Formation of unprecedented actinidecarbon triple bonds in uranium methylidyne molecules

Abstract

Chemistry of the actinide elements represents a challenging yet vital scientific frontier. Development of actinide chemistry requires fundamental understanding of the relative roles of actinide valence-region orbitals and the nature of their chemical bonding. We report here an experimental and theoretical investigation of the uranium methylidyne molecules X₃UCH (X = F, Cl, Br), F₂ClUCH, and F₃UCF formed through reactions of laser-ablated uranium atoms and trihalomethanes or carbon tetrafluoride in excess argon. By using matrix infrared spectroscopy and relativistic quantum chemistry calculations, we have shown that these actinide complexes possess relatively strong UC triple bonds between the U 6d-5f hybrid orbitals and carbon 2s-2p orbitals. Electron-withdrawing ligands are critical in stabilizing the U(VI) oxidation state and sustaining the formation of uranium multiple bonds. These unique UC-bearing molecules are examples of the long-sought actinide-alkylidynes. This discovery opens the door to the rational synthesis of triple-bonded actinidecarbon compounds.

Chemical bonding and bond order are among the most important fundamental concepts in modern chemistry since the birth of the valence theory of Lewis (1). Main-group and transition-metal compounds with multiple chemical bonds have always been fascinating to chemists because of their pivotal role in organic, inorganic, and organometallic chemistry, characteristic chemical and physical properties, and versatile applications in biological and material science (2–6). Whereas numerous organic and inorganic compounds with multiple bonds are known (7–17), f elements (lanthanides and actinides) with multiple bonds are relatively rare, except for early actinides. Such bonding has aroused great interest recently in the search for actinide complexes with multiple bonds between two actinide metals (18–21) and between actinide (An) and main-group ligands (L) (22–28).

Among the actinide complexes with AnL multiple bonds, the importance of first-row elements to bond to actinide metal centers has been highlighted by Burns (22), and molecular complexes containing metal-nitride units have been prepared recently (24–26). Uranium as the leading example forms a plethora of UO bonds and a handful of UNR and UCR₂ (R = organic groups) bonds. Considerable interest has been developed in recent years in actinide complexes with AnL double bonds, and most of these investigations have centered on organometallic systems. Examples include the compounds above with NUN linkages (26) and organoimido (AnNR) and phosphinidene (AnPR) groups (29, 30). The matrix isolation technique has revealed several inorganic uranium compounds with covalent triple bonds, including NUN, CUO, and NUO⁺ cation (31–33), which are isoelectronic with the ubiquitous uranyl dication. However, AnL multiple bonds are usually formed between hard-acidic, high-valent actinides and hard Lewis bases, particularly F⁻, O²⁻, and NR²⁻ (34, 35), and no actinide alkylidyne complexes with AnCR triple bonds are known so far. Because of the high orbital energies of carbon, L_nAnCR type of carbyne compounds is not expected to be highly stable. Well designed ligands that can stabilize the actinide center at their stable oxidation states are needed to accommodate the actinide carbon multiple bonds.

High-oxidation state transition metal alkylidene and alkylidyne complexes have received increasing attention over the past three decades owing to their importance as catalysts in a variety of synthetic organometallic processes (36). Recently, we have prepared simple methylidene and methylidyne molecules through the reaction of laser-ablated early transition metal atoms with methane or methyl halides (37). These studies were extended to the accessible actinide metal atoms Th and U for the preparation of the first actinide methylidene species HXAnCH₂ (X = H, F, Cl, Br) (38–41). Although Mo and W reactions also formed the analogous H₂XMCH methylidynes (42–46), the H₂XUCH counterparts were energetically too high to be produced in these experiments (40). However, very recent investigations with the heavy metals Zr, Hf, and Re in trihalomethane reactions have demonstrated that the highly exothermic driving force for halogen transfer from carbon to heavy metal fosters the formation of the low-energy, very stable trihalo metal carbynes (47, 48).

