Isolation of C1 through C4 derivatives from CO using heteroleptic uranium(iii) metallocene aryloxide complexes

The conversion of C1 feedstock molecules such as CO into commodity chemicals is a desirable, but challenging, endeavour. When the U(iii) complex, [(C5Me5)2U(O-2,6-tBu2-4-MeC6H2)], is exposed to 1 atm of CO, only coordination is observed by IR spectroscopy as well as X-ray crystallography, unveiling a rare structurally characterized f element carbonyl. However, using [(C5Me5)2(MesO)U (THF)], Mes = 2,4,6-Me3C6H2, reaction with CO forms the bridging ethynediolate species, [{(C5Me5)2(MesO)U}2(μ2-OCCO)]. While ethynediolate complexes are known, their reactivity has not been reported in much detail to afford further functionalization. For example, addition of more CO to the ethynediolate complex with heating forms a ketene carboxylate, [{(C5Me5)2(MesO)U}2(μ2:κ2:η1-C3O3)], which can be further reacted with CO2 to yield a ketene dicarboxylate complex, [{(C5Me5)2(MesO)U}2(μ2:κ2:κ2-C4O5)]. Since the ethynediolate showed reactivity with more CO, we explored its reactivity further. A [2 + 2] cycloaddition is observed with diphenylketene to yield [{(C5Me5)2U}2(OC(CPh2)C( 
<svg xmlns="http://www.w3.org/2000/svg" version="1.0" width="13.200000pt" height="16.000000pt" viewBox="0 0 13.200000 16.000000" preserveAspectRatio="xMidYMid meet"><metadata>
Created by potrace 1.16, written by Peter Selinger 2001-2019
</metadata><g transform="translate(1.000000,15.000000) scale(0.017500,-0.017500)" fill="currentColor" stroke="none"><path d="M0 440 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z M0 280 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z"/></g></svg>
 O)CO)] with concomitant formation of [(C5Me5)2U(OMes)2]. Surprisingly, reaction with SO2 shows rare S–O bond cleavage to yield the unusual [(O2CC(O)(SO)]2− bridging ligand between two U(iv) centres. All complexes have been characterized using spectroscopic and structural methods, and the reaction of the ethynediolate with CO to form the ketene carboxylate has been investigated computationally as well as the reaction with SO2.


Introduction
Due to the problems associated with our currently used hydrocarbon resources, there is great interest in the conversion of C1 feedstock molecules such as CO and CO 2 into larger liquid hydrocarbons and other desirable commodity chemicals. This is well established using the Fischer-Tropsch process which converts syngas mixtures (H 2 /CO) into chain hydrocarbons using heterogeneous transition metal catalysts. However, Fischer-Tropsch chemistry remains underdeveloped with respect to homogeneous catalysis, 1-3 and stoichiometric reactions involving metal complexes and CO can provide insight into CO functionalization, 4,5 especially the coupling of CO molecules, by uncovering novel reactivity and moieties.
