Unsubstituted cyclidenes - A novel family of lacunar dioxygen carriers with enhanced stability toward autoxidation: Synthesis, characterization, and a representative X-ray structure

Article abstract of DOI:10.1021/ja9624477

A major advance has been made in the incremental molecular design of long-lived cobalt(II) dioxygen carriers. Preceding mechanistic studies revealed that ionizable methyl groups trigger the autoxidation of the O₂ adduct of the cobalt(II) cyclidene. Two members of a new family of unsubstituted (no methyl groups) lacunar cyclidene dioxygen carriers have been prepared in an eight-step synthesis and a complex with a hexamethylene bridge has been structurally characterized. In contrast to previously studied cyclidenes, these materials bear no substituents on the chelated macrocyclic platform. As anticipated, the rates of autoxidation of these unsubstituted cyclidene complexes were found to be 5-8 times slower than those for the most stable previously known cyclidene derivatives. Because of the absence of Me- Me vicinal repulsion, the C6 bridge assumes a zig-zag conformation directly across the cavity. The accompanying, relatively low, dioxygen affinity is explained on the basis of electronic and steric factors. The rates of dioxygen binding to these newly prepared cobalt(II) unsubstituted cyclidenes are fast and approximately equal to the corresponding values for their Me- substituted analogs. Consequently, differences in dissociation rates are responsible for the differences in O₂ affinities. This is a clear example of an unusual steric effect for O₂ adducts.

Full text of DOI:10.1021/ja9624477

4160
J. Am. Chem. Soc. 1997, 119, 4160-4171
Unsubstituted CyclidenessA Novel Family of Lacunar
Dioxygen Carriers with Enhanced Stability toward
Autoxidation: Synthesis, Characterization, and a Representative
X-ray Structure
A. G. Kolchinski,† B. Korybut-Daszkiewicz,†,§ E. V. Rybak-Akimova,†
D. H. Busch,*,† N. W. Alcock,‡ and H. J. Clase‡
Contribution from the Department of Chemistry, UniVersity of Kansas, Lawrence, Kansas 66045,
and Department of Chemistry, UniVersity of Warwick, CoVentry CV4 7AL, United Kingdom
ReceiVed July 16, 1996X
Abstract: A major advance has been made in the incremental molecular design of long-lived cobalt(II) dioxygen
carriers. Preceding mechanistic studies revealed that ionizable methyl groups trigger the autoxidation of the O2
adduct of the cobalt(II) cyclidene. Two members of a new family of unsubstituted (no methyl groups) lacunar
cyclidene dioxygen carriers have been prepared in an eight-step synthesis, and a complex with a hexamethylene
bridge has been structurally characterized. In contrast to previously studied cyclidenes, these materials bear no
substituents on the chelated macrocyclic platform. As anticipated, the rates of autoxidation of these unsubstituted
cyclidene complexes were found to be 5-8 times slower than those for the most stable previously known cyclidene
derivatives. Because of the absence of Me-Me vicinal repulsion, the C6 bridge assumes a zig-zag conformation
directly across the cavity. The accompanying, relatively low, dioxygen affinity is explained on the basis of electronic
and steric factors. The rates of dioxygen binding to these newly prepared cobalt(II) unsubstituted cyclidenes are fast
and approximately equal to the corresponding values for their Me-substituted analogs. Consequently, differences in
dissociation rates are responsible for the differences in O2 affinities. This is a clear example of an unusual steric
effect for O2 adducts.
Introduction
molecules. Iterative molecular design is a natural consequence
of this perspective because elimination of one autoxidation
mechanism may reveal an underlying mechanism that, while
slower, may still operate at troublesome rates. The broad
realization that formation of peroxo-bridged intermediates is key
to a dominant mechanism of autoxidation3 for many familar
transition metal dioxygen carriers, especially those of iron(II),
led to the picket fence porphyrins, the lacunar complexes
described here, and other innovative molecular designs.4,1b These
advances in structure eliminated the peroxo-bridged mechanism
and were followed by studies that revealed new, albeit much
slower, autoxidation mechanisms.5 The mechanisms in the case
of the iron(II) complexes have been discussed elsewhere;1b,3c
here we will treat only those associated with lacunar cobalt(II)
complexes, with focus on those of the cyclidene ligands of the
family used in the present studies.
The preparation and investigation of synthetic reversible
dioxygen carriers has attracted substantial interest in recent
years,1 and their physico-chemical properties are sufficiently
favorable that they have been considered for applications in the
air separation and other industries.1ac Such exploitation has been
hampered by an intrinsic feature of synthetic, as well as natural,
dioxygen carriers, their irreversible oxidation by dioxygen
(autoxidation). Even the most successful biological dioxygen
carrier, hemoglobin, undergoes 1.5-3% autoxidation per day
(human Hb).2 The autoxidation process determines the potential
service lifetime of synthetic dioxygen carriers, and, therefore,
limits their utility. While those concerned with applications
have turned to other technologies,1c we see in this the great
need to understand the mechanisms of autoxidation and to direct
that knowledge toward the molecular design and synthesis of
new dioxygen carriers with enhanced functional lifetimes. In
circumstances of this kind, an ultimate test of mechanism may
be found in the guidance it provides in the design of improved
The cobalt(II) and iron(II) cyclidene dioxygen carriers (Figure
1) are among the most successful classes of non-porphyrin
dioxygen carriers.1b They include complexes with a wide range
of dioxygen affinities, solubilities, and stabilities toward au-
toxidation. Most of them contain a hydrophobic cavity (created
† University of Kansas.
‡ University of Warwick.
§ Institute of Organic Chemistry, Polish Academy of Sciences, O1-224
Warsaw, Poland.
(3) (a) Hammond, G. S.; Wu, C.-H. AdV. Chem. Series. 1968, 77, 186.
(b) Alben, J. O.; Fuchsman, W. H.; Beaudreau, C. A.; Caughey, W. S.
Biochemistry 1968, 7, 624. (c) Warburton, P. R.; Busch, D. H. Dynamics
of IronII and CobaltII Dioxygen Carriers. In PerspectiVes on Bioinorganic
Chemistry; Hay, R. W., Dilworth, J. R., Nolan, K. B., Eds.; JAI Press Ltd.:
London, 1993; Vol. 2, pp 1-79.
(4) (a) Collman, J. P.; Gagne`, R. R.; Reed, C. A. J. Am. Chem. Soc.
1974, 96, 6522. (b) Collman, J. P.; Elliott, C. M.; Halbert, T. R.; Tovrog,
B. S. Proc. Natl. Acad. Sci. U.S.A. 1977, 74, 18. (c) Collman, J. P.; Brauman,
J. I.; Collins, T. J.; Iverson, B. L.; Sessler, J. L. J. Am. Chem. Soc. 1981,
103, 2450. (d) Momenteau, M.; Reed, C. A. Chem. ReV. 1994, 94, 659.
(5) (a) Wallace, W. J.; Maxwell, J. C.; Caughey, W. S. Biochem. Biophys.
Res. Commun. 1974, 57, 4, 1104. (b) Shikama, K. Coord. Chem. ReV. 1988,
83, 73-91. (c) Wallace, W. J.; Houtchens, R. A.; Maxwell, J. C.; Caughey,
W. S. J. Biol. Chem. 1982, 257, 4966.
X Abstract published in AdVance ACS Abstracts, April 15, 1997.
(1) (a) Haggin, J. Chem. Eng. News 1996, Feb. 5, 35. (b) Busch, D. H.;
Alcock, N. W. Chem. ReV. 1994, 94, 585. (c) Li, G. Q.; Govind, R. Ind.
Eng. Chem. Res. 1994, 33, 755. (d) Martell, A. E.; Motekaitis, R. J.;
Rockcliffe, D.; Menif, R.; Ngwenya, P. M. Pure Appl. Chem. 1994, 66,
859. (e) Traylor, T. G.; Traylor, P. S. Annual ReV. Biophys. Bioeng. 1982,
11, 105. (f) Collman, J. P., Halpert, T. R.; Suslick, K. S. In Metal Ion
ActiVation of Dioxygen; Spiro, T. G., Ed.; Wiley & Sons: New York, 1980.
(g) Niederhoffer, E. C.; Timmons, J. H.; Martell, A. E. Chem. ReV. 1984,
84, 137. (h) Smith, T. D.; Pilbrow, J. T. Coord. Chem. ReV. 1981, 39, 295.
(i) Jones, R. D.; Summerville, D. A.; Basolo, F. Chem. ReV. 1979, 79, 139.
(j) McLendon, G.; Martell, A. E. Coord. Chem. ReV. 1976, 19, 1. (k) Vaska,
L. Acc. Chem. Res. 1976, 9, 175.
(2) Jaffe, E. R.; Neuman, G. Nature 1964, 202, 607.
S0002-7863(96)02447-X CCC: $14.00 © 1997 American Chemical Society

