Assessing crystal field and magnetic interactions in diuranium-μ-chalcogenide triamidoamine complexes with UIV–E–UIV cores (E = S, Se, Te): implications for determining the presence or absence of actinide–actinide magnetic exchange

Analysis of UIV–E–UIV (E = S, Se, Te) complexes reveals their behaviour is due to crystal field effects and not exchange coupling.


Introduction
There is continued interest in understanding the magnetic behaviour of actinide compounds, 1 in particular multimetallic complexes that may exhibit magnetic exchange interactions, in order to gain detailed knowledge of their electronic structure that underpins their unique chemical reactivity. 2 However, unravelling their behaviour is complicated considerably by the rich interplay of interelectronic repulsion (IER), spin orbit coupling (SOC), and crystal eld (CF) effects. 3 For uranium in particular, the strength of each of these components of electronic structure can vary signicantly depending on the oxidation state of uranium and nature of the coordinated ligands, 4 making comprehensive analysis a signicant challenge even with modern characterisation and computational techniques. Where magnetic exchange interactions involving uranium are concerned, detecting their presence by variable temperature magnetometry is difficult because of the non-trivial temperature dependence arising from the CF and SOC states of the individual magnetic centres obscuring possible magnetic exchange. Spectroscopic methods (i.e. electron paramagnetic resonance, EPR) for determination of exchange is much more reliable, 5 however this is not amenable in all cases such as when the exchange is relatively strong (J [ hn) or for non-Kramers ions like uranium(IV) where singlet states usually give EPRsilent species. 6 Although clear-cut examples of magnetic exchange in uranium complexes remain rare, the majority of documented cases involve uranium(V), [7][8][9][10][11][12] and there are few reports of magnetic exchange in diuranium(IV) complexes or uranium(IV)transition metal coupling. [13][14][15][16][17][18][19] Where magnetic exchange for actinide complexes is proposed this is usually on the basis of observing a maxima in the magnetic susceptibility (c) vs. T plot (oen referred to as the Néel point), 20,21 however care needs to be taken when using this as evidence alone. If a clear maximum is observed this is usually due to antiferromagnetic exchange, however if the maximum is obscured in any way, [22][23][24][25][26] for example by a low-level (even $1%) paramagnetic impurity which becomes prominent at low temperature, then the presence or not of magnetic coupling becomes nebulous and cannot be stated with condence; 27 however, alternative explanations are oen not immediately obvious. An alternative and rarely invoked explanation for ambiguous plateaued maxima/ shoulders in c vs. T plots could in fact, rather than owing to magnetic exchange, simply reect single-ion CF effects; this has rarely been investigated in detail since families of isostructural diuranium molecules with a systematic variation of bridging groups are few in number.
Here, we report on the synthesis and magnetism of U-E-U complexes supported by Tren TIPS , and show that shoulders that could be interpreted as local maxima in magnetometry data, which could be attributed to magnetic exchange, are in fact almost certainly the result of single-ion CF effects. This gives a greater appreciation of the CF effects that are intimately involved in unravelling the magnetic properties, and therefore electronic structure, of actinide complexes, and specically for the identication of magnetic exchange effects; this is of preeminent importance to progressing the eld of actinide molecular magnetism and improving our understanding of actinide electronic structure and correlating this to chemical reactivity.

Synthesis and characterisation of complexes 2-6
The uranium(III) complex [U(Tren TIPS )] 28 [1,Tren TIPS 3 ] reacts with Ph 3 PS (Scheme 1) in toluene to afford a yellow precipitate. The remaining solution contains signicant quantities of unreacted Ph 3 PS from a 1 : 1 reactionas ascertained by removal of solvent in vacuo and analysing the residue by 1 H NMR spectroscopybut no resonances attributable to 1 were observed. The 1 H NMR spectrum of the yellow precipitate suggested the formation of a new complex and crystallisation from THF afforded crystals of sufficient quality for a single crystal X-ray diffraction (XRD) study, which conrmed the formulation to be [{U(Tren TIPS )} 2 (m-S)] (2). The molecular structure of 2 is illustrated in Fig. 1 with selected bond lengths and angles.
