Alkylated carborane anions and radicals

New tools for chemists: Novel weakly nucleophilic anions whose salts conduct electricity well in nonpolar solvents and which form strongly oxidizing, stable radicals.

structure Weakly coordinating anions can stabilize highly electrophilic or oxidizing cations. These anions are useful in fundamental research, but they can also play a variety of practical roles, such as counterions to cationic catalysts for olefin polymerization, electrolytes in lithium ion batteries, nonlinear optical materials, and solvents for extracting radioactive cations from wastes.

Weakly nucleophilic anions—those that minimally influence their cations—have been widely studied in the past decade (1, 2). Ideally, these anions should possess the following qualities:

The icosahedral CB₁₁^– cluster is an ideal core for weakly coordinating anions. It is large (~3.4 Å average B–B diam) and very stable, with no lone pairs. The long-known closo-CB₁₁H₁₂^– anion 1^– (3) serves as a starting material for the preparation of many weakly nucleophilic anions. Replacing the relatively nucleophilic B–H vertices with the more inert B–halogen vertices has been the most common strategy, and a large family of promising halogenated anions has been developed (4–9).

Figure 1. Single-crystal X-ray structure of LiCB11Me12?toluene. Reproduced with permission from Reference 10.

Figure 1. Single-crystal X-ray structure of LiCB₁₁Me₁₂•toluene. Reproduced with permission from Reference 10.

Our approach was to avoid halogenation (and the concomitant introduction of the basic lone pairs on the strongly negatively charged halogen atoms) and to protect the central CB₁₁^– core with methyl groups. Because of the lower electronegativity of carbon, the methyl groups can be expected to have a less pronounced anionic character than halogens. We also expected methyl groups to render the anion lipophilic. This effort resulted in the preparation of CB₁₁Me₁₂^–. The single-crystal X-ray structure of its Li⁺ salt with coordinated toluene is shown in Figure 1 (10).

Developing weakly coordinating and lipophilic anions is not the only reason to study the CB₁₁ cluster. Understanding reactivity and electronic properties in simple electronic models is a hallmark of modern chemistry, but cluster chemistry lags behind many other areas in this respect. The arrow-pushing formalisms of organic chemistry cannot be used with these clusters because they are not easily described by simple valence bond models. Mechanisms are proposed infrequently for cluster reactions, and trends in reactivity are described by empirical rules. Elegant models have been developed that successfully predict the coarse properties of clusters, such as topology and electron count, but they are less useful in describing more subtle properties. Current understanding of clusters is somewhat similar to that of conjugated

-systems half a century ago, when Hückel theory was known but laborious to apply, and the perturbational molecular orbital method and Woodward–Hoffman rules were not yet developed.

Figure 2. HOMO and HOMO–1 of 1^–. A single member of each degenerate pair is shown.

To improve our understanding, we developed a procedure for visualizing the molecular orbitals (MOs) of icosahedral clusters calculated by standard abinitio programs. Figure 2 shows the highest occupied molecular orbital (HOMO) of the all- hydrogen system 1^– and the one just below it (HOMO–1); in each case, only one member of a degenerate pair of MOs is shown. Both orbital pairs are derived from the quadruply degenerate HOMO of the parent B₁₂H₁₂^2– anion. They are relatively close in energy and play an important role in understanding the properties of 1^–. The HOMO vanishes at the carbon (position 1) and the antipodal boron (position 12) atoms and only has horizontal tangential amplitudes at the proximate and the distal pentagon (positions 2–6 and 7–11, respectively). The HOMO–1 has a large amplitude on boron atom 12, and smaller vertical tangential amplitudes on borons 7–11.

The Frontier MO theory of chemical reactivity relates the sensitivity of a position to electrophilic attack to the amplitude of the HOMO at that position. In this case, the theory actually needs to take into consideration both the HOMO and the HOMO–1 orbital pairs. This accounts for

The relative reactivity of the vertices 7–11 and vertex 12 should be sensitive to HOMO and HOMO–1 energies, which in turn will depend on the substituents present on the cage. This dependence has been examined using a series of methylated derivatives whose anodic oxidation potentials correlate with MO energies in a way that is nicely rationalized by the inspection of the MO coefficients in 1^– at the position of substitution (11).

