The role of 5f-orbital participation in unexpected inversion of the s -bond metathesis reactivity trend of triamidoamine thorium( IV ) and uranium( IV ) alkyls †

We report on the role of 5f-orbital participation in the unexpected inversion of the s -bond metathesis reactivity trend of triamidoamine thorium( IV ) and uranium( IV ) alkyls. Reaction of KCH 2 Ph with [U(Tren TIPS )(I)] the cyclometallate 2 NSiPr i3 ) 2 (CH 2 CH 2 NSiPr i2 2 )}] ( 3a ) with the intermediate benzyl complex not observable. In contrast, when [Th(Tren TIPS )(I)] ( 2b ) was treated with KCH 2 Ph, [Th(Tren TIPS )(CH 2 Ph)] ( 4 ) was isolated; which is notable as Tren N -silylalkyl metal alkyls tend to spontaneously cyclometallate. Thermolysis of 4 results in the extrusion of toluene and formation of the cyclometallate [Th {N(CH 2 CH 2 NSiPr i3 ) 2 (CH 2 CH 2 NSiPr i2 C[H]MeCH 2 )}] ( 3b ). This reactivity is the reverse of what would be predicted. Since the bonding of thorium is mainly electrostatic it would be predicted to undergo facile cyclometallation, whereas the more covalent uranium system might be expected to form an isolable benzyl intermediate. The thermolysis of 4 follows well-de ﬁ ned ﬁ rst order kinetics with an activation energy of 22.3 (cid:3) 0.1 kcal mol (cid:2) 1 , and Eyring analyses yields D H ‡ ¼ 21.7 (cid:3) 3.6 kcal mol (cid:2) 1 and D S ‡ ¼ (cid:2) 10.5 (cid:3) 3.1 cal K (cid:2) 1 mol (cid:2) 1 , which is consistent with a s -bond metathesis reaction. Computational examination of the reaction pro ﬁ le shows that the inversion of the reactivity trend can be attributed to the greater f-orbital participation of the bonding for uranium facilitating the s -bond metathesis transition state whereas for thorium the transition state is more ionic resulting in an isolable benzyl complex. The activation barriers are computed to be 19.0 and 22.2 kcal mol (cid:2) 1 for the uranium and thorium cases, respectively, and the latter agrees excellently with the experimental value. Reductive decomposition of “ [U(Tren TIPS )(CH 2 Ph)] ” to [U(Tren TIPS )] and bibenzyl followed by cyclometallation to give 3a with elimination of dihydrogen was found to be endergonic by 4 kcal mol (cid:2) 1 which rules out a redox-based cyclometallation route for uranium.


Introduction
Since the search for volatile compounds for uranium isotope separation in the Manhattan project, 1,2 organoactinide chemistry has received continued interest, mainly focusing on uranium.  Early work on uranium tetraalkyls contributed to the perception that s-bonded actinide hydrocarbyls were unstable, 2 but Marks et al. showed that with suitable supporting ligands these linkages can be kinetically stabilised. [27][28][29] Moreover, Hayton et al. have recently shown that homoleptic uranium-alkyls can be kinetically stabilised by saturating the coordination sphere. [30][31][32][33] However, although these linkages can be stabilised by ancillary ligands that resist C-H activation, their intrinsically polarised, reactive nature remains and thus they are tamed but not tempered.
The extent to which f-orbital bonding interactions occur at uranium is still hotly debated, as is their inuence on reactivity and potential nuclear waste separation applications. [34][35][36][37] However, studies have shown that uranium bonding can involve moderate 5f orbital participation leading to modication of reactivity with consequences for the nature of the chemical outcome. [38][39][40] In contrast, the thorium(IV) ion is generally perceived as being unable to access the 5f-orbital manifold. Thus, its bonding is viewed as being restricted to electrostatics resulting in thorium-ligand bonds that are generally more labile than their uranium(IV) counterparts. Added to the greater ionic radius of Th(IV) (94 pm) vs. U(IV) (89 pm), 41 thorium-ligand bonds are generally expected to be more reactive compared to uranium-ligand bonds.
