Magnesium Cyanide or Isocyanide?

Abstract Preference for the binding mode of the CN− ligand to Mg (Mg−CN vs. Mg−NC) is investigated. A monomeric Mg complex with a terminal CN ligand was prepared using the dipyrromethene ligand MesDPM which successfully blocks dimerization. While reaction of (MesDPM)MgN(SiMe3)2 with Me3SiCN gave the coordination complex (MesDPM)MgN(SiMe3)2⋅NCSiMe3, reaction with (MesDPM)Mg(nBu) led to (MesDPM)MgNC⋅(THF)2. A Mg−NC/Mg−CN ratio of ≈95:5 was established by crystal‐structure determination and DFT calculations. IR studies show absorbances for CN stretching at 2085 cm−1 (Mg−NC) and 2162 cm−1 (Mg−CN) as confirmed by 13C labeling. In solution and in the solid state, the CN ligand rotates within the pocket. The calculated isomerization barrier is only 12.0 kcal mol−1 and the 13C NMR signal for CN decoalesces at −85 °C (Mg−NC: 175.9 ppm, Mg−CN: 144.3 ppm). Experiment and theory both indicate that Mg complexes with the CN− ligand should not be named cyanides but are more properly defined as isocyanides.

First reports on metal-cyanide chemistry date back to the serendipitous discovery of Prussian Blue,F e 7 (CN) 18 ,i n1 704 by the Berlin painter Diesbach. [1] Like all transition-metal cyanide complexes,t his famous blue pigment is extremely stable and can only be destroyed by strong acids or carbon monoxide,aligand isoelectronic to cyanide and one of few that can compete with cyanide in the spectrochemical series for ligand strength. [2] Although the negative charge in C N À is mainly located at the N, [3] in the vast majority of transitionmetal cyanide complexes the cyanide ligand is C-bound. This strong preference for cyanide vs.isocyanide formation is due to the HOMO (lone pair) which has alarge coefficient at the C. [4] Although d!p*b ackbonding to the negatively charged CN À ligand is less prominent than that to neutral CO, it is not negligible and increases the donor strength at the N. This explains its strong tendencyt ob ridge metals,f orming inclusion compounds with al arge variety of applications. [1,5] In contrast to the wealth of highly stable transition metal complexes stands the chemistry of s-block metal cyanides. While the badly reputed alkali-metal cyanides are important bulk chemicals,v ery little is known about Group 2m etal cyanides. [6] Thef rustrations in first attempts to isolate Mg-(CN) 2 are clearly described by Fichter and Suter. [7] Magnesium metal reacts rapidly with a2 5% solution of hydrogen cyanide in water [Eq. (1)],however, isolation of Mg(CN) 2 by evaporation of the solvent (and volatile HCN) resulted in the formation of Mg hydroxides [Eq. (2)].S ince HCN is aw eak acid, the cyanide anion is ar elatively strong base that can deprotonate water, epecially when this is acidified by coordination to as trong Lewis acid like Mg 2+ .U sing liquid ammonia as areaction medium circumvents this problem and led to first preparations of pure Mg(CN) 2 . [8] Alkali-metal cyanides form rock-salt-like structures,f or example,K CN (Phase I) crystallizes in the NaCl lattice and down to À100 8 8C, the cyanide anion rotates in acage spanned by six K + ions. [9] This essential isotropic coordinative behavior of the spinning cyanide anion is typical for ionic metal cyanides and explains its description as ap seudohalide. Calculations on MCN (M = Li, Na, K) show that an orbiting motion of M + around CN À is essentially barrier-free (< 5kcal mol À1 ). [10] More covalently bound main-group CN compounds generally prefer cyanide connectivity.F or example,o rganic nitriles (RCN) are thermodynamically more stable than the corresponding isonitriles (RNC). [11] Thec rystal structure of B(CN) 3 ·pyridine shows CN/NC disorder with am ain contribution of the cyanide form (B À CN/B À NC = 95:5). [12] Likewise,t he anion [(CF 3 ) 3 B À CN] À is 8.4 kcal mol À1 lower in energy than [(CF 3 ) 3 B À NC] À . [13] Trimethylsilyl cyanide, Me 3 SiCN,w as shown to contain small but significant quantities of Me 3 SiNC. [14] Experimental and calculation data indicate that the XÀCN/XÀNC ratio increases with increasing electronegativity of X, that is,t he cyanide isomer becomes more favorable for covalently bound CN groups (Scheme 1a). [13][14][15][16][17][18] Forionically bound CN À ,for example,LiCN,the cyanide/isocyanide energy differences become negligible while,atthe same time,the transition states for isomerization are lowered as well.
