David R.
Armstrong
,
Alan R.
Kennedy
,
Robert E.
Mulvey
* and
Stuart D.
Robertson
*
WestCHEM, Department of Pure and Applied Chemistry, University of Strathclyde, Glasgow, G1 1XL, UK. E-mail: r.e.mulvey@strath.ac.uk; stuart.d.robertson@strath.ac.uk
First published on 20th December 2012
Designed to extend the synthetically important alkali-metal diisopropylamide [NiPr2; DA] class of compounds, the first example of a hetero-alkali-metallic complex of DA has been prepared as a partial TMEDA solvate. Revealed by an X-ray crystallographic study, its structure exists as a discrete lithium-rich trinuclear Li2KN3 heterocycle, with TMEDA only solvating the largest of the alkali-metals, with the two-coordinate lithium atoms being close to linearity [161.9(2)°]. A variety of NMR spectroscopic studies, including variable temperature and DOSY NMR experiments, suggests that this new form of LDA maintains its integrity in non-polar hydrocarbon solution. This complex thus represents a rare example of a KDA molecule which is soluble in non-polar medium without the need for excessive amounts of solubilizing Lewis donor being added.
![]() | ||
Fig. 1 Pictorial representation of known molecular structures of hetero-alkali-metallic complexes of the utility amides HMDS (left) and TMP (right). |
LDA is known to exist as a helical polymer in the solid state with near linear N–Li–N units, a turn of the helix consisting of four units of alternating Li and N atoms (Fig. 2).9 The addition of THF breaks up this helix into a cyclodimer, with a central N2Li2 ring and each lithium atom solvated by a single donor molecule.10 This arrangement (N2Li2 ring with three coordinate lithium atoms) is repeated in replacing THF with TMEDA, but with the ditopic ligand acting as a monodentate (non-chelating) bridging linker between the dimeric subunits to give a different type of polymeric structure.11 Unsolvated or THF solvated molecular structures of the heavier alkali-metal diisopropylamides are currently unknown, however in the presence of TMEDA a discrete dimeric motif is witnessed for both Na12 and K13 congeners, with the TMEDA ligating in a bidentate manner to give a tetra-coordinate metal centre. The larger potassium centre also displays three agostic contacts each of less than 3 Å in length with the methyl fragments of the diisopropyl groups for an overall coordination number of seven.
![]() | ||
Fig. 2 Pictorial representation of known molecular structures of homometallic alkali-metal diisopropylamide complexes. |
![]() | (1) |
Cooling this solution to −32 °C overnight afforded a crop of X-ray quality colourless crystals in a 58% yield, their molecular structure (Fig. 3a) being determined via a single crystal diffraction experiment. This revealed 1 to exist as a discrete spirocyclic trinuclear molecule, containing a crystallographic 2-fold axis which passes through the potassium atom, the nitrogen atom of the unique DA anion (N3) and the centre of the CH2–CH2 bond of TMEDA. 1 contains a central virtually planar LiNLiNKN ring (RMS deviation from planarity = 0.0297 Å) lying approximately perpendicular [70.56(4)°] to a KNCCN ring formed by the chelate coordination of TMEDA to potassium, with lithium atoms unsolvated and thus two-coordinate. This lack of solvation at lithium is perhaps unsurprising given Collum's observation that TMEDA-solvated LDA desolvates at ambient temperature even in the absence of other donor ligands.11 Furthermore, X-ray quality crystals of polymeric unsolvated LDA were in fact obtained from a TMEDA-containing solution, albeit one having a substoichiometric quantity of the diamine.9 The overall structure of 1 can essentially be thought of as a dinuclear fragment of the LDA polymer which has trapped a monomeric fragment of KDA·TMEDA. Note no β-hydride elimination from the NiPr2 anion was witnessed unlike that recorded previously in heterometallic Mg/AM (AM = Na, K) complexes of this amide which had been refluxed in toluene/heptane.14 The optimized structure was modelled via DFT calculations and is shown in Fig. 3b for comparison, with the computed bond parameters displayed in Table 1.
