DALTON FULL PAPER J. Chem. Soc., Dalton Trans., 1999, 1741–1751 1741 Redox reactions of the boron subhalide clusters BnCln 0/~2/22 (n 5 8 or 9) investigated by electrochemical and spectroscopic methods † Bernd Speiser,*a Carsten Tittel,a Wolfgang Einholz b and Ronald Schäfer b a Institut für Organische Chemie, Universität Tübingen, Auf der Morgenstelle 18, D-72076 Tübingen, Germany. E-mail: bernd.speiser@uni-tuebingen.de b Institut für Chemie, Universität Hohenheim, Garbenstraße 30, D-70593 Stuttgart, Germany Received 23rd November 1998, Accepted 31st March 1999 The redox properties of the electron hyperdeficient boron subhalide clusters octachlorooctaborane(8), B8Cl8, and nonachlorononaborane(9), B9Cl9, were investigated in solution by cyclic voltammetry at platinum or glassy carbon electrodes, and by 11B NMR as well as ESR spectroscopy.The neutral compounds undergo a spontaneous reduction by traces of moisture usually present even in dried solvents, and the voltammetric experiment starts from B8Cl8~2 or B9Cl9~2.The radical anions were identified by ESR spectroscopy. Their formation leads to line broadening in NMR spectra of BnCln. Electrochemically, they are quasireversibly reduced to the dianions, but oxidized in an ECcat (electrochemical step, catalytic chemical step) reaction with an essentially reversible electron transfer step to the neutral compounds. The potential ordering for the two redox processes is “normal” in both clusters, being in accordance with the fact that structural changes accompanying the electron transfer are minor.The radical anion B8Cl8~2 is even more stable against disproportionation than B9Cl9~2. Introduction Multiple-stage redox systems have extensively been studied in the case of organic (see, e.g. ref. 2) and organometallic (see, e.g. ref. 3) molecules. The spacing of the redox potentials for subsequent one-electron steps is an important factor which contributes to the behavior of compounds with several oxidation states.4 Usually one would expect that oxidation or reduction becomes increasingly diYcult with increasing or decreasing redox state of the molecule.It is thus common to find the diVerence in formal potentials in eqn. (1) for two-electron transfer DE0 = E2 0 2 E1 0 (1) reactions, eqns. (2) and (3), where the superscript for all species A0 A1± ± e2 E1 0 (2) A1± A2± ± e2E2 0 (3) indicates the diVerence in redox state relative to Ao (“1” for oxidations, “–” for reductions; often, but not always, the stable starting species is one of the “extreme” oxidation states, Ao or A2±; Ao is not necessarily neutral) to be positive for oxidations and negative for reductions. The symbol |DE0| will denote E2 0 2 E1 0 for an oxidation and 2(E2 0 2 E1 0) for a reduction.Then, the equilibrium (4) is characterized by an equilibrium constant A0 1 A2± 2 A1± (4) in eqn. (5). In aprotic solvents |DE0| often attains values of Kcomp = [A1±]2 [A0][A2±] = exp F F RT |DE0|G (5) approximately 0.4–0.5 V 4 (“normal potential ordering”).However, several examples have been identified where |DE0| is † Two-electron-transfer redox systems. Part 2.1 decreased to values <0.4 V (“potential compression”) 5 or the second electron transfer even occurs thermodynamically easier than the first one (“potential inversion”).6 In systems with potential inversion the intermediate redox state A1± is unstable with respect to disproportionation (4).Potential inversion is usually accompanied by a considerable change in the structure of the molecule during the redox process, for example conformational changes 1,7–10 or changes in cluster geometry.11 One class of chemical compounds which could undergo twoelectron transfers is the series of boron subhalide clusters with a 1 : 1 stoichiometry of boron vs. halogen. The chloroborane clusters BnCln (n = 4, 8–12), of which B8Cl8 and B9Cl9 are investigated in this study, are classified as electron hyperdeficient molecules 12 and sometimes are called hypercloso clusters,13 since the number of their framework electrons is 2n.The corresponding dianions closo-BnCln 22 (n = 6, 8–12) as well as the borate clusters closo-BnHn 22 possess 2n 1 2 cage bonding electrons and follow Wade’s rules of the framework electron count to structure correlation.14 Nevertheless, the structures both of hypercloso-BnCln (n = 8, 9) and closo-BnHn 22 are based upon the same n-vertex deltahedra: dodecahedron (D2d symmetry) for B8Cl8 15–17 and B8H8 22,18 tricapped trigonal prism (D3h or C3v symmetry, respectively) for B9Cl9 19,20 and B9H9 2221 (Fig. 1). For the system B9Br9 0/22 it has recently been confirmed by X-ray crystallographic analysis and ELF (electron localization function) calculations 19 that the cluster structure remains intact upon the redox conversion while changes in atomic distances and bond angles occur.Such a behavior is in sharp contrast to Fig. 1 Geometrical shapes of the deltahedral boron subhalide clusters BnCln (n = 8 or 9).1742 J. Chem. Soc., Dalton Trans., 1999, 1741–1751 the structural rearrangements found during reduction of [Os6- (CO)18], which changes from a bicapped tetrahedral structure (neutral; 2n framework electrons) to an octahedron (dianion; 2n 1 2 framework electrons).11 As well, the 6-vertex borate clusters B6X6 22 (X = Cl, Br, I or H) 22–25 show the expected geometry of an octahedron, whereas the hypothetical neutral B6H6 is suggested by an ab initio study26 to have a capped trigonal bipyramidal (bicapped tetrahedral) structure like [Os6(CO)18].Furthermore, B4Cl4 molecules are tetrahedral with nearly Td symmetry,19,22,27 but the hypothetical B4H4 22 ion is predicted by MNDO28 and ab initio calculations 29 to exhibit a puckered D2d conformation. The reasons for these structural features can be traced to the degeneracy or non-degeneracy of the highest occupied (HOMO) and the lowest unoccupied (LUMO) molecular orbital of the polyhedrons:30–32 the 8-vertex D2d dodecahedron (B8X8 and B8X8 22) and the 9-vertex D3h tricapped trigonal prism (B9X9 and B9X9 22) have non-degenerate frontier orbitals (HOMO and LUMO), and thus can accommodate, n, n 1 1 or n 1 2 framework electron pairs.In contrast, the HOMOs and LUMOs of most of the other closo-borates BnHn 22 (n = 4–7, 10, 12) are degenerate. Removing two electrons from these clusters must result in a change of the structure according to Jahn– Teller theory.33 The 11B NMR spectra of B8Cl8 and B9Cl9 do not show two diVerent signals as would be expected by considering the molecular structure in the solid state, but only a single sharp resonance line (d11B 64.8 for B8Cl8 and 58.2 for B9Cl9, h1/2 ª 35 Hz).This eVect can be explained by the rapid fluctuation of these molecules in solution, which is described for the related eight-vertex cluster B8H8 22 by the diamond–square–diamond transformation.34 In contrast, the 11B NMR spectrum of B9Cl9 22 exhibits two peaks at d 21.5 and 25.5 in an intensity ratio of 1 : 2 representing the three boron atoms with a connectivity of 4 and the six boron atoms with a connectivity of 5 in the cage.19,35 Thus, there is no fluxional behavior or the transformation is very slow on the NMR timescale.The dianion of the eight-vertex polyhedron B8Cl8 22 is not yet known in the literature.Since the corresponding hydrogen substituted cluster B8H8 22, however, shows structural non-rigidity in solution, indicated by the appearance of only one 11B NMR signal at room temperature (d11B 25.8, doublet, JB–H = 128 Hz)34,36 we could expect the same structural features for B8Cl8 22. In earlier work, B9Cl9 was reduced chemically to both its paramagnetic mono- and its di-anion, and B9Cl9 22 oxidized by thallium(III) trifluoroacetate to the higher oxidation states.35 Bowden37 oxidized B9Cl9 22 electrochemically in CH2Cl2 and CH3CN, while Kellner 38 investigated the neutral B9Cl9 at a glassy carbon electrode in CH2Cl2.