Complexes of Cross-Bridged Tetraazamacrocycles The Proton Sponge Problem In synthetic chemistry, large gains are usually achieved by overcoming

Transcription

1 Complexes of Cross-Bridged Tetraazamacrocycles The Proton Sponge Problem In synthetic chemistry, large gains are usually achieved by overcoming substantial challenges, and that is certainly the case here. Only Cu 2+ and Ni 2+, 113 the thermodynamically strongest binding (first row, divalent) transition metal ions, 141 had been complexed to ethylene cross-bridged tetraazamacrocycles prior to this work. Initial attempts at the complexation of these ligands with Fe 2+ and Mn 2+ salts in protic solvents or in aprotic solvents using hydrated metal salts were entirely unsuccessful. It was therefore soon concluded that Cu 2+ and Ni 2+ are much better at competing with protons for the binding cavity in protic solvents than most divalent transition metal ions. This is especially true of the ions of primary interest here, the relatively hard oxyphilic ions of manganese and iron. This behavior was attributed to the proton-sponge nature of the ligands. Weisman et al. studied the protonation of B14N4 by NMR techniques, reporting that pk a2 > 24 and pk a1 = This value has been confirmed by titration in an aqueous medium (Figure 49a). The diprotonated form of free ligand B14N4 behaves as a monoprotic weak acid with a pk a1 of 9.58(3) even though stoichiometrically there are two H + bound, indicating the pk a2 must be substantially greater than The difference between the value for pk a1 from our measurements and that from the previous study probably arises from the different solvents and methods used. This difference does not change the conclusion that this ligand binds at least one proton very, very strongly, and that only the most strongly binding metal ions can compete with that proton for the cavity it occupies. 67

2 ph Equivalents Base ph Equivalents Base Equivalents Base ph Figure 49. Titration curves for a) Me 2 B14N4, b) Me 2 B13N4, c) Me 2 B12N4, and d) Me 2 B14N4Me 6 B13N4 behaves similarly to B14N4 under aqueous conditions, exhibiting a single observable deprotonation of pk a2 = 6.71(2). However, the elemental analysis and the stoichiometry of the titration point to the presence of two additional protons whose pk a s cannot be determined under normal aqueous conditions. Figure 49b shows the calculated and observed titration data for B13N4. From this result, it is clear that the first proton dissociates with pk a1 < 2 and that the third proton dissociates with pk a3 > 13. Thus, in addition to having the proton-sponge character associated with B14N4, B13N4 is also able to bind a third proton, but very weakly. 68

3 B12N4 behaves as a tribasic ligand as well. New experiments confirm this behavior initially reported by Bencini, et al 79 who have synthesized the ligand by a different route and examined its solution behavior. This previous study found only pk a2 = 5.95(1) and suggested pk a1 < 1, since deprotonation occurred below the observed ph region, and pk a3 > 13 since its deprotonation was not observed even at the maximum ph studied. New potentiometric titrations (Figure 49c) confirm that the first deprotonation occurs below the ph region normally studied in aqueous solutions. However, it was possible to observe both pk a2 and pk a3 under these conditions. According to the fit, pk a2 = 5.77(2), which is very similar to the value reported previously, but it was found that pk a3 = 11.3(2) which does not agree well with the earlier study. The reliability of pk a3 is rather low due to the high ph necessary for full deprotonation, but it is nevertheless at a value observable in many other ligand titrations. 168 Figure 50. Molecular structures of a) H 2 (Me 2 B14N4Me 6 ) 2+, and b) H 3 (Me 2 B12N4) 3+ Removal of this second constant from the fit resulted in much higher GOF values, which indicates it is actually observed. The fact that all three deprotonations occur in a normal range indicates that ligand B12N4 is not a true proton sponge and that it behaves much more like an unbridged tetraazamacrocycle. 168 This difference from ligands B14N4 and B13N4 must arise from the smaller ring size of the 12- membered parent macrocycle, producing a shallower cavity which binds more protons than B14N4, but binds them less strongly. Crystal structures of triprotonated B12N4 and diprotonated B14N4 and B14N4Me 6 have been published and demonstrate a structural relationship to this difference (Figure 50). In diprotonated B14N4 and B14N4Me 6, the two methylated nitrogens are protonated with strong hydrogen bonds to one bridgehead nitrogen each, forming six- 69