One method of functionalizing CO is through homologation. 6 The reductive coupling of CO by molecular metal complexes has received attention in the development of strategies for utilizing C1 feedstock molecules for providing insight into heterogeneous reactions. 7,8 Homologation of CO is difficult due to the high bond dissociation energy of the CO triple bond, rendering the molecule relatively inert compared to other small molecules. In addition, CO is the quintessential coordinating ligand for low-valent, electronrich transition metal complexes, which are typically more susceptible to undergoing redox chemistry. Actinides are large, electropositive metals which do not possess the ability to back bond to CO in a similar manner to transition metals 9 and uranium can span oxidation states of +1 to +6, giving a rich redox chemistry. [10][11][12] The quest for CO reductive coupling reactions began in the 1800s with Liebig and Gmelin reporting that CO reacts with molten potassium to produce (C 5 O 5 ) 2− and (C 6 O 6 ) 2− dianions. 13,14 Since then, other s-block [15][16][17][18][19][20][21][22] as well as p-block [23][24][25][26][27][28][29] complexes have been used for CO reduction over the years. There are only a handful of reports of CO reductive coupling with lanthanides. The Evans group reported that the reaction of [(C 5 Me 5 ) 2 -Sm(THF) 2 ] 30 or [{(C 5 Me 5 ) 2 La} 2 (m 2 :h 2 -N 2 )] 31 with CO both produce a similar ketenecarboxylate product. In addition, Evans and coworkers have also observed CO radical, ethynediolate, or enediolate formation 32,33 with Y(II) or treatment of trivalent lanthanide with potassium graphite in their LnZ 3 /K reactivity. While carrying out this study, Nocton and co-workers demonstrated CO homologation with a thulium(II) complex, [(1,2,4-t Bu 3 C 5 H 2 ) 2 Tm], forming an ethynediolate which can then be further functionalized into a ketenecarboxylate with additional CO, followed by insertion of CO 2 into the Tm-C bond to yield a ketenedicarboxylate. 34 The reactivity of CO with uranium complexes is well established, but mainly with U(IV) complexes in which migratory insertion occurs into uranium-element bonds. [35][36][37][38][39][40][41][42][43][44][45][46][47][48][49][50][51] These include examples in which CO inserts into a uranium-element bond, followed by C-C bond coupling with another CO molecule. The reductive coupling of CO has also been beautifully demonstrated by Cloke and co-workers by varying the steric properties of substituted-cyclooctatetrienyl, [C 8 H 6 (SiR 3 ) 2 )] 2− , R = i Pr, Me, cyclopentadienyl, (C 5 Me 4 R) 1− , R = Me, H, uranium(III) complexes to obtain different reductively coupled CO products. [52][53][54][55][56][57][58] The Arnold group was the next to observe the ethynediolate moiety with the reaction of [{(Me 3 Si) 2 N} 3 U] with CO. Heating the ethynediolate led to intramolecular C-H bond activation across the alkyne, forming a seven-membered metallocycle containing an enediolate. 59 Additionally, Arnold reported a homoleptic U(III) aryloxide complex, U(O-2,6-t Bu 2 C 6 H 3 ) 3 , 60 which upon reaction with CO also formed an ethynediolate. 61 The Liddle group isolated an ethynediolate through the reaction of U(III) ligated with the sterically encumbering Tren DMSB ligand, [U(Tren DMSB )], Tren DMSB = N(CH 2 CH 2 NSiMe 2 t Bu) 3 . 62 Thermolysis of the ethynediolate at 80°C resulted in the insertion of the ethynediolate into one of the N-Si bonds of the Tren DMSB ligands with concomitant protonation of the ethynediolate and formation of an oxo-bridge between two uranium(IV) centers. The Meyer group has also observed a unique CO bridged complex, [{(L)U} 2 (m 2 -CO)], L = 1,4,7-tris(3,5-di-tertbutyl-2-hydroxybenzylate-1,4,7-triazacyclononane. 63 To our knowledge, the only report of further functionalization of a uranium ethynediolate complex, through addition of a substrate, is a structure in the Cambridge Crystallographic Database Centre, [{(2,6-t Bu 2 C 6 H 3 O) 3 U} 2 (m 2 -OC(H)]C(BC 8 H 15 )O)], by Mansell and Arnold in which hydroboration occurs across the CC triple bond. 64 Carbon monoxide functionalization with uranium nitrides has also been reported. [65][66][67] Herein, we describe the use of two metallocene aryloxide uranium(III) complexes, [(C 5 Me 5 ) 2 (2,6-t Bu 2 -4-MeC 6 H 2 -O)U], 2,6-t Bu 2 -4-MeC 6 H 2 -O = BHT, and [(C 5 Me 5 ) 2 (2,4,6-Me 3 C 6 H 2 -O)U(THF)], 2,4,6-Me 3 C 6 H 2 -O = Mes, and their reactivity with CO. In the former, BHT derivative, only coordination takes place as observed by IR spectroscopy and structural determination by X-ray crystallography. This is only the third structure of an f element carbonyl. However, upon reductive coupling of CO to form the ethynediolate, [{(C 5 Me 5 ) 2 (MesO)U} 2 {m 2 -(OCCO)}], a C 2 moiety is found with the mesityl substituted aryloxide. Analogous to Nocton's recent ndings, the ethynediolate can react with CO at elevated temperatures to yield the C 3 ketenecarboxylate, which can react with CO 2 to yield a C 4 ketenedicarboxylate. In addition, reactions with diphenylketene and SO 2 (DABSO) were investigated which showed [2 + 2] cycloaddition reactivity.  69 Treatment of 1 atm of CO to 1 in hexamethyldisiloxane has a colour change from dark green to brown, eq 1. The liquid IR spectrum of the reaction showed a strong absorption at 1893 cm −1 indicating signicant backbonding to the CO p* orbitals. 9 With 13 CO, a stretching frequency of 1853 cm −1 is observed, consistent with the reduced mass of CO versus 13 CO. Therefore, we formulated the new compound as [(C 5 Me 5 ) 2 (2,6-t Bu 2 -4-MeC 6 H 2 -O)U(CO)], 3.