‡ In all cases, a stepwise redox reaction in the system B9Cl9 0/~2/22 was found, with all three redox states being stable in solution.The electron transfer chemistries of the smaller homologue B8Cl8 or its dianion do not seem to have been investigated. Our results of a cyclic voltammetric and spectroscopic study of nonachlorononaborane(9), B9Cl9, and octachlorooctaborane( 8), B8Cl8, are presented in this paper.Besides characterizing the redox chemistry of B8Cl8 for the first time, and determining the relative potential ordering of its formal potentials, we identified the starting species of the experiments to be diVerent from BnCln by means of rest potential measurements and ESR as well as NMR spectroscopy. Computer simulations ‡ After finishing electrochemical experimental work for the present manuscript we became aware that similar cyclic voltammetric investigations of B9Cl9 but not B8Cl8 had been conducted at the Universität Stuttgart, Germany, and that a manuscript was being prepared by the groups involved. Preliminary manuscripts were exchanged in September 1998.We refer to this version of the Stuttgart manuscript,39 which incorporates parts of the dissertation of Kellner.38 DiVerences and similarities will be discussed in the course of the present paper.of the cyclic voltammograms allowed the determination of kinetic constants. Results and discussion Overall electrochemistry of BnCln Earlier cyclic voltammetric work with B9Cl9 2237 and B9Cl9 38,39 indicates that the redox states of the nonachlorononaborane(9) cluster can be converted in two stepwise one-electron transfers. Based on these results, we expected that B9Cl9, and in analogy also B8Cl8 would be stable at suYciently positive electrode potentials E, and could be reduced to the respective dianions upon variation of E to less positive and finally negative values.Starting the voltammetric scan at rather positive potentials, both B9Cl9 and B8Cl8 in the dichloromethane electrolyte at Pt and glassy carbon (GC) electrodes indeed exhibit seemingly simple cyclic voltammograms with two separate peak couples (Fig. 2; the concentration of B8Cl8 used to record this voltammogram is only approximate due to some possible decay of the cluster during transfer to the cell).A close inspection of the current–potential curves, however, shows that at the starting potential of the voltammetric scan, where the BnCln were expected to be stable, an appreciable oxidation current flows, even though the electrode is held at this potential for a “quiet time” of 10 s before the scan is actually initiated. This indicates that at the beginning of the experiment a species is present which can be oxidized at the rather positive starting potentials.It should be noted that published voltammograms of B9Cl9 and B9Br9 exhibit the same feature.38,39 The rest potential, Erest, which is the potential at which no current flows through the working electrode in a particular electrolyte, provides a measure of the potential region where the initial species in the electrolyte is stable. Experimental determinations of Erest in the BnCln solutions immediately after dissolution of the neutral halides indeed result in values positive of the more anodic of the two peak couples in the voltammograms.However, Erest is not stable and decreases to less positive values (Fig. 3). After some time a stable state is reached, with Erest now located between the two peak couples of the respective voltammogram. Hence, the neutral clusters seem to undergo a reaction in the electrolyte to a product which is a less strong oxidant. If the cyclic voltammetric starting potential is selected close to the steady-state value of Erest, current–potential curves with a negligible current at Estart can be recorded for both clusters (Fig. 4). For the discussion below, only voltammograms recorded from such a starting potential were used. They show that the starting species is formed essentially quantitatively after dissolution and equilibration, and that it can be both oxidized and reduced in at least partially chemically and electrochemically reversible steps. In the case of B9Cl9 further, less intensive oxidation waves at more positive potentials were also observed.These will, however not be evaluated in the present paper. Fig. 2 Cyclic voltammograms of B9Cl9 (solid line) and B8Cl8 (dotted line) in CH2Cl2–0.1 M NBu4PF6 at a glassy carbon (GC) electrode with starting potentials located at values positive of both redox peak couples; c0(B9Cl9) = 1.7 mM, c0(B8Cl8) ª 2 mM, v = 0.2 V s21.J. Chem. Soc., Dalton Trans., 1999, 1741–1751 1743 Fig. 3 Temporal development of the rest potential Erest in solutions of B9Cl9 (a) and B8Cl8 (b) after dissolution at t = 0 s in CH2Cl2–0.1 M NBu4PF6 at a GC electrode; c0(B9Cl9) = 0.29 mM, c0(B8Cl8) = 0.44 mM.Fig. 4 Cyclic voltammograms of B9Cl9 [(a), c0 = 0.29 mM, platinum electrode v = 0.1 V s21] and B8Cl8 [(b), c0 ª 2 mM, GC electrode, v = 0.2 V s21] in CH2Cl2–0.1 M NBu4PF6 with starting potentials located at steady-state value of rest potential. Identification of starting species For the interpretation of the voltammograms of the boron subhalide clusters it is essential to identify the starting species formed after dissolution of the neutral compounds.The fact that these species are stable at potentials between the two respective redox waves indicates that they might correspond to a compound with an oxidation state intermediate between those of BnCln and BnCln 22, i.e. the radical anion BnCln~2. Such a radical had been prepared in the case of the nine-vertex cluster by reduction of B9Cl9 with a stoichiometric amount of NBu4I or oxidation of B9Cl9 22 with thallium(III) trifluoroacetate and its ESR spectrum was reported with g = 2.018.35 The neutral clusters, on the other hand, are diamagnetic.40,41 The analogous B9I9 cluster undergoes one-electron reduction with organic donor solvents to form B9I9~2 within minutes, but was stable in chlorinated hydrocarbon solutions.42 In contrast to earlier reports,43 ESR signals were observed in BCl3 solutions of B8Cl8 only with a very weak intensity or after addition of water, giving a diVerent g value of 2.031, and were attributed to hydrolysis products.40,41 The presence of B9Cl9~2 in the electrolyte after dissolution of B9Cl9 and equilibration is clearly shown by the ESR signal (g = 2.018, width 20 G, no hyperfine structure, Fig. 5) which is identical to the one reported earlier for the chemically prepared radical anion.35 In the case of B8Cl8 a similar ESR signal was found (g = 2.017, width 25 G, no hyperfine structure, Fig. 5). Proof Fig. 5 The ESR spectra of solutions of B9Cl9 (solid line) and B8Cl8 (dotted line) in CH2Cl2–0.1 M NBu4PF6, 30 min after dissolution, assigned to B9Cl9~2 and B8Cl8~2 respectively. that this ESR resonance is arising from the radical anion B8Cl8~2 follows from investigation of chemically prepared NBu4 1B8Cl8~2 which shows the same ESR spectrum. We thus conclude that B9Cl9 and B8Cl8 are reduced after dissolution in dichloromethane to their respective radical anions in a spontaneous redox process (6).The formation of B9Cl9~2 is BnCln 1 D BnCln~2 1 D~1 (6) observed in solutions of B9Cl9 in dichloromethane without supporting electrolyte to only a small extent, while the intensity of the ESR absorption is much stronger in the electrolyte containing NBu4PF6. After dissolution of B9Cl9 in the electrolyte, during ª30 min a deepening of the solution color to brown is observed. Simultaneously, the ESR intensity increases.After this time the intensity of the ESR signals remains essentially constant, even upon standing overnight. Note that the timescale for this development of the color and the ESR intensity coincides with that of the rest potential variation (see Fig. 3). Possibly, traces of moisture, coming either from the solvent, from the supporting electrolyte, or by diVusion of air into the electrochemical cell, are responsible for the formation of the radical anions BnCln~2. We thus investigated solutions with various concentrations of B8Cl8 in carbon tetrachloride, chloroform, or dichloromethane with diVerent contents of water by using dried and undried solvents.In each case we observed the ESR signal of B8Cl8~2. Only its intensity was varying depending on the contents of water. While in dried dichloromethane for example the intensity was low, it grew by a factor of about 15 after addition of undried, wet CH3Cl2. It is thus obvious that water is responsible for the formation of B8Cl8~2.The corresponding 11B NMR spectra of B8Cl8 solutions also reflect the influence of moisture on the half width and line shape of the B8Cl8 signal. When the dried solvent (CDCl3 or CCl4) and a relatively big amount of B8Cl8 was transferred to the NMR tube by means of vacuum or inert-gas techniques (concentration of B8Cl8 ª 0.03 M), the 11B resonance line was very sharp (h1/2 ª 35 Hz at d 64.8). When the NMR tube was opened to the atmosphere or when not well dried solvents were used the B8Cl8 signal was broadened (h1/2 = 100–200 Hz).This eVect was even stronger when using CD2Cl2 (h1/2 = 500–10001744 J. Chem. Soc., Dalton Trans., 1999, 1741–1751 Hz). Line broadening of the NMR signal can be explained by a rapid exchange of an electron between the radical anion B8Cl8~2 and the neutral cluster. When the concentration of B8Cl8 in dried CD2Cl2 was lower (0.003 M; closer to the situation as met in cyclic voltammetric experiments), no 11B NMR resonance for B8Cl8 could be detected probably because B8Cl8 was nearly quantitatively reduced to the paramagnetic anion B8Cl8~2.Only an extremely weak signal at d 58.2 (B9Cl9) was observed. This compound is probably present from the synthesis (see Experimental section). The intensity of the signal indicates that the concentration is so low that no peak in cyclic voltammograms should be visible. For B9Cl9 a similar eVect of line broadening in the 11B NMR spectrum caused by traces of water was found.By adding an excess of BCl3 under vacuum conditions the linewidth decreased. When NBu4I was added in an equivalent amount to B9Cl9 the 11B NMR signal disappeared. After condensing an excess of elemental bromine onto the mixture the sharp signal of B9Cl9 (h1/2 ª 50 Hz) in the NMR spectrum reappeared. Thus, the overall reversibility of the redox process B9Cl9 1 e2 B9Cl9~2 is proven. Since it is obvious that traces of water are responsible for the formation of the radical anions BnCln~2 (n = 8 or 9) we have to ask how this reduction process can occur.Water itself or in combination with the solvents CCl4, CHCl3 or CH2Cl2 can hardly act as an electron donor. Furthermore, there is no indication of a disproportionation of BnCln leading to BnCln~2 and BnCln~1. It is known that chloroborane clusters are cleaved by water to give B(OH)3, HCl, and H2.43 We did not, however, observe any evolution of hydrogen.Since the redox potentials E(BnCln/ BnCln~2) have rather high values (see Table 3), it could be expected that a neutral BnCln molecule should be reduced instead of H1. Since the voltammetric experiments show that most of the BnCln molecules are reduced to the anions BnCln~2 and because we could not find any other reaction products, it would be necessary that one molecule BnCln reacts completely or nearly completely with the appropriate quantity of water according to eqn. (7), so that only a small amount of BnCln will BnCln 1 3n H2O æÆ n B(OH)3 1 2nH1 1 nHCl 1 2ne2 (7) BnCln 1 e2 æÆ BnCln~2 n = 8 or 9 (8) be destroyed.At present, this hypothesis for the formation of the BnCln~2 seems to be the most reasonable one, based on the experimental facts discussed above. With this information we can explain the observations made during the reaction of B8Cl8 with CH2Cl2 which according to Morrison44 and Emery45 presumably gives the cluster molecules HB9Cl8, H2B9Cl7, and B9Cl9.They noticed that the 11B NMR spectrum does not show any resonance for B8Cl8 dissolved in dichloromethane. However, they found three signals at d 70, 63.7, and 58.5 and assigned them to H2B9Cl7, HB9Cl8, and B9Cl9, respectively along with a further signal at d 40.25 (B–H). In contrast to this, in our NMR experiments, we never observed a signal at d 40, which was supposed to indicate B–H groups. The other peaks we found as well when the least volatile fraction of B8Cl8 samples sublimed from the reaction mixture was used.The two downfield signals at d 70 and 63.7 can be assigned to B11Cl11 (d11B 69.5 44) and B10Cl10 (d11B 63.5 44) since traces of these compounds together with B9Cl9 (d11B 58.2) are present in B8Cl8 samples before adding dichloromethane if B8Cl8 is not separated well from the by-products of its synthesis. Hence, we conclude that B8Cl8 is not reacting with CH2Cl2 to give the clusters HB9Cl8 and H2B9Cl7, but that it is reduced to the paramagnetic radical anion B8Cl8~2 and this cannot be detected any more in the boron NMR spectrum. This result is in accordance with the cyclic voltammetric results, which indicate total disappearance of the neutral cluster upon dissolution in the electrolyte. The spontaneous formation of BnCln~2 from BnCln also explains the result of a bulk electrolysis experiment with B9Cl9.If B9Cl9 were the starting species and were reduced to the stable B9Cl9 22, 2 F were expected to be transferred upon reduction.Similarly, during reoxidation to B9Cl9 the charge should also correspond to 2 F. However, reduction used only ª0.7 F, while reoxidation at 11.8 V results in the transfer of a much larger charge than expected. Taking into account some loss of B9Cl9 during transfer to the cell, the reduction charge thus indicates that only a one-electron step occurs, starting from B9Cl9~2 and leading to B9Cl9 22. On the other hand, oxidation to B9Cl9 is followed by reaction (6) and reformation of the radical anion in a catalytic process (see also below, Electrochemical oxidation of BnCln~2) and