Unsubstituted Cyclidenes
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4161
sublimed before use.8 N,N′-Dimethyl-1, 8-diaminooctane was prepared
according to ref 9, with the slight modification that the amount of
sodium hydroxide added before extraction of the final product was 20
rather than 2 g (correcting a typographical error in the ref). Spectro-
scopic and electrochemical techniques, inert atmosphere manipulations,
and the technique used for dioxygen affinity determinations have been
described elsewhere.9,10
Kinetic measurements were performed at low temperatures (from
-75 to -50 °C) using a Hi-Tech Scientific (England) CryoStopped-
Flow instrument equipped with stainless steel plumbing and mixing
cell, the latter having sapphire windows, connected to a PC computer
with IS-2 Rapid Kinetic Software. The mixing cell was maintained to
(0.1 K, and mixing time is 2 ms. Kinetic measurements were
performed on acetone solutions containing 0.5-1.5 M of 1-methylimi-
dazole with various starting concentrations of the C6-bridged cobalt
complex (7 × 10-6-5 × 10-5 M) and dioxygen (7 × 10-6-1 × 10-4
M) and a more limited concentration range for the C8 bridged complex.
Dioxygen/nitrogen gas mixtures were generated using Tylan FC-260
mass flow controllers. The low concentrations of reagents necessary
to slow down the reaction limited the available range of concentrations.
Series of 8-12 shots run gave standard deviations within 10%, with
overall reproducibility within 20%.
Autoxidation measurements were conducted in a pure dioxygen
atmosphere in a sealed 1 cm quartz cell connected by graded seals to
a Pyrex tube with a prefabricated neck. For autoxidation rate
measurements, a temperature of 30 ( 0.3 °C was maintained for
extended periods of time using an Abderhalden pistol with a 2-meth-
ylbutane/pentane mixture. Precautions were taken to prevent any
possibility of acetonitrile/O2 vapor explosions during sealing.
Syntheses. 2,2′-[Propylenebis(iminomethylidyne)]-1,3-propanedi-
one (IIIa). Freshly sublimed triformylmethane8 (30g, 0.3 mol) was
dissolved in absolute ethanol (375 mL) with continuous stirring. The
solution was placed in an ice bath, and anhydrous 1,3-diaminopropane
(12.5 mL, 0.15 mol) was added dropwise with stirring at such a rate
that the temperature did not exceed 30 °C. The solution became
greenish-yellow and, after addition of approximately half of the 1,3-
diaminopropane, a crystalline precipitate started to form. After addition
was complete, the reaction mixture was refrigerated for several hours
to crystallize the main portion of the product. The yellowish precipitate
was removed by filtration, washed twice with absolute alcohol and once
with ether, and dried in Vacuo over P4O10 for several hours. The mother
liquor deposited additional product for 2 more days. Diethyl ether was
then added to precipitate more product. The overall yield was 30.3 g
(85%). The crude product is suitable for further reaction but can be
purified by recrystallization from ethanol (ca. 50 mL/g). The melting
point was 185-186 °C. Anal. Calcd for C11H14N2O4: C, 55.46; N,
11.76; H, 5.88. Found: C, 55.34; N, 11, 70; H, 6.10. FAB/TG-G
239 [M + H]+. Spectral data are given in Table 1 and discussed later
in this paper.
{2,2′-[Propylenebis(iminomethylidyne)]-1,3-propanedionato(2-
)}nickel(II) Monohydrate (IVa). Ligand IIIa (21g, 88 mmol) and
nickel(II) acetate 4-hydrate (21.8 g, 88 mmol) were stirred in 250 mL
of ethanol for 16 h. A pale lavender precipitate was filtered, washed
with ethanol, and dried in Vacuo for several hours. Yield 24.0 g (87%).
Calcd for C11H14N2O5Ni: C, 42.22; N, 8.95; H, 4.51. Found: C, 42.01;
N, 8.70; H, 4.40. FAB/MB 295 [M + H]+.
[7,15-Diformyl-1,5,9,13-tetraazacyclotetradeca-5,7,13,15-tetraenato-
(2-)]nickel(II) (VIa). Complex IVa (24.0g, 77 mmol) was slurried
in 120 mL of anhydrous 1,3-diaminopropane giving a color change
from blue to violet. On heating the mixture at 160 °C for 0.5 h, the
starting complex dissolved giving a brown solution. Approximately
one-third of the 1,3-diaminopropane escaped by distillation during the
heating. The resulting solution was poured into water (1200 mL),
resulting in immediate formation of a green precipitate V. The mixture
was then stirred for 24 h at room temperature, whereupon the precipitate
gradually changed in color from green to red. The resulting precipitate
Figure 1. Planar (a) and three-dimensional (b) representations of
cyclidene complexes; denoted below as M(R1R2R3R4cyclidene), where
M ) Ni or Co.
by a bridging group) into which dioxygen (or some other small
ligand) may enter. The opposite, and external, axial site is
occupied by an auxiliary nitrogenous base, which is usually
present in the reaction mixture in large excess. The general
cyclidene structure (Figure 1) allows extensive modification by
changing substituents R2 and R3 and/or the linking groups R1,
X, and Y. For the compounds of major interest to this paper,
the chelating groups X and Y remain unchanged (X ) Y )
(CH2)3), while bridge R1 is always (CH2)n (or simply Cn in
formulas). This paper addresses R4, which has hitherto always
been CH3, but is H in the new compounds reported here.
Cyclidene formulas will be presented in a simplified manner
as (CnR2R3R4Cyc), where n is the number of methylene groups
in the bridge R1, and R2, R3, and R4 are the substituents.
Earlier studies have provided vital mechanistic insight into
the autoxidation of cobalt(II) cyclidenes.6 A well demonstrated
and critical feature of the proposed mechanism is the deproto-
nation of the dioxygenated cobalt(II) complex to produce a
conjugate base derivative that is vulnerable to irreversible
oxidation of the ligand. The generality of this mechanistic
feature had long been apparent. When R2 is hydrogen, the
relatively acidic N-H proton leads to relatively rapid autoxida-
tion, and most of the studies on the cobalt(II) cyclidenes have
involved alkyl groups at the R2 position. Discovery that the
R3 methyl groups of those complexes having R2 ) R3 ) CH3
are acidic (crystal structure on doubly deprotonated species7)
led to the kinetic and mechanistic studies6 that established the
role of the deprotonated species. Replacement of the R3 methyl
group by phenyl results in a 20-fold retardation of the rate of
autoxidation. However, the complexes do still undergo autoxi-
dation, and the rates respond to base in a manner suggestive of
the conjugate base mechanism.6 The methyl groups at the R4
positions are common to all structures and the most likely sites
for deprotonation in those cases where NH and ionizable R3
groups have been eliminated. We have therefore proposed that
removal of this reactive site should retard the autoxidation of
these important dioxygen carriers. The results are summarized
here.
Experimental Section
Reagents and Starting Materials. The solvents and reagents used
in these studies were reagent grade or better. Solvents were purified
according to published methods. N-Methylimidazole (MeIm) was dried
over barium oxide. Methanol, acetonitrile, and MeIm used in prepara-
tion of cobalt complexes were distilled under argon and degassed by
successive freeze-pump-thaw cycles prior to use. 1,3-Diaminopro-
pane was dried by repeated distillation over sodium metal. Triformyl-
methane was prepared according to the published procedure and was
(8) Budeˇsˇinsky`, M.; Fiedler, P.; Arnold, Z. Synthesis 1989, 858.
(9) Thomas, R.; Fendrick, C. M.; Lin, W.-K.; Glogowski, M. W.; Chavan,
M. Y.; Alcock, N. W.; Busch, D. H. Inorg. Chem. 1988, 27, 2534.
(10) Korybut-Daszkiewicz, B.; Kojima, M.; Cameron, J. H.; Herron, N.;
Chavan, M. Y.; Jircitano, A. J.; Coltrain, B. K.; Neer, G. L.; Alcock, N.
W.; Busch, D. H. Inorg. Chem. 1984, 23, 903.
(6) Masarwa, M.; Warburton, P. R.; Evans, W. E.; Busch, D. H. Inorg.
Chem. 1993, 32, 3826.
(7) Goldsby, K. A.; Jircitano, A. J.; Nosco, D. L.; Stevens, J. C.; Busch,
D. H. Inorg. Chem. 1990, 29, 2523.