Results and discussion
(1) The 1 H NMR spectrum of 3 consists of resonances at −3.75 ppm and 5.57 ppm for the tert-butyl and methyl groups, respectively, as well as the (C 5 Me 5 ) 1− resonance at −5.61 ppm. When tabulating all the uranium carbonyl complexes reported, Table 1, we noticed that 3 has one of the lowest stretching frequencies reported and falls between the two structurally characterized complexes, [(C 5 Me 5 ) 3 U(CO)], 3a, at 1922 cm −1 and [(C 5 Me 4 H) 3 U(CO)], 3b, at 1880 cm −1 . Gratifyingly, brown crystals suitable for X-ray crystallographic analysis were grown from the reaction mixture, Fig. 1.
Complex 3 is stable in solution for several hours. Monitoring of the 1 H NMR spectrum showed 60% conversion of 3 to 1 over the course of 15 hours. In addition, 3 is also stable in the solidstate with a stretching frequency of 1904 cm −1 (KBr). Under reduced pressure, complex 3 converts to 1 in ∼20 minutes.
Complex 3 is only the third structurally characterized carbonyl complex with an f element, Fig. 1. Complex 3 adopts a distorted tetrahedral geometry with a centroid-U-centroid angle of 132.60°, centroid-U-CO angles of 91.27 and 90.62°, and centroid-U-O(aryloxide) angles of 113.78 and 112.96°. These metrics are similar to 1, and for other U(III) metallocene complexes. 75 The O1-U1-C36 bond angle is 96.97(11)°and 3 has a 172.7(2)°U1-O1-C(ipso) angle. The U-O(aryloxide) bond length of 2.166(2) Å is shorter than in other U(III) metallocene aryloxide complexes. For example, the U-O(aryloxide) bond  3 . 61 Commensurate with the amount of backbonding observed from the IR stretching frequencies, the U-C(CO) bond distance in 3 is 2.394(5) Å which is slightly longer than the 2.383(6) Å in 3b and much shorter than the 2.485(9) Å in 3a. Interestingly, the C-O bond distance is 1.363(4) Å which is substantially longer than that of free CO (1.128 Å), 70 and longer than those observed in 3a and 3b of 1.13(1) and 1.142(7) Å, respectively. Finally, the U-C-O bond angle is 177.8(4)°c ompared to 180°and 175.2(6)°in 3a and 3b, respectively. The electronic structure of 3 was probed using DFT calculations (B3PW91). Different spin states were considered (doublet, quartet, and sextet) and the geometry was optimized in all cases. The quartet, in line with a U(III) centre, is found to be the ground state with the doublet 13.8 kcal mol −1 higher in energy and nally the sextet 55.5 kcal mol −1 above the ground state, Table S3. † This is consistent with the UV-vis-nIR spectrum (Fig. S3 †) which shows similar features to 1. 76 The optimized geometry compares well with the experimental one with a maximum deviation of 0.03 Å on the U-C CO distance. The CO bond appears to be slightly elongated 0.03 Å with respect to free CO. This slight elongation accounts for the low CO stretching frequency (1899 cm −1 ). The bonding was thus analysed using the Natural Bonding Orbital (NBO) analysis. A U-C bond polarized toward C (75%) is observed and this is further corroborated by the low Wiberg Bond Index (WBI) of 0.48 indicating a reduced covalency in the bond (for the sake of comparison, the U-O WBI is 0.40 and the U-Cp ones are 0.11). This bond is a s bond that involves the overlap of a hybrid sp orbital on C and a hybrid spdf (15-11-55-19). The C-O bond is found to be a double bond polarized toward O (70% for the s and 76% for the p) with a WBI of 1.05, in line with the polarization of the bond. To probe the mechanism for back donation, a large core structure of 3 was optimized. Large