a large quantity of charge is transported through the electrolyte.We thus conclude that the stable starting species present in the dichloromethane electrolyte is not BnCln, but BnCln~2 which can be reduced to BnCln 22 and oxidized to BnCln in heterogeneous electron transfers at the electrode surface (Scheme 1); BnCln~2 is formed from BnCln in a homogeneous redox reaction (6).Since essentially all BnCln is transformed into BnCln~2, we can assume the concentration of the radical anion to be practically identical to the initial concentration of the neutral cluster. The loss of 5–6% due to reaction (7) can probably not be detected in electrochemical experiments, since it is within the conventionally assumed current measurement reproducibility of experiments such as those performed here. Having established the starting species and the basic reaction steps of the BnCln 0/~2/22 system in dichloromethane electrolyte, we will now separately discuss the determination of mechanistic, kinetic and thermodynamic parameters for the reduction and oxidation processes of the BnCln~2 from electrochemical experiments. Electrochemical reduction of BnCln~2 Cyclic voltammograms and chronocoulograms of both B9Cl9 and B8Cl8 in CH2Cl2–0.1 M NBu4PF6 were recorded under variation of the concentration c0 of the clusters and the scan rate v or pulse duration t, respectively, in the potential range where reduction of the radical anions was observed.Both platinum and GC electrodes were used. Cyclic voltammetry. Features of cyclic voltammograms from a typical series of experiments are given in Tables 1 and 2 for the reduction of the two boron subhalides. The peak potentials Ep red and Ep ox for the reduction and oxidation peak on the forward and reverse scans of the voltammograms, respectively, are essentially independent of the scan rate and the concentration.The peak potential diVerence DEp is independent of v and close to 58 mV in all cases, indicating a situation close to electrochemical reversibility of the redox process. Independence of DEp from c0 demonstrates that compensation of the iR drop was eVectively performed. The midpoint potential, E� , calculated as the mean value of the two peak potentials, is again independent of v and c0.The electrochemal reversibility of the process is confirmed by the fact that the peak current function ip red/÷vc0 is independent of v and c0. Furthermore, proportionality between ip red and the square root of the scan rate clearly indicates the absence of adsorption of electroactive species. Chemical reversibility, i.e. stability of the BnCln 22 species with respect to Scheme 1 Homogeneous and heterogeneous electron transfers in the system BnCln 0/~2/22.J. Chem. Soc., Dalton Trans., 1999, 1741–1751 1745 Table 1 Typical cyclic voltammetric potential and current features for the reduction of B9Cl9~2 in CH2Cl2–0.1 M NBu4PF6 at a platinum electrode c0/mM 0.34 0.67 mean n/V s21 0.01 0.02 0.05 0.1 0.2 0.5 1.0 2.0 5.0 10.0 0.01 0.02 0.05 0.1 0.2 0.5 1.0 2.0 5.0 10.0 Ep red/V 0.029 0.028 0.026 0.027 0.027 0.028 0.026 0.026 0.027 0.025 0.031 0.031 0.029 0.033 0.031 0.031 0.029 0.028 0.031 0.028 0.029 ± 0.002 Ep ox/V 0.098 0.097 0.091 0.091 0.090 0.091 0.093 0.096 0.096 0.092 0.098 0.096 0.095 0.096 0.097 0.096 0.097 0.097 0.097 0.101 0.095 ± 0.003 DEp/mV 69 69 65 64 63 63 67 70 69 67 67 65 66 63 66 65 68 69 66 73 67 ± 3 E� a/V 0.064 0.063 0.059 0.059 0.059 0.060 0.060 0.061 0.062 0.059 0.065 0.064 0.062 0.065 0.064 0.064 0.063 0.063 0.064 0.065 0.062 ± 0.002 ip red/÷vc0 b 40.5 38.9 38.3 40.8 41.4 39.2 41.1 41.4 44.0 44.0 38.9 38.6 38.6 40.8 40.8 41.1 41.1 41.4 42.0 43.3 40.8 ± 1.6 ip ox/ip red 0.80 0.98 1.03 1.00 1.01 1.05 1.02 1.02 1.01 1.04 0.94 0.99 1.00 0.98 0.98 1.00 1.00 1.01 1.08 1.05 1.00 ± 0.06 a Midpoint potential E� = (Ep ox 1 Ep red)2.b In A cm3 s1/2 V21/2 mol21. Table 2 Typical cyclic voltammetric potential and current features for the reduction of B8Cl8~2 in CH2Cl2–0.1 M NBu4PF6 at a platinum electrode c0/mM 0.21 0.28 mean v/V s21 0.01 0.02 0.05 0.1 0.2 0.5 1.0 2.0 5.0 10.0 0.01 0.02 0.05 0.1 0.2 0.5 1.0 2.0 5.0 10.0 Ep red/V 0.082 0.081 0.079 0.081 0.079 0.079 0.079 0.078 0.076 0.078 0.082 0.083 0.081 0.084 0.083 0.081 0.083 0.083 0.078 0.080 0.081 ± 0.002 Ep ox/V 0.144 0.141 0.144 0.142 0.141 0.141 0.143 0.144 0.140 0.140 0.145 0.144 0.142 0.143 0.143 0.144 0.145 0.147 0.143 0.147 0.143 ± 0.002 DEp/mV 62 60 65 61 62 62 64 66 64 62 63 61 61 59 60 63 62 64 65 67 63 ± 2 E� a/V 0.113 0.111 0.112 0.112 0.110 0.110 0.111 0.111 0.108 0.109 0.114 0.114 0.112 0.114 0.113 0.113 0.114 0.115 0.111 0.114 0.112 ± 0.002 ip red/÷vc0 b 32.3 32.9 33.5 34.8 34.8 35.1 35.4 33.2 33.2 33.5 32.3 32.3 31.6 33.2 33.8 34.2 32.3 32.3 32.6 32.9 33.2 ± 0.9 ip ox/ip red 0.99 1.00 1.03 1.06 1.02 1.03 1.02 1.05 1.06 1.12 0.92 0.98 1.10 1.02 1.02 1.02 1.01 1.04 1.06 1.12 1.03 ± 0.05 a Midpoint potential E� = (Ep ox 1 Ep red)/2.b In A cm3 s1/2 V21/2 mol21. follow-up reactions, is indicated by the values of ip ox/ip red, which are close to 1.0. Only at scan rates below 0.02 V s21 the value of this ratio drops below unity. Under the experimental conditions of this work this could be due to some non-linear diVusion (“edge”) eVects, which become increasingly important at slow scan rates.Also, additional transport by convection may play a role. The peak currents at scan rates above 0.02 V s21, however, allow the determination of the diVusion coeYcient of BnCln~2 in the electrolyte used for the experiments.46 The midpoint potentials for reduction and the diVusion coeYcients of B9Cl9~2 and B8Cl8~2 are given in Table 3 as mean values from several independent experiments.All values are independent of the electrode material used. The standard deviations of the E� results show excellent reproducibility comparable to that within individual experiments (Tables 1 and 2). On the other hand, while ip red/÷vc0 is excellently reproducible within a series of experiments in a single cell set-up, even with variation of the concentration (Tables 1 and 2), the diVusion coe Ycients vary more strongly between set-ups. These variations may be due to problems with the determination of c0 and the limited stability of the neutral boron cluster starting compounds.The diVusion coeYcient of the B8Cl8 species, however, appears consistently higher than that of the larger B9Cl9 species. Table 3 Midpoint potentials E� and diVusion coeYcients D describing electrochemical reduction and transport of BnCln 0/~2/22a Redox process B9Cl9~2 1 e2 B9Cl9 22 B9Cl9~2 B9Cl9 1 e2 B8Cl8~2 1 e2 B8Cl8 22 B8Cl8~2 B8Cl8 1 e2 E� /V 10.064 ± 0.003 10.599 ± 0.003 10.114 ± 0.002 10.959 ± 0.002 106 D/cm2 s21 2 ± 2b 2 ± 1c 1 ± 1b 1 ± 1c 4 ± 1b 3 ± 2c 4 ± 1b 4 ± 2c a Mean values from several independent experiments under variation of scan rate v, concentration c0, and electrode material.b From cyclic voltammograms. c From chronocoulograms.1746 J. Chem. Soc., Dalton Trans., 1999, 1741–1751 Fig. 6 (a) Chronocoulograms for the reduction of B9Cl9~2 (solid line) and B8Cl8~2 (dotted line) in CH2Cl2–0.1 M NBu4PF6, t = 10 s, GC electrode; c0(B9Cl9) = 0.29 mM, c0(B8Cl8) = 0.44 mM.