The above arguments suggest that the reactivity of thorium and uranium should be different, but the picture is complicated and not well understood and reports showing divergent experimental reactivity patterns of closely comparable uranium and thorium systems are rare. 38,[42][43][44][45] Intramolecular amino-alkene hydroamination reactions report a trend of thorium being more reactive than uranium; however, there are some notable exceptions for which there is no explanation, but similar trends have been observed for lanthanide complexes suggesting that f-orbital participation is not a key factor. 39,46 Studies by Marks on bis(cyclopentadienyl) actinide dialkyls showed that thorium derivatives are oen more stable than their uranium analogues. 2 However, in some cases reduction of uranium(IV) to uranium(III) provided a decomposition route for uranium not available to thorium. Whilst bond polarities and accessible oxidation states were considered, f-orbital participation was not examined. 29 Recent studies of bis(cyclopentadienyl) actinide dialkyl reactivity by Kiplinger et al. have shown varied reactivity trends for thorium vs. uranium, 43,47,48 but their origins were not resolved. Indeed, the factors that contribute to an isostructural pair of uranium and thorium complexes exhibiting similar or different reactivities are legion and a recent computational study could not derive any clear conclusion. 49 Where f-orbital participation has been probed, this has compared the differences of the f-block to d-block systems. 40 Recently, the formation of a uranium benzyne complex vs. a thorium tetraaryl derivative was reported. 38 In that report greater f-orbital participation in the bonding for the uranium product was found and it was postulated that this provided kinetic facilitation of the benzyne.
As part of our work on triamidoamine chemistry, 50-61 our attention turned to comparing f-orbital participation in the bonding of uranium and thorium and how this might affect reactivity. Since certain cyclometallated actinide complexes, [62][63][64][65][66][67][68] in particular those with triamidoamine ligands, have novel properties, 61,62 uranium and thorium metallacyclic complexes supported by N(CH 2 CH 2 NSiPr i 3 ) 3 3À (Tren TIPS ) were targeted since this ligand has stabilised reactive uranium complexes in multiple oxidation states. 50,52 It should be noted that facile cyclometallation of N-silylalkyl groups is always observed for Tren thorium and uranium alkyls because the metallocycle alkyl groups are stabilised by negative hyperconjugation to silicon as well as the entropic effects from liberating an alkane. Indeed, in a wider context out of dozens of examples of Tren N-silyl metal complexes only one benzyl derivative has been authenticated making this ligand combination very rare. 69 Uranium and thorium benzyl derivatives are known, 21,30,45,64,[70][71][72][73][74][75][76] but where stable ligand metal benzyl combinations are known this is with ligands that do not cyclometallate; where Tren N-silylalkyl ligands are concerned this is facile. 61 Here, we report the synthesis of cyclometallated and benzyl triamidoamine derivatives of uranium and thorium, which surprisingly display inverted reactivity trends. Although the thorium system would be expected to be the most ionic and hence most reactive it is the least reactive and a cyclometallate and its benzyl precursor can both be isolated. In contrast, the uranium system, which would be predicted to be most stable, is the most reactive and only the cyclometallate and not the benzyl intermediate can be isolated. These observations are surprising on two counts. Firstly, the thorium benzyl complex should be unstable with respect to cyclometallation and thus its stability is notable. Secondly, of either of the two complexes uranium should, from an a priori assessment, present the best chance for a Tren metal benzyl complex but this is not the case. We show by an experimentally calibrated computational study that: (i) decomposition involving reduction of uranium(IV) to uranium(III) with elimination of bibenzyl is not operative; (ii) the reversal of anticipated reactivity is due to the s-bond metathesis transition state for uranium being stabilised by uranium 5f-orbital participation that is not present for thorium. Thus, the additional involvement of uranium 5f orbitals actually enhances the reactivity. This study provides a rare account of how 5f-orbital participation in reactive intermediate transition states can alter the chemical reactivity of isostructural systems in counter-intuitive ways that cannot be predicted but which are important to understand.

Synthesis and structures of thorium precursors
With [U(Tren TIPS )(X)] (X ¼ Cl, 1a; I, 2a) species having previously been reported by some of us, 50,52 we targeted the analogous thorium halide complexes required to access the cyclometallated species. Complex [Th(Tren TIPS )(Cl)] (1b) was readily prepared on multi-gram scales from [Li 3 (Tren TIPS )] and [ThCl 4 (THF) 3.5 ] 77 in THF and aer work-up was isolated in nearquantitative yields as an essentially pure solid (Scheme 1). Crystals of 1b suitable for a single crystal X-ray diffraction study were obtained from toluene. The molecular structure of 1b is shown in Fig. 1 with selected bond lengths and angles.
