Characterization of an organometallic xenon complex using NMR and IR spectroscopy

Abstract

Photolysis of Re(ⁱPrCp)(CO)₂(PF₃) in liquid or supercritical Xe yields two new compounds [Re(ⁱPrCp)(CO)₂Xe and Re(ⁱPrCp)(CO)(PF₃)Xe]. Re(ⁱPrCp)(CO)(PF₃)Xe has been characterized by NMR and IR spectroscopies. The compound is an organometallic Xe complex that has been characterized by using NMR spectroscopy and is shown to be longer-lived than other organometallic Xe complexes by IR spectroscopy. ¹⁹F, ³¹P, and ¹²⁹Xe chemical shifts have been determined. The ¹²⁹Xe chemical shift of Re(ⁱPrCp)(CO)(PF₃)Xe, δ –6,179, is a Xe shift that is significantly shielded, on the order of 1,000 ppm, with respect to free Xe. The coupling constants between coordinated ¹²⁹Xe and both the ¹⁹F and ³¹P nuclei present have been extracted, confirming the identity of the compound. Observed line widths give a lower limit to the lifetime of the coordinated Xe of 27 ms at 163 K.

It is now more than four decades since the first demonstration that noble gas elements are capable of forming stable compounds (1). Since then, many compounds containing Xe and, to a lesser extent, Kr have been prepared. The formation of covalent bonds of varying stability with fluorine (Kr, Xe), nitrogen (Xe), oxygen (Kr, Xe), and more recently chlorine (Xe) (2) has been observed (3, 4). One example of a largely covalent argon compound was observed in a matrix at cryogenic temperatures (5). The great majority of this noble gas chemistry has resulted from interaction of Kr and Xe with electron-deficient, highly oxidizing reagents to create compounds containing the noble gas in positive oxidation states (6). More recently, the occurrence of atomic Xe as a discrete ligand in a selection of cationic gold and mercury inorganic complexes has been demonstrated, with the bonding occurring through a lone pair of electrons on the Xe (7–10). Theoretical studies had previously predicted high stability of such cationic gold complexes (11).

This work extends the scope of atomic Xe as a ligand by observing it bound to a neutral, organometallic, low oxidation state metal fragment in the compound Re(ⁱPrCp)(CO)(PF₃)Xe [3, ⁱPrCp = η⁵-C₅H₄-CH(CH₃)₂]. The complex 3 was detected by using a combination of NMR and IR spectroscopies. The use of solution NMR spectroscopy to characterize this class of compounds is a significant advance. The time scale over which NMR measurements are made, typically hundreds of milliseconds, means that this technique cannot be applied to very short-lived intermediates. Complexes must be sufficiently robust to withstand the rigors of the molecular collisions that are occurring in solution to be characterized by NMR. Such conditions are not encountered in the solid state, in which crystal-packing forces may also stabilize otherwise weak interactions to the point at which crystallographic methods may be applied to effect characterization. The observation of relevant chemical shifts and, in particular, indirect J couplings, can allow deduction of atomic connectivity in such molecules. Furthermore, the ability of NMR spectroscopy to observe dynamic processes allows exchange between free and bound ligands to be observed and in this case demonstrates that the lifetime of the Xe bound to the Re center is significant.

Xe has been known to act as a ligand in neutral organometallic species for some time. As early as 1975, Perutz and Turner (12) generated M(CO)₅(L) (M = Cr, Mo, or W; L = Ar, Kr, or Xe) from the photolysis of M(CO)₆ in matrices at 20 K. Simpson et al. (13) extended this work to show that Cr(CO)₅Xe could be observed at low temperatures in liquefied (l)Xe solution. Weiller (14) studied W(CO)₅Xe, which has a lifetime of ≈1.5 min at 170 K in lXe, and the Xe–W bond energy in this compound was estimated to be ΔH^‡ = 8.4 ± 0.2 kcal·mol^–1 (1 cal = 4.18 J). Bergman, Moore, and coworkers (15, 16) studied the kinetics of Rh(Cp*)(CO)₂ [Cp* = η⁵-C₅(CH₃)₅] photolyzed species in lXe and lKr and the mechanism of coordination.