4 membered H-bonded rings. In triprotonated B12N4, both methylated nitrogens are protonated with no strong hydrogen bonds and one bridgehead nitrogen is protonated with a strong hydrogen bond to the other bridgehead nitrogen. No doubt this difference in hydrogen bonding pattern and the difference observed in solution behavior arises from cavity size and flexibility. Although the crystal structure of triprotonated B13N4 has not yet been obtained, we can postulate that it has characteristics of both B14N4 and B12N4. Because of the single trimethylene chain, a six-membered H-bonded ring can form; this is the probable site of the most strongly held proton, and is responsible for the proton sponge character of B13N4. The other half of the ligand is like B12N4, containing all ethylene chains, and is probably the site of the other two, less strongly held protons. The potentiometric titration of B14N4Me 6 reveals it to behave like B14N4 rather than like B12N4. B14N4Me 6 is stoichiometrically diprotonated in aqueous solution, but like B14N4 and B13N4, exhibits only one observable pk a under normal aqueous conditions (Figure 49d). The experiment assigned pk a1 = 11.45(3), which is the highest first deprotonation of all four ligands studied. Again, pk a2 was not observed, indicating that it is greater than 13 in water. The increase in basicity of B14N4Me 6 vs. B14N4 is possibly due to the increase in rigidity associated with the presence of six methyl substituents on carbon atoms of the ring. The six methyl substituents on the ring should enhance the immobilization of the nitrogen donors. Crystal structures of diprotonated B14N4 27 and B14N4Me 6 (vide supra) show that their bicyclic skeletons have virtually identical conformations (Figures 50b). Both ligands have their methylated nitrogens protonated with strong hydrogen bonds to neighboring bridghead nitrogens, an energetically favorable arrangement apparently responsible for the exceptional proton affinities of these ligands. Having identified the obstacle, the proton sponge problem, it was then rational to overcome it by minimizing the activity of protons in the reaction system. In this 70

5 chapter, the complexes of a broad array of transition metal ions have been prepared with these ligands, often only through the use of anhydrous metal reagents with strictly deprotonated ligands in rigorously dry, aprotic solvents under a dry, inert atmosphere. 114,169 Discussion shall begin with the most easily prepared copper and zinc complexes, then proceed to the most interesting iron and manganese chemistry, and finally conclude with the cobalt and nickel complexes, which further illuminate the ligand properties through comparison with their vast literature. Synthesis and Characterization of Cu 2+ and Zn 2+ Complexes Complex Synthesis and Structure. Copper(II) complexes of the ligands B14N4, B14N4Me 6, B13N4, and B12N4 were obtained by their reaction with CuCl 2 2H 2 O in MeOH, as previously reported in the work of Weisman, et al. with similar ligands. 113 Only the complex of B14N4 could not be purified as the chloride salt - ([CuLCl]Cl) and was made into the pure PF 6 salt by anion methathesis. Zinc(II) complexes with B14N4, B13N4, and B12N4 were obtained through reaction of anhydrous ZnCl 2 with the ligands in acetonitrile under nitrogen. The reactions were performed under an inert atmosphere, not to protect the metal from oxygen (as is normally the case) but to protect the ligand from the air. The unprotonated ligands are generally viscous colorless oils that yellow and become cloudy if stored in the air but remain clear and colorless if stored under an inert atmosphere. All of the complexes have been characterized by elemental analysis and mass spectrometry and the structures of three of the Cu 2+ complexes and one with Zn 2+ have been determined by X-ray crystallography. The structures of the Cu 2+ complexes of B14N4 and B14N4Me 6 reveal pentacoordinate Cu 2+ bound to four nitrogen donors from the cross-bridged ligand and a single chloride (Figure 51, Table 6). Both ligands adopt cavity-like conformations engulfing the metal ion, as observed in two earlier structures for Cu 2+ complexes with 71

