A Cyaphide Transfer Reagent

The cyanide ion plays a key role in a number of industrially relevant chemical processes, such as the extraction of gold and silver from low grade ores. Metal cyanide compounds were arguably some of the earliest coordination complexes studied and can be traced back to the serendipitous discovery of Prussian blue by Diesbach in 1706. By contrast, heavier cyanide analogues, such as the cyaphide ion, C≡P–, are virtually unexplored despite the enormous potential of such ions as ligands in coordination compounds and extended solids. This is ultimately due to the lack of a suitable synthesis of cyaphide salts. Herein we report the synthesis and isolation of several magnesium–cyaphido complexes by reduction of iPr3SiOCP with a magnesium(I) reagent. By analogy with Grignard reagents, these compounds can be used for the incorporation of the cyaphide ion into the coordination sphere of metals using a simple salt-metathesis protocol.


■ INTRODUCTION
Along with the halide ions, cyanide (CN − ) is one of the most ubiquitous anions in chemistry. Its salts are routinely used in industrial applications including bulk chemical synthesis, electroplating, metallurgy, tanning, manufacturing of paper and plastics, photography, and as fumigants and insecticides. 1 In organic chemistry, it is an important functional group in nitriles (R−CN) and isonitriles (R−NC), many of which are produced on an industrial scale (e.g., adiponitrile, NC(CH 2 ) 4 CN, which is used to produce nylon). 2,3 By contrast, and despite the valence isoelectronic relationship between nitrogen and phosphorus, stable phosphorus-containing analogues of cyanides are much rarer; nitrile analogues, socalled phosphaalkynes (R−CP), have been known for almost 40 years and are highly reactive compounds due to the weak nature of C−P π bonds. 4,5 Isocyanide analogues (R− PC) remain unknown. 6 Unlike cyanide, which forms a multitude of stable salts, the cyaphide ion, CP − , cannot be obtained as a simple A(CP) or Ae(CP) 2 salt (where A = alkali and Ae = alkaline-earth metal). To date, the CP − ion has only ever been isolated in the coordination sphere of three metals (platinum, ruthenium, and uranium; e.g., trans-[Ru-(dppe) 2 (H)(CP)] where dppe = bis(1,2-diphenylphosphinoethane) 7−12 and an electrophilic borane. 13 While these studies demonstrate that the ion is accessible, the resulting compounds are of limited synthetic utility due to their inertness. Alkali/alkaline-earth metal salts of the cyaphide ion are a more attractive target insomuch as they should allow for the incorporation of CP − into novel molecules and solids, making use of salt metathesis protocols, a procedure that is well established for cyanides. 14 Herein we show that well-defined alkaline-earth complexes of the cyaphide ion are readily accessible and can be used as anion transfer reagents for the synthesis of novel cyaphido complexes.

■ RESULTS AND DISCUSSION
The two-electron chemical reduction of the 2-phosphaethynolate ion, PCO − , 15 to afford a uranium cyaphide complex was recently demonstrated by Meyer. 11 We reasoned that functionalization of PCO − to afford a phosphaethynolato compound (R−O−CP) would facilitate this reduction step, allowing for the straightforward generation of the cyaphide ion. A major limitation is that oxygen-functionalized phosphaethynolato compounds are rare and largely ionic in character. 11,16−21 To date, only one species with significant covalent character has been structurally authenticated. 22 In situ silylation of the [Na(dioxane) x ]PCO with tris(isopropyl)silyl trifluoromethanesulfonate in nonpolar aromatic solvents (benzene or toluene) favors silylation at the oxygen atom to afford the kinetic product i Pr 3 SiOCP (Figure 1), which ultimately rearranges to give the κ-P isomer. 23 Reduction of the former species using Jones' magnesium(I) reagent [Mg-( Dipp NacNac)] 2 24,25 cleanly affords an equimolar mixture of [Mg( Dipp NacNac)(CP)(dioxane)] (1) and [Mg( Dipp NacNac)-(OSi i Pr 3 )(dioxane)] (2) where Dipp NacNac = CH{C(CH 3 )N-(Dipp)} 2 and Dipp = 2, 6-di(isopropyl)phenyl ( Figure 1). Density functional theory (DFT) calculations predicted this reaction to be exergonic (at 298.15 K) by 52.2 kcal mol −1 with an overall energy barrier of 12.5 kcal mol −1 . The reaction proceeds via an unobserved dimagnesiated intermediate (a metalla-phosphaalkene) which rearranges by siloxyl group transfer (energy barrier of 4.4 kcal mol −1 ) to afford 1 and 2 (see the Supporting Information for the full computational analysis). Cleavage of the C−O bond in the phosphaethynolate ion necessitates a highly oxophilic two-electron reductant and significant steric protection (for example, when the less sterically encumbered magnesium(I) dimer [Mg-( Mes NacNac)] 2 (Mes = mesityl) was employed, the analogous reaction gave rise to a mixture of products including cyaphide oligomers).
