Use of a radical clock to study the photodecarboxylation of amino acidatocobalt(III) complexes

Article abstract of DOI:10.1039/a904451e

The products of steady state UV irradiation of amino acidatobis(2,2′-bipyridine)cobalt(III) complexes have been identified. The products of these photodecarboxylation reactions include a carbonyl compound derived from the amino acid side chain. The viability of an aminocyclopropylmethyl radical clock as a potential mechanistic probe for the reaction was assessed by ab initio methods. The required cyclopropylglycine containing substrate was synthesized and used to show that reaction of an α-carbon centred radical must be a fast step if it occurs in the reaction sequence leading to formation of the intermediate metallacyclic complexes. The Royal Society of Chemistry 1999.

Full text of DOI:10.1039/a904451e

View Article Online / Journal Homepage / Table of Contents for this issue
Use of a radical clock to study the photodecarboxylation of amino
acidatocobalt(III) complexes†
Richard M. Hartshorn* and Shane G. Telfer
Department of Chemistry, University of Canterbury, Private Bag 4800, Christchurch,
New Zealand
Received 4th June 1999, Accepted 23rd August 1999
The products of steady state UV irradiation of amino acidatobis(2,2Ј-bipyridine)cobalt() complexes have been
identified. The products of these photodecarboxylation reactions include a carbonyl compound derived from the
amino acid side chain. The viability of an aminocyclopropylmethyl radical clock as a potential mechanistic probe for
the reaction was assessed by ab initio methods. The required cyclopropylglycine containing substrate was synthesized
and used to show that reaction of an α-carbon centred radical must be a fast step if it occurs in the reaction sequence
leading to formation of the intermediate metallacyclic complexes.
contains a co-ordinated aminoalkyl radical, 4. The alkyl radical
is then thought to react with the cobalt() centre, yielding the
metallacyclic product 2.
Introduction
The investigation of photochemical reactions of ligands co-
ordinated to transition metals is a relatively unheralded branch
A kinetic study of this reaction, using laser flash photolysis
of inorganic chemistry,1,2 but it is one which may provide syn-
techniques, has been reported.5 In that study the rates of form-
thetic avenues to interesting or otherwise inaccessible organic or
ation of the metallacyclic products were monitored spectro-
co-ordination compounds. An example of this is the formation
of a complex containing a three-membered Co–C–N metal-
photometrically, and it was found that the observed absorbance
changes could be analysed in terms of first order kinetics. Rate
lacycle via irradiation of the bis(2,2Ј-bipyridine)glycinato-
constants of around 4 × 103 sϪ1 were calculated for the form-
cobalt() ion, [Co(gly)(2,2Ј-bpy)2]2ϩ 1a, with UV light.3 This
ation of metallacyclic complexes from a range of α-amino acid
complexes, including the glycinato example described above.
These data are consistent with any one of the three irrevers-
complex, [Co(CH2NH2)(2,2Ј-bpy)2]2ϩ 2a, is reasonably stable
and has been characterised by X-ray crystallographic methods.4
A range of similar reactions has been reported, but the stabil-
ible first order reactions in the proposed mechanism being rate
ity of the resulting cobalt–alkyl products varies consider-
determining, since the observations that were made were
ably.1a,c,3 For example, metallacycles with substituents on the
α-carbon are thermally unstable, while more robust four mem-
bered metallacycles result from the photodecarboxylation of
associated with formation of the metallacyclic product and not
with the appearance or decay of any of the intermediates. How-
ever, in the discussion of their results, the authors assigned
complexes of β-aminocarboxylates. Many of these latter
these rate constants to the final step of the reaction sequence,
metallacycles were part of polydentate frameworks.
cobalt–carbon bond formation. There was little discussion of
A mechanism for these reactions has been proposed1a
this assignment, which was presumably made on the assump-
(Scheme 1), wherein absorption of UV light leads to homolysis
tion that photochemical bond homolysis and decarboxylation
steps are likely to be fast processes.