Thea lkaline-earth-metal cyanides,A e(CN) 2 ,a re hardly explored. Being more covalent than Group 1metal cyanides, higher transition states for isomerization are expected. The crystal structure of monomeric Be(CN) 2 ·(pyridine) 2 shows BeÀCN/BeÀNC disorder with ar atio of 40:60. [12] High-level ab-initio calculations (MP4SDTQ//MP2) on Group 2m etal cyanides predict unusual features. [19] While the small and hard Be 2+ cation prefers the N-bound isocyanide structure,t he heavier Ca 2+ ,S r 2+ ,a nd Ba 2+ ions are neither cyanides nor isocyanides but instead prefer as ide-on coordination (Scheme 1b). Our group reported the first Ca-cyanide complex (I) which is stabilized for ligand exchange by the bulky ßdiketiminate ligand DIPP BDI. [20] Jones and co-workers described the formation of as imilar,T HF-free Mg complex (II). [21] Theb ridging cyanides in both trimers are statistically disordered and, like in related Al chemistry, [22] their bridging nature does not allow any conclusions regarding cyanide vs. isocyanide coordination. Surprisingly,insome reports on rare examples of terminally bound metal isocyanides,the cyanide/ isocyanide isomerism is not even subject of discussion. [23][24][25] Since there is ab road interest in metal-CN isomerism from at heoretical [26] or experimental [27] point of view,w eh ere report the synthesis and structure of aM gc omplex with at erminal CN À ligand and provide af irst comprehensive discussion on (iso)cyanide preference.
To prepare am onomeric Mg cyanide complex, we switched from the ß-diketiminate ligand (BDI) to ad ipyrromethene ligand (DPM). DPM is as ubunit of porphyrin and, although already known since 1924, [28] has only been sporadically used in Group 2metal chemistry. [29,30] TheDPM ligand is substantially more sterically demanding than the BDI ligand and noticeably encapsulates the metal by its flanking substituents that form ac avity wich prevents dimerization.
DFT calculations using the B3PW91(D3BJ)/6-311 + G** method (including D3BJ dispersion corrections) [32] on am odel system in which all mesityl substituents have been replaced by phenyl rings reproduce the crystal structure of 4 remarkably well;f or example, d(MgÀNC) = 2.049(2) (Xray) and 2.039 (DFT;F igure 2). Thec alculated MgÀCN bond length in the Mg cyanide isomer is considerably higher (2.158 ), providing further confirmation for the presence of the Mg À NC isomer in the crystal structure.T he Mg À CN isomer is also higher in energy by DG(298 K) = 1.63 kcal mol À1 .This energy difference translates to aMg ÀNC/MgÀCN ratio of 94:6, which is close to the experimentally determined ratio of 92:8 from the crystal-structure data. Interestingly,this result also compares extremely well with the 95.5:4.5 ratio for low-valent MgNC/MgCN radicals discovered in the envelope of C-rich stars for which an energy difference of 1.88 kcal mol À1 was calculated in favor of MgNC. [33]

Communications
Strong MgÀNC bonding in the isocyanide isomer is not only apparent from ashort Mg À Ndistance but also from the electron density at the bond-critical point (BCP) which is slightly higher than that for the cyanide isomer.Consequently, the CNbond in the isocyanide complex is somewhat longer and weaker than that in the cyanide isomer.Calculated NPA charges show that the isocyanideanion is extremely polarized with ahigh negative charge on the Mg-bound N(À0.98) while the cyanide anion has am uch lower charge on Mg-bound C (À0.35);calculated charges for free C N À are À0.24 (C) and À0.76 (N). Strong preference for the isocyanidei somer is therefore related to larger electrostatic contributions and polarization. Contour plots of the Laplacian of the electron density (atoms in molecules) clearly show that, although the cyanide ligand itself is much less polarized than the isocyanide ligand, the lone pair at the Cisconsiderably better polarizable than that on the N. Therefore,t he preference for the Mg À N bonding may also be explained by the hard-soft-acid-base theory (HSAB): the hard Mg atom prefers interaction with the hard Natom.
Calculations on the very simple model system HMg(NC) provided valuable insight into the complicated CN/NC isomerization process (Supporting Information, Figure S24). Similar to the LiCN/LiNC isomerization, [10] two transition states and one intermediate minimum were located. Only one transition state was found for isomerization of the larger model system ( Ph DPM)Mg(NC)·(THF) 2 ( Figure 2). Theb arrier of 12.0 kcal mol À1 for rotation of the CN À anion agrees well with that of 10.2 kcal mol À1[26c] calculated for Mg(CN) 2 and suggests that isomerization is facile.N on-classical C À H···N and CÀH···C hydrogen bonds between the CN À anion and the THF ligands contribute to the stability of the transition state.