The Th centre in 1b is ve-coordinate, adopting a distorted trigonal bipyramidal geometry with the N4 and Cl1 centres occupying the axial sites [N4-Th1-Cl1 ¼ 173.51 (6) ]. The bond lengths and angles in 1b are unremarkable, with Th-N amide , Th-N amine and Th-Cl bond lengths of 2.299 (av.), 2.668(3) and 2.693(2)Å, respectively, typical of Th(IV)-N amide Th(IV)-N amine and Th(IV)-Cl bond distances. 78 The iodide derivative of 1b, [Th(Tren TIPS )(I)] (2b), was targeted since iodide elimination from actinide centres is oen easier to achieve than elimination of chloride. 61,79,80 Accordingly, a toluene solution of 1b was treated with a slight excess of Me 3 SiI to afford [Th(Tren TIPS )(I)] (2b) and Me 3 SiCl (Scheme 1). Despite numerous attempts the structure of 2b could not be unambiguously conrmed by single crystal X-ray diffraction due to persistently poor data. However, the proposed formulation is supported by the characterisation data which additionally conrms the absence of coordinated donor solvents (full details are provided in the ESI †).

Synthesis and structures of uranium and thorium cyclometallate complexes
Cyclometallated Tren-uranium and -thorium compounds have been prepared previously 61,62 via treatment of a suitable halide precursor with one equivalent of benzylpotassium. The reaction most probably proceeds via a reactive benzyl intermediate which C-H activates the silylalkyl arm of Tren eliminating one equivalent of toluene and forming an An-C s-bond. However, these putative benzyls are very reactive since none have ever been observed with only the cyclometallated products being isolated. An analogous methodology was employed here to access uranium and thorium Tren TIPS -cyclometallates.
A green-brown toluene solution of [U(Tren TIPS )(I)] 52 (2a) reacted with KCH 2 Ph to give a dark orange-red solution. Following work-up and recrystallisation from hexane, a single crystal X-ray diffraction study was performed, which conrmed the structure of the anticipated alkyl complex to be [U{N(CH 2 CH 2 NSiPr i 3 ) 2 (CH 2 CH 2 NSiPr i 2 C[H]MeCH 2 )}] (3a). The molecular structure of 3a is illustrated in Fig. 2 with selected bond lengths and angles.
The U centre in 3a is ve-coordinate, adopting a distorted trigonal bipyramidal geometry with the carbon C2 and amine centre N4 occupying the axial positions. The U-N amide and U-N amine bond lengths of 2.282 (av.) and 2.514(15)Å, respectively, are typical of U(IV)-N amide and U(IV)-N amine bond distances. The U1-C2 bond length of 2.55(2)Å is within the reported range of U-C single bonds [2.40-2.55Å]. 78 In contrast, the cyclometallate [U{N(CH 2 CH 2 NSiMe 2 Bu t ) 2 (CH 2 CH 2 NSiMe-Bu t CH 2 )}] 62 features a longer U-C bond [2.752(11)Å], attributed to increased geometric strain relative to 3a due to its incorporation within a four-membered [U-N-Si-C] ring as opposed to the ve membered ring present in 3a. This is supported by the less acute bite angle in 3a [76.1(7) vs. 66.1(2) ].
The 1 H NMR spectrum of 3a in C 6 D 6 at room temperature consists of 27 resonances spanning the chemical shi range +47 to À35 ppm. This is consistent with the expected unsymmetrical C 1 Tren environment, which is also highlighted by the observation that three resonances are observed in the 29 Si{ 1 H} NMR Scheme 1 Experimental synthesis of thorium and uranium halide, benzyl, and cyclometallated complexes 1-4.  spectrum of 3a, and conrms the presence of a paramagnetic uranium centre. The room temperature Evans method magnetic moment of 3a in C 6 D 6 was found to be 2.90 m B , which is lower than the theoretical free ion value for the formal 3 H 4 ground state of 5f 2 U(IV) of 3.58 m B , 81 but uranium(IV) compounds usually fall within the range 2.50-3.20 m B . 14,30,82,83 The thorium(IV) analogue of 3a was targeted via the treatment of 2b with KCH 2 Ph. Following this reaction, 1 H NMR revealed the consumption of 2b along with the formation of a single new product; however the 1 H NMR spectrum of this product consisted of only 8 resonances. Following work-up and crystallisation from methylcyclohexane, colourless crystals suitable for single crystal X-ray diffraction analysis were isolated. These revealed the new compound to be the Trenthorium benzyl complex [Th(Tren TIPS )(CH 2 Ph)] (4), which is notable for being stable under ambient conditions rather than undergoing cyclometallation. The molecular structure of 4 is illustrated in Fig. 3 with selected bond lengths and angles.