The solid state molecular structure of 2 reveals a dinuclear Tren TIPS -uranium m-sulde complex with the S-centre bridging the two uranium ions. The ve-coordinate uranium centres are each coordinated by one tetradentate Tren TIPS ligand and the bridging sulde, the latter bridging the two U atoms with a linear U-S-U bond angle [179.81 (18) ]. As a result of the sterically demanding Tren TIPS ligands the uranium centres adopt distorted trigonal bipyramidal geometries and exhibit U-N bond distances that are typical of uranium(IV)-N amide and -N amine bonds. 45 The two Tren TIPS ligands in 2 adopt an eclipsed orientation when viewed along the U-S-U bond vector. The two identical U-S bond distances in 2 of 2.6903(6)Å lie at the upper end of the range of bridging U IV -S bonds in reported examples [2.588(1)-2.713(2)Å], [40][41][42][43][44] alluding to the sterically encumbered nature of Tren TIPS , but are slightly shorter than the sum of the covalent single bond radii of uranium and sulfur (2.73Å), 46 supporting the presence of a U IV -S-U IV unit.
The 1 H NMR spectrum of 2 features seven paramagnetically shied resonances in the range À22 to +7 ppm, presumably as a consequence of low symmetry from the eclipsed orientation of the Tren TIPS ligands observed in the solid state being maintained in solution on the NMR timescale. The room temperature solution magnetic moment (Evans method) of 2 in benzene of 3.79 m B is consistent with the presence of two 5f 2 uranium(IV) centres. 47 Variable temperature magnetometry carried out on a solid sample of 2 revealed a m eff value of 3.47 m B at 300 K that is in good agreement with the solution magnetic moment considering the difference of phases. This value decreases steadily to 3.25 m B at 100 K at which point the value of m eff drops more sharply to 0.61 m B at 1.8 K (Fig. 8). It is worth highlighting at this juncture that the uranium(IV) ions in 2 clearly adopt magnetic singlet ground states with modest temperature independent paramagnetism contributions at low temperature which, from a simple CF-only approximation, would be anticipated for an O h ligand eld. The trigonal bipyramidal ligand eld of these complexes predicts a magnetically active E ground state, 13 but all Tren-uranium(IV) complexes exhibit magnetisation behaviours consistent with A ground states; this suggests that SOC effects dominate over the CF, which is borne out by our modelling (see below).
The isolation of a diuranium(IV)-sulde suggests that despite the appropriate reagent stoichiometry for the anticipated twoelectron oxidation reaction, i.e. 1 + 1/8 S 8 to form "[(Tren TIPS ) U V ](S)]", kinetic trapping of the postulated terminal sulde complex by available 1 in solution with disproportionation of U V ]S/U III to U IV -S-U IV ought to be considered likely. It should be noted that Tren TIPS has proven capable of stabilising terminal uranium(V/VI)-nitrides and parent uranium(IV)-imide, -phosphinidene, and -arsinidene complexes, 5,28-35 but although these complexes suggest that stabilisation of a terminal sulde should be possible [48][49][50][51][52] with Tren TIPS , the nitride is small compared to sulde, whereas the pnictidenes are anionic formulations that would resist the formation of bridging species. Further, although dithorium bridging pnictido and pnictidiide complexes supported by Tren TIPS have been reported recently, 36,37 the analogous uranium complexes are yet to be reported which may be related to the larger size of thorium compared to uranium. In an attempt to counteract m-sulde formation, employment of a Lewis base adduct of 1 was envisaged to limit the quantity of free 1 available and thus permit isolation of a terminal sulde complex. Thus 1 was treated with one equivalent of (Me 2 N) 3 PO (HMPA, Scheme 1), to afford a dark green precipitate whose 1 H NMR spectrum indicated the formation of a new complex with threefold molecular symmetry in solution on the NMR timescale at room temperature. Crystallisation from hexanes afforded crystals of suitable quality for a single crystal XRD study, which conrmed the formulation to be [U(Tren TIPS )(HMPA)] (3). The molecular structure of 3 is illustrated in Fig. 2 with selected bond lengths and angles.