Preparation of the parent CB₁₁anions
Until recently, CB₁₁-based anions were very expensive because the high cost of the starting material 1^– severely limited their utility. The previous best route to 1^– started with the costly decaborane B₁₀H₁₄ and involved several steps (12). This motivated us to develop a new two-step synthesis of 1^–. We had found a convenient one-step synthesis of 1^– in 40% yield from the anion B₁₁H₁₄^– (3^–) (13), which is easily prepared in one step from commodity chemicals (Figure 3) (14). The insertion of a single carbon atom into the B₁₁ cage is accomplished with dichlorocarbene, generated in situ from chloroform and a base. The carbene insertion reaction appears to be general for 1-substituted derivatives of 1^– and even for carboranes of other cluster sizes.

Our original conversion of 1^– to CB₁₁Me₁₂^–(2^–) by methylation with methyl trifluoromethanesulfonate (triflate) required an expensive base, 2,6-di-tert-butylpyridine, to quench the triflic acid generated in the reaction (15). We have developed an improved synthesis that uses inexpensive calcium hydride as the only base (16). The overall procedure, starting with 1^–, provides a 95% yield of >99% pure Cs⁺2^–. In these alkylation reactions, alkyl triflates can sometimes be replaced by the cheaper bromides (17), but not in the synthesis of 2^–.

The anion 2^– can be oxizided electrochemically (1.16 V vs ferrocene) to give the stable isostructural free radical CB₁₁Me₁₂^• (2^•). This result can also be accomplished with Pb(CF₃CO₂)₄, providing 2^• in 74% isolated yield (Figure 3) (18). Unlike its ionic precursor, 2^• dissolves in oxidation-resistant nonpolar solvents (e.g., pentane, CCl₄) to give deep blue solutions. The stability of 2^• can be attributed to steric protection of the delocalized free-valence–carrying centers in the carborane icosahedron by a sheath of methyl groups.

Properties of the CB₁₁Me₁₂^– anion
The cation in the salts of 2^– can be easily exchanged by several conventional methods. These salts are stable to air, strong bases (stable for several days in saturated KOH–EtOH), weak acids (reflux in AcOH), and aqueous and other dilute strong acids (stable overnight in 5% H₂SO₄ in Et₂O), but they are destroyed after several hours in concentrated H₂SO₄ or neat CF₃SO₃H.

The CB₁₁Me₁₂^– anion is remarkably lipophilic. Although all of the salts we studied are soluble in polar organic solvents (e.g., Et₂O, CH₂Cl₂), their solubility in less polar organic solvents depends on the cation. Lithium salts are the most soluble, and 300 g/L (1.0 M) saturated benzene solutions of Li⁺2^– can be prepared. A detectable solubility was observed even in alkanes, and the replacement of a single methyl substituent with hexyl in the anion suffices to render its lithium salt very soluble in these solvents. Salts of the heavier alkali metals and of di- and trivalent cations are less soluble; nonetheless Cs⁺2^– partitions from water into toluene with a partition coefficient of 0.4. Anion 2^– also solubilizes many organic cations (e.g., polyaryl-substituted pyrylium and pyridinium) in low-polarity organic solvents.

A saturated solution of Li⁺2^– in benzene is electrically conductive. The conductivity strongly depends on the water content and increases from 1.8 × 10^–5 S/cm for the dry solution to 6.2 × 10^–3 S/cm for a solution containing 1.75 H₂O molecules per Li⁺ ion (the lithium salt can be dried easily by heating to 180 °C under reduced pressure). This conductivity implies the presence of benzene-solvated dissociated Li⁺ ions. The nature of the solvation is hinted at by the

-face interaction between Li⁺ and an aromatic ring in the crystal of LiCB₁₁Me₁₂•toluene (10).