4162 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Kolchinski et al.
was isolated by filtration, washed with water, and dried in vacuo initially
over CaCl2 and later over P4O10. The crude product was dissolved in
a minimum volume of chloroform and applied to a 5 × 25 neutral
activated alumina column (Brockmann scale activity I). The mixture
was eluted slowly with chloroform, discarding the first narrow green
band. The second red-brown band was collected and evaporated on a
rotary evaporator. The product was washed with ethanol and dried in
vacuo. Yield 8.6 g (34%). Anal. Calcd for C14H18N4O2Ni: C, 50.49;
N, 16.82; H, 5.45. Found: C, 50.40; N, 16.49; H, 5.80. FAB/MB
333 [M]+ and 665 - cluster [M2 - H]+.
[3,11-Bis(methylaminomethylidene)-1,5,9,13-tetraazacyclohexa-
deca-1,4,9,12-tetraene-k4N1,5,9,13nickel(II)] Hexafluorophosphate (IXa).
Complex VIa (1.0 g, 3 mmol) was suspended in 25 mL of dry
methylene chloride, and methyl trifluoromethanesulfonate (1 mL, 8.8
mmol) was added (caution, toxic reagent!). The complex dissolved
in a few minutes, giving a red-brown solution. The mixture was stirred
for 16 h, after which 100 mL of 1.4 M methylamine in acetonitrile
(140 mmol) was added. The mixture was stirred for 1 more h and
then evaporated to approximately 1/5 of the initial volume. Ammonium
hexafluorophosphate (2.5 g, 15 mmol) in methanol (20 mL) was added.
The solution was evaporated to a small volume and refrigerated for
several hours. The yellow crystalline product was filtered, washed with
methanol, and dried in a vacuum oven for 2 h at 60 °C. Yield 1.1g
(57%). Anal. Calcd for C16H26N6NiP2F12: C, 29.52; N, 12.91; H, 4.03.
Found: C, 29.16; N, 12.51; H, 4.35. FAB/NBA 505 [M - PF6]+ and
668 - [M + H2O].
(3,10-Dimethyl-3,10,14,18,21,25-hexaazabicyclo[10.7.7]hexacosa-
1,11,13,18,20,25-hexaene-k4N14,18,21,25)nickel(II) Hexafluorophosphate
(X) and (3,12-Dimethyl-3,12,16,20,23,27-hexaazabicyclo[12.7.7]-
hexacosa-1,13,15,20,22,27-hexaene-k4N16,20,23,27)nickel(II) Hexafluo-
rophosphate (XI). Complex VIa (2 g, 6 mmol) was suspended in 50
mL of dry methylene chloride, and methyl trifluoromethanesulfonate
(2 mL, 18 mmol) was added (caution, toxic reagent!). The complex
dissolved in a few minutes, giving a red-brown solution. The mixture
was stirred for 16 h and the solvent was then removed on a rotary
evaporator (with a KOH/MeOH trap to prevent methyl trifluo-
romethanesulfonate vapors from escaping). Copious foam formed at
the final stages of evaporation, but periodical release of vacuum with
dry air helped to prevent excessive foaming. The oily residue was
then dried at 0.1 Torr, 50 °C for half an hour and then dissolved in
100 mL of dry acetonitrile. The corresponding aliphatic R,ω-diamine
(6 mmol) was separately dissolved in 100 mL of dry acetonitrile. These
two solutions were added at a rate of 10 mL h-1, by means of a syringe
pump to 1000 mL of stirred acetonitrile maintained under nitrogen at
reflux temperature. After the addition was complete, the solvent was
removed by rotary evaporation to give a viscous dark red-brown residue.
Ammonium hexafluorophosphate (2.0 g, 12 mmol) in methanol (17
mL) was added, and the resulting solution was applied to a 2 × 25 cm
silica (70-230 mesh, 60 Å) column. The first orange bands (ca. 75
mL for X and ca. 250 mL for XI) were separated from the next brown
bands. The fractions were evaporated to small volumes, and several
milliliters of methanol were added. Solvent evaporation-methanol
addition cycles were repeated 2-3 times until crystallization occurred.
The products were filtered, washed twice with methanol, and dried in
vacuo for several hours. Yields: 0.67 g (15%) for X and 1.24 g (27%)
for XI. Anal. Calcd for C22H36N6NiP2F12: C, 36.06; N, 11.46; H,
4.95. Found: C, 35.68; N, 11.41; H, 4.99. FAB/NBA: 443 [M -
2PF6] and 587 - [M - PF6]. Anal. Calcd for C24H40N6NiP2F12: C,
37.87; N, 11.04; H, 5.30. Found: C, 37.64; N, 11.02; H, 5.45. FAB/
NBA: 470 [M - 2PF6] and 615 - [M - PF6].
(3,10-Dimethyl-3,10,14,18,21,25-hexaazabicyclo[10.7.7]hexacosa-
1,11,13,18,20,25-hexaene) Hexafluorophosphate (XII) and (3,12-
Dimethyl-3,12,16,20,23,27-hexaazabicyclo[12.7.7]hexacosa-1,
13,15,20,22,27-hexaene) Bis(hexafluorophosphate) Chloride (XIII).
Anhydrous hydrogen chloride was bubbled through 5 × 10-2
M
solutions of X or XI in acetonitrile. Higher concentrations are not
recommended since the tar-like consistency of the precipitates described
below prevents complete Ni2+ removal. After several seconds of
bubbling, the solutions have changed in color from yellow-orange to
turquoise blue. In a few more seconds a turquoise precipitate forms
and almost immediately redissolves. At this stage, bubbling of