6 Figure 51. Molecular structures of a) Cu(Me 2 B14N4)Cl + and b) Cu(Me 2 B14N4Me 6 )Cl + Table 6. Selected bond lengths (Å) and angles ( o ) for Cu(Me 2 B14N4)Cl + and Cu(Me 2 B14N4Me 6 )Cl + (i) For Cu(Me 2 (B14N4))Cl + Cu(1)-N(12) 2.093(3) Cu(1)-N(8) 2.170(3) Cu(1)-N(5) 2.108(3) Cu(1)-Cl(1) (12) Cu(1)-N(1) 2.107(3) N(12)-Cu(1)-N(5) (13) N(1)-Cu(1)-N(8) 85.30(12) N(12)-Cu(1)-N(1) 84.74(13) N(12)-Cu(1)-Cl(1) 90.60(10) N(5)-Cu(1)-N(1) 91.22(13) N(5)-Cu(1)-Cl(1) 94.16(10) N(12)-Cu(1)-N(8) 92.04(12) N(1)-Cu(1)-Cl(1) (10) N(5)-Cu(1)-N(8) 84.96(12) N(8)-Cu(1)-Cl(1) (9) (ii) For Cu(Me 2 (B14N4Me 6 ))Cl + Cu(1)-N(8) 2.114(2) Cu(1)-N(11) 2.116(2) Cu(1)-N(4) 2.128(2) Cu(1)-N(1) 2.196(2) Cu(1)-Cl(1) (10) N(8)-Cu(1)-N(4) 92.88(8) N(11)-Cu(1)-N(4) (8) N(8)-Cu(1)-N(1) 87.51(8) N(11)-Cu(1)-N(1) 92.76(8) N(4)-Cu(1)-N(1) 83.08(8) N(8)-Cu(1)-Cl(1) (6) N(11)-Cu(1)-Cl(1) 91.21(6) N(4)-Cu(1)-Cl(1) 93.42(6) N(1)-Cu(1)-Cl(1) (6) N(8)-Cu(1)-N(11) 84.39(8) related ligands. 113 The B14N4 complex (Figure 51a) is best described as tetragonal, having the chloride and three nitrogens in a near planar arrangement, while the fourth nitrogen donor is at the apex of the square pyramid. The reason for this structure rather than the expected trigonal bipyramid is unclear, although steric requirements of the ligand should not eliminate either possible geometry. The trigonal bipyramidal geometry is, however, observed for the B14N4Me 6 complex (Figure 51b), where two nitrogen donors from the macrobicycle occupy the axial positions while the other two nitrogens and a chloride fill the three equatorial sites. The difference between the two structures can be readily traced to the six C-methyl groups of the B14N4Me 6 ligand. The two sets of geminal dimethyl groups α to equatorial nitrogen donors N(1) and N(8) in Figure 51b approach the chloride ligand no closer than 3.0 Å in this trigonal bipyramidal structure. However, if a square pyramidal geometry was enforced on this complex, CAChe molecular modeling has shown the 72