![]() | ||
Fig. 3 (a) Molecular structure of complex 1 showing key atom labels with hydrogen atoms and minor disordered components of diisopropylamido anions omitted for clarity and ellipsoids drawn at the 50% probability level. Symmetry operation to generate equivalent atoms marked ′ = −x, y, 1/2 − z. (b) Computed structure of same complex, 1calccalc. |
1calccalc | 1 | 2 | |
---|---|---|---|
K1–N1 | 3.056 | 2.907(1) | 3.016(1) |
K1–N2 | 2.772 | 2.869(1) | 2.890(1) |
Li1–N2 | 1.939 | 1.912(3) | 1.950(2) |
Li1–N3 | 1.989 | 1.972(3) | 1.987(2) |
N1–K1–N1′ | 62.5 | 64.22(4) | 59.54(3) |
N1–K1–N2 | 109.8 | 111.03(4) | 119.54(3) |
N1–K1–N2′ | 129.9 | 130.19(4) | 122.62(3) |
N2–K1–N2′ | 110.0 | 107.30(4) | 107.05(3) |
K1–N2–Li1 | 95.5 | 98.4(1) | 97.9 (1) |
N2–Li1–N3 | 163.4 | 161.9(2) | 164.2(1) |
Li1–N3–Li1′ | 92.1 | 91.7(1) | 88.7(1) |
The modelled structure 1calccalc shows reasonably close agreement to 1 in the bond angles of the spirocyclic ring. However, the computed bond distances vary noticeably from those seen in the molecular structure, in particular 1calccalc predicts a shorter K–NDA distance and concomitantly longer Li–NDA distances. A subsequent effect of the shorter K–NDA distances is the prediction that TMEDA will not be able to gain as close proximity to the potassium centre, with the predicted value of 3.056 Å almost 0.15 Å longer than the experimentally determined value.
On comparing the bond parameters of complex 1 with those of its TMP analogue 2, it is noticeable that the six-membered ring of 1 has marginally shorter metal–nitrogen bond lengths. This can almost certainly be explained by the reduced steric strain imposed on this ring by the diisopropylamide anions versus the more sterically demanding TMP anions in 2. This is also manifested in the K–NTMEDA bond lengths, with the less bulky DA groups allowing the bidentate donor to approach potassium more closely in 1 [2.907(1) Å versus 3.016(1) Å in 2]. This strain imposed by the TMP anions in 2 helps explain the previously witnessed opening of the six-membered ring on substituting TMEDA with the tridentate donor PMDETA (eqn (2)).7 We note here that an analogous crystalline PMDETA solvated derivative of 1 could not be obtained in this study despite numerous attempts. Other common polydentate donors such as diglyme, O(CH2CH2NMe2)2 and Me6TREN [N(CH2CH2NMe2)3] were also examined but failed to provide an isolable product.
![]() | (2) |
Complex 1 was also investigated in the solution state for comparison with 2, which is believed to undergo an equilibrium between the trinuclear species and a dinuclear (Li/K) species along with homometallic LiTMP according to eqn (3),7 probably due to the strain imposed on the ring by the bulky secondary amide molecules.
Li2K(TMP)3·TMEDA ⇌ LiK(TMP)2·TMEDA + Li(TMP) | (3) |
With respect to its THF solvate, much less has been reported regarding the solution behaviour of diisopropylamide complexes in non-polar solvents. Collum showed via6Li NMR spectroscopy that donor-free LDA in hexane exists as a mixture of 3–5 cyclic oligomers.15 Williard noted that the solubility of LDA in warm hydrocarbon solvents was dependent on the method of its preparation, with that prepared by reaction of lithium metal with DA(H) in ether in the presence of styrene10 being more soluble than that prepared from nBuLi and DA(H) in pentane.16 Similarly it has recently come to light that the production method of LDA is important with regards to its reactivity, since the presence of minute quantities of LiCl (on the parts per million scale) can profoundly influence the rate or regioselectivity of a reaction.17 We note at this juncture that in our hands, LDA generated in situ in hexane from nBuLi and DA(H) can be stirred indefinitely without precipitating yet once precipitated it is considerably more difficult to re-dissolve.