(b) Anson plot for reduction of B8Cl8~2; “time1/2” axis corresponds to t1/2 for the forward part (upper trace) and t1/2 1 (t 2 t)1/2 2 t1/2 for the reverse part (lower trace) of the chronocoulometric experiment. Chronocoulometry. The cyclic voltammetric data were complemented with chronocoulometric results (Fig. 6). Chronocoulometry confirms the electrochemical and chemical reversibility of the reduction of the radical anions by the almost linear plots of Q vs.t1/2 [“Anson plots”; Fig. 6(b)] 47 and the charge ratio 48 Q2t/Qt = 0.41 ± 0.03 for B9Cl9~2 (t is the pulse time, i.e. the time when the potential during the chronocoulometric experiment is switched). In the case of B8Cl8~2, Q2t/Qt is slightly larger (0.63 ± 0.04) than the expected value of 0.41, but does not increase with increasing pulse time. In accordance with the interpretation of the cyclic voltammetric data, we thus exclude a chemical follow-up reaction of B8Cl8 22.The Anson plots do show only a negligible intersection with the charge axis, thus confirming that none of the redox species is adsorbed at either electrode material used. From the slopes of the Anson plots values of the diVusion coeYcients are calculated in good agreement with the results of cyclic voltammetry, but again with rather high standard deviations. The mean values from several independent experiments are given in Table 3.Simulation. The information determined from these quantitative analyses of cyclic voltammograms and chronocoulograms was subsequently used as the basis for simulations of the experimental current–potential curves. A simple quasireversible one-electron transfer under planar diVusion conditions was assumed as the mechanistic model of the reduction. For each of the compounds a single set of system parameters (formal potential E0, diVusion coeYcient D, heterogeneous electron transfer rate constant ks, and transfer coeYcient, a; Table 4) was suYcient to simulate various series of voltammograms at diVerent v, c0 and electrode material.This set was found by varying E0, D, and ks, until an optimum fit was obtained. The diVusion coeYcients of the respective neutral, mono- and di-anionic species were assumed to be identical. The value of a was fixed in the calculations to 0.5 for both compounds. Variation of a did not significantly improve the fits.Comparisons of the simulations to the corresponding experimental curves for both BnCln at various scan rates and a single c0 are shown in Fig. 7. The fit between theory and experiment is excellent, except for the smallest scan rate used, where possibly non-ideal transport eVects are already visible in the experimental data. Thus, the simulations confirm the qualitative mechanistic picture gained so far. Also, the parameters E0 and D obtained from the fitting procedure compare very well Table 4 System parameter a sets used for simulations of the process BnCln~2 1 e2 BnCln 22 Parameter E0/V 106 D/cm2 s21 ks/cm s21 a n = 9 10.067 1 0.015 0.5 n = 8 10.112 4 0.05 0.5 a Parameters describing the detailsistic reaction steps.49 to the midpoint potentials E� and diVusion coeYcients determined before.For both the nine and the eight vertex cluster, values of ks close to the limit of electrochemical reversibility (ks ª 0.1 cm s21) 50 were found.Electrochemical oxidation of BnCln~2 In analogy to their reduction, the anodic oxidation of the BnCln radical anions was investigated in CH2Cl2–0.1 M NBu4PF6. Cyclic voltammetry. Cyclic voltammetric results for the oxidation of the BnCln~2 to the BnCln are collected in Tables 5 and 6. Only the first oxidation of B9Cl9~2 was analysed, and the switching potential for the voltammograms was adjusted accordingly. As in the case of the reduction of the BnCln~2, the peak potential features for the oxidation clearly indicate a oneelectron process close to electrochemical reversibility.Both the oxidation and the reduction peak potentials are independent of c0 and v; the peak potential diVerence is independent of these experimental parameters and close to the reversible limit of 58 mV. Also, the midpoint potential does not depend either on the experimental parameters or on the electrode material used (Pt or GC; for the mean values from several independent experiments see Table 3).On the other hand, the peak current data show that, at least at slow scan rates, some additional chemical reaction of BnCln must take place: for v £ 0.02 V s21 the peak current function ip ox/ ÷vc0 starts to increase, but, moreover, ip red/ip ox clearly decreases to values below 1.0 for v £ 0.5 (B9Cl9~2) or £0.2 V s21 (B8Cl8~2). Computer simulations of the cyclic voltammograms (see below) show that the homogeneous conversion of BnCln into BnCln~2 can explain this behavior.The interpretation of these features of the current–potential curves is hampered by the fact that at scan rates above v = 1 V s21 the reproducibility of the peak current data decreases. Also, in this experimental time regime the background correction leads to artifacts, in particular at the beginning of the voltammetric scan and close to the switching potential. These problems were much more severe for the octaboron cluster as compared to the B9Cl9 system, and also more pronounced for GC as compared to Pt as the electrode material.For these reasons, only data from the limited range of scan rates 0.01 £ v £1 V s21 were evaluated. Here, however, mean values of the voltammetric potential features are reproducible both within an experiment and within several sets of cell set-ups (Tables 5 and 6, as well as Table 3, respectively). The mean values of the diVusion coeYcients as derived from the oxidation peak currents over all independent experiments are also given in Table 3.Similar reasons as given in the case of reduction of the radical anions explain the relatively high standard deviations. Chronocoulometry. Chronocoulometric oxidations of the BnCln~2 cluster species met similar problems as those in the cyclic voltammetric experiments. In particular, shortly after switching back the potential (t � t), distortions of the charge vs. time curves were observed. For pulse lengths longer thanJ.Chem. Soc., Dalton Trans., 1999, 1741–1751 1747 Fig. 7 Simulated (solid lines) and experimental (dots) cyclic voltammograms for the reduction processes of B9Cl9~2 (left; c0 = 0.63 mM, Pt, v = 0.01, 0.05, 0.1, 0.5, 1.0 V s21, from top to bottom) and B8Cl8~2 (right; c0 = 0.28 mM, Pt, v = 0.01, 0.02, 0.1, 0.5, 2.0 V s21). several s, however, still reasonably linear Q vs. t1/2 plots were obtained with negligible intersections with the charge axis. We exclude adsorption of electroactive material at the electrode surface also for the oxidation process of the cluster radical anions.The slopes of these plots were evaluated in order to estimate values for the diVusion coeYcients, and again the results are presented in Table 3. Simulation. The cyclic voltammetric curves of the oxidation processes leading from the boron subhalide radical anions to their neutral redox partners were simulated starting from the parameters determined as discussed in the previous paragraphs (Fig. 8). In this case a more complex reaction model than a simple quasireversible electron transfer was used. We retained the assumption of planar diVusion. However, the homogeneous redox process converting BnCln produced at the electrode back to BnCln~2 was