtures allowing for the differentiation of single components that would be averaged in NMR experiments. This approach is described later and suggests the presence of an equilibrating mixture of 3a and 4a.

Photolysis in n-pentane. The photolysis of 1 was investigated in a mixture of 95% n-pentane and 5% n-pentane-d₁₂ at 190 K, using a procedure similar to that described in the previous section. After photolysis, three ¹⁹F resonances were observed: a major product at δ −3.36 (dq, ¹J_FP = 1,212 Hz, 3 × J_FH = 6.33 Hz), and two low intensity signals at δ −3.32 (¹J_FP = 1,213 Hz) and δ −3.20 (¹J_FP = 1,214 Hz) for which J_FH could not be measured but is <5 Hz for δ −3.32. These three resonances were attributed to species of the formula Re(Cp)(CO)(PF₃)(1-pentane) (3b-1), and tentatively Re(Cp)(CO)(PF₃)(2-pentane) (3b-2) and Re(Cp)(CO)(PF₃)(3-pentane) (3b-3), respectively (Fig. 6). The nomenclature implies that coordination to the metal center occurs through a C-H group of the C1, C2, or C3 of n-pentane.

Fig. 6. Resolution-enhanced, 470.6-MHz 19F NMR spectrum after photolysis of 1 in n-pentane at 190 K showing resonances from complexes of the formula 3b. , secondary photoproduct.

The three J_FH splittings in the ¹⁹F resonance of the major product in n-pentane (3b-1) indicate that this product is formed by interaction of a CH₃ group with the [Re(Cp)(CO)(PF₃)] fragment, and because the three protons of the CH₃ group couple equally to the ¹⁹F nuclei, they are likely exchanging positions rapidly, too. This assignment is supported by experiments in 2,2,4,4-pentane-d₄ that produced exactly the same ¹⁹F resonance splittings (3 × J_FH), whereas use of pentane-d₁₂ results in no observable J_FH splittings and a small isotope shift. ¹⁹F NMR spectra suggest a rapid exchange between the three hydrogens of the CH₃ group even at 165 K. Exchange of complexed and uncomplexed hydrogens within the bound CH₃ group in Re(Cp)(CO)₂(1-pentane) (2b-1) is known to be fast at this temperature (11). ¹H NMR spectra were also collected (at 190 K), and these showed three low-frequency ¹H peaks due to the three different isomers of 2b (11). As in the case of cyclopentane, no ¹H resonances in the δ < 0 region that could correspond to the largest signals in the ¹⁹F spectrum were observed. However, when the temperature was increased to 205 K, in the absence of UV light, a broad signal at δ −1.67 appeared due to 3b-1. Resonances attributed to isomers of 2b rapidly disappeared at this temperature. When the temperature was lowered to 160 K in the absence of UV light, the broad signal disappeared, whereas the ¹⁹F signals did not show any major changes. A decoalescence phenomenon is clearly operative in the case of 3b-1 as well.

The experiment was repeated by using n-pentane-¹³C₁ at a higher temperature of 210 K. Two-dimensional ¹H-¹³C-{³¹P} correlation experiments reveal that this broad signal is correlated with a shielded carbon signal at δ −15.1, and a ¹³C edited ¹H NMR experiment shows J_CH = 89 ± 2 Hz. Again, this is an averaged coupling that in this case involves three hydrogens rather than two in the case of cyclopentane. If this species were a 1-pentane complex, 3b-1, the exchange of one complexed C-H with two uncomplexed C-H bonds of a coordinated CH₃ unit would be involved. Alternatively, if this species were an n-pentyl hydride, 4b, the coupling would be the average of one ²J_CH value for the hydride and two alkyl ¹J_CH values. The observed value of J_CH = 89 Hz is much lower than that of 117 Hz observed in 2b-1 and is at the upper limit of the range expected for an averaged coupling in an alkyl hydride system, predicted to be ≈71–89 Hz. Possible scenarios based on NMR data for 3b-1 are the following: (i) an alkane complex wherein the complexed C-H bond is stretched to near breaking; (ii) the system being pentyl hydride 4b with relatively high C-H coupling constants; or (iii) an equilibrium mixture of 4b and 3b where the equilibrium favors the alkyl hydride more than in the case of cyclopentane. TRIR experiments described below confirm that this third scenario is indeed the correct one.