Thei nfrared (ATR-IR) spectrum of 4 in the solid state shows asharp but relatively weak signal at 2084 cm À1 for the CN stretching vibration (Figure 3a). This fits very welll with the calculated value for 4 of 2099 cm À1 (B3PW91/6-311 + G-(2df,p), Figure S28). Am uch higher frequency of 2166 cm À1 was calculated for the alternative MgÀCN isomer.T he CN IR stretching frequencies for cyanides are generally 70-100 cm À1 higher than those for isocyanides,w hich is in accordance with their shorter CN bonds. [13] This is also in agreement with our calculations which show as horter CN bond and ahigher electron density at the BCP for the MgÀCN isomer.I nterestingly,t he solid-state ATR-IR spectrum of 4 also shows av ery small signal at 2161 cm À1 ,avalue close to that calculated for the Mg À CN isomer (2166 cm À1 ). Heating the ATRsample holder to 70 8 8Cled to intensity changes and additional signals only in the CN spectral range ( Figure S20), indicating that afast rotation of the CN ligand may take place. An IR spectrum of 4 in aKBr pellet (Figure 3b)shows the CN absorbances for MgÀNC (2085 cm À1 )a nd MgÀCN (2162 cm À1 )m ore clearly. 13 C-labeling of the CN ligand confirms their origin:signals for the isotope labeled complex appeared at the expected frequencies of 2043 cm À1 (Mg À N 13 C) and 2118 cm À1 (Mg À 13 CN;T able S6). These data are in agreement with the X-ray and DFT studies which both predict minor quantities of the cyanide isomer.S ince both isomers cannot be obtained in pure form, further quantification by IR is excluded. It should be noted, however, that exchange of 12 CN for 13 CN results in al ower Mg À NC/Mg À CN ratio.T he increased cyanide content may be explained by the stronger Mg À 13 CN bond (vs.Mg À 12 CN) while the Mg À NC bond is less affected by isotope substitution. 1 HNMR data and DOSY measurements on 4 dissolved in [D 8 ]THF confirm that the highly symmetric monomeric solidstate structure is retained in solution ( Figures S9 and S14). While all resonances in the 13 CNMR spectrum can be assigned, no signal for the CN ligand is observed. However, the 13 CN-enriched complex shows ab road resonance at 169.2 ppm which is in the typical range for isocyanide isomers: the cyanide Cresonance is typically found in the 95-145 ppm range while for isocyanides,v alues around 155-175 ppm are common. [13] Heating the solution led to signal sharpening and ashift to lower ppm values indicative of Mg À NC-to-Mg À CN isomerization. Stepwise cooling lead to signal broadening and ag radual shift of the 13 CN signal to higher ppm values.A t À85 8 8C, decoalescence is reached and asecond, much smaller, broad signal at 144.3 ppm appears,w hich is typical for ac yanide group (Figure 3c). Them ain signal, assigned to MgÀNC,isfound at 175.9 ppm, that is,atthe higher end of the range for isocyanidec omplexes.T his clearly shows that the Mg isocyanide and cyanide isomers are in fast equilibrium. Te mperature lowering increases the Mg À NC/Mg À CN ratio and results in slow exchange.D ue to different 13 Cr elaxation times in both isomers,noexact ratio has been estimated. It is, however, clear that the MgÀNC/MgÀCN ratio is large. Knowing the chemical shift of pure Mg À NC (175.9 ppm) and Mg À CN (144.3 ppm), however, enables an estimation of the Mg À NC/Mg À CN ratio at room temperature.T he 13 CNMR signal at 169.2 ppm (298 K) is the weighted average from which aM g ÀNC 13 /MgÀ 13 CN ratio of 79:21 can be deduced. Forcyanide with anatural isotope distribution, this value will be higher (see above).
We have shown that the dipyrromethene ligand Mes DPM succesfully blocks dimerization, enabling the isolation of aMg complex with at erminal CN ligand. Crystal structure determination as well as IR and NMR studies show ac lear preference for the isocyanide isomer:a t2 98 Karatio of % 95:5 is estimated. Due to ar elatively low isomerization barrier of only 12.0 kcal mol À1 ,r otation of the CN ligand within the pocket can be observed in solution as well as in the solid state. 13 CNMR studies in solution show that the isocyanide/cyanide exchange can be frozen at À85 8 8C, leading to decoalescence of 13 CNMR signals for the CN ligand. Experiment and theory both indicate that Mg complexes with the CN À ligand should not be named cyanides but rather be referred to as isocyanides.Clear preference for Mg isocyanide formation should be taken into account when discussing the mechanism of Mg-catalyzed aldehyde or ketone cyanation. [34]