The uranium centre in the solid state structure of 3 adopts a distorted trigonal bipyramidal geometry and is coordinated to the tetradentate Tren TIPS ligand and the oxygen atom of an HMPA ligand. The U-N amide and U-N amine bond distances in 3 of 2.378(4) (av.) and 2.699(4)Å lie are as expected extended by $0.05 and $0.14Å, respectively, relative to HMPA-free 1. 28 The U1-O1 bond distance of 2.498(3)Å is comparable to that The 1 H NMR spectrum of 3 features four paramagnetically shied resonances in the range À4 to +6 ppm, reecting the C 3 symmetry, and one additional resonance at 1.10 ppm assigned to the methyl groups present in the coordinated HMPA. The 31 P { 1 H} NMR spectrum reveals a single phosphorus environment at +90.0 ppm which is signicantly shied downeld from that of  (7), U1-S1-U1A 179.81 (18), S1-U1-N4 178.18 (19). uncoordinated HMPA (d 24.6 ppm) and suggests polarisation of electron density away from the P centre towards the uranium centre in 3, consistent with a [(Me 2 N) 3

PO/U(Tren TIPS )] donor interaction.
The room temperature solution magnetic moment (Evans method) of 3 in benzene of 3.02 m B compares with a value of 2.80 m B for the Tren-uranium(III) complex [U(Tren DMBS )(HMPA)] and 2.85 m B for 1. 28,53 Variable temperature magnetometry of a solid sample of 3 ( Fig. S1 †) revealed a m eff value of 2.74 m B at 300 K. These data are lower than the theoretically expected moment of 3.69 m B for a 5f 3 conguration with a 4 I 9/2 electronic ground state due to CF effects, and are consistent with reported uranium(III) complexes. 47 This value decreases steadily to $50 K at which point the value of m eff drops off more sharply to 1.55 m B at 1.8 K. This behaviour is characteristic of 5f 3 uranium(III), with magnetic moments of above 1 m B at low temperature due to the Kramers doublet ground state.
With complex 3 in hand, a terminal uranium-sulde species was targeted for comparative purposes by treatment of 3 with Ph 3 PS (Scheme 1). However, following work-up, 2 was isolated as the exclusive uranium product. We suggest that a terminal sulde complex may be formed transiently in the reaction of 3 with Ph 3 PSas additionally the production of PPh 3 is observed but it is a sufficiently strong Lewis base to displace HMPA from 3 and form the isolable bridging sulde complex 2.
Reasoning that an alternative sulfur reagent may lead to differing reactivity, 3 was treated with 0.125 equivalents of S 8 in toluene (Scheme 1), which afforded, aer work up and crystallisation, orange crystals suitable for a single crystal XRD study that conrmed the structure to be [{U(Tren TIPS )} 2 (m-h 2 :h 2 -S 2 )] (4). The molecular structure of 4 is illustrated in Fig. 3 with selected bond lengths and angles.
The molecular structure of 4 reveals two Tren TIPS -uranium units bridged by a side-on m-h 2 :h 2 -S 2 moiety. Each uranium centre is six-coordinate, ligated by the four N atoms of a tetradentate Tren TIPS ligand and two sulfur atoms and adopts a distorted octahedral geometry. The U-N amide and U-N amine bond distances in 4 of 2.264(3) (av.) and 2.709(3)Å are typical for uranium(IV)-N amide and -N amine bonds, 45 respectively, and compare favourably to those in 2. The U-S bond distances of 2.8666(12) and 2.9280(15)Å are statistically distinct and roughly lie within range of the corresponding bond lengths in uranium(IV) persulde complexes [2.7062(16)-2.9228(15)Å] which reects the large steric demands of Tren TIPS , and the S1-S1A bond distance of 2.1041(18)Å is identical to reported S-S per-sulde single bond distances. 42,43 The solid state structure of 4 adopts a symmetrical U 2 S 2 core and possesses a crystallographic inversion centre at the midpoint of the S-S bond. As a whole the structural data support the assignment of a persulde (S 2 2À ) unit bridged between two uranium(IV) centres in 4. The 1 H NMR spectrum of 4 features four paramagnetically shied resonances in the range À37 to +33 ppm, consistent with an approximate threefold-symmetry of the Tren TIPS ligands. The room temperature solution magnetic moment (Evans method) of 4 in pyridine of 3.89 m B is consistent with the presence of two 5f 2 uranium(IV) centres and compares to a value of 3.74 m B for the dinuclear bridged Tren TIPS -uranium(IV) sulde complex 2. Variable temperature magnetometry on a solid sample of 4 revealed a m eff value of 3.76 m B at 300 K ( Fig. S2 †), which is in excellent agreement with that determined from NMR. This value decreases steadily to 3.32 m B at 100 K at which point the value of m eff drops off more sharply to 1.46 m B at 1.8 K.