The lifetime of short-lived organometallic noble gas species was investigated in significant detail recently. Re(Cp)(CO)₂Xe (Cp = η⁵-C₅H₅) was shown by using time-resolved IR (TRIR) spectroscopy to be long-lived at room temperature in supercritical Xe (scXe) solution (17). Re(Cp)(CO)₂Xe was shown to have a lifetime of up to 3.5 min at 170 K in lXe, leading to the prediction that it was a realistic goal to observe such complexes by NMR spectroscopy (18). Xe, by comparison to the other noble gases, forms the most stable organometallic complexes. The reactivity of M(Cp)(CO)_nXe complexes, for instance, is in the order Re < Rh < Mn < Ta ≈ Nb (19, 20).

Materials and Methods

Xe (99.99%, natural abundance; Praxair, Danbury, CT) and ¹²⁹Xe (99.95% labeled; Chemgas, Boulogne, France) were used for the NMR studies, and Xe (research grade, natural abundance; BOC, Guildford, U.K.) and carbon monoxide (99.3%, natural abundance; BOC) were used without additional purification. The preparation of Re(ⁱPrCp)(CO)₂(PF₃) (1) from Re(ⁱPrCp)(CO)₃ is described in supporting information, which is published on the PNAS web site. Re(Cp)(CO)₂(PF₃) (4) was prepared in an analogous fashion from Re(Cp)(CO)₃.

NMR Spectroscopy. Spectra were recorded on a Bruker (Karlsruhe, Germany) DMX 500 spectrometer equipped with a 5-mm ¹H/¹⁹F-³¹P-Xe triple broad inverse probe. To study our system, we designed an NMR apparatus to allow a solution of Xe to be irradiated continuously with UV light while in the spectrometer and in the temperature range of lXe (161–165 K) (see supporting information). No lock solvent was used, because any solvent other than Xe would be a competing ligand. This lack of lock has contributed to some peak broadening in the spectra obtained (W_1/2 = 6–10 Hz). The 1D spectra shown are sums of typically 5 or 10 spectra of 64 or 32 scans each. These spectra were co-added after Fourier transformation of the individual spectra, and the peaks of the starting material were precisely aligned by using a software offset before addition of the spectra to allow for the effect of magnetic field drift.

Location of the fluorine chemical shift. In the first instance, the new compounds were identified on the basis of a new doublet in the ¹⁹F spectrum, which was directly detected by using standard pulse sequences. From this 1D spectrum, a ¹J_PF value was extracted, and this value was used in subsequent experiments. Experiments using labeled ¹²⁹Xe as a solvent allowed the extraction of the ³J_XeF value.

Location of the phosphorus chemical shift. ¹⁹F-³¹P heteronuclear single quantum coherence experiments were performed without refocusing and decoupling (21), and we used the States–time-proportional phase incrementation method to obtain pure phase spectra. Delays were optimized for a ¹J_PF value of 1,220 Hz, with 90° pulse lengths of 8.3 and 10.5 μs for ¹⁹F and ³¹P, respectively. Four scans per increment were collected in the ¹⁹F dimension with an acquisition time of 334 ms and a recycle delay of 800 ms. One hundred and twenty-eight points over a 5-ppm spectral width were collected in the ³¹P dimension. The experiment was performed initially with wider spectral widths to ensure that no folding had taken place.