In the case of 2b, there is a small thermodynamic preference for binding to the CH₂ sites over the CH₃ sites (11). Studies of species that induce C-H activation, such as [Tp′RhL] [Tp′ = Tris-(3,5-dimethylpyrazolyl)borate; L = CNCH₂CMe₃] have also shown a kinetic preference for binding at secondary sites (22). However, it is well known that there is a preference for C-H oxidative cleavage to occur at primary carbons (5, 6). In the case of [ReCp(CO)(PF₃)] interacting with n-pentane, it is possible that the availability of an energetically more accessible primary alkyl hydride 4b leads to a preference for interaction with the primary carbon, the opposite of that which occurs with 2b. An alternative explanation is that incorporation of the sterically more demanding PF₃ ligand leads to preferred binding of the less hindered primary carbon.

Photolysis in cyclohexane. The preferred conformation of a cyclohexane ring is a chair in which the axial and equatorial hydrogens within a CH₂ group are chemically distinct. At temperatures around 190 K, the rate of inversion of the chair forms, the process that can make axial and equatorial protons equivalent from an NMR perspective, is slow. By removing the equivalency of the CH₂ protons in the alkane molecule, the two dynamic processes described in the cyclopentane case, rotation of the alkane ligand and swapping of coordination of the metal center between the two CH₂ hydrogens, will never make these two protons equivalent in the ¹H NMR spectrum. Therefore, discrete resonances for axial and equatorial hydrogens of a bound cyclohexane moiety are expected.

By using a procedure similar to that used with cyclopentane, a solution of 1 in 50% cyclohexane:50% pentane-d₁₂ was cooled to 190 K. Upon photolysis, four new ¹⁹F NMR resonances were observed to grow in simultaneously. One resonance at δ −3.13 (¹J_PF = 1,213 Hz), likely with two unresolved ³J_FH couplings, both <5 Hz, was assigned to Re(Cp)(CO)(PF₃)(cyclohexane) (3c). The other resonances were due to the three isomers of Re(Cp)(CO)(PF₃)(pentane-d₁₂) (3b-d₁₂) previously assigned in the studies in n-pentane. A noticeable increase in the intensity of the ¹⁹F resonance of 3c and corresponding decrease in intensity of the signals due to 3b-d₁₂ after the cessation of photolysis was observed, indicating a thermodynamic preference for 3c over 3b-d₁₂.

¹H NMR monitoring of the photolysis showed the appearance of two new broad signals at δ −2.73 (J_PH ≈16 Hz) and δ −3.33 (J_PH unresolved) that integrated one proton each with respect to five protons of a new peak in the Cp region at δ 4.93. The signals at δ −2.73 and δ −3.33 were assigned to the equatorial and axial protons, respectively, of a bound CH₂ unit in the cyclohexane ligand of 3c. An unambiguous confirmation that these assignments are correct has not been obtained. Examination of the broad multiplet structures in the ¹H NMR spectrum suggests that the resonance at δ −3.33 contains two larger axial–axial ³J_HH splittings, consistent with it being an axial hydrogen and 2D exchange experiments favor this interpretation. A 2D ¹⁹F-³¹P correlation experiment revealed δ ³¹P = 118.3 for 3c, and the larger J_PH ≈16 Hz was retained in this experiment.