The isolation of complex 4 (and not 2) from 1 with S 8 demonstrates the sensitive nature of the reactivity prole of 1 with sulfur containing reagents, a feature that has also been observed in the reactivity of the related uranium(III)-triamide complex [U{N(SiMe 3 ) 2 } 3 ]. 42 This was further investigated by treating 4 with KC 8 in an attempt to produce a Tren-uranium complex with a terminal S functionality that could benet from additional stabilisation from coordinated K + ions. However, 4 reacts with one equivalent of KC 8 (Scheme 1) to afford a brown mixture from which a gray solid containing graphite and potassium sulde, K 2 S, was separated by ltration. A crop of orange crystals was isolated from the ltrate aer cooling to 5 C, which was identied as 2 by a crystallographic unit cell check and 1 H NMR spectroscopy. The very low yield of 2 from this reaction (7%) highlights the consistently poor crystalline yields of 2 due to the inherent sensitivity and lability of U-S and E-E bonds, 44,54,55 and also the thermodynamic favourability of U-S-U formation, given that 2 is the only product of note from this reaction.
The reactivity of 3 towards the heavier chalcogens was then explored. Treatment of 3 with one equivalent of either elemental selenium or tellurium afforded, aer work-up and crystallisation, dark orange or red crystals, respectively, suitable for single crystal XRD studies. These revealed the   (3), S1-S1A 2.1041(18), U1-S1-U1A 137.43(4), S1-U1-S1 42.57(4), S1-S1A-U1 70.27(5), S1-S1A-U1A 67.16(4). The 1 H NMR spectrum of 5 features seven paramagnetically shied resonances in the range À31 to +10 ppm, which, by analogy to 2, is suggested to be due to the eclipsed orientation of the Tren TIPS ligands on the NMR timescale. The 1 H NMR spectrum of 6, however, features only three broad resonances, which by contrast to 2 and 5 is ascribed to the staggered orientation of the Tren TIPS ligands observed in the solid state structure being maintained in solution on the NMR timescale. The room temperature solution magnetic moments (Evans method) of 5 and 6 in benzene of 3.88 and 4.10 m B are each consistent with the presence of two 5f 2 uranium(IV) centres and are similar to that measured for 2. These data are reected in variable temperature magnetometry experiments on powdered 5 and 6, which reveal magnetic moments of 3.5 and 3.75 m B at 298 K, respectively, that fall monotonously reaching values of $0.7 m B at 1.8 K and tending to zero (Fig. 8). Notably, the magnetic data for 2, 5, and 6, whether powdered or in solution, exhibit magnetic moments that are ordered 6 > 5 > 2.
Electronic structure and magnetic analysis of 2, 5, and 6 The electronic ground state of uranium(IV) is well dened by the [Rn]5f 2 conguration, with the next electronic conguration [Rn]5f 1 6d 1 lying at ca. 100 000 cm À1 . Following the Russell-Saunders (RS) scheme for angular momentum coupling, IER splits the [Rn]5f 2 conguration into terms, the ground one being 3 H given by Hund's rules (Fig. 6, le). SOC then splits these terms into total angular momentum states J ¼ |L À S|, |L À S| + 1, ., L + S À 1, L + S (Fig. 6, right), with the 3 H 4 SO multiplet lying lowest ca. 5000 cm À1 below the 3 F 2 SO multiplet arising from the rst excited 3 F term. When the uranium(IV) ion is incorporated into a molecular complex, the electronic states are split owing to the formation of molecular orbitals (MO); for transition metal complexes this is oen described as the effect of the CF and we use the same nomenclature here. For all three complexes, 2, 5 and 6, and despite the large radial extent of the 5f orbitals compared to 4f orbitals, the CF generated by the Tren TIPS and E ligands is rather small compared to both the IER and SOC (Fig. 7 cf. Fig. 6). Thus, the low-lying electronic states of each uranium(IV) ion in complexes 2, 5 and 6 are well-described by the 3 H 4 SOC multiplet split by the CF; this is the familiar case for lanthanide ions, and thus uranium(IV) resembles the [Xe]4f 2 Pr III ion in this case.