Unsubstituted Cyclidenes
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4163
hydrogen chloride was stopped. The solvent was removed by rotary
evaporation at room temperature. The resulting gummy residues were
carefully washed several times with water to obtain a negative test for
nickel(II) with dimethylglyoxime. The off-white precipitates were dried
over CaCl2 in Vacuo. Yields: 68% for XII and 56% for XIII. Anal.
Calcd for C22H39N6P3F18: C, 32.13; N, 10.22; H, 4.78. Found: C,
32.66; N, 10.18; H, 5.01. FA B/NBA: 385 [M - 3PF6]; 531 [M -
2PF6] and 675 - [M - PF6]. Anal. Calcd for C24H43N6P2F12Cl: C,
38.90; N, 11.34; H, 5.85. Found: C, 39.78; N, 11.63; H, 5.90. FAB/
NBA: 413 [M - 2PF6 - Cl] and 559 - [M - PF6 - Cl].
(3,10-Dimethyl-3,10,14,18,21,25-hexaazabicyclo[10.7.7]hexacosa-
1,11,13,18,20,25-hexaene-k4N14,18,21,25)(methanol)cobalt(II) Hexafluo-
rophosphate (XIV) and (3,12-Dimethyl-3,12,16,20, 23,27-hexaaza-
bicyclo[12.7.7]hexacosa-1,13,15,20,22,27-hexaene-k4N16,20,23,27)-
(methanol)cobalt(II) Hexafluorophosphate (XV). These preparations
were carried out in an inert atmosphere glovebox using degassed dry
methanol. The ligand salt XII or XIII (1 mmol) was slurried in
methanol (20 mL), and the mixture was heated to boiling. A solution
containing cobalt(II) acetate 4-hydrate (0.249 g, 1 mmol) and sodium
acetate 3-hydrate (0.136 g, 1 mmol) in 10 mL of hot methanol was
added to the suspension. After several minutes the ligand salt dissolved,
and the color changed to a deep orange. The volume of the solution
was reduced to approximately one-half, and the reaction mixture was
left to crystallize. Crystallization was complete in about 30 min for
XIV and in several hours for XV. The products were washed with
methanol and dried in the glovebox. Crystals of XV gradually turn
opaque upon removal from the mother liquor due to loss of methanol.
Yields: 0.46 g (63%) for XIV and 0.47 g (62%) for XV. Anal. Calcd
for C23H40N6OCoP2F12: C, 36.09; N, 10.98; H, 5.27. Found: C, 36.38;
N, 11.68; H, 5.20. FAB/NBA: 443 [M - 2PF6] and 588 - [M -
PF6]. Anal. Calcd for C25H44N6OCoP2F12: C, 37.84; N, 10.59; H,
5.59. Found: C, 37.57; N, 10.64; H, 5.60. FAB/NBA: 471 [M -
2HPF6] and 616 - [M - PF6].
Table 2. Selected Bond Lengths (Å) and Angles (deg) for
[Co(C6MeHH[16]cyclidene)‚CH3OH](PF6)2 (XIV)
Co(1)-N(1)
Co(1)-N(9)
Co(1)-O(30)
N(1)-C(16)
N(5)-C(6)
1.951(14)
1.916(14)
2.209(12)
1.321(21)
1.279(29)
1.475(26)
1.308(22)
1.502(29)
1.484(23)
1.496(27)
1.508(27)
1.498(31)
1.501(23)
1.428(29)
1.399(26)
1.455(32)
1.464(26)
1.391(26)
Co(1)-N(5)
Co(1)-N(13)
N(1)-C(2)
1.958(15)
1.921(14)
1.490(20)
1.493(22)
1.284(28)
1.516(29)
1.280(25)
1.511(35)
1.290(21)
1.509(29)
1.394(25)
1.367(29)
1.529(27)
1.405(24)
1.460(49)
1.469(38)
1.456(23)
N(5)-C(4)
N(9)-C(8)
N(9)-C(10)
N(13)-C(14)
N(18)-C(19)
N(26)-C(25)
N(26)-C(28)
C(3)-C(4)
N(13)-C(12)
N(18)-C(17)
N(18)-C(20)
N(26)-C(27)
C(2)-C(3)
C(6)-C(7)
C(7)-C(8)
C(7)-C(27)
C(11)-C(12)
C(15)-C(16)
C(20)-C(21)
C(22)-C(23)
C(24)-C(25)
C(10)-C(11)
C(14)-C(15)
C(15)-C(17)
C(21)-C(22)
C(23)-C(24)
O(30)-C(31)
N(1)-Co(1)-N(5)
N(5)-Co(1)-N(9)
89.5(7) N(1)-Co(1)-N(9)
88.9(7) N(1)-Co(1)-N(13)
165.1(7)
91.4(7)
86.9(7)
97.4(5)
N(5)-Co(1)-N(13) 166.8(7) N(9)-Co(1)-N(13)
N(1)-Co(1)-O(30)
N(9)-Co(1)-O(30) 100.4(6) N(13)-Co(1)-O(30) 95.6(6)
94.6(6) N(5)-Co(1)-O(30)
Co(1)-N(1)-C(2)
C(2)-N(1)-C(16)
Co(1)-N(5)-C(6)
Co(1)-N(9)-C(8)
C(8)-N(9)-C(10)
121.4(10) Co(1)-N(1)-C(16)
118.2(14) Co(1)-N(5)-C(4)
120.7(12) C(4)-N(5)-C(6)
121.1(14) Co(1)-N(9)-C(10)
120.2(12)
118.2(13)
120.9(16)
122.1(13)
116.4(15) Co(1)-N(13)-C(12) 122.2(11)
Co(1)-N(13)-C(14) 119.6(14) C(12)-N(13)-C(14) 118.2(15)
C(17)-N(18)-C(19) 122.1(21) C(17)-N(18)-C(20) 124.4(19)
C(19)-N(18)-C(20) 112.6(17) C(25)-N(26)-C(27) 130.1(19)
C(25)-N(26)-C(28) 111.9(14) C(27)-N(26)-C(28) 117.8(15)
N(1)-C(2)-C(3)
N(5)-C(4)-C(3)
C(6)-C(7)-C(8)
C(8)-C(7)-C(27)
111.9(15) C(2)-C(3)-C(4)
109.8(13) N(5)-C(6)-C(7)
115.2(20) C(6)-C(7)-C(27)
128.5(16) N(9)-C(8)-C(7)
114.5(17)
125.1(20)
114.0(19)
122.9(17)
X-ray Structural Analysis. Crystal Data: C23H40N6OCoP2F12, M
) 765.5, monoclinic, P21/c, a ) 19.053(13), b ) 12.419(7), c )
14.635(7) Å, â ) 109.95(5)°, U ) 3255 Å3, Z ) 4, Dc ) 1.56 g cm-3
Mo KR radiation, λ ) 0.71069 Å, µ (Mo KR radiation) ) 0.72 mm-1
T ) 200 K.
,
,
N(9)-C(10)-C(11) 112.1(14) C(10)-C(11)-C(12) 114.3(14)
N(13)-C(12)-C(11) 105.2(18) N(13)-C(14)-C(15) 124.6(16)
C(14)-C(15)-C(16) 120.7(16) C(14)-C(15)-C(17) 123.0(16)
Data were collected with a Siemens R3m four circle diffractometer
in ω-2θ mode. The crystal was held at 200 K with an Oxford
Cryosystems Cryostream Cooler.11 Maximum 2θ was 45° with scan
C(16)-C(15)-C(17) 115.6(18) N(1)-C(16)-C(15)
123.5(18)
N(18)-C(17)-C(15) 128.3(22) N(18)-C(20)-C(21) 100.3(20)
C(20)-C(21)-C(22) 114.7(27) C(21)-C(22)-C(23) 113.7(22)
C(22)-C(23)-C(24) 113.1(19) C(23)-C(24)-C(25) 114.2(17)
range (0.7°(ω) around the Ka1-Ka2 angles, scan speed 3-15°(ω) min-1
,
N(26)-C(25)-C(24) 113.9(17) N(26)-C(27)-C(7)
Co(1)-O(30)-C(31) 126.0(10)
126.6(16)
depending on the intensity of a 2 s prescan; backgrounds were measured
at each of the scans for 0.25 of the scan time. hkl ranges were -1/15;
-1/13; and -19/19. Three standard reflections were monitored every
200 reflections and showed a slight decrease (2%) during data
collection. The data were rescaled to correct for this. Unit cell
dimensions and standard deviations were obtained by least-squares fit
to 15 reflections (7° < 2θ < 16°). Reflections were processed using
profile analysis to give 4264 unique reflections (Rint ) 0.046), of which
1496 were considered observed (I/σ(I) > 2.0). These were corrected
for Lorentz, polarization, and absorption effects (by the analytical
method using ABSPSI12); minimum and maximum transmission factors
were 0.88 and 0.91. Crystal dimensions were 0.23 × 0.31 × 0.17
mm.
) 0.08 Å2. Those defined by the molecular geometry were inserted at
calculated positions and not refined; methyl groups were treated as
rigid CH3 units, with their initial orientation based on a staggered
configuration. Final refinement was on F by least squares methods
refining 426 parameters. Largest positive and negative peaks on a final
difference Fourier synthesis were of height 0.7 and -0.5 el. Å-3
.
A weighting scheme of the form W ) 1/(σ2(F) + gF2) with g )
0.000 24 was used and shown to be satisfactory by a weight analysis.
Final R ) 0.085, Rw ) 0.076, S ) 2.11; R for all reflections ) 0.257.
Maximum shift/error in final cycle was 0.05. The relatively high
R-value is readily understandable in the light of the weak scattering
and the severe disorder of the anions.
Systematic reflection conditions: h01, 1 ) 2n; 0k0, k ) 2n indicate
space group P21/c. The structure was solved (with difficulty on account
of the weak data) by direct methods using SHELXTL (TREF). Further
light atoms were then found by E-map expansion and successive Fourier
syntheses. One PF6 group was extensively disordered and was
approximately modeled by two sets of fluorine atoms with occupancies
0.7 and 0.3. It was necessary to constrain the C-C distances along
the polymethylene chain, to keep the refinement and anisotropic
displacement parameters stable (refining the 1-2 and 1-3 C-C
distances). Anisotropic displacement parameters were used for all
non-H atoms (apart from the minor fluorine atoms and one carbon atom
which proved unstable on refinement and was held isotropic). Hy-
drogen atoms were given fixed isotropic displacement parameters, U
Computing was with SHELXTL PLUS13 on a DEC Microvax-II.
Unusually, refinement on all reflections with SHELXL proved unstable,
apparently because of anomalously strong weak reflections; this is
attributed to scattering from minor twin components of the crystal
selected. Scattering factors in the analytical form and anomalous
dispersion factors were taken from ref 14 (stored in the program). Final
atomic coordinates are given in supporting information and selected
bond lengths and angles in Table 2.
(13) Sheldrick, G. M. SHELXTL PLUS User’s Manual; Nicolet Instr.
Co.: Madison, WI, 1986.
(14) International Tables for X-ray Crystallography; Kynoch Press:
Birmingham, 1974; Vol. IV (Present distributor Kluwer Academic Publish-
ers, Dordrecht).
(11) Cosier, J.; Glaser, A. M. J. Appl. Cryst. 1986, 19, 105.
(12) Alcock, N. W.; Marks, P. J. J. Appl. Cryst. 1994, 27, 200.

4164 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Kolchinski et al.
Scheme 1. Selected Synthetic Transformations in Cyclidene Chemistry
Results and Discussion
forms of the molecule. Interconversion of these forms causes
substantial broadening of the NMR peaks.
Synthesis and Characterization. A general synthetic scheme
showing the numbering of the materials and the sequential steps
is given in Scheme 1. Newly prepared materials were routinely
characterized and identified by analysis, 13C and 1H NMR, FAB
MS, and IR (Table 1 and procedures). In general, the synthetic
route to unsubstituted cyclidenes was patterned after that for
the substituted analogs,9,10,15 (Scheme 1), but differences abound,
and they start with a change in the initial starting material.
3-(Ethoxymethylene)-2, 4-pentanedione (I) (Scheme 1) used for
all previous cyclidene syntheses necessarily gives products with
R3 ) R4 ) CH3. Triformylmethane, which is the essential
reagent, has recently been prepared and characterized (sublimed
white solid, mp ) 104-106 °C).8 Triformylmethane mainly
exists in solution in the enol form (II) with a structure similar
to that of 3-(ethoxymethylene)-2,4-pentanedione (I).
The physical properties of the open-chain unsubstituted
nickel(II) complex (IVa), prepared by reaction of ligand IIIa
with nickel(II) acetate 4-hydrate, differ substantially from those
of its methyl substituted analogs IVb and IVc.15a,16 The latter
are typical diamagnetic orange or brown, pseudosquare planar
nickel(II) complexes. IVa has a rather pale lavender color and
a magnetic moment of 3.17 µB which are well-known charac-
teristics of octahedral nickel(II). These properties, its low
solubility, and a composition that does not allow other ligands
all point to an oligomeric structure involving bridging by the
exocyclic aldehyde groups, in analogy to the previously
described polymeric benzoyl substituted copper complexes.17
Another significant difference between the preparation of the
macrocyclic nickel(II) complex VIa and those of its substituted
analogs (VIb, for example) is the formation of the intense green
side-product/intermediate V (transformation 5). This material
precipitates on addition of water to the 1,3-diaminopropane
solution and undergoes slow hydrolysis to VIa (transformation
6). Continuing the uniqueness of the unsubstituted systems,
the macrocyclic Ja¨ger-type complex (VIa) was obtained, after
lengthy workup, as a bright-red microcrystalline solid in contrast
with its yellow orange congenors.
The first ligand produced by condensation, 2,2′-[propylenebis-
(iminomethylidyne)]-1,3-propanedione (IIIa), prepared in anal-
ogy to its well-known counterpart, 3,3′-[ethylenebis(iminome-
thylidene)]di-2,4-pentanedione (IIIc),15a was found by FAB/
TG-G mass-spectra ([M + H] peak m/z ) 239) to contain the
macrocyclic byproduct VIII (m/z ) 279). Conveniently, both
IIIa and its impurity VIII give the desired macrocyclic complex
1
VIa via transformations 3, 5, and 6 (Scheme 1). 13C and H
NMR spectra suggest that solutions of IIIa are mixtures of at
Methylation of the carbonyl group of the Ja¨ger complex VIb
(transformation 7) is a critical step that facilitates the synthesis
of the macrobicyclic lacunar ligands. The reaction commonly
makes use of an excess of a powerful methylating agent (methyl
trifluoromethanesulfonate, methyl fluorosulfate, or trimethyl-
oxonium tetrafluoroborate), followed by subsequent decomposi-
tion of the excess methylating agent with methanol.15b,c The
least two different tautomers containing either aldehyde or enol
(15) (a) Riley, P. R.; Busch, D. H. Inorg. Synth. 1978, 18, 36. (b) Cairns,
C. J.; Busch D. H. Inorg. Synth. 1990, 27, 261. (c) Busch, D. H.; Olszanski,
D. J.; Stevens, J. C.; Schammel, W. P.; Kojima, M.; Herron, N.; Zimmer,
L. L.; Holter, K. A.; Mocak, J. J. Am. Chem. Soc. 1981, 103, 1472. (d)
Busch, D. H.; Christoph, G. G.; Zimmer, L. L.; Jackels, S. C.; Grzybowski,
J. J.; Callahan, R. C.; Kojima, M.; Holter, K. A.; Mocak, J.; Herron, N.;
Chavan, M.; Schammel, W. P. J. Am. Chem. Soc. 1981, 103, 5107. (e)
Busch, D. H.; Jackels, S. C.; Callahan, R. C.; Grzybowski, J. J.; Zimmer,
L. L.; Kojima, M.; Olszanski, D. J.; Schammel, W. P.; Stevens, J. C.; Holter,
K. A.; Mocak, J. Inorg. Chem. 1981, 20, 2834.
(16) Ja¨ger, E. Z. Chem. 1968, 8, 392.
(17) Ja¨ger, E. Z. Anorg. Allg. Chem. 1966, 346, 76.