The 1H NMR spectrum of 1 in deuterated cyclohexane solution at ambient temperature displayed only one principal set of resonances corresponding to the diisopropylamido anions at 3.06 and 1.02 ppm [Fig. 4, cf. DA(H) 2.86 and 0.96 ppm], with only a minute amount of DA(H) (presumably from unavoidable hydrolysis), suggesting that the two distinct DA environments within the molecular structure of the crystalline species are equivalent in NMR terms, although some form of rapid equilibrium which is faster than the NMR timescale cannot be unequivocally ruled out. A comparison with the corresponding 1H NMR spectra of LDA both in the presence and absence of TMEDA suggested that complex 1 was not cleaving into its homometallic constituent parts (Fig. 4). It is highly unlikely that a stable TMEDA solvated LDA species would arise from cleavage of 1 given the previous observations of Collum et al. of the poor affinity of LDA for substoichiometric quantities of this donor. As shown in Fig. 4 and in accord with previous research, there appears to be more than one DA− component in a hydrocarbon solution of unsolvated LDA.
![]() | ||
Fig. 4 Part of the 1H (top) and the full 7Li NMR (bottom) spectra of various NiPr2 containing species relevant to this study. A comparison of the 7Li NMR spectra of 1 at both high and low concentration confirmed there was no concentration dependencies. |
The 1H NMR resonances corresponding to TMEDA appear at 2.29 and 2.19 ppm for the methylene and methyl groups respectively, very close to those of free TMEDA which appear at 2.30 and 2.14 ppm, suggesting that this bidentate donor is perhaps weakly bound or not bound at all to the potassium centre given that it is typical for TMEDA resonances to be considerably shielded when bound to an alkali-metal amide moiety.18 Furthermore, it has previously been noted that the relative positioning of the two TMEDA resonances is inverted upon coordinating to a metal12 yet that is not the case here. We do note however that while such inversion is prevalent in C6D6 solutions, far less is known on alkane solutions and thus it is perhaps premature to draw any firm conclusions based on such a small sample set.
The 7Li NMR spectrum of 1, unlike that of 2, displayed only a single resonance at 2.90 ppm in non-polar cyclohexane solution. Again, a comparison with LDA (both with and without TMEDA present) suggests that 1 is not simply extruding a homometallic LDA moiety in solution. Collum, Williard and co-workers have shown a TMEDA concentration dependence on the 6Li chemical shift of LDA in hexane solution, ranging between almost 3 ppm (for no TMEDA) to slightly greater than 2 ppm (for 8.0 M TMEDA).11 Corroborating our 1H NMR spectra (vide supra), unsolvated LDA contains more than one lithium environment, in agreement with the findings of Collum et al. A low temperature 7Li NMR spectrum of 1 (in hexane solution) was then recorded at 210 K and compared with the corresponding room temperature spectrum (Fig. 5). The latter spectrum was similar to that collected in C6D12, namely a singlet with a very small shoulder on the upfield side. Lower temperature affected the chemical shift of the resonance (and resolved the shoulder marginally better) but despite some broadening, the resonance did not split.
An interesting feature worthy of mentioning here is the excellent solubility of 1 in this aliphatic hydrocarbon solvent, which is in contrast to the poor solubility of homometallic KDA complexes. This hints at the prospect of utilising 1 as a soluble source of KDA in organic reactions which are to be carried out in such a medium.