We report here an integrated experimental and theoretical study of the actinide-methylidyne species, namely F₃UCH, Cl₃UCH, Br₃UCH, F₂ClUCH, and F₃UCF, which render the long-sought UC triple bonds in methylidyne compounds. Detailed bonding analysis based on a variety of relativistic quantum chemistry calculations indicates that the UC triple bonds are composed of one (df-sp)σ bond and two (df-p)π bonds. Interestingly the UC bond length and bond strength are tunable by changing the electronegativity of the neighboring atoms attached to the UC bond, making these complexes attractive reactive intermediates. This discovery of the AnC triple-bonded complexes provides important insights and is expected to assist in the rational synthesis of these compounds in larger quantities, which might be stabilized by electron-withdrawing substituents and steric protection of the UC bonds with bulky ligands.

Results and Discussion

Reactions of laser-ablated U atoms were done with fluoroform, chloroform, bromoform, carbon tetrafluoride, and several isotopic modifications (CHX₃, ¹³CHX₃, CDX₃ [X = F, Cl, Br], and CF₄) as precursors in excess argon (0.5%, 1%, or 2% concentrations) during condensation onto an 8 K cesium iodide window, as described (49, 50). Depleted high-purity uranium metal targets (ORNL) were filed to remove surface contaminants. Common bands in uranium experiments with different precursors are limited to oxides and nitrides (38–41, 49). Infrared spectra were recorded on a Nicolet 550 spectrometer after sample deposition, after annealing, and after irradiation by using a 175 W mercury arc street lamp. Infrared spectra from reactions of fluoroform or carbon tetrafluoride reagents with uranium are shown in Fig. 1, and the results from reactions with chloroform and bromoform are detailed in Fig. 2.

Fig. 1.

Infrared spectra in the 590–490 cm⁻¹ region for laser-ablated U atoms codeposited with fluoromethanes in excess argon at 8 K. U and 1% CHF₃ in argon codeposited for 1 h (a), after >290 nm irradiation (b), and after >220 nm irradiation (c). U and 1% CDF₃ in argon codeposited for 1 h (d), after >290 nm irradiation (e), and after >220 nm irradiation (f). U and 1% CF₄ in argon codeposited for 1 h (g), after >290 nm irradiation (h), and after >220 nm irradiation (i). Precursor absorptions are labeled P.

Fig. 2.

Infrared spectra in the 550–410 cm⁻¹ region for laser-ablated U atoms codeposited with chloroform in excess argon at 8 K. U and 0.5% CHCl₃ in argon codeposited for 1 h (a), after λ > 290 nm irradiation (b), after λ > 220 nm irradiation (c), and after annealing to 30 K (d). U and 0.5% ¹³CHCl₃ in argon codeposited for 1 h (e), and after λ > 220 nm irradiation (f). U and 0.5% CDCl₃ in argon codeposited for 1 h (g), and after λ > 220 nm irradiation (h). U and 2% CHBr₃ in argon codeposited for 1 h (i), after λ > 290 nm irradiation (j), after λ > 220 nm irradiation (k), and after annealing to 30 K (l).

The geometries, vibrational frequencies, and electronic structures of the potential uranium product complexes were calculated by using relativistic density functional theory (DFT) with the generalized gradient PW91 approach (51) as implemented in ADF 2006.1 (52). Inasmuch as the 6s and 6p semicore orbitals are important for actinide bonding, they are included explicitly in the variational space along with the 5f, 6d, 7s, and 7p valence orbitals, whereas frozen-core approximation was applied to the U [1s²-5d¹⁰] atomic core. Slater basis sets with the quality of triple-zeta plus two polarization functions (TZ2P) were used. The zero-order regular approximation was used to account for the relativistic effects (53). We also performed ab initio calculations at the level of coupled-cluster with single, double, and perturbative triple excitations [CCSD(T)] (54) on X₃UCH (X = H, F) with use of the Stuttgart quasi-relativistic pseudopotential and valence basis set for U and 6-31+G* basis sets for C, F, and H (55, 56). The optimized CCSD(T) UC distances (1.926 Å in H₃UCH) lie in the same range as those from DFT calculations, indicating that the later are applicable for evaluating these close-shell triple-bond actinide systems.