Unsubstituted Cyclidenes
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4165
Figure 2. Hypothetical U-Z isomerization (Lin, W.-K.; Alcock, N. W.; Busch, D. H. J. Am. Chem. Soc. 1991, 113, 7603) of the dianion IXa.
Figure 3. The HETCOR spectrum of [Ni(C6MeHH[16]cyclidene)](PF6)2 (X) in CD3CN. Dots that appear in front of the strong acetonitrile signal
at about 120 ppm and are not connected by dotted lines appeared due to the leakage of radio frequency.
previously known methoxyethylidene substituted macrocyclic
complexes (e.g., VIIb) are crystalline, reasonably stable materi-
als. In contrast, for the unsubstituted complex VIIa, the addition
of methanol leads to decomposition, even though the presence
of the bis-methylated complex VIIa was confirmed by FAB/
NBA. An unseparated mixture of the methylated products was
therefore used in the next stage, after distilling off the excess
of methyl trifluoromethylsulfonate (transformation 10).
visualized in Figure 2. Earlier X-ray structural and molecular
mechanics studies have shown that the saddle or U-shaped
conformation is generally preferred for neutral substituted [16]-
cyclidene complexes. However, deprotonation of the two
methylamine moieties produces a dianionic ligand, and the
Z-conformation may be more stable than the U-conformation.
Obviously the Z-isomer cannot undergo a bridging reaction with
an R,ω-ditosylalkane.
Final ring closure also provided its challenges. For substituted
cyclidenes the second ring closure could be performed either
by reaction of the bismethylamino-derivative IXb with R,ω-
ditosylalkanes (transformation 9) or by the direct reaction of
bis(methoxyethylidene) macrocycle VIIb with diamines (trans-
formation 10).15b,c For the unsubstituted complexes, the first
approach was initially investigated, since complex IXa was
available, but extensive studies gave exclusively unbridged
products with hexamethylene or decamethylene chains attached
on only one side. A rationale for this difference in behavior is
Nickel cyclidenes X and XI, containing the final desired
ligand, were prepared in 15 and 27% yields, respectively, using
an alternative cyclization route based on the reaction of aliphatic
hexamethylenediamine or octamethylenediamine with crude
VIIa (transformation 10). The monomeric bicyclic structures
of the products were confirmed by mass spectrometry and NMR
results (e.g., Figure 3).
Demetalation approximated the known procedure (transfor-
mation 11),15b giving XII as an off-white solid tris(hexafluo-
rophosphate) salt and XIII as a mixed chloride-hexafluoro-

4166 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Kolchinski et al.
angles, the conformation found for the unsubstituted complex
contains a majority of low energy anti torsional angles (Figure
5b).
Differences in the twist of the CH3N(CH2-) fragment around
the nitrogen-vinyl carbon bond (ꢀ, Table 3), seem to be the
source of the conformational differences between the bridges
for these two complexes, with values of 0° and 22° for [Co-
(C6MeHH[16]cyclidene)]2+ and [Co(C6MeMeMe[16]cycli-
dene)]2+, respectively (Table 3). This difference can be
attributed to the absence or presence of steric repulsion between
the R2 and R3 methyl groups. This repulsion forces the CH3N-
(CH2-) units to rotate inwards, toward the cavity, reducing the
overall bridge length. The absence of this repulsion favors a
straighter bridge featuring low energy anti torsion angles.
Therefore, it may be anticipated that elimination of the R2/R3
steric repulsion in [Co(C6MeHH[16]cyclidene)]2+ makes bridge
folding more difficult and generates more strain in the oxygen-
ated complex.
Electrochemistry. Comparison of electrochemical data for
nickel and cobalt cyclidenes (Table 4) provides a useful
distinction between electronic and steric effects arising from
ligand modifications. Earlier work19 has shown that the redox
potentials (M(III)/M(II)) of the nickel cyclidene complexes are
insensitive to bridge length and only dependent on electronic
factors, while the redox potentials of the cobalt complexes
respond to both. This difference arises because of the tendency
of cobalt to bind solvent molecules, especially in the trivalent
state where 6-coordination is dominant. Since the ligand
providing the sixth donor atom must go into the cavity,
formation of the six-coordinate cobalt(II) complex depends on
bridge length and cavity size. For cobalt complexes with
identical sets of substituents R2-R4 but different bridge lengths
(Table 4, entries 1, 5, 9, 14, 18 or 4, 8, 12, 17) redox potential
gradually decreases as bridge length increases. For nickel,
replacement of an electron donating methyl groups at R2 and/
or R3 by hydrogen or phenyl causes an increase in redox
potential. The unsubstituted cobalt cyclidenes, [Co(C6MeHH-
[16]cyclidene)]2+ and [Co(C8MeHH[16]cyclidene)]2+, display
irreversible voltammograms with poorly defined reductive
processes, and redox potentials of ∼0.53 and >0.41, respec-
tively, indicating weak sixth ligand binding.19 The high redox
potentials are clearly associated with the replacement of the
electron donating methyl group by hydrogen, but they almost
certainly also reflect a decreased ability to bind the sixth axial
ligand because of the distinctive bridge conformation.
Figure 4. X-ray structure and atom numbering for [Co(C6MeHH[16]-
cyclidene)(CH3OH)] cation (XIV).
phosphate, compositions consistent with earlier work.15b The
ultimate products, the unsubstituted cobalt(II) complexes XIV
and XV, were prepared by the procedure for the corresponding
methyl substituted cobalt(II) cyclidenes and differed only in that
the crystals contained one or three molecules of coordinated
methanol.
Structure of Complex [Co(C6MeHH[16]cyclidene)(CH3-
OH)](PF6)2 (XIV). The complex exhibits (Figure 4) the typical
saddle shaped lacunar structure that has been well documented
for the 16-membered cyclidenes.1b,18 Comparison with the
structures of [Co(C6MeMeMe[16]cyclidene)]2+ and related
compounds reveals a profound example of structural changes
caused by vicinal repulsion (Figure 5a). In the structure of [Co-
(C6MeHH[16]cyclidene)]2+ coordinated methanol leads to the
displacement of the cobalt atom from the plane of the four
nitrogens by 0.23 Å (compared to 0.06 Å in the substituted
species). Also, methanol coordination causes both saturated
chelate rings to assume the chair conformation, a behavior typical
of cyclidenes with axial ligands.1b The macrocyclic backbones
of [Co(C6MeHH[16]cyclidene)]2+ and [Co(C6MeMeMe[16]-
cyclidene)]2+ have approximately the same dimensions.18b
Substantially different widths occur between exocyclic nitrogens
and an even greater variation in the distances between bridge
terminal carbons (T) (Table 3). Data for the dioxygen adduct
of [Co(C6MeMeMe[16]cyclidene)]2+ are included to emphasize
the fact that two bridge conformations exist for the subsituted
(R4 ) methyl) complex, depending on whether the cavity is
filled or vacant.
ESR Spectra. The unoxygenated, unsubstituted cobalt(II)
cyclidenes in acetonitrile/MeIm solution display an axial ESR
pattern typical of many cobalt(II) cyclidenes (Figure 6).20-22
Comparison of ESR spectral parameters for oxygenated unsub-
stituted cobalt(II) cyclidenes with previously published data
supports the generalization that values for g are extremely
similar for dioxygen adducts of all cobalt cyclidenes (Table 5).
Values for g|, A , and A| are somewhat smaller than those for
cyclidenes alkyl substituted at R2 and/or R3 and correspond to
the spectral parameters of cyclidenes which lack electron
donating substituents at these positions.22 Thus, the electro-
chemical and ESR data both indicate a decrease in electron
The conformations of the polymethylene bridges (drawn in
Figure 5b-d) constitute the greatest and most important
differences between [Co(C6MeHH[16]cyclidene)]2+, [Co-
(C6MeMeMe[16]cyclidene)]2+. Both conformations are un-
symmetrical; a symmetrical conformation for polymethylene
bridges with an even number of links (n) is possible only if the
central carbon-carbon bond has a high energy, completely
eclipsed conformation.18b In contrast to the folding of the bridge
into its own cavity for [Co(C6MeMeMe[16]cyclidene)]2+
(Figure 5c), a conformation featuring mostly gauche torsional
(19) Chavan, M. Y.; Meade, T. J.; Busch, D. H.; Kuwana, T. Inorg.
Chem. 1986, 25, 314.
(20) Busch, D. H.; Stephenson, N. A. Inclusion Compounds Volume 5:
Inorganic and Physical Aspects of Inclusion; Atwood, J., Davies, E.,
MacNicol, D., Eds.; Oxford University Press: Oxford, 1991; pp 276-310.
(21) Drago, R. S.; Corden, B. B. Acc. Chem. Res. 1980, 13, 353.
(22) Busch, D. H. In Oxygen Complexes and Oxygen ActiVation by
Transition Metals; Martell, A. E., Sawyer, D. T., Eds.; Plenum Publishing
Corporation: 1988.
(18) (a) Alcock, N. W.; Lin, W.-K.; Jircitano, A.; Mokren, J. D.; Corfield,
P. W. R.; Johnson. G.; Novotnak, G.; Cairns, C.; Busch, D. H. Inorg. Chem.
1987, 26, 440. (b) Alcock, N. W.; Lin, W.-K.; Cairns, C.; Pike, G. A.;
Busch, D. H. J. Am. Chem. Soc. 1989, 111, 6630.