This heterometallic species was also studied in solution via DOSY spectroscopy (Fig. 6) in an attempt to glean more information on its solution state constitution.19
![]() | ||
Fig. 6 DOSY NMR spectrum of complex 1 in C6D12 at 27 °C in the presence of the standards of decreasing molecular weight tetraphenylnaphthalene (TPhN), phenylnaphthalene (PhN) and tetramethylsilane (TMS). |
This technique, which is gaining in popularity for identifying solution structures of both homo-20 and heterometallic alkali-metal complexes,21 can separate components according to their diffusion coefficient (and therefore indirectly to their size – akin to NMR chromatography). This revealed that the principal diisopropylamido-anion containing species has a molecular weight (MWDOSY) of approximately 382, noticeably less than the molecular weight of the crystalline sample (MW = 469.7). However, as can clearly be seen in Fig. 6, the TMEDA component (MWDOSY = 242) does not have the same molecular weight as the DA− component. If TMEDA were completely dissociated from 1 in solution it should have a MWDOSY equal to its true MW (116). This difference appears consistent with a rapid coordination–decoordination event occurring in solution with MWDOSY giving a value intermediate between that of free and bound TMEDA and explains why MWDOSY of the DA anions is intermediate between that of solvated and unsolvated Li2K(DA)3 (eqn (4)). This phenomenon has been observed previously in the solution behaviour of the related solvated amide species [Li(TMP)·THF]2.20c
![]() | (4) |
We note here that varying the Li:
K ratio (for example 1
:
1, or with an excess of K) within the reaction mixture did not result in a different complex being prepared. To probe this observation further, we compared the formally 1
:
1 reaction with the stoichiometrically precise 2
:
1 reaction (matching that in the formula of 1) via DFT calculations according to eqn (5) and (6) respectively. In each case we commenced with a cyclotetramer of LDA as the lithium starting material22 and dimeric TMEDA-solvated KDA13 as the potassium starting material/Lewis donor source.
![]() | (5) |
![]() | (6) |
These calculations supported our assertion that the lithium rich constitution 1 is the energetically preferred product as the 1:
1 reaction yielding a dinuclear product was calculated as being moderately endothermic (by +0.54 kcal mol−1) while the 2
:
1 reaction yielding a trinuclear product was exothermic by a more substantial value of −3.16 kcal mol−1.
NMR spectra were collected on a Bruker AV400 MHz spectrometer operating at 400.13 MHz for 1H, 155.47 MHz for 7Li and 100.62 MHz for 13C. All 13C NMR spectra were proton decoupled.
1H NMR (D12-cyclohexane, 300 K): 3.06 (sept, 6H, 2JH–H = 6 Hz, CH(CH3)2), 2.29 (s, 4H, TMEDA CH2), 2.19 (s, 12H, TMEDA CH3), 1.02 (d, 36H, 2JH–H = 6 Hz, CH(CH3)2) ppm.
13C NMR (D12-cyclohexane, 300 K): 58.4 (TMEDA CH2), 51.2 (CH(CH3)2), 46.2 (TMEDA CH3), 28.2 (CH(CH3)2) ppm.
7Li NMR (D12-cyclohexane, 300 K): 2.90 ppm.
Crystallographic data were collected at 123(2) K on an Oxford Diffraction Instrument using MoKα (λ = 0.71073 Å) radiation.24 Structure was solved using SHELXS-97 and refined to convergence against F2 against all independent reflections by the full-matrix least-squares method using the SHELXL-97 program.25 The isopropyl arms of the unique DA anion were modelled as being disordered over two sites in a 81:
19 ratio, as was one of the methyl arms of the other DA anion. CCDC 901793 contains the supplementary crystallographic data for this paper.
Theoretical calculations were carried out using the Gaussian 03 package.26 Geometry optimization was undertaken at the HF/6-31G*27 level, followed by a frequency analysis. The geometry was then refined by further calculation at the B3LYP28/6-311G**29 level. The structural parameters reported were taken from the DFT calculations, whereas the total energy abstracted from the DFT calculations was adjusted by inclusion of the zero-point energy value from the HF calculation modified by the factor 0.91.
Footnote |
† CCDC 901793. For crystallographic data in CIF or other electronic format see DOI: 10.1039/c2dt32204h |
This journal is © The Royal Society of Chemistry 2013 |