Other CB₁₁ derivatives
Substituents can be introduced selectively at all four unique atom positions in 1^– (Figure 4). Substitution on carbon is most easily controlled. The CH-group acidity is roughly comparable to that of acetylene, and reactions with electrophiles after deprotonation by a strong base yield 1-X-CB₁₁H₁₁^– derivatives. Most attention has been focused on X = Me or Et because these groups lack lone pairs and are compatible with reactive cations (5), but we also have prepared the 1-halogen derivatives (X = F, Cl, Br, I) (19, 20). Three methods of preparing 1-aryl derivatives are known. We found that the Pd-catalyzed Negishi cross-coupling of PhOTf with 1-ClZnCB₁₁H₁₁^– produces 1-PhCB₁₁H₁₁^– (19, 20); Kennedy and co-workers published a multistep synthesis of this material from decaborane (21); and most recently, we found that inserting parent and para-substituted phenylchlorocarbenes into 3^– provides 1-phenyl substituted carboranes (13).

Substitution at boron requires differentiation among the three inequivalent vertices. This can be provided by prior substitution, and the Pd-catalyzed Kumada cross-coupling of 12-I-CB₁₁H₁₁^– with RMgX (R = Me, Et, n-Bu, n-hexyl, and Ph) yields 12-alkylated and 12-arylated derivatives (22). This coupling is sensitive to steric hindrance and fails when five methyl groups are present in positions 7–11.

All the boron vertices of 1^– react with electrophiles, but as indicated in the earlier discussion of the HOMO and HOMO–1 orbitals, they do so at different rates. This is best illustrated for halogenation (9). The antipodal position 12 is the most reactive and requires the mildest conditions for selective substitution. The distal positions 7–11 are less reactive but can be halogenated under harsher conditions. The proximate positions 2–6 are less reactive still, and their halogenation requires forcing conditions. Permethylation with methylating agents can be achieved under mild conditions because incorporating the methyl groups activates the cluster for further methylation (15). Other alkylations are possible, for instance, the introduction of tropylium as a substituent (22).

The mechanism of these electrophilic substitutions has not been studied in detail. A tentative way of rationalizing the observed processes is to imagine that an electrophile E, such as a proton, is transferred to a B–X bond, forming a three-center two-electron bond (in reality, the protonation may be occurring on an edge or a triangular face of the cluster). Then an electrophile is lost with the re-formation of a two-electron exocyclic bond.

An example of such a process is the “electrophile-promoted nucleophilic substitution” reaction (23), in which an acid HY induces nucleophilic substitution on a boron vertex of the CB₁₁ anion, converting B–H or B–Me to B–Y, presumably with the formation of molecular hydrogen or methane (X = H or Me, E = H⁺, Nu = Y^–). Thus, 2^– reacts with HF to yield 70% 12-F-CB₁₁Me₁₁^– and 30% 7-F-CB₁₁Me₁₁^– (18). Also, a base must be present during the electrophilic methylation of 1^– with MeOTf (15) to avoid a similar reaction of the TfOH byproduct with 2^–.

Protecting group strategies combined with an understanding of intrinsic reactivity differences among the vertices permit access to a variety of substitution patterns (11). A bulky i-Pr₃Si-group can be easily introduced into position 1 to protect adjacent positions 2–6 from substitution. It can be later removed with F^–. Similar protection is provided by 1-F substitution. An iodine substituent protects its boron vertex from methylation and can be removed later by dissolving-metal reduction.

In our laboratories, the fundamental reactions described in Figure 4 have been combined in various ways to prepare ~50 variously substituted CB₁₁ anions with fivefold symmetry in the substitution pattern. The cyclic voltammetry of many of these compounds has been studied; most undergo reversible one-electron oxidation.