Unsubstituted Cyclidenes
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4167
Table 3. Structural Parameters for Co(C6MeMeMe[16]cyclidene)2+, Its MeIm/O2 Adduct, and Co(C6MeHH[16]cyclidene)2+
chain conformations
D, Å T, Å width, Å height, Å C1-C2-C3-C4-C5-C6
b
R0
â0
δ0
ꢀ0
CoC6MeMeMe[16]cyclidene2+ a
98.8 44.5 -23.43
-22.32 6.63 4.92
6.74
6.76
6.99
5.13
5.91
5.48
g1-a-g2-g2-g2
(g2)-e2-e1-g1-a
a-(a)-a-(e2)-g2
27.80
13.23
CoC6MeMeMe[16]cyclidene (MeIm)(O2)2+ a 110.6 55.8
-6.68 -114.2
75.90
6.97 4.76
6.65 5.49
-23.18
-0.4
20.8
CoC6MeHH[16]cyclidene2+
93.3 44.4 -29.43
25.33
a Reference 18b. b The idealized labeling system is defined as follows, in the order torsional angle, label, type: 60, g1, gauche; 120, e1, eclipsed;
180, a, anti; -120, e2, eclipsed; -60, g2, gauche. Parentheses indicate substantial deviation from ideal angle. Structural parameters defining the
cyclidene unit: R, angle between opposite [N1N2C1C2] planes; â, angle between opposite [C1C2C3] planes; δ, torsion angle C2C3C4N3; ꢀ, torsional
angle C3C4N3CH2; D, distance C3-C3′; width, N3-N3′ distance; T, distance between bridge terminal CH2 groups.
Figure 5. Bridge conformations from the X-ray structures: (a) Comparison of CoC6MeHH (XIV) (solid line) and the corresponding substituted
species CoC6MeMeMe (dashed line) view from below; (b) [Co(C6MeHH[16]cyclidene)(CH3OH)] cation; (c) [Co(C6MeMeMe[16]cyclidene)] cation;
and (d) [Co(C6MeMeMe[16]cyclidene)MeIm(O2)] cation.
density at the cobalt atom due to the absence of electron
donating R3 and/or R4 substituents.
Dioxygen Affinities. The spectral changes used in the
determination of dioxygen affinities (Figure 7) are completely
reversible, and they show two isosbestic points, at 385 and 490
nm, suggesting a simple equilibrium involving only two
absorbing species. Both of the unsubstituted cobalt(II) cyclid-
enes reported here have unusually low dioxygen affinities in

4168 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Kolchinski et al.
Table 4. Selected Redox Potentials for Nickel and Cobalt
Cyclidenesa
R1R2R3R4
E1/2 (M ) Ni)b E1/2 (M ) Co)b
refs
1
2
3
4
5
6
7
8
9
C4MeMeMe
C4HMeMe
C4HPhMe
C4MePhMe
C5MeMeMe
C5HMeMe
C5HPhMe
0.77
0.84
0.90
0.91
0.76
0.85
0.92
0.91
0.775
0.895
0.97
0.92
0.87
0.74
0.82
0.92
0.90
0.78
0.84
0.95
0.87
0.34
19, 15e
15e
10
10, 20
15e, 19
15e
0.50
0.28
10
C5MePhMe
C6MeMeMe
0.44
0.20
10, 20
15e, 19
15e
10 C6HMeMe
11 C6MeHMe
12 C6MePhMe
13 C6MeHH
14 C7MeMeMe
15 C7HMeMe
16 C7HPhMe
17 C7MePhMe
18 C8MeMeMe
19 C8HMeMe
20 C8HPhMe
21 C8MeHH
10
0.34
0.53
0.08
10, 20
this paper
15e, 19
15e
10
0.24
0.05
10, 20
15e, 19
15e
10
this paper
>0.41
a In acetonitrile solution containing 0.1 M Nbu4BF4. b For the M(III)/
M(II) redox couple.
Table 5. ESR Parametersa for Square Pyramidal, Low-Spin
Cobalt(II) Complexes of Unsubstituted Cyclidenes and Their
Dioxygen Adducts
Figure 6. ESR spectra in acetonitrile solution at -196 °C of the square
pyramidal cobalt(II) complexes: (a) [Co(C6MeHH[16]cyclidene)MeIm]-
(PF6)2 and (b) [Co(C6MeHH[16]cyclidene)MeIm(O2)](PF6)2.
R1R2R3R4
g|
g
A| (G)
A (G)
aN
Square Pyramidal Complexes
C6MeHH
C8MeHH
2.009
2.012
2.312
2.319
91.0
96.0
14
14
Dioxygen Adducts
C6MeHH
C8MeHH
2.084
2.083
2.001
2.001
16.8
16.1
8.6
8.2
a Estimated standard deviations: g| and g (0.002. A| and A (0.5
G. Conditions: acetonitrile solvent, 1.5 M MeIm axial base, frozen
glass at -196 °C.
comparison to corresponding substituted species (Figure 8,
Tables 6 and 7). CoC6MeHH binds dioxygen approximately
400 times less strongly than its methyl substituted counterpart,
CoC6MeMeMe, giving equilibrium constants close to those of
CoC4MeMeMe. The dioxygen affinity of CoC8MeHH falls
between those of CoC5MeMeMe and CoC6MeMeMe. These
dramatic changes can be explained on the basis of the electronic
and steric changes accompanying the replacement of the alkyl
substituents.
The substantial influence on dioxygen affinity exerted by the
electron donating properties of the substitutents has been
discussed in much detail.1b Both the nature of the substitution
and the position at which it occurs are important. Replacement
of the methyl group by hydrogen at R2 in CoC6MeMeMe causes
an approximately 10-fold decrease in dioxygen affinity, while
the same transformation at R3 causes a 130-fold decrease. In
the present examples, the methyl groups at both the R3 and R4
positions are replaced by hydrogen. Not surprisingly the
decrease in dioxygen affinity is greater than that observed for
CoC6MeHMe. Attributing a lowering of the dioxygen affinity
to electronic factors is also consistent with the high Co(III/II)
redox potentials of unsubstituted cyclidenes. However, elec-
tronic factors are not solely responsible for the low dioxygen
affinity of the unsubstituted cyclidenes: (a) While the redox
potentials of nickel unsubstituted cyclidenes fall in the range
typical of all nickel cyclidene complexes (Table 4), the redox
potentials for cobalt unsubstituted cyclidenes seem to be
exceptionally high. This is indicative of substantially hindered
Figure 7. Electronic spectra of 5 × 10-5 M acetonitrile/1.5 M MeIm
solution of [Co(C8MeHH[16]cyclidene)](PF6)2 at different partial
pressures of dioxygen at 0 °C. Partial pressures: ∼0; 1.3; 2.6; 6.4; 13;
19; 26; 38; 77; 153; 306; 460; 613; 766 Torr. Inset: Kinetic traces at
410 and 350 nm superimposed with first-order (one-exponential) curve
fits obtained in acetone/1.5 M MeIm solution at -75 °C; initial
concentration of Co complex 7 × 10-6 M, initial concentration of O2
2 × 10-5 M.
solvent molecule coordination under the bridge. Since the C6
and C8 bridges in CoCnMeMeMe provide enough space for
axial coordination of a sixth ligand, restrictions in axial
coordination for CoCnMeHH are clearly connected to the bridge
conformation. (b) Changing the bridge length from C6 to C8
for the CoCnMeMeMe cyclidenes results in a 4-5-fold increase
in oxygen affinity (entries 5 and 12, Table 7). The same bridge
length increase for the CoCnMeHH cyclidenes results in almost
30-fold increase in oxygen affinity (entries 8 and 14, Table 7)
indicating more substantial structural changes from the stand-
point of accessibility of the sixth site by a ligand.
X-ray crystallographic information clarifies this matter. The
folded C6 bridge in CoC6MeMeMe cyclidene has the ability
to flip out, away from the cavity, and provide a convenient
coordination site for dioxygen binding (Figure 5). The low