Location of the Xe chemical shift. First, several 1D experiments to locate the chemical-shift window containing the ¹²⁹Xe resonance were performed. A modified version of the heteronuclear spin echo search experiment (22) in which 180° pulses on the X channel (¹²⁹Xe in this case) were replaced with broadband inversion pulses (BIPs) (23) was used to this end. The experiment uses BIPs to excite as broad a chemical-shift range of the X nucleus as possible. This approach was necessary before undertaking 2D ¹⁹F-¹²⁹Xe correlation experiments, because the possible chemical-shift range (≈8,000 ppm) is many times the effective bandwidth (≈200 ppm) of the rectangular Xe pulses that are used in the correlation experiments and the ¹²⁹Xe shift of 3 proved to be well outside of the previously known shift range. A 156-μs BIP-1382-250-15 pulse (23) was used as a 180° pulse on ¹²⁹Xe. The bandwidth of the BIP pulse under these conditions is 125 kHz, giving an effective bandwidth of ≈900 ppm, allowing the Xe chemical-shift range to be scanned in 900-ppm regions. Delays that, in theory, should be 1/[4 × (³J_XeF)] were set to a shorter value of 15.6 ms to offset the effects of relaxation. Once the spectral range containing the ¹²⁹Xe shift was known, ¹⁹F-¹²⁹Xe heteronuclear multiple quantum coherence experiments were recorded by using a standard pulse sequence (24) with the States–time-proportional phase incrementation f₁ phasing method. No decoupling of ¹²⁹Xe was used. Delays that, in theory, are 1/[2 × (³J_XeF)] were set to a smaller value of 45.5 ms. The experiment was carried out twice, once with a large (5,015-ppm) spectral width in the Xe dimension and once with a reduced (1,000-ppm) spectral width to ensure that no folding had occurred. Two hundred fifty-six increments in t₁ of four scans each were recorded. The 90° pulses on ¹²⁹Xe had a 9-μs duration. A spectral width of 33.8 kHz with 8,192 data points was acquired in the ¹⁹F dimension, giving an acquisition time of 122 ms. A recycle delay of 800 ms was used. The ¹²⁹Xe shift was referenced relative to external XeOF₄ by using an absolute referencing method with a value of Ξ= 27.810184 MHz (25).

IR Spectroscopy. The TRIR apparatus has been described in detail elsewhere (26). In these experiments, a Q-switched Nd:YAG laser (Spectra-Physics Quanta-Ray GCR-S12) operating at 266 nm initiated the photochemical reactions in a high-pressure optical cell containing a solution of 1 in scXe. A continuous-wave IR diode laser (Mütek MDS 1100) and mercury–cadmium–telluride detector were used to monitor transient IR absorptions. The change in IR transmission at one particular IR frequency was measured after excitation by the Nd:YAG laser. Transient IR spectra were built up on a “point-by-point” basis by repeating this measurement at different IR frequencies. Heating of the high-pressure IR cell was achieved with an external aluminum jacket containing cartridge heaters.

The high-pressure, low-temperature cell used for the Fourier transform IR (FTIR) experiments in liquefied noble gases has been described in detail elsewhere (27). The high-pressure, low-temperature cell was charged with compound 1 and sealed; the cell then was attached to a cryogenic cooling system (Displex, APD Cryogenics, Allentown, PA) and evacuated to remove atmospheric gases and finally pressurized with the noble gas(es). A medium-pressure Hg arc lamp (Philips HPK 125 W, Eindhoven, The Netherlands) equipped with a water filter was used to irradiate the noble gas solution with broadband UV and visible light.

Results and Discussion

This work describes the characterization of an organometallic noble gas complex by both IR and NMR spectroscopy. This work builds on our use of NMR to characterize short-lived organometallic alkane complexes such as Re(Cp)(CO)₂(C₅H₁₀) (28) and TRIR studies (18). Low-temperature FTIR studies suggested that Re(Cp)(CO)₂Xe, produced by photolyzing Re(Cp)(CO)₃ in lXe, should be sufficiently long-lived to allow characterization by NMR. However, this proved particularly problematic because of the limited solubility of Re(Cp)(CO)₃ in lXe at low temperature, resulting in spectra with insufficient signal-to-noise ratios to be of practical use. There is also a lack of NMR handles that would indicate complexation of the Xe in this compound. We therefore designed Re(ⁱPrCp)(CO)₂(PF₃) [ⁱPrCp = η⁵-C₅H₄-CH(CH₃)₂] (1) as a suitable precursor complex because it offers several benefits. First, two excellent NMR-active nuclei, ¹⁹F and ³¹P, are introduced. These nuclei have large chemical-shift ranges, and shifts that are likely to be more sensitive to changes in other ligands at the metal center than cyclopentadienyl protons. It is significant that the PF₃ may allow the coordination of Xe atoms to be confirmed, because there is potential for these two nuclei to couple to the ¹²⁹Xe isotope. Second, both the PF₃ ligand and the isopropyl group on the Cp enhance the solubility in lXe compared to Re(Cp)(CO)₃, allowing acquisition of NMR spectra on a manageable time scale. Finally, it was reasoned that given the closeness of donor and acceptor properties of PF₃ and CO as ligands, the lifetime of the target complex Re(ⁱPrCp)(CO)(PF₃)Xe (3) should be sufficient for NMR characterization.