The asymmetric unit of the solid state structure of 4 comprises four crystallographically independent molecules, but in each case the thorium centre is 5-coordinate adopting a distorted trigonal bipyramidal geometry where the amine resides trans to a terminal h 1 -benzyl group. The mean Th-N amide bond lengths within each independent molecule of 4 range from 2.294(6) to 2.311(7)Å and the four Th-N amine bond lengths are statistically invariant with a mean value of 2.706Å. The Th-C bond lengths in 4 range from 2.536(8) to 2.622(7)Å (mean Th-CÀC ispo : ¼ 123.2 ). All the Th-N amide , Th-N amine and Th-C bond lengths in 4 lie within the reported ranges for Th(IV) complexes. 78 In addition the 1 H NMR spectrum of 4 in C 6 D 6 at room temperature consists of 8 resonances located in the chemical shi range of 0-10 ppm and the 13 C{ 1 H} NMR spectrum of 4 comprises 9 resonances with the signal for the benzyl-CH 2 group observed at 113.7 ppm.
With 4 in-hand, it was envisaged that C-H activation to produce the desired Tren-thorium metallacycle might require thermolytic activation since a benzyl complex has been proposed to be an intermediate en route to reported Tren-uranium cyclometallates. 61,62 It was observed that warming NMR samples of 4 at 80 C produced a markedly different 1 H NMR spectrum aer 2 hours consistent with: (i) complete consumption of 4; (ii) production of a new Tren-containing product; and (iii) elimination of one equivalent of toluene. This reaction was reproduced on a bulk scale and following work-up and recrystallisation from hexane, colourless single crystals were subjected to an X-ray diffraction analysis that revealed the new complex to be [Th{N(CH 2 CH 2 NSiPr i 3 ) 2 (CH 2 CH 2 NSiPr i 2 C[H]MeCH 2 )}] (3b). The molecular structure of 3b is illustrated in Fig. 4 with selected bond lengths and angles.
Complex 3b is essentially isostructural to 3a, with any signicant deviation between the metal-ligand bond distances presented by the two complexes ascribed to the smaller ionic radius of U(IV) versus Th(IV). 41 Similar to 3a, complex 3b exhibits a highly unsymmetrical ligand environment with its room temperature 1 H NMR spectrum revealing 16 resonances spanning the narrow range 0.68 to 3.90 ppm. The large reduction in the number of signals in the 1 H NMR spectrum of 3b compared to 3a almost certainly arises from the large difference in chemical shi range of 3a vs. 3b leading to multiple overlapped resonances in the latter. The metallated carbon exhibits the expected downeld chemical shi (105.7 ppm) in the 13 C{ 1 H} NMR spectrum of 3b and like 3a three resonances are observed in the 29 Si{ 1 H} NMR spectrum of 3b.
Given the generally electrostatic bonding of Th(IV) and the larger ionic radius leading to an overall increase in thoriumligand bond lability, it would be expected that 4 would be more unstable toward cyclometallation than its hypothetical uranium congener. Counter-intuitively, what is experimentally observed is the reverse of this where 4 is a stable crystallisable  intermediate, albeit one that will convert to the metallacycle upon thermolysis.