core structures are obtained with f-in-core Relativistic Core Potentials (RCPs) where the f electron conguration is xed and adapted to a given oxidation state, in this case 5f 3 for U(III). The large core calculations do not allow any back donation from the metal since the f electrons are not explicitly treated. The large core structure gave a stretching frequency of 2098 cm −1 , only 12 cm −1 lower than that of free CO. In contrast, for the small core structure, in which the f electrons are treated explicitly, a stretching frequency of 1899 cm −1 was found. Therefore, backbonding observed in 3 is primarily due to the SOMO-2 orbital (20% 5f, 55% 6d), Fig. S18, † with a minor contribution of the (C 5 Me 5 ) 1− to CO interaction, Fig. S19. † Next, we examined the reactivity of CO with a less sterically crowded aryloxide, i.e., a mesityl group. Addition of 1 atm of CO to 2 in pentane at ambient temperature led to a colour change from dark green to black. A red powder was isolated in excellent (89%) yield. The solid-state structure was determined by X-ray diffraction analysis to reveal the bridging ethynediolate  A weak broad absorption in the IR spectrum at 2010 cm −1 is observed for 4, consistent with an alkyne which is asymmetric in the solid-state. The 1 H NMR spectrum showed resonances ranging from −31 to +2.54 ppm, characteristic of a paramagnetic complex. While the solid-state structure had only one set of unique bonds and angles, the 1 H NMR spectrum revealed an asymmetric mesityl group with resonances for the orthomethyl groups at −17.76 and −30.98 ppm, the para-methyl at 0.05 ppm, and the meta-hydrogens at 1.48 and 2.54 ppm. The reaction was also conducted with 13 CO which produced a resonance at 310 ppm, very similar to the 314.2 ppm observed in [{{C 8 H 6 (Si i Pr 3 ) 2 }(C 5 Me 5 )U} 2 (m-OCCO)]. 54 The formation of 4 was investigated computationally at the DFT level. The reaction (Fig. 3) begins by the formation of the so-called key intermediate where a doubly-reduced CO molecule is sandwiched in between two uranium centres. This diuranium complex is slightly destabilized by 9.4 kcal mol −1 in enthalpy (7.7 kcal mol −1 in Gibbs Free energy) with respect to the separated reactant. Different spin states were considered for this intermediate to check the degree of reduction of the CO molecule. The ground state is a quintet spin state, in line with the presence of two U(IV) and therefore a doubly-reduced CO. This reduction of the CO bond is further highlighted by the CO bond length (1.26 Å) which is elongated by 0.11 Å with respect to 3. The activated CO can then react with another CO molecule, via a low-lying transition state (TS) with an associated barrier of 9.3 kcal mol −1 . This C-C coupling TS is better described as a CO insertion reaction onto the U-C bond with a classical 4-member ring TS, as opposed to a zig-zag like moiety. At the TS, the C-C bond is not yet formed (1.88 Å) while the C-O bond of the insertion CO is elongated to 1.22 Å with a U-O distance of 2.34 Å. Following the intrinsic reaction coordinate, it yields complex 4 whose formation is exothermic by 61.4 kcal mol −1 . While the solid-state structure of 4 showed the mesityl groups on opposite sides to each other, calculations indicate that the isomer with the mesityl groups on the same side is only 0.4 kcal mol −1 higher in energy in the gas phase, Fig. S21. † However, only the energy states were calculated, and a variable temperature NMR experiment did not show the other isomer.