There are two anomalies that arise as a result of this assign-
ment of the rate determining step. First, intermolecular reac-
tions between cobalt() complexes and methyl radicals have
second order rate constants of greater than 107 MϪ1 Ϫ1.6 We
s
consider it remarkable that similar intramolecular reactions
should have rate constants as small as 103 sϪ1. Secondly, the
measured rates would imply that the cobalt()–aminoalkyl
radical intermediate 4 has a lifetime of the order of milli-
seconds. The aminoalkyl radical is a monodentate ligand at this
point in the reaction sequence, and therefore relatively easily
exchanged for other ligands. Indeed, given that cobalt() com-
plexes can undergo ligand exchange reactions on the micro-
second timescale in aqueous solution, one might expect that
essentially all of the radical ligands would be lost before a
reaction with the cobalt() ion at ≈103 sϪ1 could occur.
Scheme 1 a: gly, R1 = R2 = H; b: ala, R1 = CH3, R2 = H; c: val,
R1 = CH(CH3)2, R2 = H; d: cpg, R1 = cyclopropyl, R2 = H; e: aib,
R1 = R2 = CH3.
We consider, on the basis of these inconsistencies, that there
is some doubt concerning the assignment of the rate determin-
ing step in the proposed mechanism. Indeed, it may be that it is
the mechanistic proposal itself which is incorrect.
of the cobalt–oxygen bond, affording a cobalt()-bound
aminoacyloxy radical intermediate 3. This complex is proposed
to undergo a decarboxylation reaction to give a species which
Our approach to the further investigation of this system has
been to attempt to measure the relative lifetime of the putative
aminoalkyl radical intermediate, 4, with the aid of a radical
clock. In particular, we proposed to employ a cyclopropyl-
methyl radical clock, in which cyclopropane ring opening com-
† Supplementary data available: results of ab initio calculations. For
direct electronic access see http://www.rsc.org/suppdata/dt/1999/3565/,
otherwise available from BLDSC (No. SUP 57632, 5 pp.) or the RSC
Library. See Instructions for Authors, 1999, Issue 1 (http://www.rsc.org/
dalton).
J. Chem. Soc., Dalton Trans., 1999, 3565–3571
This journal is © The Royal Society of Chemistry 1999
3565

View Article Online
petes with the radical reaction of interest. To this end we have
synthesized a complex of cyclopropylglycinate, 1d, and explored
its photochemistry.
and cooled overnight. The red plate-like crystals and mother-
liquor were treated as in ref. 9, to yield pentaammine(2-
cyclopropyl-2-oxo-ethanoato)cobalt() perchlorate and then
tetraammine(2-cyclopropyl-2-iminoethanoato)cobalt() per-
chlorate (1.91 g, 4.4 mmol, 34% from LiBOP).
Knowledge of the rate of the radical clock rearrangement
reaction is required before the results of a radical clock
experiment can be used to infer anything about the reaction of
interest. Unfortunately, no kinetic information was available for
the rearrangement reaction of the aminocyclopropylmethyl
radical, despite numerous approaches, both theoretical and
experimental, to this problem for many related radicals. Cer-
tainly, there was no information available on the rates of
rearrangement of such radicals when they are co-ordinated to
metal ions. We therefore set out to obtain an estimate for the
rate constant for the ring opening of the aminocyclopropyl-
methyl radical and its protonated analogue, according to the
methods of Radom7 and of Newcomb and co-workers.8 These
results, together with the trends relating the nature of the sub-
stituent and the rate constants for various related radicals, can
be used to make a prediction for the rate of ring opening of
the amino substituted radical.