Mechanistic studies
Since 4 converts to 3b in solution under thermolytic conditions, the mechanism of this cyclometallation was probed by a kinetic NMR study. The reaction was found to follow well-dened rst order kinetics with respect to 4, with rate constants showing the expected increase from 2.25 AE 0.1 Â 10 À5 s À1 at 40 C to 5.13 AE 0.3 Â 10 À4 s À1 at 70 C, Fig. 5. Arrhenius analysis revealed an activation energy of 22.3 AE 0.1 kcal mol À1 , which lies within the range of those reported for thermal cyclometallation reactivity proles of thorium complexes, e.g. [ThCp* 2 (R) 2 ] (Cp* ¼ h 5 -C 5 Me 5 , R ¼ alkyl). 29 Eyring analysis determined DH ‡ to be 21.7 AE 3.6 kcal mol À1 , which is in line with the elimination of one molecule of toluene per Th centre. DS ‡ was found to be À10.5 AE 3.1 cal K À1 mol À1 , which is consistent with this type of reaction; 84-86 the value for DS ‡ is negative and of moderate magnitude, inferring that the transition state is sterically constrained and well-dened, and is consistent with the exclusive product selectivity observed experimentally. Such a transition state would be consistent with a four-membered [Th-C-H-C] ring during s-bond metathesis as observed for lanthanides and early or high oxidation state transition metals, which are incapable of supporting classical oxidative addition-reductive elimination sequences. 87 Closely related transition metal Tren complexes readily cyclometallate, presumably due to the availability of d-orbitals and their much greater ability to engage in covalent interactions. 69,[88][89][90] The observation that 4 is isolable in the reaction from 2b to 3b raises the question of whether "[U(Tren TIPS )(CH 2 Ph)]" is stable. However, variable-temperature NMR measurements of the reaction between 2a and KCH 2 Ph always proceed smoothly to give 3a with no intermediates observable over the temperature range 193 to 298 K and thus it was not possible to obtain any experimental mechanistic data on the conversion of 2a to 3a.  3 were also unsuccessful but it should be noted that benzylpotassium is not capable of deprotonating the N-silyl alkyl groups of Tren compounds.
Marks and coworkers showed that bis(cyclopentadienyl) uranium(IV) dialkyls can decompose to uranium(III) derivatives, 29 so we considered the possibility that addition of benzyl potassium to 2a affords "[U(Tren TIPS )(CH 2 Ph)]" which then reductively decomposes to [U(Tren TIPS )] and half an equivalent of bibenzyl; the former could in principle then cyclometallate with concomitant elimination of dihydrogen to form 3a. However, [U(Tren TIPS )] has been prepared by us previously and it does not cyclometallate to give 3a, even at elevated temperatures. Also, we did not observe any resonances in the 1 H NMR spectra from reactions of 2a with benzyl potassium that would be consistent with the formation of bibenzyl or dihydrogen.
We considered the possibility that steric effects could be driving the observed reactivity since uranium is smaller than thorium. However, we note that [U(Tren TMS )(Cl)(THF)] [Tren TMS ¼ N(CH 2 CH 2 NSiMe 3 ) 3 ], which is open enough to leave space for a THF molecule to coordinate to uranium, also spontaneously cyclometallates when treated with benzyl potassium 61 so we conclude that steric effects are not a factor (see below).
Since a redox-based reaction mechanism seemed highly unlikely, the observed contrast between the reactivity modes for uranium and thorium was probed computationally to determine the origin of this discrepancy since a s-bond metathesis reaction would intuitively be the most likely mechanism by which cyclometallation of the N-silyl alkyl groups could occur.

Computational investigations
In order to propose a plausible reaction prole, DFT (at B3PW91 level of theory) calculations were carried out. The calculations show that although the reductive extrusion of benzyl from "[U(Tren TIPS )(CH 2 Ph)]" to give [U(Tren TIPS )] and half a molar equivalent of bibenzyl is thermodynamically favorable (À12 kcal mol À1 ), subsequent cyclometallation to give 3a and half a molar equivalent of dihydrogen is unfavorable (16 kcal mol À1 ) and thus this route, which has no experimental support, is disfavoured by 4 kcal mol À1 and can thus be ruled out. From a thermodynamic standpoint, there is little preference for the formation of the benzyl complexes (labelled 4a and 4b here-aer) from complexes 2a and 2b (Scheme 2). Thus, the experimental conversion of 2a to 3a but isolable sequence of 2b to 4 to 3b should originate from kinetic factors that derive from orbital facilitation present for uranium but not thorium.