The reaction of 2 with CO can be heated to 80°C for 3 days, resulting in the formation of 5, Scheme 1. Complex 5 can be obtained from isolation of 4, followed by treatment with 1 atm CO. Orange crystals suitable for X-ray diffraction analysis were grown from a saturated pentane solution at −25°C. The solid-state structure of 5 revealed a ketenecarboxylate bridging two uranium(IV) metal centers in which one oxygen of the carboxylate is bound k 1 to one uranium (U1) with a U-O2 bond distance of 2.137(6) Å, while the other oxygen (O3) is coordinated to U2 with a distance of 2.461(6) Å, Fig. 4. The U2-C31 bond length of 2.639(9) Å is long for a U-C bond. Therefore, the geometry about U1 is pseudo-tetrahedral while U2 is a distorted trigonal bipyramidal. The O1-U1-O2 bond angle is 102.2(2)°, the C31-U2-O3 angle is 51.8(2)°, and the C31-U2-O5 and O3-U2-O5 angles are 131.8(2)°a nd 81.67 (19)°, respectively. Consistent with a ketene moiety, the C31-C32 bond distance is 1.295(13) Å, consistent with the C-C double bond, with a short C32-O4 length of 1.193(11) Å, assigned as a carbonyl. A longer C31-C30 distance of 1.442(11) Å, a C-C single bond, and C30-O2 and C30-O3 lengths of 1.233(8) and 1.297(8) Å, respectively, are consistent with a delocalized   Fig. 5. Again, this is the same moiety obtained by Nocton and co-workers with thulium. Although the crystal structure renement of most light atoms in 6 is severely affected by the non-ideal nature of the crystal, there is evidence that the ketenedicarboxylate itself is not severely impacted by error. In particular, the unusually wide OOC-C-COO bond angle of 131.7(6) o observed in the Tm complex is reproduced in 6 (130.4(2) o ). The U/U distance (7.824(1) Å) is also very comparable to the reported Tm/Tm distance (7.6583(9) Å) and slightly longer than U/U distances in alpha-dicarboxylates (7.167 to 7.678 Å), which were previously only reported for U(VI). The 1 H NMR spectrum of 6 shows the (C 5 Me 5 ) 1− resonance at −0.89 ppm and the ortho-methyl groups at 17.44 and 44.24 ppm. However, the para-methyl groups could not be located. We did not observe the two asymmetric ketene stretching vibrations in the IR spectrum. This might be due to that the C-C bond distance of the ketene moiety in 6 is 1.40(3) Å compared to 1.343(10) Å in the analogous thulium complex. We do observe the absorption band at 1474 cm −1 that we assign for the carboxylate groups.
We next investigated electrophilic substrates capable of [2 + 2] cycloaddition such as diphenylketene and SO 2 . Reaction of 4 with Ph 2 C]C]O results in [{(C 5 Me 5 ) 2 U} 2 (OC(CPh 2 )C(=O)CO)], 7, a 1,2dioxy-4,4-diphenylcyclobut-2-en-1-one bridge between two U(IV) centres with concomitant formation of the ligand distribution product, [(C 5 Me 5 ) 2 U(OMes) 2 ], 8, Scheme 1. This is a rare example of a cycloaddition reaction from CO homologation, and rst with an ethynediolate, and, like complex 5, increases the carbon chain to a C4 product. Complex 7 is also an unusual example of [2 + 2] cycloaddition with a metal complex that does not result in a metallocycle. Complex 8 is readily obtained from the reaction of [(C 5 Me 5 ) 2 UCl 2 ] with two equivalents of KOMes.
The structure of 7 has a pseudo-tetrahedral geometry about each uranium centre with an inversion making only one set of unique bond distances and angles, Fig. 6. The C21-C22 distance of 1.359(5) Å can be compared to the C21-C24, C22-C23, and C23-C24 bond lengths which are 1.541(5) Å, 1.436(5), and 1.568(5) Å,   respectively. Therefore, C21-C22 can be assigned as a C-C double bond, while the others are consistent with C-C single bonds, although the 1.436(5) Å is in between a double and single bond. The C-O bond distances, C21-O1 and C22-O2 are 1.295(4) and 1.330(4) Å, respectively, assignable as C-O single bonds while C23-O3 is a carbonyl with a C-O bond distance of 1.207(4) Å.