(b) 2-Cyclopropylglycine was produced and liberated from
the cobalt centre of the tetraammine(2-cyclopropyl-2-imino-
ethanoato)cobalt() perchlorate (1.1 g, 2.5 mmol) by treat-
ment with an excess of NaBH4 (0.25 g) in a Na2CO3/NaHCO3
buffer (pH 10, 90 ml). The solution was stirred for four minutes
and then acidified with dilute HCl to pH 3. A black by-product
deposited with concomitant generation of H2 gas. The reaction
mixture was filtered through Celite, and the pale pink filtrate
loaded on to Dowex (Hϩ form, 8 × 3 cm). The column was
washed with 0.1 M HCl (300 ml), then eluted with 0.5 M NH3
solution. Fractions of 30 ml were taken and the solvent
removed on a rotary evaporator. The cyclopropylglycine, as
1
identified by H NMR, eluted after 90 ml. The hydrophobic
amino acid was recrystallised twice by layering a supersaturated
aqueous solution with PriOH and cooling in a freezer. A white
powder precipitated which was washed with PriOH and diethyl
ether, then air dried. Yield 82 mg, 59%. 1H NMR (dilute DCl):
δ 0.40–0.85 (complex peak, cyclopropyl CH2, 4 H), 1.14 (m,
cyclopropyl CH, 1 H) and 3.13 (d, α-H, 1 H, J = 10.7 Hz). 13C
NMR (D2O): δ 5.38, 6.05 (cyclopropyl CH2s), 14.06 (cyclo-
propyl CH), 61.40 (α-C) and 175.97 (COO). Calc. for
C5H9NO2: C, 52.16; H, 7.88; N, 12.17. Found: C, 51.94; H, 7.88;
N, 11.96%. EI-MS: m/z 116 ([C5H10NO2]ϩ).
The results from the theoretical study and from the photo-
chemical experiments are then brought together in a discussion
of the mechanism of the photochemical reaction.
Experimental
Materials
Reagent grade reagents were used without further purification
for all experimental work. ,-Alanine, ,-valine and 2-amino-
isobutyric acid were obtained from Aldrich Chem. Co., glycine
and 2,2Ј-bipyridine from BDH, deuterium oxide (D2O), deuter-
ium chloride (DCl) and dimethyl-d6 sulfoxide (dmso-d6) from
Aldrich.
2. Synthesis and spectroscopy of the amino acidatobis(bipyridyl)-
cobalt(III) complexes
These complexes were prepared by the general method of Tate-
hata.10 The amino acid (2.92 mmol) was added to a suspension
of [CoCl2(2,2Ј-bpy)2]Clؒ3H2O11 (0.5 g, 2.92 mmol) in dry
methanol (30 ml). Sodium acetate (0.2 g) was added and the
mixture warmed at 45–50 ЊC with stirring. After one hour the
orange solution was diluted with distilled water (300 ml) and
loaded on to SP-Sephadex(C25) (Naϩ form, 3 × 20 cm). An
orange band eluted with 0.2 M NaCl, preceded by a red-orange
band and followed by a yellow band. The orange eluate was
concentrated on a rotary evaporator and desalted on Sephadex
G-10. This solution was concentrated once again, and an
orange solid precipitated upon the addition of saturated
NaClO4 solution. Recrystallisation of the crude solid from
hot dilute HClO4 gave orange crystals suitable for the micro-
analyses and electronic spectra.
Measurements
1
The H NMR spectra were taken at 300 MHz on a Varian
Unity-300 spectrometer at 23 ЊC. For those recorded in dmso-d6,
the dmso-d5 line (δ 2.50, multiplet) was used as a reference.
Sodium 3-trimethylsilylpropanesulfonate (TMPS, δ 0, singlet)
was used as in internal reference for those spectra recorded
in D2O. A Varian XL-300 spectrometer was employed for the
13C NMR spectra at 75 MHz and all spectra were proton
decoupled. The spectra were recorded at 23 ЊC and referenced
to the dmso-d6 peak (δ 39.5) or, in D2O, to 1,4-dioxane (internal
standard, δ 67.4). The signals are described as singlets (s), doub-
lets (d), triplets (t), or multiplets (m), and broad (br) where
appropriate. A GBC-920 spectrophotometer was used to record
the UV-VIS spectra in water and the data are reported as λmax
(εmax). Microanalyses were performed by the Campbell Micro-
analytical Laboratory, University of Otago. FAB Mass spectral
data were obtained on a Kratos MS80RFA mass spectrometer
equipped with an IonTech ZN11NF atom gun using xenon as
the reagent gas and 3-nitrobenzyl alcohol as the matrix. Elec-
tron impact spectra were obtained at 70 eV.