Repeating the experiment by using uniformly labeled cyclohexane-[¹³C]₆ reveals values of ¹J_CH of 93 ± 3 and 107 ± 3 Hz for the resonances at δ −2.73 and δ −3.33, respectively. A ¹H-¹³C correlated experiment indicates that both of these ¹H resonances are from the same bound CH₂ moiety that has δ ¹³C = −17.6. This can be compared with a ¹J_CH = 125 ± 0.5 and 96.5 ± 0.5 Hz for the equatorial and axial protons, respectively, observed for the bound CH₂ unit of the cyclohexane ligand in Re(Cp)(CO)₂(cyclohexane) (2c), which has δ ¹³C = −22.4 (12). For further comparison, free cyclohexane at the same temperature has ¹J_CH = 128 Hz (both hydrogens) and δ ¹³C = 26.8. In the case of 2c, there was a clear preference for the complexation of axial over equatorial bonds, leading to a significantly more shielded ¹H NMR shift and lower ¹J_CH value for the axial proton that complicates the comparison with 3c. It is possible that there is a steric factor playing a role here, the bulkier PF₃ ligand tending to disfavor the coordination of the sterically more encumbered axial sites relative to the situation in 2c with the result that there appears to be almost no thermodynamic preference for axial or equatorial sites in 3c. Regardless of which are the axial and equatorial protons, the average of the two J_CH couplings observed in 3c (99.5 Hz) is significantly less than the corresponding average in the alkane complex 2c (110.7 Hz). It is likewise clear that the average J_CH coupling of the bound CH₂ in 3c (99.5 Hz) is significantly larger than for the bound CH₂ in cyclopentane complex 3a (75 Hz). This finding suggests that in 3c, the complexed C-H bonds may be more stretched in comparison with those found in 2c but less stretched than in 3a. Alternatively (or simultaneously), there may be a contribution from alkyl hydride isomers, Re(Cp)(CO)(PF₃)(cyclohexyl)H 4c, that are in fast exchange with alkane complex isomers 3c. The relative amount of alkyl hydride isomers would be significantly lower for 3c than 3a. This proposed equilibrium between 3c and 4c is described in Fig. 7. The variety of possible orientations and geometries of the alkane, alkyl, or hydride ligands with respect to the PF₃ and CO ligands means that the observed signals may be the average of numerous subtypes of different isomers.

Fig. 7. Possible equilibria between four different cyclohexane and cyclohexyl hydride complexes. Keq appears to be close to 1 in complex 3c.

The value of J_CH for the axial hydrogen (107 Hz) is higher compared with the equatorial hydrogen (93 Hz) in bound cyclohexane and to a C-H in bound cyclopentane (75 Hz). This finding suggests that complexed axial C-H bonds retain a mostly, if not completely, alkane complex character. By comparison, complexed equatorial C-H bonds in cyclohexane and bound C-H in cyclopentane are becoming either progressively more stretched or have progressively greater contributions from the corresponding alkyl hydride isomer (i.e., K″ > K′ in this case; see Fig. 7).

TRIR Studies.

Pentane. The TRIR spectrum obtained after photolysis of 1 in n-pentane under CO (2 atm) is shown in Fig. 8. It is clear that the parent ν(CO) bands (1,941 and 2,004 cm⁻¹) are bleached, and three transient bands can clearly be observed at ≈1,887, 1,923, and 1,953 cm⁻¹. The transient bands are formed within the time resolution of the apparatus (20 ns). Previous TRIR studies on the photolysis of 1 in scXe (17) have shown both Re(Cp)(CO)₂Xe and Re(Cp)(CO)(PF₃)Xe are formed. The band at 1,923 cm⁻¹ is assigned to Re(Cp)(CO)(PF₃)(pentane) (3b). The band of 3b is not stable and decays monoexponentially [k_obs = 9.1 (± 0.1) × 10⁰ s⁻¹, 292 ± 1 K] in the presence of CO to reform 1. Comparison with the ν(CO) band positions for known Re(Cp)(CO)₂(alkane) complexes allows the transient bands at 1,887 and 1,953 cm⁻¹ to be assigned to Re(Cp)(CO)₂(pentane) (2b). However, the bands at 1,887 and 1,953 cm⁻¹ show different kinetic behavior (Fig. 9a). The low-frequency band at 1,887 cm⁻¹ decays monoexponentially [k_obs = 1.01 (± 0.01) × 10² s⁻¹, 292 ± 1 K]. The band at 1,953 cm⁻¹ decays with biexponential kinetics. The faster component is in good agreement with the decay of 1,887 cm⁻¹ [k_obs = 9.6 (± 0.2) × 10¹ s⁻¹, 291 ± 1 K], and this confirms the band's assignment to 2b. Further evidence for this assignment is that in the presence of CO, these bands decay to form Re(Cp)(CO)₃ (5). The presence of two components in the decay of the 1,953 cm⁻¹ band indicates that a second ν(CO) band is heavily overlapped with the 2b band at 1,953 cm⁻¹, and the slower decay is very similar to the band at 1,923 cm⁻¹ [k_obs = 9.4 (± 0.9) × 10⁰ s⁻¹, 291 ± 1 K]. To obtain more information regarding this mystery peak, we monitored the photolysis of 5 in n-pentane by TRIR to determine precisely the ν(CO) bands of Re(Cp)(CO)₂(pentane) (1,887 and 1,952 cm⁻¹), and the ratio of bands is clearly different from those obtained in the fitting of the Re(Cp)(CO)₂(PF₃) experiment (Fig. 8).