The magnesium−cyaphido complex (1) exhibits a resonance in its 31 P{ 1 H} NMR spectrum at 177.2 ppm and a diagnostic singlet resonance in the 1 H NMR spectrum corresponding to the Dipp NacNac γ-proton at 4.78 ppm. A doublet resonance corresponding to the cyaphide ligand was observed in the 13 C{ 1 H} NMR spectrum at 270.97 ppm ( 1 J C−P = 34.0 Hz). Fractional crystallization of the reaction mixture allowed for the isolation and structural determination of compounds 1 and 2.
In situ generated mixtures of 1 and 2 can be used to transfer the cyaphide ion to metal complexes (vide infra) via a salt metathesis protocol, in a manner reminiscent of Grignard reagents. 28 However, because of the similar solubility of 1 and 2 in common laboratory solvents, the isolation of compositionally pure samples of 1 is only possible in low yields (∼20%).
Thus, we sought strategies to modify the solubility of 1. Quantitative dioxane displacement was achieved by using THF-d 8 to form [Mg( Dipp NacNac)(CP)(THF-d 8 )] (3; see Scheme 1); however, this adduct is equally difficult to separate from the siloxymagnesium side-product. In addition, it was observed for both solvent adducts 1 and 3 that exposure to vacuum initiated decomposition of the target compounds, evidenced by broadening of NMR spectra ( Figure S11). We hypothesize that initial cleavage of the Mg−solvent interaction forms the base-free analogue [Mg( Dipp NacNac)(CP)] x (4), which then rapidly decomposes. Employing dioxane-free Na(OCP) during the generation of i Pr 3 SiOCP subsequently led to the specific formation of solvent-free analogue 4, evidenced by a 31 P{ 1 H} NMR singlet resonance at 246.7 ppm. The solubility of this desolvated analogue is sufficiently lower than 2 to facilitate efficient separation by precipitation; however, in the solid state 4 is unstable, decomposing rapidly once isolated (see the Supporting Information for further details). The structure of 4 is currently unknown, but the downfield shifted 31 P NMR resonance suggests it is oligomeric; the related solvent-free cyanido complex, [Mg( Dipp NacNac)-(CN)] 3 , is a cyclic trimer. 29 Addition of dioxane or THF-d 8 to solutions of 4 resulted in the formation of the corresponding solvated adducts 1 or 3, respectively.
Spectroscopically 5 and 6 do not differ greatly from compound 1 and exhibit comparable NMR shifts [e.g., 31 P{ 1 H} NMR: 162.9 (4); 174.9 ppm (5)]. Both NHC adducts were structurally authenticated by single-crystal X-ray diffraction ( Figure 2) and confirm the expected association of the NHC with the magnesium metal center. The Mg−CP (5: 2.166(2); 6: 2.144(3) Å) and C−P bonds (5: 1.550(2); 6: 1.531(3) Å) for both compounds are in line with those observed for 1. Interestingly, it was found that while the IMes carbene associated with the magnesium metal center in an "abnormal" fashion (i.e., through the alkenic backbone), the I i Pr carbene adduct coordinates as expected in the solid state, an observation we put down to the increased steric bulk of IMes. 30 However, in solution the I i Pr moiety of 6 fluctuates between normal and abnormal coordination. The 31 P{ 1 H} NMR spectrum of 6 at room temperature features a particularly broad singlet signal (υ 1/2 ≈ 224 Hz) which when   Table S6). Calculations further indicate that the difference in energy between these two isomers is negligible. 1 does not react with IMes Me and I i Pr Me , analogous NHCs featuring methylated backbones where abnormal coordination is blocked.
Compound 5 cocrystallizes with one stoichiometric equivalent of IMes as a cocrystal, which is also seen in the 1 H NMR spectrum of the bulk product. Probing further, we calculated no energy payoff for the displacement of the dioxane by IMes (1−3 kcal mol −1 , within error), with neither compound being thermodynamically favored between −100 and 100°C. The calculated thermodynamic data are indicative of an equilibrium. However, even when just one equivalent of IMes is added to a solution of 1, 5·IMes can be isolated from the reaction mixture by crystallization (albeit in lower yields). This indicates that the additional molecule of IMes present in the lattice is critical to isolate crystalline samples of 5.