In the fluoroform spectra stable binary uranium fluorides give rise to very weak absorptions at 496 and 584 cm⁻¹ for UF₃ and UF₅, respectively (57), which shows that uranium abstracts fluorine from the precursor molecule. Three new bands marked with arrows in Fig. 1a are observed at 576.2, 540.2, and 527.5 cm⁻¹ in the infrared spectrum recorded after the initial reaction of U and CHF₃. These bands increase by 30% on UV irradiation (λ > 290 nm) and another 20% on further UV irradiation (λ > 220 nm). A more dilute ¹³CHF₃ sample gave only the most intense absorption shifted to 539.2 cm⁻¹, which demonstrates clearly that carbon is involved in the product species. The reaction with CDF₃ shown in Fig. 1d gives the same upper band, with the strong band shifting to 535.9 cm⁻¹ and the lower band shifting to the low 400 cm⁻¹ poor-signal region. Because CF₄ is less reactive, on reagent codeposition (Fig. 1g) weaker bands are detected at 578.7 and 536.4 cm⁻¹, which increase slightly during the first irradiation (λ > 290 nm), but undergo major growth in the second irradiation (λ >220 nm) (Fig. 1i). These bands are stable on annealing the sample to 30 K, and they increase slightly on further UV irradiation.

The new absorptions in the UF stretching region are due to a reaction product trapped in solid argon, which is a uranium species other than binary UF_x (x = 1–6) (57). After our previous work on the titanium reaction with CF₄, which produced the electron-deficient methylidyne complex F₃TiCF (58), and analogous work with heavy metal atoms (47, 48), which produced the F₃MCH molecules, we consider these absorptions for assignment to the F₃UCH and F₃UCF uranium-bearing methylidyne compounds. Relativistic quantum chemistry calculations have been performed on various structural isomers, isotopomers, and different electronic states of these molecules to determine the thermodynamic stability and to validate the experimental assignments. The optimized geometries of X₃UCH (X = F, Cl, Br) and F₃UCF by using the PW91 approach for the singlet ground-state molecules are depicted in Fig. 3, and their vibrational frequencies and IR intensities are listed in Tables 1 and 2.

Fig. 3.

Optimized molecular structures of F₃UCH (a), Cl₃UCH (b), Br₃UCH (c), and F₃UCF (d) by using PW91. The UC bond lengths are in angstroms.

Table 1.

Observed and calculated fundamental vibrational frequencies for the C_3v F₃U≡CX (X = H, D, F) molecules^*

Mode description	F3U≡CH		F3U≡13CH		F3U≡CD		F3U≡CF
	Obs.	Calc.	Obs.	Calc.	Obs.	Calc.	Obs.	Calc.
C-X str, a1		2,979 (2)		2,969 (2)	—‡	2,200 (1)	—‡	1,268 (312)
U≡CX str, a1	—‡	747 (46)	—†‡	721 (42)	—‡	717 (41)	— §	441 (34)
UF sym str, a1	576.2	585 (122)	—†	585 (123)	576.2	586 (123)	578.7	589 (118)
UF antisym str, e	540.2	561 (284)	539.2	559 (280)	535.9	541 (207)	536.4	544 (177)
U≡C-X bend, e	527.5	508 (34)	—†	506 (24)	— §	412 (49)	—	311 (28)

*Vibrational frequencies (cm⁻¹) and intensities (km/mol, in parentheses) are calculated by using PW91/TZ2P (see text). Three real lower-frequency bending modes (a₁, e, e) are not listed. Absorptions are observed in argon matrix.

^†Sample too dilute to observe weaker bands.

^‡Absorption masked by very intense precursor band.

^§Region noisy because of low detector response at the instrumental limit.

Table 2.

Observed and calculated fundamental vibrational frequencies for the C_3v X₃U≡CH (X = Cl, Br) molecules^*

Mode description	Cl3U≡CH		Cl3U≡13CH		Cl3U≡CD		Br3U≡CH
	Obs.†	Calc.	Obs.	Calc.	Obs.	Calc.	Obs.	Calc.
CH str, a1		3,001 (2)		2,991 (2)		2,219 (7)		3,005 (3)
U≡CX str, a1	—‡	770 (69)	—‡	744 (65)	—‡	738 (64)	—‡	777 (70)
U≡C-H bend, e	527.2	522 (224)	522.8	518 (218)	415.9	410 (216)	527.6	527 (178)
UX sym str, a1		339 (29)		339 (29)		339 (29)		225 (84)§
UX antisym str, e		329 (140)	—	329 (140)	—	326 (100)		216 (10)§

*Vibrational frequencies (cm⁻¹) and intensities (km/mol, in parentheses) are calculated by using PW91/TZ2P (see text). Three real lower-frequency bending modes (a₁, e, e) are not listed.