Monitoring the Photolysis of Re(ⁱPrCp)(CO)₂(PF₃) (1) in Xe Solvent. The photolysis of 1 was initially monitored by using ¹⁹F NMR spectroscopy in lXe at 163 K (Fig. 1). The appearance of the new doublet at δ –5.1 labeled “3” in Fig. 1 stimulated additional studies using NMR and IR, and the doublet was later shown to be caused by Re(ⁱPrCp)(CO)(PF₃)Xe (3) (see below). The appearance of free PF₃ at δ –31.3 suggested that PF₃ may also be photochemically lost from 1, leading to the formation of Re(ⁱPrCp)(CO)₂Xe (2), which obviously cannot be observed directly by ¹⁹F NMR. Because both Re-containing species 2 and 3 can be observed directly by IR spectroscopy, this technique is preferred for the study of the kinetics of formation and reactivity of the accessible Xe complexes in this system, and hence the IR studies are presented first.

Fig. 1.

¹⁹F NMR spectra of Re(ⁱPrCp)(CO)₂(PF₃) (1) in lXe at 163 K. (Lower) Before photolysis. (Upper) Spectrum obtained during prolonged photolysis showing new peaks corresponding to Re(ⁱPrCp)(CO)(PF₃)Xe (3) and free PF₃, respectively.

TRIR Studies at Room Temperature in scXe. We recently demonstrated that the combination of TRIR spectroscopy and supercritical fluids is an effective method of characterizing organometallic noble gas complexes in solution at room temperature (29). Fig. 2a shows the transient IR spectrum obtained 20 μs after irradiation of 1 in scXe [2,300 psi (1 psi = 6.89 kPa)] in the presence of CO (32 psi) at room temperature. It is clear that parent bands are bleached and new transient bands are produced at 1,953, 1,921, and 1,887 cm^–1 (see Table 1).†† The presence of three new ν(CO) bands indicates that at least two transient species have been generated. The bands at 1,953 and 1,887 cm^–1 both decay exponentially at the same rate (k_obs = 5.6 × 10² s^–1), and they can readily be assigned to Re(ⁱPrCp)(CO)₂Xe (2) by comparison with previous TRIR experiments (17). The band at 1,921 cm^–1 is longer-lived and decays at the same rate (k_obs = 1.3 × 10² s^–1), as the parent is partially reformed and can be assigned to Re(ⁱPrCp)(CO)(PF₃)Xe (3). The rate of decay of both transient complexes increases at higher CO concentration (see Fig. 3 b and d), which indicates that most of 2 and 3 react with CO to generate Re(ⁱPrCp)(CO)₃ or reform 1, respectively. In addition, the rate of decay of the intermediates was determined as a function of temperature, and this dependence was used to estimate the activation energy for the reaction of the Xe complexes with CO in scXe§§ (see Fig. 3 a and c Insets). The activation energies obtained for 2 (11.7 ± 0.4 kcal·mol^–1) and 3 (12.4 ± 0.4 kcal·mol^–1) were found to be very close to the value for the previously investigated Re complex Re(Cp)(CO)₂Xe (11.9 ± 0.4 kcal·mol^–1) (18) and greater than in Re(Cp*)(CO)₂Xe (11.1 ± 0.4 kcal·mol^–1) (30).