(a) [Co(gly)(2,2Ј-bpy)2][ClO4]2ؒH2O. 1H NMR (dmso-d6):
δ 3.15–3.32 and 3.75–3.85 (m, α-H, 2 H), 6.16 (br m, NH, 1 H),
6.64 (br m, NH, 1 H), 7.18 (d, 1 H), 7.62 (m, 3H), 8.22 (m, 2 H),
8.41 (m, 2 H), 8.57 (d, 1 H), 8.73 (m, 2 H), 8.89 (t, 2 H), 9.03 (d,
2 H) and 9.20 (d, 1 H). 13C NMR (dmso-d6): δ 44.46 (α-C),
124.84, 125.70, 126.14, 128.81, 129.07, 139.93, 130.08, 142.17,
142.89, 143.22, 149.54, 150.99, 152.32, 153.48, 155.94, 156.38,
156.70, 156.95 and 180.67 (COO). Calc. for [C22H26Cl2-
CoN5O10ؒH2O: C, 39.90; H, 3.35; N, 10.57. Found: C, 39.67; H,
3.00; N, 10.34%. UV-VIS: 486 nm (93 MϪ1 cmϪ1). MS (FAB):
m/z 544 ({[Co(gly)(2,2Ј-bpy)2]ClO4}ϩ).
CAUTION: perchlorate salts of metal complexes containing
organic ligands are potentially explosive and should be handled
with care and only in small quantities.
1. Preparation and isolation of 2-cyclopropylglycine (Hcpg)
(b) [Co(ala)(2,2Ј-bpy)2][ClO4]2ؒH2O. Two diastereoisomers
(a) Tetraammine(2-cyclopropyl-2-iminoethanoato)cobalt()
perchlorate was prepared by a modification of the method of
Lawson et al.9 Lithium 5-bromo-2-oxopentanoate (LiBOP)
(2.95 g, 13 mmol) in ethanol (40 ml) was added in portions over
were found in the crude product. A COSY spectrum aided the
assignment of peaks in the H NMR spectrum for the minor
1
isomer as the signal of the α-H was fully obscured by the
residual H2O peak. This assignment was confirmed by obtain-
ing a spectrum in D2O which revealed the resonances for both
two hours to
a solution of pentaammineaquacobalt()
1
perchlorate (7.40 g, 16 mmol) in water (60 ml) at 60 ЊC. The
reaction mixture was kept at this temperature, with stirring for
two hours while the pH was kept at 5.25–5.5 by the addition of
Na2CO3. The solution was then acidified to pH 2 with HClO4
α-H. H NMR (dmso-d6): major isomer, δ 1.18 (d, CH3, 3 H,
J = 6.8 Hz), 3.85 (q, α-H, 1 H), 5.82 (br t, NH, 1 H), 7.11–7.18
(d, bpy H overlapping with NH, 2 H), 7.59–7.73 (br m), 8.15–
8.21 (m), 8.38–8.45 (m), 8.68–8.73 (m), 8.90 (m), 8.98–9.05 (m)
3566
J. Chem. Soc., Dalton Trans., 1999, 3565–3571

View Article Online
and 9.18 (d, 1 H); minor isomer, δ 1.39 (d, CH3, 3 H, J = 7.3
Hz), 3.36 (α-H, obscured by H2O peak), 6.17 (br m, NH, 1 H)
and 6.63 (br m, NH, 1 H); most of the bipyridine resonances
overlapped those of the major isomer, however the following
peaks could be seen clearly, δ 8.49 (d, 1 H, J = 5.4) and 9.28 (d,
1 H, J = 5.4 Hz). 13C NMR (dmso-d6): major isomer, δ 19.34
(CH3), 52.06 (α-C), 124.95 (2C), 125.72, 126.36, 128.89, 129.14,
130.01, 130.19, 142.31 (2C), 142.98, 143.41, 148.85, 150.99,
152.73, 153.58, 156.37, 156.52, 157.19, 157.26 and 182.12
(COO); minor isomer, δ 18.74 (CH3), 51.82 (α-C) and 182.48
(COO); again most of the bipyridine signals were obscured
by the major isomer, however nine peaks were identifiable,
δ 125.12, 125.72, 129.25, 129.71, 143.09, 149.69, 151.42, 156.58
and 157.10. Calc. For [C23H22Cl2CoN5O10ؒH2O: C, 40.85; H,
3.58; N, 10.36. Found: C, 40.80; H, 3.30; N, 10.04%. UV-VIS:
482 nm (92 MϪ1 cmϪ1). MS (FAB): m/z 558 ({[Co(ala)(2,2Ј-
bpy)2]ClO4}ϩ).