Substitution that breaks fivefold symmetry is harder to control. A 2-Br derivative is easily available from the carbene insertion reaction (10), and some reactions that introduce bulky substituents can be constrained to monosubstitution. A curious example is the replacement of a methyl group by an aryl, which is accomplished by heating a lithium salt of a methylated CB₁₁ anion in an aromatic solvent; the substitution can occur in position 7 or 12. This reaction possibly proceeds by the lithium cation performing a second-order electrophilic substitution (S_E2) on the carbon of a methyl group to give methyllithium and a boronium ylide, which then performs an electrophilic substitution on the arene present.

Electrochemical uses of CB₁₁Me₁₂–
The conductivity of the solutions of Li⁺2^– suggested several possible applications. For instance, the lithium cation has long been recognized as an effective Lewis-acid catalyst for pericyclic reactions. The catalysis of Diels–Alder reactions by solutions of LiClO₄ in ether has received the most attention (24), but the explosive nature of this reaction medium is a problem. In benzene solution, lithium ions promised to be even more active than those solvated by ether. Indeed, we found remarkably high rates for pericyclic rearrangements in solutions of Li⁺2^–in benzene and 1,2-dichloroethane, particularly under benchtop conditions, when these solutions always contain a little moisture (25). We first observed that at 67 °C the Claisen rear rangement of phenyl allyl ether proceeded with a half-life of 21 h in a saturated benzene solution of Li⁺2^–, but no reaction was detected in 5 M LiClO₄ in Et₂O. The uncatalyzed reaction requires heating to 200 °C. We then examined the rearrangement of several strained hydrocarbons devoid of any specific Lewis-base sites. Room-temperature isomerization of quadricyclane to norbornadiene, of cubane to cuneane, and of diademane to triquinacene in saturated Li⁺2^–in benzene proceeded with remarkable ease.

Moisture has a dramatic effect on the rate constant of these isomerizations. The very dry solution is a poor catalyst, the addition of up to 0.1 mol of water/mol of Li⁺ greatly enhances catalysis, and additional water attenuates it. Solvation of Li⁺ by water apparently favors the dissociation of the LiCB₁₁Me₁₂ tight ion pairs and enhances conductivity, but excessive hydration of dissociated Li⁺ makes it a poorer catalyst. Only in the narrow region where there is sufficient water to favor the dissociation of the ion pairs, but not enough to attenuate the catalytic activity of Li⁺, is LiCB₁₁Me₁₂ optimally active, and this ideal situation appears to be approximated by standard benchtop conditions.

On the basis of these results, we wanted to examine other additives and solvents that might enhance ion-pair dissociation but not bind the Li⁺ ion strongly. A cursory survey of scrupulously dry solvents showed that solutions in 1,2-dichloroethane were especially effective. In an attempt to find out whether catalytic amounts of Li⁺2^– would be sufficient, we conducted a series of reactions with ~0.03 M Li⁺2^– in C₂D₄Cl₂ under anhydrous conditions. In these dry and dilute solutions, the catalysis is less pronounced than in the saturated benchtop solutions in benzene, but it is still very significant. For instance, at 25 °C, the quadricyclane to norbornadiene rearrangement, which normally requires heating to 150 °C, proceeds with a half-life of 4.2 h (vs 11 min in a saturated benchtop solution of Li⁺2^– in benzene).

We have now observed catalysis of other reactions as well, such as Diels–Alder additions and allyl and silyl ether solvolyses. Much optimization still remains to be done, but judging by these preliminary observations and given the now low cost of 2^–, the CB₁₁Me₁₂^– salts of Li⁺ and other cations, possibly including chiral ones, promise to become practical catalysts.

Nonaqueous solvents are used widely in studies of redox reactions of organic and coordination compounds (26). There are many reasons to perform voltammetry and other electrochemical measurements in such solvents: increased solubility of nonpolar compounds, elimination of proton transfer reactions, suppression of analyte adsorption onto the electrode, and effects of solvent ligation. Until now, aromatic hydrocarbons have found limited use as solvents in electrochemistry because they do not dissolve the supporting electrolytes required for achieving a reasonable electrical conductivity. Whereas considerable effort has been made to use aromatic solvents or polyaromatic melts in the electrodeposition of various metals (27), routine electrochemical experiments with conventional electrodes, concentrations, and scan rates have rarely been attempted. The electrical conductivity of Li⁺2^–C₆H₆ is comparable with that of common nonaqueous electrochemical media and offers an attractive alternative for performing electro chemical reactions. We therefore decided to examine the pos sible utility of Li⁺2^– as a supporting electrolyte for electrochemical measurements in nonpolar solvents (28).