added to the one-electron oxidation (“catalytic” follow-up reaction; ECcat mechanism 46,51,52). Values of E0, D, ks, and the rate constant k for the homogeneous electron transfer step were varied until an optimum fit between experiment and theory was found for 0.01 £ v £ 10 V s21 and two concentrations of BnCln~2.The homogeneous step was assumed to follow first-order kinetics in this model. Again, a = 0.5 was used throughout. Close to the switching potential the fit between experimental and simulated curves is less satisfactory as compared to that for the reductive voltammograms discussed before. However, the changes in the shapes of the voltammograms at slow scan rates (decrease of reverse peak intensity, flattening of forward peak) which are characteristic for the ECcat mechanism46,51,52 are modelled adequately. Table 7 lists the optimum values of the system parameters as found from the best fitting simulations.The results for the E0 compare excellently to the E� determined from the peak potential analysis (Table 3). The diVusion coeYcients for the particular B9Cl9 experiments evaluated in the simulation are somewhat smaller than the mean values in Table 3, but we again attribute this to the low reproducibility of the concentration owing to the reactivity of the starting material.In the case of the B9Cl9~2 oxidation the fitting procedure of DigiSim converged on a value of 1400 cm s21 for the heterogeneous electron transfer rate constant. This indicates that the electron transfer is indeed fully diVusion controlled, i.e. electrochemically reversible. The numerical value, however, is not regarded as significant, since at such large rate constants the features of the cyclic voltammograms do no longer change with ks, and, consequently, the sensitivity of the curves with respect to this parameter becomes close to zero.53 The fitting routine will select a numerical value for this parameter which is strongly influenced by random errors in the data and is expected to have a large statistical uncertainty.54 The DigiSim software does not1748 J.Chem. Soc., Dalton Trans., 1999, 1741–1751 Table 5 Typical cyclic voltammetric potential and current features for the oxidation of B9Cl9~2 in CH2Cl2–0.1 M NBu4PF6 at a platinum electrode c0/mM 0.34 0.67 mean v/V s21 0.01 0.02 0.05 0.1 0.2 0.5 1.0 2.0 5.0 10.0 0.01 0.02 0.05 0.1 0.2 0.5 1.0 2.0 5.0 10.0 Ep ox/V 0.629 0.627 0.630 0.628 0.628 0.628 0.631 0.629 0.628 0.631 0.629 0.627 0.628 0.627 0.627 0.628 0.630 0.630 0.627 0.630 0.628 ± 0.001 Ep red/V 0.560 0.563 0.561 0.565 0.565 0.565 0.562 0.561 0.561 0.560 0.559 0.561 0.564 0.565 0.568 0.567 0.566 0.565 0.563 0.558 0.563 ± 0.003 DEp/mV 69 64 69 63 63 63 69 68 67 71 70 66 64 62 59 61 64 65 64 72 66 ± 2 E� a/V 0.595 0.595 0.596 0.597 0.597 0.597 0.597 0.595 0.595 0.596 0.594 0.594 0.596 0.596 0.598 0.598 0.598 0.598 0.595 0.594 0.596 ± 0.001 ip ox/÷vc0 b 42.4 37.3 39.5 37.3 37.9 36.7 36.0 ——— 40.5 37.6 35.7 37.0 35.1 35.1 34.5 ——— 36.6 ± 1.4 c ip red/ip ox 0.62 0.81 0.85 0.93 0.94 0.99 1.01 ——— 0.60 0.74 0.86 0.87 0.93 0.97 0.98 ——— — a Midpoint potential E� = (Ep ox 1 Ep red)/2.b In A cm3 s1/2 V21/2 mol21. c From values for v > 10 mV s21. Table 6 Typical cyclic voltammetric potential and current features for the oxidation of B8Cl8~2 in CH2Cl2–0.1 M NBu4PF6 at a platinum electrode c0/mM 0.21 0.28 mean v/V s21 0.01 0.02 0.05 0.1 0.2 0.5 1.0 2.0 5.0 10.0 0.01 0.02 0.05 0.1 0.2 0.5 1.0 2.0 5.0 10.0 Ep ox/V 0.991 0.996 0.992 0.987 0.987 0.988 0.987 0.998 0.994 0.993 0.997 0.996 0.994 0.989 0.991 0.991 0.987 0.990 0.996 0.996 0.992 ± 0.003 Ep red/V 0.925 0.924 0.0.927 0.926 0.927 0.925 0.930 0.923 0.926 0.924 0.926 0.929 0.928 0.929 0.932 0.926 0.926 0.924 0.927 ± 0.002 DEp/mV 66 72 63 61 60 62 60 64 64 70 71 72 68 60 63 62 55 64 70 72 65 ± 5 E� a/V 0.958 0.960 0.961 0.957 0.957 0.957 0.957 0.957 0.962 0.958 0.962 0.960 0.960 0.959 0.960 0.960 0.960 0.958 0.961 0.960 0.959 ± 0.002 ip ox/÷vc0 b 45.9 42.7 41.1 33.8 33.2 32.6 37.6 ——— 43.0 41.1 38.9 37.6 37.6 32.6 34.8 ——— 35.0 ± 2.2 c ip red/ip ox 0.59 0.72 0.72 0.91 0.96 1.04 0.96 ——— 0.62 0.78 0.82 0.87 0.88 1.07 1.07 ———— a Midpoint potential E� = (Ep ox 1 Ep red)/2.b In A cm3 s1/2 V21/2 mol21. c From values for v > 50 mV s21. provide direct quantitative measures of the standard deviations of the parameter values estimated by fitting. Consequently, no further analysis is possible. The heterogeneous electron transfer rate constant determined by the fitting procedure for the oxidation of B8Cl8~2 is still rather close to the reversibility/quasireversibility border.The characteristic changes in the shapes of the voltammograms at slow scan rates allow the determination of the rate of the homogeneous redox process (Table 7). Its value is similar for both clusters. Thus, simulation of these voltammograms confirms the reaction mechanism proposed and the system parameters derived from the peak features and chronocoulograms. Furthermore, it allows determination of the rate constants of the homogeneous redox processes under the assumption of first-order kinetics.Formal potentials and stability of the radical anions BnCln~2 In previous paragraphs we have shown that both the reduction and (at large scan rates) the oxidation of B9Cl9~2 and B8Cl8~2 are chemically reversible processes with rates of the electron transfers close to or in the region of electrochemical reversibility. Under such conditions and under the assumption of equal diVusion coeYcients for the three redox partners, respectively, the midpoint potentials, calculated as the mean values of the peak potentials, are good approximations for the formal potentials of the respective electron transfer processes.This is confirmed by the simulations which result in optimum E0 values identical to the E� within one standard deviation. The formal potentials thus determined for the BnCln systems in CH2Cl2 show a normal ordering, i.e. they increase with the oxidation state involved.The relative position of the E0 for the redox processes of B9Cl9~2 is very similar in CH2Cl2 (|DE0| = 0.533, this work; 0.51;38 0.53 V 39) and CH3CN (|DE0| = 0.540 V37). On the other hand, the absolute values determined here and in the work of other authors are not comparable due to the use of diVerent reference standards and the possibility of eVects of the halogenated substrates on the potential of the Ag–AgCl electrode used in the earlier work.37,38 If indeed the cluster radical anions are produced by hydrolysis of some of theJ.Chem. Soc., Dalton Trans., 1999, 1741–1751 1749 Fig. 8 Simulated (solid lines) and experimental (dots) cyclic voltammograms for the oxidation processes of B9Cl9~2 (left; c0 = 0.98 mM, GC, v = 0.01, 0.05, 0.1, 0.5, 1.0 V s21, from top to bottom) and B8Cl8~2 (right; c0 = 0.28 mM, Pt, v = 0.01, 0.02, 0.1, 0.5, 2.0 V s21). subhalide molecules, as formulated in eqns. (7) and (8), chloride ions are liberated which will shift the reference potential.Our reference system should not be aVected by such processes. Only the data in ref. 39 (E0 1 = 1 0.10 V and E0 2 1 0.63 V vs. Fc/Fc1) seem to have been determined with careful exclusion of such eVects. They diVer from our values by less than 40 mV. From the formal potentials the equilibrium constants Kcomp of