156.66, 156.81, 157.31, 157.78 and 184.49 (COO). Calc. for
[C24H24Cl2CoN5O10ؒH2O: C, 41.76; H, 3.80; N, 10.14. Found: C,
41.58; H, 4.07; N, 9.86%. UV-VIS: 482 nm (99 MϪ1 cmϪ1). MS
(FAB): m/z 572 ({[Co(aib)(2,2Ј-bpy)2]ClO4}ϩ).
3. Steady state photolysis
These experiments were performed using a 200 W high pressure
mercury arc lamp equipped with a 254 nm Pyrex filter (Corning
7-54). The D2O solutions, in NMR tubes or UV cells, were
1
cooled in a quartz ice–water bath during irradiation. The H
NMR spectra were run following 60–70 min of photolysis.
(a) [Co(ala)(2,2Ј-bpy)2][ClO4]2ؒH2O. 1H NMR spectrum
following photolysis (DCl): free alanine, δ 1.59 (d, 3 H, CH3,
J = 7.3 Hz) and 4.12 (m, 1 H, α-H); acetaldehyde, δ 2.23 (br s,
3 H, CH3) and 9.66 (br s, 1 H); hydrated acetaldehyde, δ 1.31 (d,
3 H, CH3) and 5.23 (d, 1 H); free bipyridine, δ 7.98 (m, 2 H),
8.52 (m, 4 H) and 8.91 (m, 4 H); [Co(2,2Ј-bpy)3]3ϩ, δ 7.51 (d, 6
H, J = 5.8), 7.83 (t, 6 H, J = 7.3), 8.58 (t, 6 H, J = 7.8) and 8.88
(d, 6 H, J = 7.8 Hz).
(c) [Co(val)(2,2Ј-bpy)2][ClO4]2.
A mixture of diastereo-
isomers was present in the crude product. A diastereoisomeric-
ally pure product was obtained by recrystallising the crude
1
product from dilute HClO4 several times. H NMR (dmso-d6):
(b) [Co(val)(2,2Ј-bpy)2][ClO4]2. 1H NMR spectrum following
photolysis (DCl): free valine, δ 0.97–1.04 (m, 6 H, both
CH(CH3)2), 2.26 (m, 1 H, CH(CH3)2) and 4.04 (d, 1 H, α-H,
J = 2.5 Hz); 2-methylpropanal, δ 0.89 (d, CH(CH3)2), 1.71 (m,
CH(CH3)2) and 9.57 (d, 1 H, J = 1.4 Hz); hydrated 2-methyl-
propanal, δ 1.08 (d, CH(CH3)2), 2.61 (m, 1 H, CH(CH3)2) and
4.71 (d, 1 H, J = 5.4 Hz); free bipyridine, δ 7.98 (m, 2 H), 8.52
(m, 4 H) and 8.91 (m, 4 H); [Co(2,2Ј-bpy)3]3ϩ, δ 7.51 (d, 6 H),
7.83 (t, 6 H), 8.58 (t, 6 H) and 8.88 (d, 6 H).