The range of potentials in which a gold electrode and 0.5 M Li⁺2^– in benzene can be used is conveniently wide: It is limited by oxidation of 2^– at +0.7 V to the neutral radical 2^• and by reduction of Li⁺ at –2.0 V. Ferrocene oxidation proceeds reversibly at a half-wave potential of +0.04 V. The separation of anodic and cathodic peaks is 59 mV at a scan rate up to 4 V/s, as expected for a reversible one-electron redox process. Even a saturated solution of Li⁺2^– in silicone oil [(PhMeSiO)_n] gives a good ferrocene wave, but only at elevated temperatures (the viscosity is too high at room temperature).

A solution of Ag⁺2^– in 0.5 M Li⁺2^– in benzene yields a well-developed quasireversible cathodic peak and a correspon ding anodic dissolution of deposited silver metal. When the voltage scan is reversed at a value more negative than –1.55 V, the observed anodic wave indicates the formation of Li–Ag alloys. Forming amalgams of alkali metals is an industrially important process that permits reduction of alkali metals at potentials 1 V less negative than their equilibrium potentials (29).

Uses of CB₁₁Me₁₂^– in SPEs
The electrical conductivity of Li⁺2^– in benzene hinted at another possible application of Li⁺2^–, namely as a component of solid polymer electrolyte (SPE) materials. The liquid electrolyte batteries presently used have serious disadvantages (30). Liquid electrolytes are not entirely stable chemically (e.g., they react with lithium anodes and cathode materials) or electrochemically (i.e., they can undergo electrolysis). Potential leakage makes liquid electrolyte batteries unreliable and environmentally unsafe. Their processibility is limited because neither very desirable thin-film production nor a variety of shapes can be achieved.

Lithium rechargeable batteries based on SPE technologies have been proposed to replace electrolytes in a wide variety of applications. Thus far, however, the SPE-based batteries have exhibited several important performance limitations (31). Most SPE materials development has focused on modifications in polymers and additives, and relatively few salts have been investigated. Yet the nature and concentration of the incorporated salt may have a major influence on the properties of an SPE. In particular, the anion strongly affects phase composition and conductivity, as has been shown for such commonly used anions as perchlorate and triflate (32).

Highly alkylated or peralkylated carboranyl anions may improve some major characteristics of SPE materials. These anions will have low mobility in the polymer solutions, increasing cation transference numbers by decreasing concentration gradients. Incorporating 2^– into polymeric structures would maximize the cation transference number and could greatly improve conductivities because of the very low nucleophilicity of this anion and thus a significantly weaker anion–cation interaction. Finally, 2^– and its analogues could provide high thermal, chemical, and electrochemical stability to SPE materials. These ideas were confirmed when we found that Li⁺2^– is soluble in polymers such as poly(dimethylsiloxane) and poly(methylphenylsiloxane) at concentrations up to 20% (33). Even before any optimization, the conductivity of the very viscous high-concentration solutions was 5 × 10^–5 S/cm at 25 °C, which is comparable to that of other SPE materials produced to date (34).

We started by making a permethylated anion 4^– (Figure 5) with a polymerizable 5-hexenyl substituent instead of a methyl group in position 1 of 2^–. We also incorporated 2^– directly into phenyl-containing polymers by heating Li⁺2^– with polystyrene or poly(phenylmethylsiloxane) under vacuum for 3 days at 160 °C to yield 5^–, relying on the phenyl–methyl substitution reaction described above.