Given the ionic nature of the Mg−CP bond in 1, we reasoned that salt metathesis reactions between this species and main-group or metal halides would allow for cyaphide group transfer. This hypothesis was probed by addition of chlorotrimethylsilane to a C 6 D 6 solution of 1 formed in situ. The 31 P{ 1 H} NMR spectrum showed a single resonance at 97.9 ppm, corresponding to the known phosphaalkyne Me 3 SiCP. 31 This clean, quantitative transfer of the CP − ion is, to our knowledge, the first instance of such reactivity. Encouraged by this finding, we targeted novel cyaphide−metal complexes. It is worth noting at this stage that comparable cyaphide transfer reactions are also possible with compounds 5 and 6.
Moving to heavier group 14 elements, addition of [Ge-( Dipp NacNac)Cl] to an in situ generated mixture of 1 and 2 leads to rapid consumption of 1 and a new 31 P{ 1 H} NMR signal at 106.4 ppm (Scheme 2). A new singlet signal in the corresponding 1 H NMR spectrum at 5.08 ppm, within the characteristic region for γ-H protons, indicates a new Dipp Nac-Nac environment. Also evident was that siloxy byproduct 2 remained unreacted ( Figure S20). Over the course of a few hours, this deep-red solution, presumably containing [Ge-( Dipp NacNac)(CP)] (7), changed to a dark-green color, and NMR spectroscopy showed decomposition of the metal− cyaphide complex into multiple phosphorus-containing compounds, a process which was accelerated by any physical manipulation.
Reaction of 1 with [Sn( Dipp NacNac)Cl] led to the formation of [Sn( Dipp NacNac)(CP)] (8), which can be isolated by fractional crystallization. Cyaphide transfer was first indicated by 31 P{ 1 H} NMR spectroscopy which revealed a new resonance with P−Sn coupling satellites at 122.4 ppm ( 2 J P−Sn = 69. 8 Hz) and confirmed in the solid-state structure ( Figure  3). The CP bond is intact and comparable (1.542(4) Å) to those of 1, 5, and 6. At 2.216(4) Å, the Sn−C bond is relatively long, and the Sn−CP unit is practically linear (179.16°). No resonance could be found in the 13 C NMR spectrum of 8 corresponding to the cyaphide group which we attribute to broadening due to coupling to two adjacent NMR active nuclei. A weak band was observed in the IR spectrum of 8 at 1321 cm −1 , which is consistent with the predicted value (1327   (15) Å). 36 The C−P bond distance in 9 is 1.552(6) Å, which is comparable to the other cyaphide complexes discussed thus far (cf. 1.553(2) Å for 1). The IR spectrum of compound 9 reveals a band at 1342 cm −1 , which is higher than the value reported for trans-[Ru(dppe) 2 (H)(CP)] (1229 cm −1 ) and thus indicative of little π-back-bonding; however, it is worth noting that this vibrational mode is heavily coupled with the Au−C carbene stretch on account of the linear coordination geometry of 9.
Finally, in an effort to illustrate the broad synthetic utility of the cyaphide transfer reagent 1, we sought to synthesize the first example of a 3d metal−cyaphide complex. For this purpose we reacted a mixture of 1 and 2 with [( Dipp PDI)CoCl] ( Dipp PDI = 2,6-{2,6-i Pr 2 C 6 H 3 NCMe} 2 C 5 H 3 N). The reaction results in the clean quantitative formation of a new product, [( Dipp PDI)Co(CP)] (10), which exhibits a single resonance in its 31 P{ 1 H} NMR spectrum at 345.4 ppm. This is notably downfield from all of the other known cyaphide complexes, presumably due to a greater paramagnetic contribution to the NMR shielding constant (σ). Our calculations support this, predicting a δ value of 341 ppm for the 31 P NMR chemical shift. The 1 H and 13 C{ 1 H} NMR spectra for 10 are in line with the presence of a single Dipp PDI ligand. Notably, we were unable to observe the NMR resonance for the cyaphide ligand in the 13 C{ 1 H} NMR spectrum due to coupling with the 31 P and quadrupolar 59 Co nuclei. [(PDI)Co(R)] complexes have previously been attributed biradical character which explains anomalous 1 H NMR shifts, which we also see in [Co( Dipp PDI)-(CP)] (e.g., the imine NCCH 3 protons are upfield shifted to −0.23 ppm). 37 The IR spectrum of compound 10 reveals a band at 1306 cm −1 , in between the values observed for 9 (1342 cm −1 ) and trans-[Ru(dppe) 2 (H)(CP)] (1229 cm −1 ), suggesting a moderate degree of π-backbonding (as expected for a first row transition metal). Extremely air-and moisture-sensitive blue crystals of 10 were obtained from a concentrated toluene solution at −35°C. The structure of 10 was unequivocally confirmed by single-crystal X-ray diffraction (Figure 4) which reveals a square-planar cobalt(I) compound bonded to a cyaphide ligand. Generation of i Pr 3 SiOCP was adapted from the previously reported synthesis. 23 Na[PCO(dioxane) 5.6 ], 38 43 were synthesized according to previously reported synthetic procedures. Triisopropylsilyl trifluoromethanesulfonate (Sigma-Aldrich) and chlorotrimethylsilane (Sigma-Aldrich) were used as received. Hexane (hex; Sigma-Aldrich, HPLC grade) and toluene (Sigma-Aldrich, HPLC grade) were purified by using an MBraun SPS-800 solvent system. C 6 D 6 (Aldrich, 99.5%) was dried over CaH 2 and degassed prior to use. THF (Sigma-Aldrich, HPLC grade) and THF-d 8 (Sigma-Aldrich, 99.5%) were distilled over sodium/benzophenone. All dry solvents were stored under argon in gastight ampules. Additionally, solvents were stored over activated 3 Å molecular sieves.