Unsubstituted Cyclidenes
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4169
oxygen 85 times more strongly than the complex with hydrogens
at the corresponding R3 positions CoC6MeHMe. R2R3 vicinal
repulsion in the phenyl substituted complex forces the bridge
to fold, opening the cavity to O2, whereas CoC6MeHMe, with
no R2R3 vicinal repulsion probably, has a confining straight
bridge like CoC6MeHH. The decreases in the ratios of the
dioxygen affinities for CoCnMeHMe compared to CoCnMe-
MeMe is approximately 10-, 9-, 6-, and 3.5-fold, respectively,
for n ) 5, 6, 7, and 8, respectively. For the longer bridges, the
response to the removal of vicinal repulsion becomes smaller,
because completely straight conformations are no longer achiev-
able.
For most of the previously studied cyclidenes, variations in
∆H and ∆S are relatively small (Table 7, Figure 8); ∆H ranging
from -15 to -18 kcal/mol, and ∆S varying between -37 and
-44 e.u. However, for the unsubstituted cyclidenes, oxygen-
ation of CoC8MeHH (∆H ) -20.9 kcal/mol) is distinctly more
exothermic than the average for cobalt cyclidene complexes
(about -17 kcal/mol); while the corresponding value of ∆H
for CoC6MeHH is unusually small (-7.6 kcal/mol). The latter
value is comparable only to the value for Co(C2CMe2C2)-
MeMeMe cyclidene, a material bearing a highly hindered,
moderately short C5 bridge (ca. -9 kcal/mol).23 These results
suggest that the bridges in both CoC6MeHH and CoC2CMe2C2-
MeMeMe restrict access to the cobalt atom and weaken Co-
dioxygen interaction.
Figure 8. van’t Hoff plots for equilibrium constants of dioxygen
binding with CoC6MeHH and CoC8MeHH.
Table 6. Dioxygen Affinities for the Unsubstituted Cobalt(II)
Cyclidenes at Different Temperaturesa
R1R2R3R4
KO (Torr-1) [(°C)]
2
C6MeHH
C8MeHH
0.0099 [-20.0]; 0.0047 [-5.0]; 0.0020 [5.0]; 0.0012 [23.5]
0.376 [-10]; 0.092 [0.0]; 0.026 [10]; 0.006 [20]
Kinetics of Dioxygen Binding. The equilibrium constant
for dioxygen binding with cobalt complexes is determined by
the rate constants of O2 binding and dissociation:
a In acetonitrile/1.5 M methylimidazole.
k
Table 7. Selected Dioxygen Affinities for the Cobalt(II)
on
Cyclidenesa
Co(L)(B) + O2 a Co(L)(B)(O2) KO ) kon/koff (1)
2
k
off
KO
∆H,b
kcal/mol
∆S,b
eu
2
R1R2R3R4
C4MeMeMe
(Torr-1
)
ref
Historically, steric hindrance deriving from the superstructure
of a macrocyclic ligand has been found to affect kon, causing
decreases in binding rates as the cavity shrinks.3c,4d This effect,
traceable to the entropy of activation, is attributed to limited
access of the dioxygen molecule to the cobalt(II) ion: only a
restricted fraction of the possible O2 and complex orientations
permit Co-O2 interaction. In contrast, electronic effects (e.g.,
changing the axial base B) lead to significant changes in the
dissociation rates (the weaker the Co-O2 bond, the faster the
dissociation).3c,5d These trends were also observed for CoCn-
MeMeMe cyclidenes (n ) 4-6).24
The kinetics of oxygenation of unsubstituted cyclidenes were
studied at low temperature (-75 °C) by a cryogenic stopped-
flow technique with spectrophotometric monitoring of concen-
tration changes, because (1) at low temperatures the process is
slow enough to permit the precise recording of kinetic traces
and (2) equilibrium (1) is shifted toward dioxygen adduct
formation at low temperatures, enhancing precision. Further,
the reverse reaction is suppressed, and a simple second-order
rate law applies:
1
0.0020
1b
(-40 °C)
0.064
0.0066
2
3
4
5
6
7
8
9
C5MeMeMeb
C5HMeMe
-16.2
-15.4
-42 1b, 23
-44 1b, 23
(C2CMe2C2)MeMeMe 0.027
-9.2c -18c 1b, 23
C6MeMeMe
C6HMeMe
C6MeHMe
C6MeHHb
C6MePhMe
1.3
-17.2
-17.6
-15.0
-7.6
-17.8
-18.3
-17.3
-17.3
-40 1b, 23
-45 1b, 23
-41 1b, 23
-17 this paper
-42 1b, 23
-41 1b, 23
-41 1b, 23
-37 1b
0.14
0.010
0.0032
0.85
4.5
0.80
5.92
1.7
10 C7MeMeMe
11 C7HMeMe
12 C8MeMeMe
13 C8HMeMe
14 C8MeHH
1b, 23
0.092
-20.8
-58 this paper
a In acetonitrile/1.5 M methylimidazole, 0 °C. b Calculated from
temperature dependence. In order to get the entropy values in cal/K
mol, we assumed the oxygen solubility in acetonitrile to be 8.1 mM
(Achord, J. M.; Hussey, C. L. Anal. Chem. 1980, 52, 601), and the
temperature dependence of O2 solubility was ignored. The entropy
values in cal/K Torr which were reported earlier (ref 23) can be obtained
from these numbers by subtracting 22.9. c Two data points are available.
energy, straight conformation of the bridge in CoC6MeHH is
unique and obviously inhibits O2 binding; no relatively open
low energy conformation can compete. The straight conforma-
tion of the bridge in CoC6MeHH must either substantially
restrict the approach of dioxygen to the metal center or lead to
unfavorable geometry within the dioxygen adduct itself. This
is an issue that is addressed in the studies on rates of binding
and dissociation.
V ) k[Co(L)(B)][O2]
(2)
This simplification of the rate law was very important in our
case, because variations in the concentrations of the reagents
were confined by rapid rates and the sensitivity of the
instrumentation. Unfortunately, this strategy prevented direct
measurement of the rates of dioxygen dissociation.
The changes in bridge conformation in going from Co-
C6MeMeMe to CoC6MeHH are traceable to the elimination
of vicinal R2R3 methyl-methyl repulsions, and this observation
is readily extended to other cases. Despite the electron-
withdrawing effect of phenyl groups, CoC6MePhMe binds
(23) Busch, D. H.; Jackson, P. J.; Kojima, M.; Chmielewski, P.;
Matsumoto, N.; Stevens, J. C.; Wu, W.; Nosco, D.; Herron, N.; Ye, N.;
Warburton, P. R.; Masarwa, M.; Stephenson, N. A.; Christoph, G.; Alcock,
N. W. Inorg. Chem. 1994, 33, 910.
(24) Rybak-Akimova, E. V.; Masarwa, M.; Marek, K.; Warburton, P.
R.; Busch, D. H. Chem. Commun. 1996, 1451.