7 distance between chloride and one of the geminal dimethyl groups would have to be about 2.4 Å, clearly an unfavorable proximity. The two previously published 16 Cu 2+ structures of related R 2 B14N4 ligands having similar bulk to B14N4 find Cu 2+ in geometries also distorted towards square pyramidal and away from trigonal bipyramidal. One has a pseudo-octahedral geometry where the sixth ligand involves an agostic interaction with a benzyl hydrogen at a distance of 2.74 Å and the other is five coordinate and significantly closer in structure to a square pyramid (equatorial angles = 87.8(2), 122.0(1), and 150.2(2) ) than is [Cu(B14N4Me 6 )Cl] + (equatorial angles = 87.51(8), (6), and (6) ). From these four related structures it may be concluded that for a non-sterically crowded ligand with the cross-bridged N 4 Cl donor set (ie; B14N4) Cu 2+ apparently favors a square pyramidal geometry, but that the steric bulk of B14N4Me 6 forces the trigonal bipyramidal geometry on copper. Both structures with ligand B12N4 reveal hexacoordinate metal ions bound to four nitrogen donors from the cross-bridged ligand and to two labile, monodentate Figure 52. Molecular structures of a) Cu(Me 2 B12N4)(MeCN) 2 2+, and b) Zn(Me 2 B12N4)Cl 2 ligands, chloride in the case of the Zn 2+ complex and acetonitrile in the Cu 2+ case (Figure 52, Table 7). The ligand in both structures adopts the same cavitylike conformation engulfing the metal ion as seen in the above larger ligands. The structures with B12N4 also confirm that the smaller 12-membered ring system behaves similarly to the 14-membered system of B14N4 and its analogues in transition metal coordination, contrary to an earlier report 79 which suggested this small ligand could only coordinate transition metal ions weakly, in an externally bound fashion. The Zn 2+ structure is as expected with a pdeudo octahedral geometry completed by two 73

8 Table 7. Selected bond lengths (Å) and angles ( o ) for Cu(Me 2 B12N4)(MeCN) 2+ and Zn(Me 2 B12N4)Cl 2 2+ (i) For Cu(Me 2 B12N4)(MeCN) 2 Cu1-N (2) Cu1-N (4) Cu1-N (2) N4-Cu1-N4# (13) N1-Cu1-N (12) N4-Cu1-N (9) N4-Cu1-N01# (10) N4-Cu1-N1# (9) N1-Cu1-N01# (11) N1-Cu1-N1# (14) N01-Cu1-N01# (18) N4-Cu1-N (11) (ii) For Zn(Me 2 B12N4)Cl 2 Zn(1)-N(11) 2.217(2) Zn(2)-N(21) 2.222(2) Zn(1)-N(17) 2.237(2) Zn(2)-N(27) 2.227(2) Zn(1)-N(110) 2.248(2) Zn(2)-N(210) 2.235(2) Zn(1)-N(14) 2.260(2) Zn(2)-N(24) 2.246(2) Zn(1)-Cl(11) (7) Zn(2)-Cl(21) (6) Zn(1)-Cl(12) (7) Zn(2)-Cl(22) (7) N(11)-Zn(1)-N(17) 78.51(7) N(21)-Zn(2)-N(27) 78.32(7) N(11)-Zn(1)-N(110) 77.28(7) N(21)-Zn(2)-N(210) 76.69(7) N(17)-Zn(1)-N(110) 78.06(7) N(27)-Zn(2)-N(210) 80.59(8) N(11)-Zn(1)-N(14) 78.18(8) N(21)-Zn(2)-N(24) 80.78(7) N(17)-Zn(1)-N(14) 77.65(8) N(27)-Zn(2)-N(24) 76.75(7) N(110)-Zn(1)-N(14) (8) N(210)-Zn(2)-N(24) (7) N(11)-Zn(1)-Cl(11) 94.04(6) N(21)-Zn(2)-Cl(21) 92.97(5) N(17)-Zn(1)-Cl(11) (6) N(27)-Zn(2)-Cl(21) (5) N(110)-Zn(1)-Cl(11) 99.51(6) N(210)-Zn(2)-Cl(21) 96.05(6) N(14)-Zn(1)-Cl(11) (6) N(24)-Zn(2)-Cl(21) (5) N(11)-Zn(1)-Cl(12) (6) N(21)-Zn(2)-Cl(22) (5) N(17)-Zn(1)-Cl(12) 94.22(6) N(27)-Zn(2)-Cl(22) 94.36(5) N(110)-Zn(1)-Cl(12) (6) N(210)-Zn(2)-Cl(22) (5) N(14)-Zn(1)-Cl(12) 99.24(6) N(24)-Zn(2)-Cl(22) 97.27(6) Cl(11)-Zn(1)-Cl(12) 93.24(3) Cl(21)-Zn(2)-Cl(22) 94.36(2) chloro ligands. The acetonitrile ligands and the six coordinate structure of the copper complex are unique, however, as all previous Cu 2+ structures with ethylene cross-bridged ligands displayed the aforementioned five coordinate Cu 2+ with a single chloride ligand. The crystals used to determine this structure were obtained by ether diffusion into an acetonitrile solution of [Cu(B12N4)Cl]Cl H 2 O containing excess PF 6 -. It was expected that only the unbound chloride would be replaced and that the mono hexafluorophosphate salt would be obtained. Instead, the structure of the crystallized compound contained two bound acetonitrile ligands and two hexafluorophosphate anions. Clearly, Cu(B12N4) 2+ has an affinity for acetonitrile that allows facile replacement of the bound chloro ligand, even though the charge attraction to chloride would seem to dictate otherwise. To gain information about the structures of the complexes in solution, the molar conductances of the Cu 2+ complexes were determined in various solvents 74