The 1 H NMR spectrum was taken in CD 2 Cl 2 due to the poor solubility in C 6 D 6 of 7 and showed a (C 5 Me 5 ) 1− resonance at 5.47 ppm. The phenyl protons were paramagnetically shied from −11.41 to 5.06 ppm. The 1 H NMR spectrum of 8 has a (C 5 Me 5 ) 1− resonance at 3.59 ppm, ortho-methyl resonances at −8.79 and −3.16 ppm, and para-methyl at 2.47 ppm. A weak absorption at 1752 cm −1 is attributed to the carbonyl group. The structure of 8 was also determined and shows similar metrical parameters to other U(IV) metallocene bis(aryloxide) complexes. 83 Finally, while not extending the carbon chain, another substrate capable of [2 + 2] addition, i.e., electrophile SO 2 , was obtained using half an equivalent of 1,4-diazabicyclo [2,2,2]octane bis(sulfur dioxide) adduct, DABSO, Scheme 1. Very few molecular examples of SO 2 chemistry are reported with f elements. [84][85][86][87][88][89] To our surprise, the product, 10, involves the cleavage of a S-O bond to form a thiocarbonyl, 90,91 a rare accomplishment and testament to the highly nucleophilic nature of the ethynediolate, to yield an unprecedented [O 2 CC(O)(SO)] 2− ligand bridging between two [(C 5 Me 5 ) 2 (MesO)U] 1+ motifs. Further, thiocarbonyl (SO) is rarely isolated in the reactivity of SO 2 but oen invoked in its reduction.
While the quality of the data is not ideal and displays large error limits, we can extract enough metrics to determine the correct connectivity of the atoms, Fig. 7 93 For example, The C30-C31 bond distance that was formally a C-C triple bond at 1.223 (7) 94 and nearly identical to that observed in sulnates formed by SO 2 insertion into Zn-C bonds. 95,96 From these metric parameters, we conclude that the two chelates to each uranium are both fully delocalized, like the coordination of an oxalate dianion. 97 The formation of complex 9 from complex 4 ( Fig. 8) begins by the formation of a van der Waals adduct of SO 2 which is marginally stabilized by 1.7 kcal mol −1 . From this adduct, the system undergoes a [2 + 2] cycloaddition with an associated low barrier of 9.8 kcal mol −1 . The [2 + 2] cycloaddition nature of the reaction is highlighted by the orientation of the SO 2 molecule at the TS, which lies in a plane parallel to the equatorial plane of the ethynediolate in 4. In such an orientation, the reaction implies the p system of the two molecules (4 and SO 2 ). At the TS, the C-S bond is almost formed (1.87 Å) while the C-O one remains long (2.24 Å). Following the intrinsic reaction coordinate, it yields the formation of a cycloaddition (4-membered ring) intermediate which, although stable by 24.0 kcal mol −1 , readily evolves by breaking the S-O bond involved in the 4-member ring. The associated barrier is 2.0 kcal mol −1 , indicating a very facile reaction. This low-lying TS allows the formation of the very stable complex 9 (−83.8 kcal mol −1 ).

Conclusions
In summary, we have examined the reactivity of two heteroleptic metallocene aryloxide uranium(III) complexes with CO. In the case of the larger aryloxide, only a coordination complex was observed and structurally characterized. However, no further reactivity was observed indicating that the steric bulk of the aryloxide did not allow for homologation to occur. With the sterically less bulky mesityl group, homologation to form an ethynediolate occurred. The reactivity of the ethynediolate complex was probed with additional CO, Ph 2 C]C]O, and SO 2 (DABSO). While there was a precedent for the CO reactivity, the [2 + 2] cycloaddition obtained with Ph 2 CCO and the S-O bond cleavage seen with SO 2 afforded novel and unusual reactivity.  This establishes the potential for uranium to be involved in carbon chain growth chemistry directly from CO, as well as the ability for the ethynediolate, formed by CO homologation, to subsequently undergo unique transformations.

Conflicts of interest
There are no conicts to declare.

Data availability
Experimental procedures, spectroscopic data, information on the theoretical calculations and crystallographic details can be found in the ESI. †