Because direct methylation of the 5-hexenyl derivative of 1^– with methyl triflate methylates the double bond as well as the cage, 4^– was prepared by introducing a 6-chlorohexyl group into position 1 of 1^–, permethylating, and dehydrochlorinating with a Schwesinger base. Anion 4^- was polymerized with a zirconocene-based catalyst to a mixture of low– molecular-weight oligomers with a conductivity of 2.2 × 10^–3 S/cm. The 2^–-doped poly(phenylmethylsiloxane) showed a conductivity of 1.62 × 10^–4 S/cm. These preliminary results show that SPE materials based on 2^– can be made relatively easily and that they promise to provide solid polymer electrolytes with superior properties.

Additional uses of the anion
Another possible use of the anion 2^–, or its even more lipophilic analogues with longer alkyl groups, is removal of radioactive cesium from waste products of nuclear reactions by extraction from an aqueous solution into a solvent such as toluene, or into a resin containing bound anions similar to 2^–. Release of the Cs⁺ cations is induced by reversible oxidation of the anions to radicals of the type 2^•. The practical utility of such procedures depends on factors such as the Cs⁺/Na⁺ extraction ratio, stability of the radicals in air, and the radiation stability of the carborane anions, all of which have not yet been assessed. Neat 2^• is severely degraded in air after a few days, and its solution is much more air sensitive, but substituted derivatives might be more stable.

Yet another possible use for 2^–is in nonlinear optical materials. The ylides obtained by placing a tropylium substituent at the 12-position of 1^– (22) and 2^– (11) have higher nonlinear activity than the p-nitroaniline standard. These ylides’ strong absorption also occurs at shorter wavelengths than that of the standard, perhaps making carborane anions generally useful in this context.

Uses of the CB₁₁Me₁₂^• radical
The 2^•radical is a strong oxidant, comparable to Ce(IV), but it is soluble in all types of oxidation-resistant organic solvents (18). In pentane, for instance, 2^• oxidizes triphenylamine, but not perylene, to its radical cation, and it induces polymerization of styrene. In the more polar methylene chloride, it oxidizes perylene as well as tetracene and acenaphthene. Oxidation of [Fe(CO)₂Cp]₂ in toluene affords solvated Fe(CO)₂Cp⁺2^–. This reaction is a general and clean method for the preparation of highly electrophilic or oxidizing cation salts of weakly coordinating anions in alkanes and other poorly coordinating solvents.

To illustrate the utility of 2^• in this respect, we will describe the preparation of R₃E⁺ salts of 2^–, where E is an atom of a group 14 element (35). Normally, it is difficult to prepare and crystallize group-14 R₃E⁺ salts because adducts form even with weakly nucleophilic solvents, and the solubility is often too low in sufficiently inert ones. Reaction of 2^• with a neutral R₃E-containing precursor in an inert solvent to produce the R₃E⁺ salt of the solubilizing and only weakly nucleophilic anion 2^– suggests a general solution to this problem.

The white solid n-Bu₃Sn⁺2^– (6) was prepared this way in dry pentane from n-Bu₆Sn₂ and two equivalents of 2^•.

Figure 6. Single-crystal X-ray structure of n-Bu3Sn+2?. Hydrogen atoms and one component of the disordered butyl groups are omitted for clarity. Sn atoms are red, C atoms are black, and B atoms are green. Reproduced from Reference 36.

Figure 6. Single-crystal X-ray structure of n-Bu₃Sn⁺2^–. Hydrogen atoms and one component of the disordered butyl groups are omitted for clarity. Sn atoms are red, C atoms are black, and B atoms are green. Reproduced from Reference 36.

We obtained an X-ray diffraction structure of its single crystal grown from hexane (36). The structure revealed a tri-n-butylstannyl cation weakly coordinated to methyl groups of two 2^– anions in a trigonal bipyramid (Figure 6). Each anion interacts with two cations through two antipodal methyl groups in infinite columns of alternating n-Bu₃Sn⁺ and 2^– ions. The presence of these weak interactions is obvious from

Whereas the chemical shift is much higher than that of other known trialkylstannyl cations, confirming the weakness of the coordination, it is still far lower than the value expected for an isolated cation, >1500 ppm. Metal cation–alkane interactions are of considerable research interest (37), and crystal structures with a cationic transition metal atom coordinated to a methyl group have been known for more than a decade (38).