In such cases where SOC > CF, it is more appropriate to consider the action of the CF on the ground J manifold. Thus, for ligand environments with trigonal bipyramidal symmetry such as these, the J ¼ 4 state will be split into three singlets and three doublets. However as uranium(IV) is a non-Kramers ion (even number of unpaired electrons), the CF can entirely remove the degeneracy of the 3 H 4 multiplet, resulting in 9 singlet states in low symmetry; importantly, these singlets are highly anisotropic and will be affected by a magnetic eld and thus uranium(IV) species are not diamagnetic at low temperature, and rather show temperature independent paramagnetism (TIP). Indeed, complete active space self-consistent    (Tables S1-S3 †). While the splitting of the ground multiplet for the three complexes is broadly quite similar, there is a systematic increase in CF splitting from 2, 5 and 6, despite the systematic lengthening of the U-E bond, suggesting that the donor strength increases as S < Se < Te. Trends are also observed in the LoProp charges on the bridging chalcogenide atom and in the percentage of the active space made up of E-based atomic orbitals (AOs) (Table S4 †), suggesting that the increase in eld strength is due to increased charge accumulation on the chalcogenide atom whilst the covalency may actually be decreasing as S > Se > Te.
The temperature dependence of the magnetic moment of polycrystalline 2, 5 and 6 are all broadly similar (Fig. 8); m eff has values of 2.46-2.78 m B (cT ¼ 0.75-0.96 cm 3 mol À1 K) per uranium at room temperature, in good agreement with that expected for uranium(IV), 47 that decrease slowly upon cooling until 100 K when they drop more rapidly reaching 0.43-0.53 m B (cT ¼ 0.02-0.04 cm 3 mol À1 K) per uranium at 1.8 K (Fig. 8). The magnetic susceptibilities plateau at around 50 K with values of ca. 0.01 cm 3 mol À1 for all three complexes, before increasing rapidly below 10 K. The observation of such plateaus is direct evidence of a singlet ground state, in agreement with the CASSCF-SO calculations; if the ground state were a pseudodoublet then the susceptibility would be Curie-like and behave as c f 1/T at all temperatures.
Plateaus such as these are but one of three experimentally observed low temperature proles for mono and dimetallic uranium(IV) species. In addition to plateaus, 48,56 continual Curie-like rising of the susceptibility 14,27,48,52 and Néel-type maxima 14,15 are also observed. Given that c vs. T proles are sometimes employed to directly infer the presence or absence of magnetic interactions between uranium ions, 14,15,27 it would be helpful to have a guide to aid in such interpretations; hence, we seek to simulate a set of magnetic traces for mono and dimetallic U IV species, to aid in the identication and characterisation of magnetic interactions. While the CF will be different for each molecule, we employ a simplied CF model to highlight the differences between singlet and (pseudo-) doublet ground states for individual uranium(IV) centres. Thus our model consists of a single axial CF term for each uranium centre, along with an exchange interaction of the Lines type (Hamiltonian eqn (1) and (2)). 57 We performed our simulations with PHI 58 using the |J ¼ 4, m J i basis for each uranium(IV) centre, where B 0 2 are the axial CF parameters,Ô 0 2 are the Stevens operators, J is the Lines exchange parameter, g J is the Landé gfactor, and the exchange term is treated using a Clebsch-Gordan decomposition. We choose our CF parameter to be B 0 2 ¼ AE30 cm À1 (positive for a singlet m J ¼ 0 ground state and negative for a doublet m J ¼ AE4 ground state), chosen to generate CF splitting on the order of magnitude calculated for 2, 5 and 6, and x g J ¼ 0.8 from the free-ion 3 H 4 SO multiplet. The magnetic susceptibility is simulated from 1.8-300 K in a eld of 0.1 T using the expression c z M/B in order to match the most common experimental conditions.
For a monometallic species where inter-uranium magnetic interactions cannot occur (excluding through-space dipolar couplings that are expected to be negligible for magnetic measurements, though detectable with EPR 5 ), a plateau in the susceptibility must arise from isolation of a singlet ground state due to the CF (Fig. S3a †); indeed this feature has been observed previously for monometallic uranium(IV) complexes. 56,[59][60][61][62] In such cases, any further increase at the lowest temperatures can be condently attributed to a small paramagnetic impurity which can be very difficult to avoid when dealing with airsensitive sampleseven 1% impurity can easily be seen by sensitive magnetic measurements (Fig. S5 †). If on the other hand the susceptibility for a monometallic uranium(IV) species appears Curie-like at all temperatures, this is indicative of a (pseudo-)doublet ground state (Fig. S3b †).
