Identifying the factors which determine the remarkable variety of bonding modes in the coordination of CO 2 to metal centers [1] is central to controlling electron transfer and ultimately the reactivity. [2][3][4] In this respect, it seems legitimate to state that d electrons are of critical importance since CO 2 coordination was never observed in d 0 -or main-group-metal systems. [1] Furthermore, the CO 2 oxygen atoms do not possess sufficient basicity to form stable adducts with Lewis acids. The sole case of a genuine end-on-bound CO 2 complex [5] involves a very strongly reducing trivalent uranium center, for which it seems likely that partial electron transfer plays a role. Nevertheless, CO 2 is only weakly bound in this complex. As a rule with no exception until very recently, only low-and medium-valent metals have been used for CO 2 coordination. For these reasons, the recent findings [6] describing the robust end-on coordination of CO 2 to magnesium cations with formation of the unprecedented Mg/Al-containing cluster
, certainly appear to be a major breakthrough. Not only do compounds 1 represent the first case of CO 2 coordination to a non-transition-metal center in an unprecedented bridging end-on fashion, but also the irreversible fixation occurs in THF, the oxygen atom of which is normally regarded as a far better donor than the oxygen atoms of CO 2 .
We have revisited the experimental data and, on the basis of additional observations, propose herein that the triatomic units that bridge the magnesium atoms end-on, believed to be CO 2 , are in reality isoelectronic NCO anions.
The formation of complexes 1 a,b was originally rationalized by assuming the attack of CO 2 at only one silazanate group. [6] It was not explanined why the second silazanate group did not follow the same fate in the presence of excess CO 2 . Also, no conclusive evidence was provided for the identification of the bridging imido group generated by the attack of Me 3 Al on the intermediate [Mg(OSiMe 3 ){N-(SiMe 3 ) 2 }]. The presence of disorder between the imido and silanolate group with equal occupancy was claimed to account for the crystallographic equivalency of the two donor atoms bridging the Al and Mg centers. This interpretation was the only possibility to have the charges balanced within the complex if the triatomic unit bridging the three Mg atoms has to be CO 2 . The NMR spectra unexplainably showed only one resonance for the silanolate and silylimido groups together.
To our surprise, geometry optimization [7] of the full complex 1 a as well as several simplified model compounds invariably led to fragmentation of the complex and release of CO 2 , which thus appears not to be strongly bound at all (Figure 1 a). This finding in turn made us question the assumptions behind the assignment of the structure of the complex. Figure 1. a) Partially optimized structure of 1 a; [7] CO 2 units are dissociating. b) Optimized structure of 1 a with N and O atoms exchanged.