In a similar fashion, the oxidation of Me₃E-containing precursors, Me₃E–EMe₃, Me₄E, or (t-Bu₃E)₂Hg (E = Ge, Sn, and Pb), with 2^• yielded the salts Me₃E⁺2^– (35). Using NMR and extended X-ray absorption fine structure (EXAFS) measurements and ab initio and density functional theory (DFT) calculations, we found that the coordination of the Me₃E⁺ cations with the methyl group of the 2^– anion grows stronger with increasing exothermicity of the S_E2 reaction, in which the methyl is transferred from boron to the metal atom as one goes up column 14 of the periodic table. Along the series, the equilibrium geometry changes as would be expected for the path of such a backside S_E2 reaction, and the results provide a nice example of the Bürgi–Dunitz analysis of reaction paths (39). In the case of silicon, such a displacement apparently actually takes place, because the oxidation reaction yields a tetraalkyl silane and a novel internally charge-compensated boronium ylide CB₁₁Me₁₁, stable only up to –60 °C. This result would be expected if the now even more exothermic substitution of the methyl group by the R₃Si⁺ cation proceeds to completion. An attempt to generate the salt t-Bu⁺2^– led to similar results.

The boronium ylide CB₁₁Me₁₁ is an extremely potent electrophile. It reacts rapidly at low temperatures with alcohols, ethers, aromatics, and other nucleophiles to yield 12- substituted derivatives of 2^–. As indicated earlier, we suspect that it is a general intermediate in the acid-promoted nucleophilic substitution on 2^–. The oxidation of the stannylene [(Me₃Si)₂N]₂Sn with 2^• or 1-n-Bu-CB₁₁Me₁₁^• radicals produces salts of a cyclic stannylene with a positively charged quaternary nitrogen atom (35).

The scope of this new approach to crystalline salts of highly reactive cations is limited by the salts’ solubility in inert solvents and by the coordinating ability and chemical reactivity of the anion 2^–. In the following section, we describe an attempt to overcome the reactivity limitation.

Preparation and properties of CB₁₁(CF₃)₁₂^–We expected that perfluorination would suppress the sensitivity of CB₁₁Me₁₂^– to strong acids and oxidants, and we therefore decided to prepare the “perflate” anion, CB₁₁(CF₃)₁₂^– (7^–) (40).

Fluorination in CFCl₃ of Cs⁺2^– adsorbed on silica gel using an excess of 10% F₂ in N₂ gave a mixture of incompletely fluorinated anions. Repeated attempts at perfluorination using fluorine with and without irradiation, at various pressures, temperatures, and stirring rates were unsuccessful and provided only incompletely fluorinated mixtures. Unlike 2^–, the partially fluo rinated mixture is unaffected by anhydrous hydrogen fluoride; hence treatment with Bartlett’s reagent (K₂NiF₆) in liquid HF was possible. It provided the perflate salt in 25% overall isolated yield (96% for each of 36 successive substitutions).

The anion 7^– meets all of the usual requirements for a useful weakly coordinating anion.

Two shortcomings make 7^– less than ideal. The first is its tendency to be disordered in the crystalline state. Numerous attempts at single-crystal X-ray diffraction analysis with a variety of salts failed. However, electron density maps did reveal that 7^– is essentially spherical, with an ~8.0-Å outer F-sphere diameter and a 3.38-Å inner CB₁₁-sphere diameter, in agreement with the 8.12-Å and 3.42-Å average diameters in a DFT-optimized structure. Paradoxically, the criteria for a good weakly coordinating anion are also the criteria for anions that will tend to be disordered: large anions that do not exhibit any specific interactions with their cations.