Fig. 2.

Infrared spectra obtained after photolysis of 1.(a) TRIR spectrum 20 μs after laser flash of a solution of 1 in scXe, doped with CO, at room temperature, showing the bleached absorptions of 1 (negative) and the transient absorptions of 2 and 3 (positive). (b) FTIR spectrum of this solution. (c and d) FTIR spectra in lXe solution at 166 K with no added CO before and after 1, 6, 12, and 36 min of irradiation with broadband UV light. The upper spectra (c) show an expansion of the region 1,960 to 1,870 cm^–1.

Table 1. Wavenumbers of the rhenium complexes.

Complex	Condition	Wavenumber, cm-1	Source
Re(Cp)(CO)3	scXe, 298 K	2,035 and 1,946	Ref. 17
	IXe, 170 K	2,032 and 1,941	Ref. 37
Re(Cp)(CO)2(PF3) (4)	IXe, 166 K	2,001 and 1,940	This work
Re(iPrCp)(CO)2(PF3) (1)	scXe, 298 K	2,000 and 1,941	This work
	IXe, 166 K	1,996 and 1,934	This work
Re(Cp)(CO)2Xe (5)	scXe, 298 K	1,957 and 1,894	Ref. 17
	IXe, 170 K	1,953 and 1,888	Ref. 37
Re(iPrCp)(CO)2Xe (2)	scXe, 298 K	1,953 and 1,887	This work
	IXe, 166 K	1,947 and 1,882	This work
Re(Cp)(CO)(PF3)Xe (6)	IXe, 166 K	1,924	This work
Re(iPrCp)(CO)(PF3)Xe (3)	scXe, 298 K	1,921	This work
	IXe, 166 K	1,920 and 1,916	This work

Fig. 3.

TRIR decay traces of 2 (c and d) and 3 (a and b) recorded after irradiation of 1 in scXe in the presence of CO. (a and c) Layers of normalized traces recorded at different temperatures, with inset plots of k_obs vs. 1/T, [CO] = 8.6 × 10^–2 mol·dm^–3.(b and d) Dependence of the decay of 2 (d) and 3 (b) on [CO] (8.6 × 10^–2 and 2.5 × 10^–1 mol·dm^–3).

FTIR Studies at Low Temperature in lXe. Irradiation of 1 in lXe (1,000 psi) at 166 K generates three bands at 1,947, 1,916, and 1,882 cm^–1, which are assigned to 2 and 3 in lXe (Fig. 2d). Continued photolysis, which was monitored in intervals by FTIR spectroscopy, revealed the presence of a fourth band resulting in a shoulder at 1,920 cm^–1. The two overlapped bands, 1,920/1,916 cm^–1, increase in the course of photolysis, and the ratio between the areas of these bands remained constant (see Fig. 2c). The amount of 2 increased initially but then began to decrease after extended irradiation. After the photolysis was stopped, the reaction was monitored by using FTIR spectroscopy and these spectra showed that the bands of 2 at 1,947 and 1,882 cm^–1 decayed relatively quickly (t_1/2 ≈ 2.2 ± 0.5 h), whereas the bands at 1,920 and 1,916 cm^–1 remained nearly unchanged (t_1/2 ≈ 24 ± 2 h). After 16 h, 2 had decayed completely. Then the temperature was raised to 186 K, which allowed for the accelerated decay of the bands at 1,920 and 1,916 cm^–1 to be monitored. Both bands decayed at the same rate (t_1/2 = 1.1 ± 0.04 h) following a second-order rate law. At the same time, the bleached bands of 1 recovered. During the course of decay, the shape of the overlapped band remains unchanged, which indicates that they either arise from the same species or from two species that are in rapid equilibrium, which may cast a little doubt on the assignment of 3 in scXe (see above). One possibility is that one or both bands are caused by intramolecular interactions of C–H bonds on the isopropyl group with the metal atom. Irradiation of Re(Cp)(CO)₂(PF₃)(4) in lXe at 166 K resulted in depletion of the bands of 4 and the production of Re(Cp)(CO)₂Xe (5) (17) and a second transient at 1,924 cm^–1, which is assigned to Re(Cp)(CO)(PF₃)Xe (6). The ν(CO) band of 6 has no shoulder, which clearly shows that the splitting of the band observed at 1,920/1,916 cm^–1 in the experiments carried out for 1 in lXe is caused by the presence of the isopropyl group. Irradiation of 1 in lKr (1,100 psi) even at a much lower temperature (128 K) resulted only in destruction of bands of 1 and no new transient bands. Photolysis of 1 in lKr doped with Xe (≈33:1 mol/mol at an overall pressure of 350 psi) at 166 K resulted in depletion of the bands of 1 and the production of two bands characteristic for the formation of 2 in the mixed Xe-Kr solvent. Two bands at 1,926 and 1,921 cm^–1 were also observed analogous to the experiment in pure Xe (see Table 1). Interestingly, the half-life of 3 was reduced under those conditions to only 0.62 ± 0.03 h. This result indicates that the extra band observed at 1,920 cm^–1 is not caused by an agostic interaction, because we would expect it to have a similar half-life in the Xe-Kr solvent, but it strongly suggests that these bands are caused by coordination of a Xe atom to the metal center. This result is in agreement with the NMR studies described below, and we assign the bands at 1,920 and 1,916 cm^–1 observed in the lXe experiments to arise from a pair of isomers of 3, which are in rapid equilibrium.