Fig. 8. TRIR spectra showing 2b, 3b, and 4b. (a) FTIR of 1 in n-pentane under CO (2 atm). (b) TRIR difference spectra of the same solution 50 μs after photolysis (266 nm). (c) Expansion (1,965–1,875 cm−1) of TRIR spectrum of 1 in n-pentane. (more ...)

Fig. 9. Kinetic traces of ν(CO) bands at ≈1,883–1,887 and 1,948–1,952 cm−1 after photolysis of 1 in pentane at 293 ± 1 K (a), cyclopentane at 292 ± 1 K (b), and cyclohexane 293 ± 1 K (c), all under (more ...)

Multilorentzian fitting of the TRIR spectrum obtained after irradiation of 1 reveals the presence of an extra band at 1,955 cm⁻¹. We have recently shown that photolysis of Re(Cp)(CO)₃ in scCH₄ generates Re(Cp)(CO)₂(CH₄) and Re(Cp)(CO)₂(CH₃)H that exist in a rapid equilibrium at room temperature in solution (14). We assign the overlapped band at 1,955 cm⁻¹ to Re(Cp)(CO)(PF₃)(pentyl)H (4b). The similar lifetimes of 3b and 4b also indicate that both may be present in a rapid equilibrium. The assignments of 3b and 4b have been further supported by density functional theory calculations in which we have computed the IR frequencies (see the SI).

Cyclopentane/cyclohexane. NMR experiments discussed above have indicated that the nature of the alkane solvent has a significant effect on the balance between coordination and activation of the alkane solvent. Experiments in cyclopentane have suggested the presence of a significant proportion of Re(Cp)(CO)(PF₃)(cyclopentyl)H (4a), whereas in cyclohexane only a small fraction, if any, of Re(Cp)(CO)₂(cyclohexyl)H (4c) was indicated to be present.

To investigate these results further, we carried out analogous experiments to those described above for pentane in cyclopentane and cyclohexane (see the SI). In both experiments, Re(Cp)(CO)(PF₃)(alkane) and Re(Cp)(CO)₂(alkane) (alkane = cyclopentane or cyclohexane) were observed. It was not possible to establish whether Re(Cp)(CO)(PF₃)(alkyl)H was formed from inspection of the band ratios of Re(Cp)(CO)₂(alkane) as described above for Re(Cp)(CO)(PF₃)(pentyl)H (4b). Re(Cp)(CO)(PF₃)(alkane) reacts with CO to reform the parent [cyclopentane: k_obs = 3.3 (± 0.1) × 10⁰ s⁻¹, 292 ± 1 K; cyclohexane: k_obs = 4.2 (± 0.2) × 10⁰ s⁻¹, 293 ± 1 K], slower than the corresponding reaction of Re(Cp)(CO)₂(alkane) to form Re(Cp)(CO)₃. We found no evidence from the kinetics of the ν(CO) band at ≈1,950 cm⁻¹ of Re(Cp)(CO)₂(cyclohexane) (2c) for the presence of another species. NMR studies suggest that that the amount of cyclohexyl hydride species 4c present is much less than in the case of pentane or cyclopentane. Any possible concentration of 4c would be small and therefore difficult to detect due to the overlapping of the IR spectral features. In the TRIR experiment in cyclopentane, we examined the decay of the two ν(CO) bands of Re(Cp)(CO)₂(cyclopentane) (2a) and both bands decay at a similar rate [1,883 cm⁻¹: k_obs = 4.4 (± 0.2) × 10¹ s⁻¹; 1,951 cm⁻¹: k_obs = 3.8 (± 0.1) × 10¹ s⁻¹, both at 292 ± 1 K]. The slightly increased lifetime of the band at 1,951 cm⁻¹ may be due to a small concentration of 4a, having ν(CO) bands overlapped with those of 2a again consistent with NMR studies. However, this assignment cannot be made solely on the TRIR kinetics.