reaction (4) follow through eqn. (5) as 1.1 × 109 (n = 9, in close agreement with Kcomp = 1.2 × 109 in ref. 39) and 1.9 × 1014 (n = 8).Both equilibria are strongly shifted to the side of the radical anions, which are thus rather stable with respect to disproportionation. Results for the B9Br9,37–39 B9I9,37,39 and B10Cl10 37 systems show a similar picture. It appears that the Table 7 System parameter a sets for simulations of the process BnCln~2 2 e2 BnCln Parameter E0/V D/cm2 s21 ks/cm s21 a k/s21 n = 9 10.600 6 × 1027 —b 0.5 0.09 n = 8 10.959 5.5 × 1026 0.13 0.5 0.07 a Parameters describing the details of the mechanistic reaction steps.49 b Electron transfer fully diVusion controlled.smaller cluster radical anion is even more stable in this respect than B9Cl9~2. Conclusion The electrochemical investigation of two electron hyperdeficient boron subhalides shows that both B9Cl9 and B8Cl8 and their respective radical anions and dianions can be interconverted at an electrode in a dichloromethane electrolyte through one-electron processes, well separated in potential.The reduction of the neutral clusters to the radical anions proceeds at rather positive potentials at the electrode, and additionally spontaneously with an electrolyte component. In view of the hypothetical rationalization of the “potential inversion” 6 phenomenon, the stepwise manner of electron transfer in the clusters investigated nicely mirrors the fact that probably all oxidation states attain the deltahedral closo type structure without drastic geometrical changes accompanying the redox process.Possibly, more pronounced structural rearrangements would be noticeable in smaller clusters of this series, such as the tetrahedral B4Cl4, where chemical reduction with trimethylstannane leads to the butterfly-shaped arachno-B4H10.44 However, reduction of B4Cl4 without simultaneous transfer of hydrogen has not yet been observed. It is thus planned to investigate the1750 J. Chem. Soc., Dalton Trans., 1999, 1741–1751 electrochemical reduction of such boron subhalide clusters of smaller size in future work.Experimental Solvents and supporting electrolyte Dichloromethane (Burdick & Jackson) was distilled to separate the stabilizing cyclohexene and dried by standing for several hours over activated basic Al2O3 (activation procedure: 4 h at a temperature of 400 8C and a pressure of 2 × 1023 mbar). Tetran- butylammonium hexafluorophosphate, NBu4PF6, was prepared from NBu4Br and NH4PF6 (Fluka) as described before.55 It was used in a concentration of 0.1 M.The electrolyte was degassed by three freeze–pump–thaw cycles before transferring it into the electrochemical cell under argon. Carbon tetrachloride (p.a., Merck) and the deuteriated solvents were dried over molecular sieves; NBu4I (puriss.) was purchased from Fluka. Syntheses The synthesis and all manipulations of the chloroboranes B8Cl8 and B9Cl9 were carried out by using standard high-vacuum or inert-atmosphere techniques as described by Shriver and Drezdzon.56 The compound B2Cl4 was obtained by the reaction of BCl3 with copper vapor 57 and purified by fractional condensation until it showed a vapor pressure of 59 mbar at 0 8C. Nonachlorononaborane(9).The compound B9Cl9 was prepared by heating B2Cl4 at 450 8C for 5 min under vacuum according to the procedure reported by Morrison.44,58 The product was purified by fractional sublimation into a long glass tube connected to a high-vacuum line. Octachlorooctaborane(8).The compound B8Cl8 was prepared similar to the synthesis described by Morrison.44,59 In a typical experiment, a solution of 3.7 g B2Cl4 (22.6 mmol) in 12.5 g of CCl4 was heated in a 100 ml flask under argon at 95 8C for 14 d. After evaporation of all volatile material (BCl3, B2Cl4, CCl4) at 0 8C (1024 mbar) a black residue remained, which contained, according to the 11B NMR spectrum, B8Cl8 (d 64.8 in CDCl3, 93 mol%), B9Cl9 (d 58.2, approximately 7 mol%), as well as traces of B10Cl10 (d11B 63.2, cf.ref. 44; 63.5) and an unidentified boron compound with d11B 51.7. The compound B8Cl8 was separated from the reaction mixture and purified by fractionated sublimation under vacuum (1024 mbar). Yield: 100 mg (0.27 mmol, 10% based upon B2Cl4). It should be noted that thick layers of B8Cl8 are nearly black, whereas thin layok dark green and become purple upon contact with traces of air. Electrochemical experiments All electrochemical experiments were performed with a Bioanalytical Systems (BAS, West Lafayette, IN, USA) 100 B/W electrochemical workstation controlled by a standard 80486 processor based personal computer (control program version 2.0).For electroanalytical experiments a BAS platinum or glassy carbon electrode tip was used as the working electrode. The electroactive area of the disks was determined from cyclic voltammograms, chronoamperograms, and chronocoulograms of ferrocene in dichloromethane under the assumption of a diVusion coeYcient D(Fc) = 2.32 × 1025 cm2 s21.60 The counter electrode was a platinum wire (diameter: 1 mm).A Haber–Luggin double reference electrode61 was used. The resulting potential values refer to Ag–Ag1 (0.01 M in CH3CN–0.1 M NBu4PF6). Ferrocene was used as an external standard.62 Its potential was determined by separate cyclic voltammetric experiments in CH2Cl2. All potentials reported in this paper are rescaled to E0(Fc–Fc1) = 10.226 V (vs.the Ag–Ag1 reference) and thus given vs. the Fc–Fc1 redox potential. For cyclic voltammetry, chronoamperometry and chronocoulometry a gas-tight full-glass three-electrode cell as described before 55 was used. It was purged with argon before being filled with the electrolyte. Background curves were recorded before adding the substrate to the electrolyte. The background currents were later subtracted from the experimental data measured in the presence of substrate. The uncompensated resistance in the cell was determined by the built-in procedure of the BAS 100 B/W instrument.For each scan rate a series of cyclic voltammograms was recorded with 70, 80, and 90% feed-back compensation of the iR drop. This was repeated for at least a second concentration in the same cell set-up. The resulting current–potential curves were compared and optimum compensation was assumed if the peak potential separation did not increase with concentration. The instability of the boron subhalides with respect to oxygen and traces of water required special precautions during weighing of the compounds and transfer of the samples to the electrochemical cell.Weighing was performed under argon. A concentrated stock solution was prepared with the degassed electrolyte and defined volumes of this solution were added to the blank electrolyte in the cell. After registration of all necessary voltammograms and chronocoulograms, further portions of the stock solution were added without changing the electrode arrangement.In this way at least two series of curves were recorded in each experiment with diVerent concentrations but otherwise identical conditions. For electrolysis experiments (bulk electrolysis), working and counter electrodes were Pt/Ir 90/10 nets (Degussa, Hanau, Germany), separated by a glass frit. The bulk electrolysis cell was also gas-tight and its temperature was controlled