Isolation of Elusive HAsAsH in a Crystalline Diuranium(IV) Complex

The HAsAsH molecule has hitherto only been proposed tentatively as a short-lived species generated in electrochemical or microwave-plasma experiments. After two centuries of inconclusive or disproven claims of HAsAsH formation in the condensed phase, we report the isolation and structural authentication of HAsAsH in the diuranium(IV) complex [{U(TrenTIPS)}2(μ-η2:η2-As2H2)] (3, TrenTIPS=N(CH2CH2NSiPri3)3; Pri=CH(CH3)2). Complex 3 was prepared by deprotonation and oxidative homocoupling of an arsenide precursor. Characterization and computational data are consistent with back-bonding-type interactions from uranium to the HAsAsH π*-orbital. This experimentally confirms the theoretically predicted excellent π-acceptor character of HAsAsH, and is tantamount to full reduction to the diarsane-1,2-diide form.

Dipnictenes REER (E = N, P, As,Sb, Bi;R= H, alkyl, aryl) are af undamental class of molecules that have played acentral role in the development of main-group chemistry. [1] Diazenes have been known for decades and are most prevalent, and though diphosphenes,d iarsenes,d istibenes, and dibismuthenes have all been reported in the past thirty years,their numbers rapidly decrease down the group. [1] This reflects the difficulties of stabilizing multiple bonds between increasingly large nuclei, the importance of dispersion forces, [1a] and the reduced tendency of heavier p-block elements to catenate,and thus sterically demanding substituents are required to stabilize these linkages. [1] However,i tis fundamentally appealing to study parent REER molecules (R = H), free of structural distortions caused by bulky stabilizing groups,t om ore clearly probe their potential pacceptor properties when bonded to metal centers.However, the combination of ad ouble bond and lone pairs renders dipnictenes increasingly reactive;HNNH is only found in the solid state when coordinated to metals, [2,3] and only three metal-HPPH complexes are known. [4] Conspicuously,t here are no examples of structurally authenticated HEEH (E = As, Sb,a nd Bi)i na ny charge state and therefore little is known about these parent molecules.
Where HAsAsH is concerned, generation in electrochemical-IR and microwave plasma-IR experiments has been proposed, [5] but the assignments,w hilst consistent with As À Hb onds,w ere not conclusive regarding the precise nature of these transient, surface-absorbed hydrides.I nt he routine condensed phase,H AsAsH was first proposed as areaction product in 1810 by Davy [6] and ayear later by Gay-Lussac and ThØnard. [7] In 1924, Weeks and Druce claimed that the action of stannous chloride on arsenic trichloride in the presence of hydrochloric acid produced brown, amorphous solids formulated as HAsAsH. [8] However,i n1 957, Jolly,Anderson, and Beltrami showed that these products are ostensibly arsenic with adsorbed sub-stoichiometric arsenic hydrides. [9] Thus,HAsAsH has eluded capture,a nd has most likely never actually been made,w hich probably reflects the absence of synthetic methods to construct HAsAsH and prevent subsequent decomposition in the absence of bulky arsenic-bound stabilizing groups.H ere,m ore than two centuries after it was first proposed, we report the synthesis and structural authentication of HAsAsH in ac rystalline diuranium(IV) complex.
To prepare 2,the KAsH 2 reagent has to be finely ground, otherwise intractable product mixtures are obtained. [15] However,o no ne occasion as mall crop of dark brown crystals of [{U(Tren TIPS )} 2 (m-h 2 :h 2 -As 2 H 2 )] (3)was obtained in about 1% yield. Deducing that 3 is likely formed due to sluggish KAsH 2 reactivity when not ground, therefore resulting in localized excesses of KAsH 2 deprotonating 2 when formed, we repeated the reaction with different ratios of 1:KAsH 2 varying from 1:1.1 to 1:2w here KAsH 2 was ground. We determined 1:1.4 to be the optimal ratio,w hich reproducibly affords 3 in circa 50 %c rude yield, as determined by 1 HNMR spectroscopy using 2,4,6-Bu t 3 C 6 H 3 as an internal standard (Scheme 1). [18] Recrystallization reproducibly affords 2 in 8% pure crystalline yield, reflecting the instability of HAsAsH.
Whilst it seems certain that the KAsH 2 deprotonates 2 to give [U(Tren TIPS )(AsHK)],and presumably AsH 3 ,insitu, it is unclear how it promotes oxidative homocoupling to give 3,as all attempts to identify by-products have been inconclusive. However,w en ote that AsH 3 has precedent for forming MAsH 2 and H 2 (M = Na or K) from M-containing substrates, and that KAsH 2 is known to decompose to "KAs" and H 2 , which might provide the redox path to 3. [16] We also note that although 3 is obtained most conveniently by treatment of 1 with excess KAsH 2 ,rather than isolating 2 and reacting with further KAsH 2 ,the latter method is effective,suggesting that the HAsAsH unit may be formed by coupling and subsequently remains isolated between two uranium centers.W e investigated alternative methods of producing 3,bypreparing [U(Tren TIPS )(AsHK)] and treating it with oxidants to effect homocoupling;h owever, adding stoichiometric iodine,l ead-(II) iodide,T EMPO,p yridine-N-oxide,4 -morpholine-Noxide,t rimethylamine-N-oxide,s ilver tetraphenylborate, and copper(I) iodide all gave intractable products.W eh ave also separately refluxed and photolyzed 2 to see if dihydrogen elimination to give 3 occurs,b ut only decomposition occurs under these conditions.T hese observations underscore the fragile nature of HAsAsH and hence why it was elusive. Although 3 is obtained in poor crystalline yield or moderate yield in crude form, the synthesis is reproducible.