From the combined FTIR and low-temperature IR work, it can be concluded that photolysis of 1 leads to cleavage of CO or PF₃ ligands, producing the Xe complexes 2 and 3, and 3 reacts with CO to reform the parent complex 1. The photochemistry of 1 in lXe is summarized in Scheme 1.

Scheme 1.

NMR Studies at Low Temperature in lXe. When a solution of 1 (≈4 × 10^–3 M) in lXe (natural-abundance isotope ratio) at 163 K is irradiated, a new doublet appears at δ –5.1, ¹J_PF = 1,216 Hz, in the ¹⁹F NMR spectrum with no resolved ¹²⁹Xe satellite splitting and is assigned to 3 (see Fig. 1). Shortly after, another doublet becomes visible at δ –31.3, ¹J_PF = 1,402 Hz, which corresponds to free PF₃. Doublet 3 gradually disappears with time when the irradiation is stopped. Under conditions of continuous photolysis, which typically lasts for 1–2.5 h, the starting material 1 is slowly consumed and an approximately steady-state concentration of doublet 3 is maintained, whereas the concentration of PF₃ steadily increases. The consumption of 1 and the increasing concentration of the PF₃ limits the overall experiment time. These results are consistent with the production of 2 and 3 in the NMR experiment. Performing the same photolysis experiment by using Re(Cp)(CO)₂(PF₃) (4) as precursor resulted in a new species appearing, which we assign as Re(Cp)(CO)(PF₃)Xe (6), with ¹⁹F NMR parameters (δ –5.1, ¹J_PF = 1,218 Hz) that are almost identical to those of species 3 (shown in Fig. 1). The similarity of the ¹⁹F NMR data for 3 and 6 suggests that the isopropyl group does not affect the coordination of the Xe ligand in this case and supports the assertion that the two overlapped bands observed for 3 in the IR spectrum are caused by different rotamers of the isopropyl group rather than an agostic interaction. The lowest-energy conformation of the isopropyl group has the unique hydrogen in the plane of the ⁱPrCp ring, with methyl groups up and down. Complex 3 is chiral, because four different groups are attached to the Re center. Therefore, two different orientations of the isopropyl group relative to the rest of the molecule are possible, each with the unique hydrogen in the plane of the ⁱPrCp ring. These different orientations give rise to two diastereomeric rotamers. It is possible that such diastereomers would be in fast exchange on the NMR time scale but not on the IR time scale. We are currently undertaking detailed modeling studies to address this issue. Monitoring the photolysis of 4 in lXe with ¹H NMR results in a reduction in intensity of the resonance of the Cp moiety in the starting material (δ 5.20) and growth of a new resonance at δ 4.84, which is likely caused by the Cp protons in 6 and probably 5 as well, because the Cp resonances in Re(Cp)(CO)₃ and Re(Cp)(CO)₂(PF₃) (4) differ by only 0.032 ppm and the corresponding Xe complexes 5 and 6 may have similarly close Cp proton resonances that are not separated in our relatively poorly resolved spectra. We note that the amount of PF₃ produced in the NMR experiments cannot be used as a measurement of the amount of 2 or 5 produced, because decomposition of metal-containing species will also lead to liberation of PF₃. Monitoring the photolysis of 1 by using ¹H NMR leads to the observation of a new set of peaks attributable to 3, most notably including two resonances (δ 1.25 and δ 1.14) for the methyl groups of the isopropyl unit, which are diastereotopic, consistent with a chiral Re center in 3 (spectra are shown in supporting information). These methyl resonances were shown to be part of the same isopropyl group by correlation to the same methine proton (δ 2.56) in a COSY NMR experiment. Four different cyclopentadienyl protons are expected for complex 3, with three being observed at δ 5.10, δ 4.83, and δ 4.67, the former having an intensity consistent with the overlap of two resonances.