^†Absorptions are observed in argon matrix.

^‡Absorption masked by very intense precursor band.

^§Mode symmetries reversed.

The upper bands in Fig. 1 are assigned to the symmetric UF stretching mode, which shows very little change between these two F₃UCR (R = H, F) molecules. The strongest absorption is due to the very strong degenerate UF stretching mode, which shifts slightly from F₃UCH to F₃U¹³CH, to F₃UCD and to F₃UCF. Finally, the weak 527.5 cm⁻¹ band is assigned to the degenerate HCU bending mode, which is weak in this molecule probably, in part, because of intensity gained by the nearby degenerate UF stretching mode. In the deuteriated species, this mode shifts into a region of low signal near our limit of detection.

This assignment is confirmed in chloroform and bromoform experiments, where the HCU bending mode gains substantial infrared intensity and shifts slightly to 527.2 and to 527.6 cm⁻¹, respectively, and the UX stretching frequencies are much lower because of the heavier halogen mass. The lack of a significant difference between the chloroform and bromoform reaction product absorptions demonstrates clearly that the halogen atoms in the product structure are farther separated from the hydrogen atom than in the precursor haloform molecules. Notice in Fig. 2 that the strong new band with chloroform at 527.2 cm⁻¹ increases on UV irradiation, and shifts to 522.8 cm⁻¹ with ¹³CHCl₃ and to 415.9 cm⁻¹ with the CDCl₃ precursor. The PW91 calculation of frequencies (Table 2) for isotopic Cl₃UCH molecules reveals one strong absorption at 522 cm⁻¹, which exhibits almost exactly the same isotopic shifts as the observed new product absorption and defines the bending mode. We note that the frequency of the HCU bending mode increases as the CU bond becomes stronger and shorter in the X₃UCH series, and the reaction of excited U atom with chloroform and bromoform is even more favorable than the reaction with fluoroform.

The CH stretching modes are too weak to be observed here, and the CF stretching mode is masked by very intense CF₄ precursor absorption. Finally, the UCH stretching mode is unfortunately in the region of very strong haloform precursor bands, and the UCF stretching mode is predicted at the low end of our region of detection where the signal-to-noise is diminished. However, in the molecule of lower symmetry, F₂ClUCH, prepared from the reaction with Freon 22, the UCH stretching mode is found as a weak band at 671.3 cm⁻¹, which is shifted to 648.7 cm⁻¹ by using ¹³CHF₂Cl, confirming the vibrational assignment of UCH stretching mode. Note that the large 22.6 cm⁻¹ carbon-13 shift for the UCH stretching mode agrees well with theoretical calculations; our PW91 calculation predicts the UCH mode in F₂ClUCH at 682.4 cm⁻¹ with an isotopic shift of 23.6 cm⁻¹.

Our calculations also indicate that the X₃UCH (X = F, Cl, Br) and F₃UCF molecules prefer a singlet ground state, with the triplet and quintet states being much higher in energy. The methylidynes are indeed more stable than their methylidene and methyl uranium halide counterparts, and the predicted C_3v molecular symmetry, vibrational frequencies, infrared absorption intensities, and isotopic frequencies are all in excellent agreement with the experimental observations, as shown in Tables 1 and 2. This agreement confirms our vibrational assignments and the identification of uranium methylidyne molecules.

Additional chemical support for this exothermic α-halogen transfer reaction of the uranium systems is found in our subsequent investigations with Re, Mo, and W atoms (48, 59). These metals react with haloforms to give analogous X₃MCH molecules with C_3v symmetry for M = Mo and W. In the case of Re, Jahn–Teller distortion lowers the symmetry to C_s, which increases the IR intensity of the CH stretching absorption due to polarization of the triple bond, and high CH stretching frequencies characteristic of sp hybridization are also observed. Like U the heavy group 6 metals Mo and W also support hexavalency, leading to the formation of the symmetric X₃MCH molecules. In addition to the MX₃ stretching and HCM deformation fundamentals, high CH and MC stretching frequencies are observed to further characterize these methylidyne molecules (59).