Characterization Techniques. NMR spectra were acquired on Bruker AVIII 500 MHz ( 1 H 500 MHz, 13 C 126 MHz) and Bruker AVIII 400 MHz NMR spectrometers ( 31 P 162 MHz) at 295 K unless otherwise stated. 1 H and 13 C NMR spectra were referenced to residual protic solvent resonance ( 1 H NMR C 6 D 6 : δ = 7.16 ppm; 13 C NMR C 6 D 6 : δ = 188.06 ppm). 31 5.6 ]PCO (167 mg, 0.29 mmol). The resulting suspension was stirred for 4 h to generate i Pr 3 SiOCP, with occasional washing of the walls of the vial to ensure complete consumption of the starting materials. The resulting mixture was Journal of the American Chemical Society pubs.acs.org/JACS Article filtered through a glass paper filter, and the solids were washed with a small amount of toluene. [Mg( Dipp NacNac)] 2 (195 mg, 0.22 mmol) was added as a solid to the resulting yellow solution, causing it to darken to orange. Reaction completion was confirmed by 31 P{ 1 H} NMR spectroscopy. The solution can be used as an in situ supply of 1 with an equimolar amount of 2 also present; a representative 1 H NMR of such a solution can be found in Figure S1. Concentration of the solution (taking care to avoid evaporation to dryness) afforded a red oil. The residue was extracted into hexane (1 mL) and filtered.
Cooling the orange solution to −35°C overnight yielded 1 as yellow crystals suitable for X-ray diffraction (29 mg, 22% yield). Further concentrating the solution, or cooling for longer periods, resulted in mixtures of 1 and 2. Synthesis of [Mg( Dipp NacNac)(CP)(IMes)] (5). 1 (∼129.5 mg, 0.22 mmol) was generated in situ, as outlined above. To this, IMes (143.9 mg, 0.47 mmol, 2.1 equiv) was added as a solution in toluene (0.5 mL), and the resulting mixture was stirred overnight. The volatiles were removed, and the residue was extracted with pentane (5 mL). After filtration, the solution was concentrated to 1 mL and recrystallized in a single crop over several days at −35°C. The product was isolated as a pale orange solid (134.4 mg, 0.12 mmol, 54% yield). Anal. Calcd (%) for C 72 H 89 MgN 6 P: C, 79.06; H, 8.20; N, 7.68. Found: C, 78.10; H, 8.18; N, 7. 28 ). Cyaphide carbon not observed up to 350 ppm, likely due to broadening by the two adjacent nuclei. 31  Synthesis of [Co( Dipp PDI)(CP)] (10). 1 (∼65 mg, 0.11 mmol) was generated in situ as described above. To this toluene solution was added [Co( Dipp PDI)Cl] (33 mg, 0.06 mmol) as a solid with stirring. The reaction was stirred for 4 days. The solvent was removed, and the residue was washed with hexane (2 × 5 mL) and then extracted with toluene (1.5 mL). After filtration, the resulting deep blue solution was concentrated and stored at −35°C for 1 week to form blue crystals of 10. The supernatant solution was concentrated and stored at −35°C to yield a second crop of crystals (17 mg, 0.03 mmol, 51% yield). 1  The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.1c04417.