9 (Table 8). The electrolyte type 166 and extent of dissociation observed for the Table 8. Molar Conductance of copper(ii) complexes. Complex Molar Conductivity in Indicated Solvent MeCN Methanol Water [Cu(Q14N4)Cl]PF [Cu(Q13N4)Cl]Cl H 2 O [Cu(Q12N4)Cl]Cl H 2 O [Cu(iso-Q14N4)Cl]Cl H 2 O : : : Reference 166 contains the theoretical values found in the last three rows. complexes parallels the solvent dielectric constant. Specifically, in water all four complexes behave as 2:1 electolytes, indicating replacement of the chloro ligand by water. In methanol, three of the complexes behave as 1:1 electrolytes indicating the chloride ligand is still bound, while the complex of B14N4 approaches the behavior of a 2:1 electrolyte. Finally, in acetonitrile three of the complexes behave as 1:1 electrolytes, indicating the chloro ligand is not dissociated. However, [Cu(B12N4)Cl]Cl H 2 O is clearly a 2:1 electrolyte in which the chloro ligand is replaced by acetonitrile. This behavior is consistent with the replacement of chloride in acetonitrile indicated by the crystal structure of [Cu(B12N4)(CH 3 CN) 2 ][PF 6 ] 2 (vide supra). This compound may differ in coordination number and labile ligand preference from the other Cu 2+ analogues due to its rather strained geometry and non-ideal bond angles caused by the small ligand cavity. For example, the axial N(4)-Cu-N(4#1) bond angle is only (13) o, describing a rather distorted octahedral complex. The B14N4 and B14N4Me 6 complexes are much less distorted with corresponding angles of (13) o and 75

10 175.14(8) o, respectively. The relative complementarities of these coordination geometries can be invoked to explain the labile ligand preferences of the complexes. 2 Electronic Structure. The magnetic moments of the copper complexes were obtained on the solid samples at ambient temperatures. All four complexes exhibit typical magnetic moments 158 for d 9 Cu 2+ : µ = 2.24 B.M. for [Cu(B14N4)Cl]PF 6, µ = 2.21 B.M. for [Cu(B14N4Me 6 )Cl]PF 6, µ = 1.85 B.M. for [Cu(B13N4)Cl]Cl H 2 O, and µ = 1.98 B.M. for [Cu(B12N4)Cl]Cl H 2 O. Cyclic voltammetry at a scan rate of 200 mv/s in 1 mm acetonitrile (Figure 53) showed one quasi-reversible to irreversible reduction process (Cu 2+ /Cu + ) for each Cu 2+ complex, at V for Cu(B14N4)Cl +, at V for Cu(B13N4)Cl +, at V for Cu(B12N4) 2+, and at V for Cu(B14N4Me 6 )Cl + versus SHE. The reduction of each Cu 2+ complex to Cu + is unremarkable, but the return oxidation wave is sensitive to scan rate in all Potential (V) vs SHE -0.5 Figure 53. Cyclic voltammograms of a) Cu(Me 2 B14N4Me 6 )Cl +, b) Cu(Me 2 B12N4)Cl +, c) Cu(Me 2 B13N4)Cl +, and d) Cu(Me 2 B14N4)Cl + a) b) c) d) -1 cases. At high scan rates, the Cu + Cu 2+ oxidation is broadened, indicating coupled chemical events. However, at slow scan rates, the oxidation wave narrows. This behavior indicates that the reduced complexes may exist as mixtures of at least two species, which quickly equilibrate. From the crystal structure of a related Cu + complex, (vide infra) it appears possible for these Cu + complexes to exist in four-coordinate forms. Therefore these systems may consist of equilibrium mixtures of four- and fivecoordinate Cu + complexes. The B14N4Me 6 complex appears to undergo this 76