To understand the relative roles of the U 6p, 5f, and 6d orbitals and their chemical bonding, a series of theoretical analyses have been performed. Table 3 lists the natural charges, natural localized molecular orbitals (NLMOs), natural hybrid orbitals, and the bond orders calculated for X₃UCH and F₃UCF molecules. The orbital contributions to the σ- and π-orbitals in X₃UCH are listed in Table 4 for comparison with those of the formal triple bonds in uranyl. The natural population analysis (60) reveals that U carries highly positive charges in these U(VI) complexes, which helps to form stable molecules by both covalent and ionic interactions. The DFT calculations predict that the UC distances in H₃UCH (a model compound), Br₃UCH, Cl₃UCH, F₃UCH, and F₃UCF are 1.901, 1.905, 1.910, 1.941, and 2.007 Å, respectively. These UC distances are slightly longer than the sum of the U and C triple-bond radii (17), but correlate well with what is expected from the experimentally measured distances for UC (2.29 Å) and UC (2.54 Å) bonds (61). The UC triple bonds in the X₃UCR complexes are further established by theoretical calculations from different methodologies. As shown in Fig. 4, the canonical molecular orbitals from DFT calculations also reveal one σ- and two π-orbitals with six electrons, consistent with orthodox triple-bond models (1, 2).

Table 3.

Natural charges (q_N), natural localized molecular orbital (NLMO) compositions, natural hybrid orbitals, and bond U-C orders

	qU	qC	NLMO	BOW	BOM	BOGJ
H3U≡CH	2.45	−1.17	22% U(s0.05p0.05d1.03f) + 78% C(sp1.06)	2.51	2.47	3.06
			39% U(d0.34f) + 61% C(p)
F3U≡CH	2.87	−0.95	18% U(s0.10p0.05d0.96f) + 82% C(sp0.97)	2.48	2.38	2.88
			47% U(d0.19f) + 53% C(p)
Cl3U≡CH	2.42	−0.91	20% U(s0.09p0.03d1.05f) + 80% C(sp0.96)	2.53	2.40	2.90
			48% U(d0.19f) + 52% C(p)
Br3U≡CH	2.35	−0.93	21% U(s0.09p0.03d1.07f) + 79% C(sp0.97)	2.53	2.39	2.90
			47% U(d0.21f) + 53% C(p)
F3U≡CF	2.78	−0.33	11% U(s0.45p0.03d1.27f) + 89% C(sp0.45)	2.14	2.18	2.57
			51% U(d0.19f) + 49% C(p)

Only one of the two π-orbitals is listed as they are equivalent. BO_W, natural Wiberg bond order; BO_M, Mayer bond order; BO_GJ, Gophinatan–Jug bond order.

Table 4.

PW91 MO percentages (%) of U valence and semicore orbitals in uranyl and X₃UCH (X = H, F) molecules

	U							C/O
	MO	6 s	6p	5f	6d	7 s	7p	2 s	2p
UO22+	σu		8	57			2		33
	σg				14	2		7	77
	πu		1	34					64
	πg				19				79
H3U≡CH	π			43	7				38
	σ		1	15	5	11		7	21
F3U≡CH	π			39	9				46
	σ		6	12	13	2		12	26

*Orbital contribution <1% is not shown.

Fig. 4.

Comparison of the σ- and π-molecular orbitals of ethyne HCCH and the uranium-methylidyne F₃UCH and F₃UCF complexes. (Isosurface value = 0.05 atomic unit.)

The effective bond orders calculated by using different formalisms are also listed in Table 3 which agree well with triple-bond covalency. A further support of the UC triple bonds in these complexes are obtained from calculations of the characteristic topology of the electron localization function (ELF), which measures the spatial distribution of paired electrons and exhibits maxima at the most probable positions of localized electron pairs (62, 63). Fig. 5 illustrates the ELF isosurfaces of HCCH, F₃UCH, Cl₃UCH, Br₃UCH, and F₃UCF. The polarized ring attractors between U and C indicate that all of these actinide molecules possess triple bonds as is the case for acetylene. In addition, the calculated NLMOs (60) reveal the formation of the UC triple bonds by one σ-bond between the U dfσ and C spσ hybrid orbitals and two π-bonds between the U dfπ and C pπ orbitals, thus providing unequivocal support to the formation of triple bonds in these methylidyne molecules. As shown by the NLMO results and the orbital contributions listed in Tables 3 and 4, the threefold symmetry of the X₃UCR molecules allows significant hybridization or orbital mixing between U 6d and 5f orbitals, which enhances the bonding strengths of the UC triple bonds. Table 4 also reveals that the σ-bond of F₃UCH molecule involves nonnegligible U 6p hybridization with 5f/6d orbitals, similar to the “pushing from below” 6p-5f mixing in uranyl (64).