When considering the presence of a magnetic interaction for dimetallic species the picture is much less clear (Fig. S3, S4 and S6-S8 †). In fact, there are only two cases where the magnetic trace can unequivocally dene magnetic interactions under visual inspection: (1) c vs. T shows a Néel-type maximum at low temperature, corresponding to a pair of uranium(IV) ions with (pseudo-) doublet ground states with an antiferromagnetic interaction; (2) m eff or cT vs. T shows an increase at low temperature, corresponding to a pair of uranium(IV) ions with (pseudo-) doublet ground states with a ferromagnetic interaction.
A trace of any other type could easily be assigned to a number of different situations, and thus in such situations it is inappropriate to assign the presence of a magnetic interaction on this data alone. Furthermore, it is also possible that a small yet not insignicant magnetic impurity may be obscuring the true low temperature behaviour; this is particularly pertinent for uranium magnetic data given the wide range of room temperature magnetic moments reported and overlaps between oxidations states. 47 Therefore, using these guidelines, the magnetic data for compounds 2, 5 and 6 do not indicate signicant magnetic interactions, though they cannot be absolutely ruled out. As both sites are crystallographically equivalent and the CASSCF-  SO calculations suggest a well-isolated singlet for the individual U IV ions, we interpret the plateaus in c at low temperatures as conrmation of this feature, while the rise below 10 K is attributed to a small paramagnetic impurity. The CF parameters for the 3 H 4 SO multiplet obtained directly from the CASSCF-SO calculations (Table S5 †) lead to simulations of the magnetic susceptibility that have extremely similar proles to the magnetic data per uranium for 2, 5 and 6 ( Fig. S9-S11 †), however the absolute magnitude is incorrect. These simulations employ the free-ion |J ¼ 4, m J i basis, while the true magnetic orbitals are MOs with non-negligible ligand character; a common approach to account for this is to include an orbital reduction parameter, k. 5, 63 We include this parameter by allowing g J to be reduced from 0.80 (indeed this is also suggested by CASSCF-SO, Table S4 †); using g J as a variable, and incorporating a small S ¼ 1 Curie-like impurity, an excellent t of the experimental data is obtained (Fig. 8, S12 and S13, † and Table 1). By re-arranging the equation for the Landé g-factor, 64 the resulting g J values for uranium(IV) can be re-cast as orbital reduction parameters with the expression k ¼ 2/6 + 5g J /6; these are found to be between 0.85 and 0.92, which is of a similar magnitude to those determined for the [(Tren TIPS )U V (N)] À anion of between 0.88 and 0.97. 5 The trend of decreasing covalency across the series as suggested by the percentage of E-based AOs in the CASSCF active space as S > Se > Te is also reected experimentally by the increase in k in the same sense; however, we note that k, whilst accommodating the effects of covalency on the magnetic properties, does not provide an unequivocal measure of covalency and therefore should not be compared between different molecular series where other factors, such as oxidation state and CF, will play a major role in the electronic structure and bonding.

Conclusions
To conclude, we have prepared three diuranium m-chalcogenide complexes along with one persulde. Complexes with U-E-U (E ¼ Se, Te) cores are consistently formed, but for sulfur U-S-U and U(S 2 )U could be selectively obtained. The essentially linear U-E-U cores provide an opportunity to study the magnetism of these linkages where plots of c vs. T present shoulders that could be interpreted as evidence of uranium-uranium magnetic exchange. However, a detailed study of the electronic structure of these uranium complexes reveals that the magnetic properties of these systems can be simply correlated to single-ion CF effects that vary as the nature of the chalcogen varies.