The identity of 3 was confirmed by using a combination of ¹²⁹Xe labeling and 2D NMR experiments. When 99.95% isotopically enriched ¹²⁹Xe is used, each line of the resonance at δ –5.1 is further split into two with a coupling constant of 5.1 ± 0.8 Hz due to ³J_XeF (Fig. 4). The small magnitude of this coupling explains why ¹²⁹Xe satellites could not be seen in a natural-abundance Xe sample (26% ¹²⁹Xe), because the peak width at half height is typically 6–10 Hz and the signal from molecules containing other isotopes masks the satellites. No extra splitting is observed for the resonance of the PF₃, implying that we are not observing an instrumental artifact giving rise to extra splitting in the resonances of 3.

Fig. 4.

¹⁹F NMR spectra of Re(ⁱPrCp)(CO)₂(PF₃) (1) in liquid Xe at 163 K obtained during prolonged photolysis and expansions of the highlighted region that corresponds to one of the resonances from 3. (d) Using ¹²⁹Xe as solvent. (c) Using unlabeled Xe. (b) An expansion of d using ¹²⁹Xe as solvent. A small splitting caused by ³J_XeF can be seen. (a) An expansion of c.

Fig. 5 shows a 2D ¹⁹F-³¹P heteronuclear single quantum coherence shift-correlated spectrum of ¹²⁹Xe-labeled Re(ⁱPrCp)(CO)(PF₃)Xe (3) recorded at 163 K on the same sample in Fig. 4 b Lower. The expanded region clearly consists of two cross peaks, offset in both dimensions, caused by doublet splitting by the ¹²⁹Xe spin ½ isotope of both ³¹P (²J_XeP = 41.8 ± 1 Hz) and ¹⁹F. Compounds containing phosphorus and Xe have been reported but without any NMR data (31). The magnitudes of the ¹⁹F-¹²⁹Xe and ³¹P-¹²⁹Xe couplings seem to be reasonable for three and two bond interactions, respectively. The size of ³J_XeF is an order of magnitude smaller in comparison with such couplings found in other compounds containing Xe and F (32), but the predictably weak nature of the Re–Xe interaction likely results in inefficient transmission of coupling information in comparison to the molecules containing more conventional covalent bonds to the Xe.

Fig. 5.

2D ¹⁹F-³¹P heteronuclear single quantum coherence shift-correlated spectrum of ¹²⁹Xe-labeled Re(ⁱPrCp)(CO)(PF₃)(¹²⁹Xe) (3) recorded at 163 K. No refocusing or decoupling was used. (Upper) An expansion of the highlighted region.