On one occasion, after isolating 3,asmall crop of light brown crystals deposited from the mother liquor in less than 1% yield. These were identified as [{U(Tren TIPS )} 2 (m-h 2 :h 2 -As 2 )] (4). [18] Although the low yield of 4 has prevented its characterization, its structure serves to support the formulation of 3 by virtue of their metrical differences,a nd underscores the complex dehydrogenative chemistry that operates for these redox active molecules with polar bonds. [16] Once 3 is crystalline,ithas very low solubility in non-polar solvents and it decomposes in polar solvents,soreliable UV/ Vis/NIR spectra could not be obtained. The 1 HNMR spectrum exhibits two very broad resonances at circa 5.3 (Pr i )a nd circa 6.2 ppm (CH 2 ); we attribute this to the dinuclear nature of 3 and the absence of AsÀHr esonances to their close proximity to the paramagnetic uranium ions.
TheA TR-IR spectrum of 3 exhibits one weak AsÀH absorption at 2029 cm À1 (2052 and 2031 cm À1 for 2), [15] and this compares to As À Ha bsorbances at 2040 and 2000 cm À1 assigned as HAsAsH generated in situ deposited on GaAs surfaces, [5a,b] but is significantly different to AsÀHstretches of 2306 and 2298 calculated for gas-phase HAsAsH. [19] An analytical frequencyc alculation predicts symmetric and asymmetric AsÀHs tretches in the IR spectrum of 3 at 2049 and 2028 cm À1 ,r espectively,w hich for the latter compares well to the experimentally observed value.The As À Hstretch at 2029 cm À1 can thus be assigned as the asymmetric stretching mode,because due to selection rules the symmetric stretch cannot be IR active as 3 exhibits an inversion center (see below). Thes ymmetric stretch should be observable in the Raman spectrum of 3,b ut samples of 3 decompose in the beam, or the inherently weak As À Hs tretch cannot be observed at low-/mid-power settings or in dilute samples,s o this data remains unobtainable.T he ATR-IR data for 3 rule out the presence of the Z isomer because,lacking an inversion center, it would exhibit both symmetric and asymmetric AsÀ Hs tretches,w hich is not observed experimentally.T o examine this aspect further we prepared [{U(Tren TIPS )} 2 (mh 2 :h 2 -As 2 D 2 )] (3D), using previously unknown KAsD 2 . [18] As anticipated the ATR-IR spectrum of 3D does not exhibit the absorbance at 2029 cm À1 ,b ut the AsÀDs tretch could not be observed because from reduced-mass considerations this absorbance falls in the fingerprint region where as trong and broad absorbance (1410-1490 cm À1 )r esides.
Them olecular structure of 3 is shown in Figure 1; [18] the salient feature is the presence of HAsAsH bridging two [U(Tren TIPS )] units.I nt he solid state, 3 crystallizes over an inversion center between the two arsenic ions.Although this

Angewandte
Chemie is consistent with the presence of the E isomer, in the crystal that was examined the hydride is disordered over two sites,so the presence of the Z isomer,a so pposed to two averaged E isomers of opposite "hands", could not initially be discounted. However,the ATR-IR data rule out the presence of the Z isomer,which is consistent with the greater prevalence of E dipnictenes compared to the corresponding Z isomers. [1,20] Theu ranium-amide and uranium-amine bond lengths in 3 are typical of such distances. [21] TheU À As distances of 3.1203(7) and 3.1273 (7) in3 are longer than the sum of the single bond covalent radii for uranium and arsenic (2.91 ), [22] but are only slightly longer than the formal UÀAs covalent s-bond in 2 (3.004(4) ). [15] TheAs ÀAs bond length in 3 of 2.4102 (13) i sc onsistent with as ingle rather than double bond, [23] the latter of which tends to be about 2.2 , [1] and rules out the presence of an (As 2 )unit that when trapped between two transition metals exhibits As À As bond lengths of circa 2.2-2.3 ( see 4 below). [17, 23a, 24] When diarsenes with sterically demanding substituents are bonded to transition metals the AsÀAs bond tends to lengthen as aresult of backbonding,f or example to 2.365 i n[ (CO) 4 Fe(h 2 -As 2 Ph 2 )], [25] which suggests as ignificant uranium to diarsene backbonding-type interaction in 3 (see below), which would also be consistent with ad iuranium(IV) formulation. Further support for the formulation of 3 comes from the crystal structure of 4, [18] which crystallizes in adifferent crystal habit to 3.In4 the As=As distance of 2.2568 (14) isshorter than the analogous distance in 3 and is characteristic of As 2 . [23a] Also,t he U À As distances are shorter than in 3 at 3.0357(7) and 3.0497(8) that is consistent with the high charge load of As 2 .