Elucidating the presence of magnetic coupling between actinide ions is an area fraught with difficulty and ambiguity, and although a maximum in c vs. T is usually good evidence for magnetic exchange, there are numerous examples where it would be tempting to suggest magnetic exchange occurs (i.e. chemically plausible scenarios) when in reality the behaviour is instead actually the manifestation of CF effects. The present study highlights this caveat and places the assignment of CF effects, rather than magnetic exchange, on a rmer footing thus enabling assignments to be made with more condence more broadly. Method A. A solution of 1 (0.85 g, 1.0 mmol) in toluene (10 ml) was added slowly to a cold (À78 C) stirring solution of triphenylphosphine sulde (0.29 g, 1.0 mmol) in toluene (10 ml). The mixture was allowed to warm to room temperature whilst stirring over 16 h, affording a yellow precipitate. The solid was collected by ltration, washed with hexanes (3 Â 2 ml) and dried in vacuo. The solid was recrystallised from hot (50 C) THF at 5 C which yielded yellow crystals that were isolated by ltration, washed with toluene (3 Â 2 ml) and dried for 30 minutes. Yield: 0.33 g, 35%.

Experimental
Method B. Toluene (10 ml) was added slowly to a stirring mixture of 3 (0.51 g, 0.5 mmol) and triphenylphosphine sulde (0.15 g, 0.5 mmol) at À78 C. The mixture was allowed to warm to room temperature whilst stirring over 16 h, which afforded a yellow precipitate. The solid was collected by ltration, washed with hexanes (3 Â 2 ml) and dried in vacuo. The solid was recrystallised at 50 C from hot THF which yielded yellow crystals of 2 which were isolated by ltration, washed with toluene (3 Â 2 ml) and dried for 30 minutes. Yield: 0.15 g, 33%.
Method C. THF (15 ml) was added to a cold (À78 C) stirring mixture of 4 (0.10 g, 57 mmol) and KC 8 (15 mg, 113 mmol) and the resulting brown suspension allowed to warm to ambient temperature and stirred at this temperature for 16 h to afford a dark yellow solution and a black solid. The solution was ltered and solvent was removed in vacuo to afford a yellow solid that was dissolved in warm (60 C) toluene (5 ml) and reduced in volume to ca. 0.5 ml. Crystalline material was obtained by storage of this solution at room temperature overnight and the yellow crystals obtained were isolated by ltration, washed with toluene (3 Â 2 ml) and dried for 30 minutes. Yield: 7 mg (7%). 1

Preparation of [U(Tren TIPS )(HMPA)] (3)
A solution of HMPA (0.09 g, 0.5 mmol) in hexanes (2 ml) was added dropwise to a dark blue stirring solution of 1 (0.42 g, 0.5 mmol) in toluene (10 ml) at À78 C. The dark green solution produced was allowed to warm to room temperature whilst stirring over 16 h. The resulting precipitate was isolated by ltration and washed with hexanes (2 Â 5 ml) to yield 3 as a very dark green solid. Complex 3 was recrystallised from a saturated solution of hexanes at 5 C. Yield: 0.40 g, 78%. 1

Preparation of [{U(Tren TIPS )} 2 (m-h 2 :h 2 -S 2 )] (4)
A green-black solution of 3 (0.61 g, 0.6 mmol) in toluene (10 ml) was added dropwise to a cold stirring (À78 C) suspension of S 8 (19 mg, 75 mmol) in diethyl ether (10 ml) over ten minutes. The resulting brown-black mixture was allowed to warm to ambient temperature over 1 hour aer which time a colour change to orange-brown was observed accompanied by the consumption of any residual sulfur. Aer stirring at ambient temperature for a further 1 h, removal of solvent in vacuo and suspension of the residue in hexanes (10 ml) afforded an orange-yellow solid that was isolated by ltration, washed with hexanes (3 Â 2 ml) and dried for 30 minutes. This product is essentially pure but may be recrystallised from a saturated toluene solution at À30 C, although crystalline yields are low. Yield: solid 0.18 g (35%) or crystalline 50 mg (10%). 1

Preparation of [{U(Tren TIPS )} 2 (m-Se)] (5)
A green-black solution of 3 (0.61 g, 0.6 mmol) in toluene (5 ml) was added dropwise to a cold stirring (À78 C) suspension of selenium (47 mg, 0.6 mmol) in diethyl ether (10 ml) over ten minutes. The resulting dark brown-black mixture was allowed to warm to ambient temperature over 1 h aer which time a colour change to green-black was observed. A further gradual colour change to yellow-brown was observed by stirring the mixture at ambient temperature for a further 16 h. Removal of solvent in vacuo and suspension of the residue in toluene (10 ml) afforded an orange-yellow solid that was isolated by ltration, washed with hexane (3 Â 2 ml) and dried for 30 minutes.