11 equilibration most rapidly as would be expected for the sterically more crowded structure. The B13N4 complex has the most reversibility with peak separations of only 97 mv. Interestingly, only the complexes of B14N4 and B13N4 show an observable oxidation (Cu 2+ /Cu 3+ ), at V and V, respectively, while this wave is not seen in the other two complexes under these conditions. Ligand B13N4 better supports Cu 3+, probably because its smaller size better matches the small trivalent ion, than ligand B14N4, as seen in the large difference in oxidation potentials. The smallest ligand B12N4 appears unable to support Cu 3+ in acetonitrile, even though it might be expected to best stabilize the small Cu 3+ ion within its small cavity. As noted above in the discussion of crystal structures, B12N4 is so small that its Cu 2+ geometry is quite distorted. It may be similarly too small to support Cu 3+, which requires strong bonds to stabilize the high valent state. B14N4Me 6 may not stabilize Cu 3+ due to its sterically imposed 5-coordinate, trigonal bipyramidal coordination geometry. The electronic spectra of the Cu 2+ complexes in acetonitrile present the expected ligand field transitions for d 9 Cu The electronic spectra of Cu(B14N4)Cl +, and Cu(B14N4Me 6 )Cl + in acetonitrile are shown in Figure 54. Cu(B14N4)Cl + exhibits a charge transfer band at λ max = 292 nm (ε = 5830 M -1 cm -1 ) and a d-d transition at λ max = a) nm (ε = 100 M -1 cm -1 ). (M -1 cm -1 ) b) Cu(B13N4)Cl + has its chargetransfer band at λ max = 290 nm (ε = 6080 M -1 cm -1 ) and a d-d band at λ max = 650 nm (ε = 110 M -1 cm -1 ) Wavelength (nm) Figure 54. Electronic spectra of a) Cu(Me 2 B14N4)Cl +, and Cu(Me 2 B14N4Me 6 )Cl + in MeCN Cu(B12N4) 2+ exhibits a chargetransfer band at λ max = 298 nm (ε = 3970 M -1 cm -1 ) and a d-d band at 77

12 λ max = 711 nm (ε = 90 M -1 cm -1 ). Cu(B14N4Me 6 )Cl + has its charge-transfer band at λ max = 311 nm (ε = 5140 M -1 cm -1 ) and a d-d band at λ max = 750 nm (ε = 240 M -1 cm -1 ). These bands have proven useful in the study of the kinetic stabilities of the complexes (vide infra). The EPR spectra of the these same complexes (Figure 55) feature much greater differences than is found in their electrochemistry or electronic spectra. [Cu(B14N4)Cl]PF 6, exhibits a signal typical in appearance for Cu 2+. Rudimentary simulation studies show it to most likely be a rhombic signal, but Magnetic Field (G) Figure K EPR spectra of a) Cu(Me 2 B14N4Me 6 )Cl +, and b) Cu(Me 2 B14N4)Cl + a) b) accurate Hamiltonian parameters have not yet been calculated. In contrast, [Cu(B14N4Me 6 )Cl]Cl H 2 O, in the same solvent system, shows a much more complex spectrum. Simple simulation studies indicate that this is a reversed spectrum in which g > g //, which is consistent with the trigonal bipyramidal geometry shown in the crystal structure. 170 The simulations also show that the signal is likely due to an overlap of two different signals, one of minor contribution only. Complete assignment of the parameters for these spectra are beyond current capabilities and are perhaps best left to a later study. Kinetic Stability in Acidic Solution. Probably the most interesting property of these cross-bridged tetraazamacrocycle copper complexes is their remarkable kinetic stability under classically harsh conditions. 140 Copper(II), known for forming the most thermodynamically stable yet most kinetically labile divalent transition metal complexes, 141 provides a fascinating insight into the stabilities of the metal complexes of these ultra rigid ligands. The UV-Vis spectrum of a 0.1 mm solution of 78