Fig. 5.

Isosurfaces of the electron localization function (ELF) calculated at value 0.8 for HCCH, F₃UCH, Cl₃UCH, Br₃UCH, and F₃UCF (from left to right). The ring attractors in all of these molecules clearly show the triple bonds.

Our calculations on the series X₃UCR (X, R = H, F, Cl, Br, I, CH₃, SiH₃, CO) also reveal that the UC bond distances and bond strengths depend on the ligands and the coordination pyramidality around the U and C atoms. In particular, the strong inductive effect of electronegative fluorine supports a large positive charge on the central U atom, which helps to retain the most stable U(VI) oxidation state and to push the U 7s orbital to high energy to facilitate strong bonding between U and CR. For the halogen substitutions on U, the less electronegative ligands decrease the UC bond length by increasing the effective overlap between U 6d and C 2p orbitals. Hence, among the molecules with halogen substituents, the triple bond in I₃UCH is the strongest and it has the shortest UC distance (1.895 Å), close to the calculated distance in (SiH₃)₃UCH (1.893 Å). The calculations on F₃UCR (R = F, Cl, Br, I, H, Me) also demonstrate that the UC bonds are shortest for less electronegative R such as H atom.

The experimental results suggest that laser-energy-excited atomic uranium is a vigorous reducing agent. The reactions occur as U cleaves the strong XC (X = F, Cl, Br) bond and then two successive α-halogen transfers to uranium occur, forming the U(VI) products identified in this work. Consistent with this notion, theoretical calculations show that halogen transfer from C to U atoms is strongly favored thermodynamically, with each step being strongly exothermic because of the difference in XU and XC bond energies. The overall reaction (1) requires initiation by using electronically excited uranium (noted U*) from laser ablation or UV irradiation, and the reaction with fluoroform is calculated to be exothermic by 183 kcal/mol with approximate treatment of the spin-orbit-coupling effects. The solid argon matrix soaks up this excess energy and stabilizes the final methylidyne product.

Through a combined experimental and theoretical effort, we have hereby shown that laser-ablated uranium atoms are quite reactive, and even the unreactive CF₄ molecule falls prey and yields fluorine atoms to the active uranium metal center, leading to the formation of a relatively strong UC triple bond. Although the activation of C-halogen bonds itself is an important research field based on the need to dispose of environmentally hazardous halocarbons, the most significant aspect of this discovery is that X₃UCH (X = F, Cl, Br), F₂ClUCH, and F₃UCF are examples of the long-sought actinide-alkylidyne molecules that have been prepared and characterized. Although the chemistry of these species prepared and isolated in a low-temperature noble-gas matrix is difficult to investigate experimentally, these unique diamagnetic X₃UCR molecules might be synthesized by coordination of U with highly electron-withdrawing ligands and steric protection of the UC bonds with bulky ligands, possibly with pseudo-threefold local symmetry. The ability of fluorine-substituted or sterically encumbering ligands (25) to generate new types of metal–ligand multiple bonds might be extended to the still-elusive actinidecarbon multiple-bond alkylidyne and arylidyne systems. The discovery of these actinide complexes with AnC triple bonds is potentially important for organosynthesis, catalysis, interstellar chemistry, atmospheric chemistry, and actinide chemistry under extreme conditions.

Extending the well known methylidyne chemistry of the transition metals (36) to actinide metals has proven to be a difficult challenge. Although this may simply be because actinide 6d and 5f valence orbitals do not behave like transition-metal nd orbitals (65), we indeed see glimpses of transition-metal-like behavior in the early actinides. This and our earlier research work on Th and U methylidenes (38–41) provide clues to the existence and nature of actinidecarbon, multiple-bonded species.