The cyclometallation transition states (TS) were located on the potential energy surface and are depicted in Fig. 6. The geometries of the TSs are essentially identical to the classical s-bond metathesis arrangement, 91 with the hydrogen atom that is ultimately transferred lying between the two alkyl groups as reported in previous studies of s-bond metathesis for U-Th catalyzed C-H activation. 92 In particular, the M-H distance is the shortest interatomic distance and the C-H-C angle is close to 180 (176 for U and 172 for Th). Interestingly, Natural Bond Orbital (NBO) analyses at the transition states indicate that the metal-benzyl bonds are already broken (no interactions are found either at the rst nor at the second order donor-acceptor level) and that a M-CH 2 bond is formed (found at the rst order NBO level). The activation barriers were computed to be 19.0 and 22.2 kcal mol À1 for the uranium and thorium cases, respectively. It is germane to note that the calculated activation barrier for the Th case is in excellent agreement with the experimental value of 22.3 kcal mol À1 and compares well to computed f-block s-bond metathesis reaction activation barriers of $20 kcal mol À1 . [93][94][95] In order to rationalise the slight kinetic preference for U over Th, natural population analysis (NPA) was carried out (Fig. 6). The charge is slightly lower at the uranium centre than at thorium, which is consistent with a slightly more ionic TS for the latter than for the former. This is further highlighted by scrutinising the natural charges of the carbon centres of the alkyl groups that are somewhat higher in the thorium case than for uranium. Overall, the description of the charge distribution shows the characteristic hallmarks of a s-bond metathesis TS. 94 It has been proposed that the observed difference between uranium and thorium can be attributed to the lack of 5f involvement in the bonding for the latter, hence inducing a difference in covalency between the two metals (with uranium being more covalent). 38 To further test the hypothesis of the difference in ionicity/covalency due to the 5f electron involvement at the TS, the molecular orbitals (MOs) were analyzed (see ESI †). In the uranium case, four occupied orbitals describe metal-alkyl interactions (mainly employing 5f orbitals) whereas only two are found in the thorium case (mainly involving 6d orbitals). This is in-line with reports 38,40 of the involvement of 5f orbitals in bonding but there is no clear correlation between this and a difference in covalency between thorium and uranium. The cyclometallation reaction step was computed to be slightly exothermic for uranium by 2.8 kcal mol À1 overall whereas it is almost athermic for Th overall (0.8 kcal mol À1 ).
This reactivity might be explained by either the steric hindrance around the metal centre (sterically induced reactivity) or the participation of the 5f orbitals. In order to test the hypothesis of the importance of 5f orbital participation and their effect on the observed reactivity, we computed the enthalpy proles for both metals without including the 5f Scheme 2 Enthalpy energy profiles (at 298 K) for the gas phase cyclometallation reaction for the complexes 2a (uranium, blue profile) and 2b (thorium, black profile) computed at the B3PW91 level of theory. electrons and orbitals in the calculation using large core relativistic effective core potentials (RECPs) that were thus treated as 6d transition metals. The use of large core RECPs has been demonstrated to be appropriate to describe U(IV) and Th(IV) chemistry 95-97 and it has even been possible to show previously that the major effect was the core polarisation rather than the 5f involvement. When f-orbitals and electrons are excluded from the calculation, the proles for uranium and thorium are found to be the same, with the same activation barrier (see ESI †). Moreover, the barrier for thorium is exactly the same with both approaches, in-line with a lack of 5f involvement for this atom. For uranium, the barrier is approximately 3 kcal mol À1 higher without 5f electrons than when these electrons are included in the calculation. In this case, the charge at the uranium centre is higher and similar to the thorium example. This suggests that the lack of 5f orbitals increases the charges and thus makes the bonds more ionic. Furthermore, calculations were carried out on the hypothetical Zr and Hf complexes. The computed barriers were also found to be around 24-26 kcal mol À1 , which is higher than for uranium. As the steric congestion is really enhanced in the two latter complexes compared to the thorium or uranium ones, these results rule out sterically induced reactivity. Thus, the calculations suggest that 5f orbital participation in the bonding indeed induces a clear change in the bonding, which can supercially be interpreted as covalency, [98][99][100] and this in turn has a profound effect on the experimentally observed reactivity.

Conclusions
To conclude, we have reported the synthesis of cyclometallated and benzyl triamidoamine derivatives of uranium and thorium, which surprisingly exhibit reactivity trends that are the reverse of what would be predicted. Although the thorium system would be anticipated to be the most ionic and therefore reactive, it is the least reactive and a cyclometallated complex and its benzyl precursor can both be isolated. In contrast, the uranium system, which would be expected to be less reactive, is more reactive and only the cyclometallated complex can be isolated. Experimentally supported calculations suggest that this reversal of expected reactivity can be attributed to the s-bond metathesis transition state for uranium being stabilised by 5f orbital participation in the interatom interactions in the TS, which thus enhances the reactivity. In contrast, the thorium case lacks 5f orbital stabilisation and thus the transition state is more ionic and kinetically less accessible and therefore this system is surprisingly less reactive. The calculations also suggest that a cyclometallation mechanism for uranium involving redox chemistry or steric effects can be discounted. This study represents an example of how 5f orbital participation can steer chemical reactivity in an unexpected yet profound manner.