to be 17 8C. It was purged with argon prior to being filled with electrolyte.Rest potential measurements were performed using the standard experimental protocol of the BAS 100 B/W electrochemical workstation. Data analysis and simulations Cyclic voltammetric and chronocoulometric data were background corrected and evaluated with the BAS 100 B/W control program. Peak current ratios were determined according to Nicholson’s procedure.63 All error measures given in this paper are standard deviations. For simulations of the cyclic voltammograms the commercial simulator DigiSim 64 (Version 2.0) was used with standard numerical options.ESR and NMR spectra A Bruker ESP 300 spectrometer was used to record the ESR spectra. Preparation of the solution was similar to that for the electroanalytical experiments. Spectra were taken at various times after dissolution. For the determination of the g values the spectrometer was calibrated with Bruker “strong pitch” of g = 2.0028. The 11B NMR spectra at 80.25 MHz were obtained on a Bruker WM 250 spectrometer. All 11B NMR chemical shifts are referred to external F3B?OEt2 in CDCl3 or CD2Cl2, respectively.Investigation of B8Cl8 and NBu4 1B8Cl8~2 solutions. Solutions of B8Cl8 in CDCl3 or CCl4 were prepared by condensing the solvent (which was dried before with molecular sieves) onto B8Cl8 under vacuum. The solution was transferred under argon with a syringe to an NMR tube equipped with a polytetra- fluoroethylene (PTFE) valve.For the ESR measurements, B8Cl8 and the purified and dried solvent (CDCl3 or CH2Cl2) were condensed under vacuum into an ESR glass tube equipped with a PTFE valve or the components were condensed together in a flask connected to the vacuum line and transferred under argon with a syringe into the ESR tube. The NBu4 1B8Cl8~2 wasJ. Chem. Soc., Dalton Trans., 1999, 1741–1751 1751 prepared according to the synthesis of NBu4 1B9Cl9~235 by reduction of B8Cl8 with the equivalent amount of NBu4I in dried CDCl3. Acknowledgements The authors thank Paul Schuler for recording the ESR spectra, Stefan Dümmling for technical assistance and the Fonds der Chemischen Industrie, Frankfurt/Main, Germany, for financial support.References 1 B. Speiser and S. Dümmling, Part 1: DECHEMA-Monogr., in the press. 2 K. Deuchert and S. Hünig, Angew. Chem., 1978, 90, 927; Angew. Chem., Int. Ed. Engl., 1978, 17, 875. 3 N. G. Connelly and W. E. Geiger, Adv. Organomet.Chem., 1984, 23, 1. 4 J. Phelps and A. J. Bard, J. Electroanal. Chem. Interfacial Electrochem., 1976, 68, 313. 5 D. H. Evans, Acta Chem. Scand., 1998, 52, 194. 6 D. H. Evans and K. Hu, J. Chem. Soc., Faraday Trans., 1996, 3983. 7 K. Hu and D. H. Evans, J. Electroanal. Chem. Interfacial Electrochem., 1997, 423, 29. 8 K. Hu, M. E. Niyazymbetov and D. H. Evans, J. Electroanal. Chem. Interfacial Electrochem., 1995, 396, 457. 9 K. Hu and D. H. Evans, J. Phys. Chem., 1996, 100, 3030. 10 B. Speiser, M. Würde and C. Maichle-Mössmer, Chem. Eur. J., 1998, 4, 222. 11 B. Tulyathan and W. E. Geiger, J. Am. Chem. Soc., 1985, 107, 5960. 12 J. A. Morrison, in Advances in Boron and the Boranes, eds. J. F. Liebman, A. Greenberg, R. E. Williams, D. P. Loker and K. B. Loker, VCH, Weinheim, 1988, vol. 5, ch. 8, pp. 151–189. 13 R. L. Johnston and D. M. P. Mingos, Inorg. Chem., 1986, 25, 3321. 14 K. Wade, Adv. Inorg. Chem. Radiochem., 1976, 18, 1. 15 R. A. Jacobson and W. N.Lipscomb, J. Am. Chem. Soc., 1958, 80, 5571. 16 M. Atoji and W. N. Lipscomb, J. Chem. Phys., 1959, 31, 601. 17 R. A. Jacobson and W. N. Lipscomb, J. Chem. Phys., 1959, 31, 605. 18 L. J. Guggenberger, Inorg. Chem., 1969, 8, 2771. 19 W. Hönle, Y. Grin, A. Burkhardt, U. Wedig, M. Schultheiss, H. G. von Schnering, R. Kellner and H. Binder, J. Solid State Chem., 1997, 113, 59. 20 M. B. Hursthouse, J. Kane and A. G. Massey, Nature (London), 1970, 228, 659. 21 L. J. Guggenberger, Inorg.Chem., 1969, 7, 2260. 22 J. Thesing, J. Baurmeister, W. Preetz, D. Thiery and H. G. von Schnering, Z. Naturforsch., Teil B, 1991, 46, 800. 23 W. Preetz and J. Fritze, Z. Naturforsch., Teil B, 1984, 39, 1472. 24 A. Heinrich, H.-L. Keller and W. Preetz, Z. Naturforsch., Teil B, 1990, 45, 184. 25 R. SchaeVer, Q. Johnson and G. S. Smith, Inorg. Chem., 1965, 4, 917. 26 M. L. McKee, Inorg. Chem., 1999, 38, 321. 27 M. Atoji and W. N. Lipscomb, Acta Crystallogr., 1953, 6, 547. 28 M. J. S. Dewar and M. L. McKee, Inorg. Chem., 1978, 17, 1569. 29 A. Neu, T. Mennekes, U. Englert, P. Paetzold, M. Hofmann and P. von R. Schleyer, Angew. Chem., 1997, 1094, 2211; Angew. Chem., Int. Ed. Engl., 1997, 36, 2117. 30 M. E. O’Neill and K. Wade, Inorg. Chem., 1982, 21, 461. 31 M. E. O’Neill and K. Wade, J. Mol. Struct., 1983, 103, 259. 32 M. A. Fox and K. Wade, in The Borane, Carborane, Carbocation Continuum, ed. J. Casanova, Wiley, New York, 1998, ch. 2, pp. 62–64. 33 R. W. Rudolph and W. R. Pretzer, Inorg. Chem., 1972, 11, 1974. 34 D. A. Kleier and W. N. Lipscomb, Inorg. Chem., 1979, 18, 1312. 35 E. H. Wong and R. M. Kabbani, Inorg. Chem., 1980, 19, 451. 36 F. Klanberg, D. R. Eaton, L. J. Guggenberger and E. L. Muetterties, Inorg. Chem., 1967, 6, 1271. 37 W. Bowden, J. Electrochem. Soc., 1982, 129, 1249. 38 R. Kellner, Ph.D. Thesis, Universität Stuttgart, 1997. 39 H. Binder, R. Kellner, K. Vaas, M. Hein, F. Baumann, M. Wanner, R. Winter, W. Kaim, W. Hönle, Y. Grin, U. Wedig, M. Schulthesis, R. K. Kremer, H. G. von Schnering, O. Groeger and G. Engelhardt, personal communication. 40 G. F. Lanthier and A. G. Massey, J. Inorg. Nucl. Chem., 1970, 32, 1807. 41 G. F. Lanthier, J. Kane and A. G. Massey, J. Inorg. Nucl. Chem., 1971, 33, 1569. 42 E. H. Wong, Inorg. Chem., 1981, 20, 1300. 43 E. P. Schram and G. Urry, Inorg. Chem., 1963, 2, 405. 44 J. A. Morrison, Chem. Rev., 1991, 91, 35. 45 S. L. Emery, Ph.D. Thesis, University of Illinois, Chicago, 1985. 46 R. S. Nicholson and I. Shain, Anal. Chem., 1964, 36, 706. 47 F. C. Anson, Anal. Chem., 1966, 38, 54. 48 A. J. Bard and L. R. Faulkner, Electrochemical Methods. Fundamentals and Applications, Wiley, New York, 1980, p. 201 V. 49 B. Speiser, Anal. Chem., 1985, 57, 1390. 50 J. Heinze, Angew. Chem., 1984, 96, 823; Angew. Chem., Int. Ed. Engl., 1984, 23, 831. 51 J. M. Savéant and E. Vianello, Adv. Polarography, 1960, 2, 367. 52 J. M. Savéant and E. Vianello, Electrochim. Acta, 1965, 10, 905. 53 L. K. Bieniasz and B. Speiser, J. Electroanal. Chem., 1998, 441, 271. 54 L. K. Bieniasz and B. Speiser, J. Electroanal. Chem., 1998, 458, 209. 55 S. Dümmling, E. Eichhorn, S. Schneider, B. Speiser and M. Würde, Curr. Sep., 1996, 15, 53. 56 D. F. Shriver and M. A. Drezdzon, The Manipulation of Air- Sensitive Compounds, 2nd edn., Wiley, New York, 1986. 57 P. L. Timms, Adv. Inorg. Chem. Radiochem., 1972, 14, 121. 58 T. Davan and J. A. Morrison, Inorg. Chem., 1986, 25, 2366. 59 S. L. Emery and J. A. Morrison, J. Am. Chem. Soc., 1982, 104, 6790. 60 J. B. Cooper and A. M. Bond, J. Electroanal. Chem. Interfacial Electrochem., 1991, 315, 143. 61 B. Gollas, B. Krauß, B. Speiser and H. Stahl, Curr. Sep., 1994, 13, 42. 62 G. Gritzner and J. Ku° ta, Pure Appl. Chem., 1984, 56, 461. 63 R. S. Nicholson, Anal. Chem., 1966, 38, 1406. 64 M. Rudolph, D. P. Reddy and S. W. Feldberg, Anal. Chem., 1994, 66, 589A. Paper 8/09134J