major isomer, δ 0.60 (d, CH(CH3)2, 3 H, J = 7.0), 0.96 (d,
CH(CH3)2, 3 H, J = 7.0 Hz), 2.08 (br m, CH(CH3)2, 1 H), 3.77
(br m, α-H, 1 H), 5.04 ( br t, NH, 1 H), 7.03 (br m, NH coinci-
dent with bpy H, 2 H), 7.60–7.71 (m, 3 H), 8.15 (br s, 2 H),
8.40–8.46 (br m, 3 H), 8.68–8.72 (m, 2 H), 8.87–9.00 (m, 3 H),
9.07 (d, 1 H) and 9.27 (d, 1 H). 13C NMR (dmso-d6): major
isomer, δ 15.80, 18.06 (both CH(CH3)2), 30.54 (CH(CH3)2),
61.34 (α-C), 124.67, 124.81, 125.52, 125.96, 128.48, 128.96,
129.42, 129.79, 142.08 (2C), 142.72, 143.23, 149.14, 150.73,
152.94, 153.55, 156.44, 156.67, 156.94, 157.70 and 180.54
(COO). Calc. for [C25H26Cl2CoN5O10: C, 43.75; H, 3.82; N,
10.20. Found: C, 43.44; H, 3.90; N, 9.86%. UV-VIS: 483 nm (99
(c) [Co(cpg)(2,2Ј-bpy)2][ClO4]2ؒH2O. 1H NMR spectrum
following photolysis (DCl): free cpg, δ 0.40–0.85, 1.14 and 3.13
(peaks assigned in section 1(b)); cyclopropanecarbaldehyde,
δ 1.38 (br m, cyclopropyl CH2, 4 H), 1.95 (br m, 1 H, cyclo-
propyl CH) and 8.79 (d, 1 H, J = 7.4 Hz); free bipyridine, δ 7.98
M
Ϫ1 cmϪ1). MS (FAB): m/z 586 ({[Co(val)(2,2Ј-bpy)2]ClO4}ϩ).
(d) [Co(cpg)(2,2Ј-bpy)2][ClO4]2ؒH2O. Both diastereoisomers
(m, 2 H), 8.52 (m, 4 H) and 8.91 (m, 4 H); [Co(2,2Ј-bpy)3]3ϩ
,
were found in the crude product. Again, a COSY spectrum
helped in the interpretation of the 1H NMR spectra. 1H NMR
(dmso-d6): major diastereoisomer, δ 0.40–0.70 (br m, cyclo-
propyl CH2s, 4 H), 0.75–0.85 (br m, cyclopropyl CH, 1 H),
3.01–3.09 (m, α-H, 1 H), 6.00–6.06 (br m, NH, 1 H), 7.08
(d,1H), 7.13–7.19 (br m, NH, 1 H), 7.59–7.67 (m, 3 H), 8.11–
8.21 (m, 2 H), 8.36–8.43 (m, 2 H), 8.49 (d, 1 H), 8.63–8.74 (m, 2
δ 7.51 (d, 6 H), 7.83 (t, 6 H), 8.58 (t, 6 H) and 8.88 (d, 6 H). The
D2O photolysate was extracted with CDCl3. 1H NMR (CDCl3):
δ 1.08 (complex peak, 4 H), 1.84 (br m, 1 H) and 8.91 (d, 1 H,
J = 5.8 Hz). 13C NMR (CDCl3): δ 7.29, 22.66 and 201.33.
(d) [Co(aib)(2,2Ј-bpy)2][ClO4]2ؒH2O. 1H NMR spectrum
following photolysis (DCl): free aib, δ 1.61 (s, 6 H); acetone,
δ 2.23 (s); free bipyridine, δ 7.98 (m, 2 H), 8.52 (m, 4 H) and 8.91
(m, 4 H); [Co(2,2Ј-bpy)3]3ϩ, δ 7.51 (d, 6 H), 7.83 (t, 6 H), 8.58
(t, 6 H) and 8.88 (d, 6 H).
1
H), 8.86–8.91 (m, 2 H), 8.96–9.04 (m, 2 H), 9.12 (d, 1 H); H
NMR (dmso-d6): minor diastereoisomer, δ 0.40–0.70 (br m,
cyclopropyl CH2), 1.18 (br m, cyclopropyl CH, 1 H), 2.62 (m,
α-H, 1 H), 6.38 (br m, NH, 1 H) and 6.62 (br m, NH, 1 H); most
of the bipyridine resonances were coincident with those of
the major isomer except δ 9.35 (d, 1 H, J = 5.3 Hz). 13C NMR
(dmso-d6): major isomer, δ 2.91, 4.93 (both cyclopropyl CH2s),
14.95 (cyclopropyl CH), 60.45 (α-C), 124.63, 124.72, 125.48,
126.06, 128.55, 128.92, 129.89, 129.92, 142.03 (2C), 142.65,
143.09, 148.72, 150.66, 152.57, 153.40, 156.33 (2C), 156.90,
156.26 and 180.12 (COO); minor isomer, a few peaks were dis-
tinguishable from those of the major diastereoisomer, δ 5.70
(cyclopropyl CH2), 14.12 (cyclopropyl CH) and 180.65 (COO).