Theassignment of uranium(IV) ions in 3,suggested by the solid-state metrical data, is also supported by magnetic measurements (Figure 2). Ap owdered sample of 3 exhibits am agnetic moment of 4.0 m B at 298 K, which decreases monotonously to amoment of 1.2 m B at 2Kand tends to zero as would be expected for uranium(IV), which at low temperature is am agnetic singlet with residual temperature-independent paramagnetism. Themagnetic moment per uranium ion in 3 at 298 K(2.7 m B )islower than the theoretical value of 3.58 m B for uranium(IV), but this is common for uranium-(IV). [26] To probe the nature of the U À As interactions in 3,w e calculated the electronic structure of the full model using density functional theory (DFT). [18] With the Z isomer experimentally ruled out, our discussion focuses on the E isomer. [27] Thegeometry-optimized structure agrees well with experiment, with bond lengths and angles predicted to within 0.05 a nd 28 8,r espectively;t he DFT model can thus be considered to present aq ualitative description of the electronic structure of 3.T he calculated MDC q charges and MDC m spin densities at each uranium average + 3.20 and À2.31, respectively,w hich suggests modest net donation of electron density to uranium(IV) from the ligands. [28] The arsenic MDC q charges average À1.12, which is consistent with the HAsAsH fragment carrying af ormal À2c harge overall, which is ar equirement of being bonded to two uranium(IV) [U(Tren TIPS )] + cations for charge neutrality.T he calculated AsÀAs Mayer bond order is 0.97, consistent with the AsÀAs single bond suggested by the X-ray diffraction data, whereas the U À As Mayer bond orders average 0.34 and suggest polarized interactions;for comparison, calculated As À H, U À N amide ,a nd U À N amine Mayer bond orders average 0.92, 0.82, and 0.22, respectively.
Thet op four most energetic electrons in 3 are of essentially pure,n on-bonding 5f character and constitute the top four quasi-degenerate (0.05 eV spread) a-spin highest occupied molecular orbitals (HOMOs), which are each singularly occupied. HOMOÀ4inthe a-and b-spin manifolds comprise the principal UÀAs interactions,a nd represent formal back-bonding from uranium to the p*-orbital of HAsAsH ( Figure 3). As nitrogen-based orbital coefficients intrude into HOMOÀ4o f3,n atural bond orbital (NBO) analyses were performed to obtain alocalized, clear description of the U À As interactions.N BO analyses reveal highly polarized UÀAs interactions that comprise an average of  91.3 %A sa nd 8.7 %Ucharacter,a nd these interactions are also found at the second-order level of perturbation. The arsenic components are essentially pure 4p character,whereas the uranium contributions are 53.2 %5 fa nd 45.3 %6 d character with no meaningful 7s or 7p contributions.
When considering the bonding of the HAsAsH fragment to two [U(Tren TIPS )] fragments,a napriori treatment yields two bonding extremes.Onone hand, HAsAsH could donate electron density purely from its filled p-orbital to vacant orbitals on each uranium center,which would be assigned as formally trivalent, with no back-bonding and thus retain the As=As double bond. Alternatively,e ach uranium could formally engage in ab ack-bond-type interaction into the vacant p*-orbital of HAsAsH, leading to reduction to give two uranium(IV) centers and aHAsAsH dianion with an As À As single bond. Interestingly,all attempts to computationally model 3 as diuranium(III) with af ormally neutral HAsAsH met with failure or converged instead to ad iuranium(IV) HAsAsH-dianion spin-state formulation. Previous calculations on HAsAsH have predicted it to be an excellent pacceptor ligand, [19,29] and the combined characterization data for 3 clearly support the latter bonding picture,t hat is,i n3 HAsAsH can be considered as ad iarsane-1,2-diide resulting from extensive electron transfer from uranium.
In summary,b yc areful control of reaction conditions we have been able to isolate the highly reactive HAsAsH unit between two sterically demanding uranium fragments,t hus confirming the synthesis of amolecule first proposed over two centuries ago.The characterization data for 3 uniformly point to the HAsAsH unit being formally reduced to its dianionc form by the two uranium centers,i n-line with the predicted excellent acceptor properties of HAsAsH. This study highlights the capacity of an f-block element, uranium, to bond in amanner that is reminiscent of d-block metals,though at one bonding extreme with highly polarized U À As bonding interactions.C omplex 3 is an isoelectronic model for a palkene complex of uranium, which is aclass of complex yet to be realized under any experimental conditions. [30]