The second significant shortcoming of 7^– is that it is explosive and unsafe. When ignited, its Cs⁺ salt burns vigorously. Scratching 300 mg of Cs⁺7^– contained in a Pyrex flask with a metal spatula caused an explosion that broke glassware 2 m away. We believe it is likely that other salts of 7^– will also be explosive and they should be treated with extreme care. Principal products of the explosive decomposition of Cs⁺7^– in O₂ include BF₃, BF₄^–, CO₂, and soot. The calculated heat of explosion of 7^– is 1272 kcal/mol, or 1.32 kcal/g (TNT is 1.05 kcal/g). The high energy content is mainly due to the greater strength of the B–F bond (154 kcal/mol in BF₃) compared with the C–F bond (116 kcal/mol).

According to calculations, the perfluorinated analogue 7^• of the stable neutral radical 2^• is stable, with a structure very similar to that of 7^–. It is also calculated to be an unusually potent chemical oxidant, –2.9 V above the 2F^– → F₂ couple (40). It remains to be seen whether it can be made and used to produce extremely highly oxidized states of matter.

To sum up
Weakly coordinating methylated carborane anions can be prepared in a few steps from commodity chemicals. Several substitution methods provide access to a broad range of derivatives. These anions are promising components of electrolytes and counterions for reactive cations. The CB₁₁Me₁₂^– anion undergoes reversible oxidation to give a stable free radical, CB₁₁Me₁₂^•, which in turn is a powerful oxidant and provides a new general method to prepare salts of reactive cations in weakly coordinating solvents. Perfluorination of CB₁₁Me₁₂^– provides the nearly chemically inert but explosive weakly coordinating anion, CB₁₁(CF₃)₁₂^–.

Benjamin T. King is a National Institutes of Health postdoctoral fellow at the University of California, Berkeley (kingbt@socrates.berkeley.edu), working with Robert G. Bergman on molecular recognition in organometallic systems. He received his B.S. in chemistry from Northeastern University, Boston. After 2 years in industry, he obtained his Ph.D. in chemistry with Josef Michl at the University of Colorado, Boulder. He was the 1997 recipient of an ACS Graduate Fellowship.

Ilya Zharov is a Beckman Fellow at the University of Illinois, Urbana–Champaign (zharov@staff.uiuc.edu), working with Steven C. Zimmerman on the preparation of dendrimers and hyperbranched polymers for molecular recognition. He received his Dipl. Chem. (Honors) from Chelyabinsk State University, Chelyabinsk, Russia, and his M.Sc. from Technion–Israel Institute of Technology, Haifa. He obtained his Ph.D. in chemistry with Josef Michl at the University of Colorado. He was the 1999 recipient of a Link Foundation Energy Fellowship.

Josef Michl is a professor of chemistry at the University of Colorado (Department of Chemistry and Biochemistry, Boulder, CO 80309-0215; michl@eefus.colorado.edu). He received his M.S. in chemistry at Charles University, Prague, Czechoslovakia, and his Ph.D. with Rudolf Zahradník at the Czechoslovak Academy of Sciences, also in Prague. He left Czechoslovakia in 1968. He did postdoctoral work at the University of Houston, TX; the University of Texas, Austin; Aarhus University, Denmark; and the University of Utah, Salt Lake City, where he stayed, became a full professor in 1975, and served as chairman from 1979 to 1984. From 1986 to 1990 he held the M. K. Collie-Welch Regents Chair in Chemistry at the University of Texas, Austin, then moved to the University of Colorado. He has been the editor of Chemical Reviews since 1984. He has held many visiting professorships and named lectureships and has won numerous awards. He is a member of the National Academy of Sciences, the American Academy of Arts and Sciences, and the International Academy of Quantum Molecular Science. His current research interests are the development of a molecular-size “Tinkertoy” construction set for molecular electronics, photochemistry, chemistry of silicon and boron, preparation and study of reactive organic and main-group organometallic molecules, and the use of quantum chemical and experimental methods for better understanding of electronic excited states.