It is crucial to note that the observation of the couplings to Xe indicates that the Xe ligand is not exchanging rapidly with free Xe. Rapid exchange of bound and free Xe would effectively randomize the ¹²⁹Xe spin state sensed by the ¹⁹F and ³¹P nuclei and no coupling would be observed. Because the small ³J_XeF of ≤5.9 Hz is not coalesced, the upper limit for the free and bound Xe exchange rate is 37 s^–1, and it is likely significantly less than this. The existence of a small coupling between ¹⁹F and ¹²⁹Xe allows for the indirect detection of the ¹²⁹Xe chemical shift with a ¹⁹F-¹²⁹Xe heteronuclear multiple quantum coherence experiment, the result of which is shown in Fig. 6.

Fig. 6.

2D ¹⁹F-¹²⁹Xe heteronuclear multiple quantum coherence shift-correlated spectrum of ¹²⁹Xe-labeled Re(ⁱPrCp)(CO)(PF₃)(¹²⁹Xe) (3) recorded at 163 K.

The Xe chemical shift was determined as δ –6,179 ± 3 at 163 K. It represents a compound in which the Xe nucleus is significantly shielded compared to the free Xe atom. By comparison, our measurements indicate a shift of δ –5,135 for liquid Xe; the shift of gaseous Xe dissolved in nC₆F₁₄ is δ –5,331 at ambient temperature (33) and the shift of an isolated Xe atom is calculated to be at δ –5,460 (34). Hence, Xe in complex 3 is shielded in the region of 719–1,044 ppm compared with an unbound Xe atom, and the known chemical-shift range for ¹²⁹Xe has been increased by ≈800 ppm (or nearly 10%) to ≈8,400 ppm (32). Smaller shifts to high field (up to 50 ppm) have been observed for gaseous Xe trapped in Cu⁺- or Ag⁺-exchanged NaX or NaY zeolites (35), for which it is postulated that the increased shielding is caused by back donation from the d¹⁰ Cu or Ag centers that occurs in transient interactions with the Xe. Theoretical calculations on Cu⁺-Xe and Ag⁺-Xe species predict these upfield shifts and show their source to be a mixing of the filled metal d shell with the 5p, 4p, and 3p orbitals on the Xe (36). A general trend correlating increasing Xe chemical shifts with increasing oxidation number is well known (32), and other compounds in which the Xe is formally in an oxidation state of zero are obviously rare. Noteworthy is the fact that a ¹²⁹Xe chemical shift has been reported for the coordinated Xe atoms in the cationic complex [Au(AsF₃)Xe]⁺ dissolved in HF/SbF₅ (9). In this case, the ¹²⁹Xe chemical shift was found at δ –5,150, which is within 1 ppm of the free Xe present in excess in this sample. It is unclear to us why the Xe chemical shift should change by 800–1,000 ppm on complexation in compound 3 but only by such a small amount in this gold compound, and there is clearly still much to learn about organometallic noble gas complexes.

Conclusions

An organometallic complex of Xe, Re(ⁱPrCp)(CO)(PF₃)Xe (3), has been characterized successfully by NMR spectroscopy. The lifetime of 3 is longer than in all other related complexes such as Re(Cp)(CO)₂Xe and Mn(Cp)(CO)₂Xe. The lifetime of the Xe ligand with respect to exchange with free Xe (more than ≈27 ms) is significant at 163 K. Such lifetimes in solution, in which the frequent molecular collisions are likely to lead to ligand displacement or decomposition, indicate that it is more appropriate to consider the Xe as a discrete ligand in these systems rather than a weakly solvating atom. Key NMR parameters associated with the ¹²⁹Xe nucleus that reflect the nature of the metal–Xe bonding interaction, namely δ(Xe) = –6,179 ± 3 ppm, ²J_XeP = 41.8 ± 1 Hz, and ³J_XeF = 5.1 ± 0.8 Hz, were determined. Methods for calculating NMR parameters such as shifts and coupling constants have advanced significantly recently, and unusual molecules such as 3 present an interesting challenge for the latest theoretical methods. The combination of low-temperature NMR with in situ photolysis is proving to be another valuable tool, complementary to established methods such as TRIR spectroscopy, in studying short-lived complexes such as those of Xe and alkanes.