13 Figure 56. The electronic spectra of 0.1 mm Cu(Me 2 B14N4) 2+ (a) in 1 M HClO 4 and (b) its change (at 300nm) over 1000 h at 40 o C Cu(B14N4) 2+ in 1 M HClO 4 remains essentially unchanged (Figure 56) over 1000 h at 40 C; from estimated errors, the lower limit of the half-life for ligand dissociation is > 6 years (pseudo first order rate constant of dissociation = 3.5 x 10-9 s -1 ). 167 This is to be compared with the classic experiments 140 with Cu(meso-14N4Me 6 ) 2+. A blue isomer, believed to contain meso-14n4me 6 in the folded form, similar to that forced by our cross-bridged ligands, lost its ligand in 6.1 M HCl with a half-life of about 3 minutes. Replacement of the six C-methyls in meso- 14N4Me 6 by 2 N-methyl groups and the short 2-carbon cross-bridge in B14N4 has increased the kinetic stability of the corresponding complex by something like 6 orders of magnitude, or more. Even the square planar red isomer of Cu(meso-14N4Me 6 ) 2+ is much more labile than the folded bridged cyclam complex, showing a half-life of 22 days in 6.1 M HCl, making it at least 100 times more labile than Cu(B14N4) 2+. Another appropriate comparison to illustrate the kinetic stabilization imparted by the cross-bridge is that of Cu(B14N4) 2+ with Cu(tmc) 2+ (tmc = 1,4,8,11-tetramethylcyclam) whose unbridged ligand is most similar to Me 2 B14N4 (14-membered tetraazamacrocycle with all tertiary nitrogens). The half-life of this complex at 25 in 1 M HNO 3 (calculated from the experimentally determined rate law 171 ) is only 2 seconds, or eight orders of magnitude smaller than that of Cu(B14N4) 2+. Further, it must be emphasized that 6 years is the 79

14 lower limit of the half-life for the loss of ligand from Cu(Me 2 (B14N4)) 2+ in 1 M HClO 4 ; the actual half-life is almost certainly substantially longer. A parallel experiment carried out on the Cu(B14N4Me 6 ) 2+ complex under the same conditions over 530 h demonstrated a similar stability. In this case, the half-life can be estimated at > 8 years (pseudo-first order rate constant of dissociation = 2.5 x 10-9 s -1 ). The Cu(B13N4) 2+ complex under the same conditions demonstrated a similar stability with a half-life of > 8 years (pseudo first order rate constant of dissociation = 2.6 x 10-9 s -1 ). Despite the absence of exhaustive experiments, we conclude that the half-lives of these compounds, in media that would instantly destroy most typical Cu 2+ complexes, is of the order of years. But the ligand B12N4 complex was much less kinetically stable, having a half-life of only 30 h (pseudo first order rate constant of dissociation = 6.4 x 10-6 s -1 ). This ligand is somewhat smaller than the others and these results confirm the conclusion that it is less complementary for the Cu 2+. The zinc complexes presented above were used in similar experiments that were monitored by 13 C NMR. The decrease in the intensity of carbon signal that corresponds to the metal-bound ligand, and the intensifying signals for the protonated, uncomplexed ligand were observed to determine the rate constants for ligand dissociation (Figure 57). These experiments were performed in DCl, because DClO 4 media caused precipitation of the protonated ligand salt at the concentration (0.1 M) required for efficient 13 C NMR spectral acquisition. Figure C NMR of Zn(Me 2 B14N4) 2+ in 1M DCl at 298K The half-lives of the complexes at 298 K in 1 M DCl are 3.9 h 80