4170 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Kolchinski et al.
Table 8. Kinetic Parameters for Dioxygen Binding to Cobalt(II)
perature dependencies, we can justify the extrapolation of
oxygenation rates beyond the very low temperature range for
the purpose of making comparisons.
Cyclidenesa
q
kon, M-1 s-1
(-75 °C)
∆Hon
,
compd
kcal/mol
ref
The most general conclusions from these rate studies are very
clear. Despite substantial differences in dioxygen affinities
between unsubstituted cyclidenes and their methyl-substituted
analogs (see above), the rates of dioxygen binding of CoCn-
MeMeMe and CoCnMeHH are essentially identical (for the
same bridge length n). This indicates that, at the transition state
for O2 binding, there is no substantial difference between
methyl-substituted and unsubstituted cyclidenes. Further, the
very small activation enthalpy suggests that the transition state
occurs early in the Co-O bond-forming process, a scenario
consistent with the identity of the rates for the two rather
different cobalt compounds.
1
2
3
4
5
6
CoC4MeMeMe 5 × 103
CoC5MeMeMe 2 × 105
CoC6MeMeMe (5 ( 1) × 106
2.2
0.7
4b
calcd from 25
calcd from 25
25, this paper
this paper
this paper
this paper
CoC6MeHH
(4 ( 1) × 106
3b
CoC8MeMeMe (2 ( 0.5) × 107
4c
CoC8MeHH
(2 ( 0.5) × 107
4d
a In acetones1.5 M 1-methylimidazole, at -75 °C. b (1 kcal/mol.
c Estimated from three data points. d Estimated from two data points.
In contrast, it was shown above that the dramatic difference
in dioxygen affinity between CoC6MeMeMe and CoC6MeHH
can be explained by a combination of steric and electronic
factors. Now, the kinetic data tell us that the nature of the steric
factor is different from that usually observed (usually, kon is
affected).27 From the equilibrium constants and the rate
constants for binding, we can conclude that the rate constants
for dissociation must account for the differences in dioxygen
affinities. Clearly, the difference in the bridge conformation
for CoC6MeHH accelerates O2 departure (when compared to
CoC6MeMeMe), indicating that the Co-O linkage is weakened.
For both the unsubstituted and methyl-substituted complexes,
the C8 bridge permits dioxygen binding that is approximately
five times faster than that for the C6 bridged derivative. A
similar rate/bridge length trend has been reported for a series
of C4-C6 methyl-substituted cobalt cyclidenes.24 The increase
in rate constants for O2 binding with increasing cavity size is
attributable to the activation entropy and is associated with entry
of the dioxygen into the lacuna. The difference in dioxygen
affinities between CoC6MeMeMe and CoC8MeMeMe (five
times) is completely assignable to different binding rates, while
differences in both binding and dissociation rates contribute to
the difference in O2 affinities between CoC6MeHH and
CoC8MeHH (30 times).
Autoxidation. Previous studies on the kinetics and mech-
anism of autoxidation of cobalt(II) cyclidene-dioxygen ad-
duct6,28 revealed a conjugate base mechanism in which depro-
tonation may occur either in the rate determining step or in a
preequilibrium step. In previous studies, substitution for
relatively acidic protons at the R2 and R3 positions had given
orders or magnitude decreases in rates of autoxidation, but the
same conjugate mechanism persisted, albeit at much retarded
rates. In view of the well demonstrated role of a methyl group
at the R3 as the site for acceleration of autoxidation due to
deprotonation, suspicion fell on the ubiquitous R4 methyl groups.
The synthesis of the so-called unsubstituted cyclidenes followed
and is reported here. Preliminary rate data on the autoxidation
Figure 9. Temperature dependence of the second-order rate constants
for the oxygenation of cobalt(II) cyclidenes (1.5 M 1-MeIm/acetone).
Absorbance changes upon mixing in the stopped flow
experiments correspond very well to those observed during
spectrophotometric titrations (Figure 7), and the rate constants
determined at different wave lengths are identical. In acetone,
rates do not depend on the axial base concentration over the
range studied (0.5-1.5 M 1-methylimidazole), but they do
depend on the concentrations of cobalt(II) complex and dioxygen
(Table 8).25 For the two complexes having C6 bridges,
significant variations of reagent concentrations (about 1 order
of magnitude) and pseudo-first-order conditions were accessible.
For the C8 bridged complexes,26 the reactions are faster, and
reasonable kinetic traces could be determined only for a
relatively narrow range of reagent concentrations, limiting the
second-order rate constants to uncertainties of about 50%.
The activation energies for oxygenation reactions were
estimated from Arrhenius plots (Figure 9). Because of high
reaction rates, measurements were limited to the temperature
range from -75 to -50 °C for C6 bridged compounds and from
-75 to -65 °C for C8 bridged compounds. While it is obvious
that such temperature ranges are not adequate for precise
determination of activation enthalpies and entropies, these
limited data clearly show that the activation energies are
relatively small and very similar for all of the complexes under
investigation (Figure 9). Because, this general result requires
the oxygenation rates for CoC6MeMeMe, CoC8MeMeMe,
CoC6MeHH, and CoC8MeHH to display almost parallel tem-
fast
CoLH(O2)2+ + B salow CoL(O2)2+ + BH a products (3)
of these new compounds completes the work presented here.
Since the suspected acidic methyl groups R4 have been
eliminated from the structures of the unsubstituted cyclidenes,
increased stability toward autoxidation is expected. The results
of the most preliminary of experiments showed that this was
indeed the case since the autoxidation reactions are very slow
under conditions used in previous studies of the many R4 )
(25) O2 solubility in acetone at 25 °C (temperature at which dioxygen
solutions were prepared for our kinetic measurements) is reported to be
11.5 mM (Linke, W. F. Solubilities of Inorganic and Metal-Organic
Compounds; American Chemical Society: Washington, DC, 1965), or 11.0
mM (Achord, J. M.; Hussey, C. L. Anal. Chem. 1980, 52, 601). An average
value of 11.2 mM has been used in our calculations of O2 concentration.
(26) CoC8MeMeMecyclidene (hexafluorophosphate salt) was synthesized
according to ref 15b.
(27) Distal side steric effects usually derive from accessibility of the
metal ion and appear in kon (refs 3c and 4d).
(28) Chia, P. S. K.; Masarwa, M.; Warburton, P. R.; Wu, W.; Kojima,
M.; Nosco, D.; Alcock, N. W.; Busch, D. H. Inorg Chem. 1993, 32, 2736.

Unsubstituted Cyclidenes
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4171
design through which so-called ideal dioxygen carrier structures
are being approached incrementally. A very slow rate of
autoxidation, or none at all, is a requisite property of the ideal
dioxygen carrier. Previously, mechanistic studies have shown
that ionizable protons support a conjugate base mechanism for
the autoxidation of the cobalt(II) cyclidene dioxygen adducts,
and with the completion of the present work, all substituents
capable of contributing to that process have now been elimi-
nated. The results have been striking; the rate of autoxidation
has been retarded by some 4 orders of magnitude by controlling
that structural weakness alone. Unfortunately, the complexes
still undergo autoxidation, albeit, at the slowest rate so far
observed for compounds of this class. If this slow autoxidation
rate is associated with the familar conjugate base mechanism,
then the likely source of deprotonation involves the trimethylene
groups of the saturated six-membered chelate rings. Improved
molecular design will obviate this possibility in the future.
Figure 10. Absorbance changes in the course of CoC8MeHH
autoxidation (5 × 10-5 M solution of complex in 1.5 M 1-MeIm/
acetonitrile, 760 Torr O2). Reactions run at 30 °C, spectra recorded at
20 °C for nonoxygenated complex and 0; 4; 8; 20; 32; 44; and 72 h
after oxygenation. Inset: linear first-order kinetic curve (measured at
348 nm).
Removal of the methyl substituents from the unsaturated
portion of the parent cyclidene ring produced an unexpected
decrease in O2 affinity, especially in the case of the hexameth-
ylene bridged complex. While this is attributable in part to
decreased electron density at the metal atom, a profound change
in the conformation of the bridging hexamethylene chain is
mainly responsible. Exploring this phenomenon has led to
increased understanding of the manner in which the bridging
group controls O2 affinity and an exceptionally clear example
of a steric effect in which the weakening of the Co-O bond
causes an increase in the rate of dissociation of O2. This further
increases the number of structural relationships available to fine
tune dioxygen affinity.
Table 9. Observed Pseudo-First-Order Rate Constants for the
Autoxidation of Co(II) Cyclidene Complexes in 1.5 M
1-Methylimidazole-Acetonitrile, 30 °C, 760 Torr of Dioxygen
complex
kobs, s-1
ref
1
2
3
4
5
6
7
8
9
CoC3MeMeMe[16]2+
CoC4MeMeMe[16]2+
CoC5MeMeMe[16]2+
CoC6MeMeMe[16]2+
CoC7MeMeMe[16]2+
CoC8MeMeMe[16]2+
CoC8HMeMe[16]2+
CoC6MePhMe[16]2+
CoC6MeHH[16]2+
2.0 × 10-5
2.0 × 10-5
2.0 × 10-4
4.0 × 10-4
1.0 × 10-3
8.0 × 10-4
1.0 × 10-2
2.0 × 10-5
4.0 × 10-6
2.5 × 10-6
6
6
6
6
6
6
6
6
this paper
this paper
10
CoC8MeHH[16]2+
methyl complexes; this required special techniques for the
determinations reported here (see Experimental Section; Figure
10). The data in Table 9 indicates a substantial decrease (5-8
times) in autoxidation rates for unsubstituted cyclidenes when
compared with the most stable cobalt(II) cyclidenes previously
investigated, confirming the prediction. The high redox po-
tentials of the unsubstituted cyclidenes may also contribute to
their slow rates of autoxidation. Detailed investigation of the
kinetics and mechanism of autoxidation of the unsubstituted
cyclidenes is the subject of further research. The simple
measurements reported here merely confirm the hypothesis that
removal of the suspected R4 groups does indeed lead to dioxygen
carriers that autoxidize more slowly. Clearly that is true. It
will be of interest to learn if the same mechanism still persists,
perhaps involving the methylene protons of the saturated chelate
rings, or if some entirely new autoxidation mechanism has
appeared.
Acknowledgment. This material is based upon work sup-
ported by the NSF under Grants numbered OSR-9255223 and
CHE-9118455 and matching support from the state of Kansas.
The collaboration between The University of Kansas and The
University of Warwick has been supported by a NATO travel
grant. We wish to thank Dr. Todd Williams, Mr. Bob Drake,
and Ms. Homigol Biesiada of the University of Kansas for the
efforts in acquiring mass spectra. The authors also greatly
appreciate the contributions to the acquisition of 2-D NMR data
by Dr. Martha Morton. Dr. Tho Nguyen of the University of
Kansas supplied elemental analysis. These contributions are
deeply appreciated.
Supporting Information Available: Tables listing aniso-
tropic thermal parameters, H-atom coordinates, full bond lengths
and angles, and atomic coordinates (6 pages). See any current
masthead page for ordering and Internet access instructions.
Conclusions
The family of cyclidene ligands continues to be remarkably
flexible and appropriate for the process of iterative molecular
JA9624477