Combined NMR and TRIR data reveal that the species produced by the interaction of n-pentane with [Re(Cp)(CO)(PF₃)] are an equilibrium mixture of Re(Cp)(CO)(PF₃)(pentane) (3b) and Re(Cp)(CO)(PF₃)(pentyl)H (4b). This contrasts with the [Re(Cp)(CO)₂] fragment, which only forms alkane complexes with n-pentane and has a slight preference for complexation of secondary C-H bonds. The relative concentrations of 3b and 4b may be temperature-dependent, with lower temperatures leading to an increase of the relative concentration of alkyl hydride, but accurate equilibrium constants are difficult to extract. Data for the products from the interaction of cyclopentane with [Re(Cp)(CO)(PF₃)] again suggest that an equilibrium between the cyclopentyl hydride 4a and the isomeric cyclopentane complex 3a exists. What is clear is that there is a trend in both the NMR and TRIR data on going from n-pentane to cyclopentane to cyclohexane with a progressive favoring of the proportion of alkane complex present. Hence, in the case of cyclohexane, even at 180 K there is only a small amount, if any, of the cyclohexyl hydride form 4c present. The NMR data suggest that there is little preference for the interaction of [Re(Cp)(CO)(PF₃)] with either axial or equatorial hydrogens in cyclohexane but that interaction with the equatorial site is more likely to result in the formation of 4c, based on the lower J_CH and higher J_PH values in the equatorial case. Alternatively, the J_CH and J_PH values may be indicative of a more stretched C-H interaction in the case of binding of the equatorial hydrogen. All of the alkane complex-alkyl hydride equilibria in Figs. 4, 5, and 7 are fast on the NMR time scale at 215 K, but decoalescence phenomena are observed at lower temperatures for 3a/4a and 3b/4b, possibly due to a slowing of rotation and/or swapping of complexed and uncomplexed hydrogens in an alkane ligand.

Although CO and PF₃ ligands are considered similar in terms of their π-acceptor properties, there is a general shift toward products that display a higher degree of oxidative cleavage when [Re(Cp)(CO)(PF₃)] reacts with alkanes in comparison with [Re(Cp)(CO)₂], which favors alkane complexation without activation. The complexes incorporating a PF₃ ligand have greater thermal stability and react more slowly with CO.

Re(Cp)(CO)₂(PF₃) (1) was synthesized by using the literature procedure (17).

NMR/Photolysis Experiments. NMR samples of 1 (≈1 mg) in a mixture of appropriate alkane and pentane-d₁₂ (530 μl) were prepared. These samples were precooled (160–223 K) in the probe of a Bruker (Rheinstetten, Germany) DPX 300, DMX 500, or DMX 600 NMR spectrometer. Light from a 100-W Hg arc lamp was delivered to the top of the solution by using a single-core fiber-optic as described in refs. 10 and 11. Progress of reactions was monitored by using ¹⁹F and ¹H NMR spectroscopy either on separate, similar samples or by alternating experiment type. Conventional ¹⁹F spectra were obtained and were ¹H-coupled. ¹H NMR spectra were obtained by using an excitation sculpting scheme to suppress the resonances of the free alkane (11). Collected NMR spectra and further experimental details are given in the SI.

IR Experiments. Re(Cp)(CO)₃ (Strem Chemicals, Newburyport, MA; 99%) was used as received. Cyclopentane (Fluka, Buchs, Switzerland; HPLC grade), pentane (Lancaster, Ward Hill, MA; >99%), and cyclohexane (Aldrich, St. Louis, MO; HPLC grade, >99.9%) were distilled from CaH₂ and degassed before use. All experiments were carried out by using a CaF₂ cell (Harrick, Ossining, NY; pathlength of 0.5–1.5 mm) under CO (2 atm) at room temperature. Fresh solution was flowed into the cell after every UV laser shot. Details of the diode laser-based TRIR apparatus are described in ref. 23. Briefly, the IR source is a continuous-wave IR diode laser (MDS 1100; Mütek, Herrsching, Germany). In these experiments, the change in IR transmission at one IR frequency was measured by a fast MCT detector, after UV excitation of the sample by a pulsed Nd:YAG laser (Quanta-Ray GCR-12; Spectra Physics, Mountain View, CA; 266 nm), which initiates the photochemical reactions. A spectrum was built up on a point-by-point basis in custom software, by repeating this measurement at different IR frequencies.

This work was supported by FUJIFILM Imaging Colorants (A.J.C.), the European Union (P.P.), the Engineering and Physical Sciences Research Council (M.W.G.), SASOL (C.M.B.), the Australian Research Council (G.E.B.), the University of Nottingham (A.J.C.), the University of New South Wales (G.E.B.), and a J. W. T. Jones Fellowship from the Royal Society of Chemistry (to J.P.R.).