15 for Zn(B14N4)Cl 2 (pseudo first order rate constant of dissociation = 4.9 x 10-5 s -1 ), 20 min (pseudo first order rate constant of dissociation = 5.7 x 10-4 s -1 ) for Zn(B13N4)Cl 2, and 48 min (pseudo first order rate constant of dissociation = 2.4 x 10-4 s -1 ) for Zn(B12N4)Cl 2. In view of the behavior of the copper complexes (vide supra), it is surprising that the complex with the smallest ligand, B12N4, did not have the fastest ligand dissociation for the series of Zn 2+ derivatives. In this case the most easily removed ligand is B13N4, the one of intermediate size. It is noted, however, that the difference with the B12N4 complex is small. Zn 2+ and Cu 2+ are similar in size and charge, so these factors alone do not control the stabilities of the complexes with the ethylene cross-bridged ligands. Clearly, the copper(ii) complexes benefit from a large thermodynamic stability with these cross-bridged ligands that the d 10 zinc(ii) does not receive. This benefit is most clearly seen in the two complexes of B13N4, which have the longest and shortest half-lives in acidic conditions for the Cu 2+ and Zn 2+, respectively. Why Zn 2+ is most stable in its complex with B12N4, and why Cu 2+ is more stable in its complex with B13N4 is still unclear. Obviously, other factors besides size are important, but the demonstrated size selectivities show us that matching size between ligand and metal ion is crucial. Ligand flexibility, i.e., rigidity constraint, may select against B14N4 being the most stabilizing ligands for these ions. This ligand has two trimethylene chains that should make it markedly more flexible than B12N4, which has only dimethylene chains. Size, i.e., complementarity, in combination with flexibility, selects between B13N4 and B12N4 depending on the metal ion. A pair of comparisons from the literature is useful. The Springborg group has produced kinetically inert Cu 2+ complexes of a trimethylene cross-bridged cyclen (Figure 58a). 93,172 The crystal structure shows a coordination geometry similar to the present complexes, a macrobicycle engulfed copper ion with a labile fifth ligand and a geometry imtermediate between trigonal bipyramidal and square pyramidal. Kinetic 81

16 a) b) N NH HN N H 3 C N N (CH 2 ) n N N D = O, S, NH, NCH 3 n = 2, 3 CH 3 D (CH 2 ) n Figure 58. Springborg s (a) and Bencini s (b) cross-bridged ligands studies in 5M HCl reveal a first-order dissociation constant for the macrobicycle in CuLCl + of 1.48 x 10-6 s -1 (or a halflife of over 5 days). Though an impressive number, it is still some three orders of magnitude less stable than the dimethylene cross-bridged cyclam complex, but somewhat better than the copper complex of ligand B12N4. The likely reasons for the improved stability for Cu(B14N4) 2+ are better size complementarity for the Cu 2+ ion by the larger parent macrocycle and greatly improved rigidity of the bound ligand, apparently due to decreasing the length of the cross-bridge by one carbon atom, from trimethylene to dimethylene. Yet, making the macrobicycle too small, as is the case for B12N4, sacrifices some complementarity and, consequently, weakens its ability to hold onto the metal ion. Inescapably, size does matter; the addition or removal of a single methylene group causes vast stability differences. Extensive studies have been reported on bridged cyclen derivatives in which the bridge is a chain that contains a fifth donor atom (Figure 58b). The corresponding five coordinate Cu 2+ complexes also exhibit exceptional stabilities under strongly acidic conditions, although dissociation rates have not been quantified. 81,82,84,85 Zn 2+ complexes of these pentadentate ligands have much more kinetic stability in strongly acidic solution than the Zn 2+ complexes of B14N4, B13N4, and B12N4, often surviving in strong acid for weeks or months. Even though the bridging group is longer (providing less rigidification) in these examples, addition of a fifth donor, and the associated extra chelate rings, could explain why these complexes are so stable. 82

17 In summary, new Cu 2+ and Zn 2+ complexes of cross-bridged ligands derived from cyclam, 14N4, 13N4, 12N4, and racemic-14n4me 6 have been synthesized. 83