Calc. for [C25H24Cl2CoN5O10ؒH2O: C, 42.75; H, 3.73; N,
9.97. Found: C, 42.71; H, 3.75; N, 10.01%. UV-VIS: 479 nm
(105 MϪ1 cmϪ1). MS (FAB): m/z 584 ({[Co(cpg)(2,2Ј-bpy)2]-
ClO4}ϩ).
4. Theoretical procedures
Ab initio calculations were performed with GAUSSIAN94.12
Reactant and transition state structures were optimised at the
B3LYP level of theory using the 6-31G* basis set. Vibrational
frequencies were also calculated with these methods, and scaled
with appropriate factors (1.0013 for the vibrational partition
function and 0.9806 for the zero point energy).13 The partition
functions and, subsequently, the rate constants were evaluated
at three different temperatures, and numerical differentiation of
ln(k) with respect to TϪ1 yielded the Arrhenius parameters, A
and Eact. All vibrational modes were treated as harmonic oscil-
lators for the evaluation of the vibrational partition functions.
Errors associated with this approximation will be small as the
vibrational partition functions for the transition state and the
reactant, Q‡vib and Qvib, are treated in the same manner.14
(e) [Co(aib)(2,2Ј-bpy)2][ClO4]2ؒH2O. 1H NMR (dmso-d6):
δ 1.01, 1.35 (both s, CH3, total 6 H), 6.10 (d, 1 H, NH), 6.61 (d,
1 H, NH), 7.03 (d, 1 H), 7.63 (t, 1 H), 7.70 (t, 1 H), 7.86 (d, 1 H),
8.16 (t, 1 H), 8.24 (t, 1 H), 8.39–8.45 (m, 3 H), 8.72 (m, 2 H),
8.89 (m, 2 H), 9.00 (d, 1 H), 9.08 (d, 1 H) and 9.52 (d, 1 H). 13
C
Results
Ab initio calculations
NMR (dmso-d6): δ 27.50, 28.38 (both CH3), 58.75 (α-C),
124.73, 124.94, 125.59, 126.31, 128.52 (2C), 128.91, 129.54,
142.07, 142.13, 143.01, 143.33, 149.26, 150.82, 152.42, 153.59,
Although the calculations in this study were initiated before
J. Chem. Soc., Dalton Trans., 1999, 3565–3571
3567

View Article Online
Table 1 Arrhenius parameters for the rearrangement of the unproton-
ated and protonated aminocyclopropylmethyl (acp) radicals
log A
Eact/kJ molϪ1
k/sϪ1 (298 K)
acp ring opening
acp ring closing
13.1
12.4
13.0
12.2
39.6
49.6
21.9
29.0
1 × 106
5 × 103
2 × 109
1 × 107
Hacpϩ ring opening
Hacpϩ ring closing
density on C2 implies that the radical has a significant inter-
action with the cyclopropyl ring.
The Arrhenius parameters predicted for the rearrangement
reactions of the amine substituted radical (acp) and its proton-
ated form (Hacpϩ) are shown in Table 1. The inclusion of an
unprotonated amine substituent at the radical centre is pre-
dicted to decrease the rate of the ring opening reaction from
that seen for the unsubstituted case. The ring-closing reverse
reaction is predicted to have a rate constant two hundred
times smaller than that for the ring-opening reaction. The rate
constants for the ring-opening and -closing reactions of the
protonated forms are both significantly larger than for the
unprotonated species.
Synthesis and characterisation
R,S-Cyclopropylglycine was prepared from lithium 5-bromo-2-
oxopentanoate via a template synthesis on cobalt(), extending
the work of Lawson et al.9 In that work a cobalt() complex of
cyclopropylglycinate was prepared via the reduction of a chel-
ated 2-cyclopropyl-2-iminoethanoate ligand. We employed
more vigorous conditions to reduce the cobalt centre to the ϩ2
oxidation state at the same time, thereby liberating the amino
acid from the metal centre. The cyclopropylglycine was then
purified by ion-exchange chromatography in order to avoid con-
tamination from NH4ϩ and cobalt() species, and recrystallised
to give a microanalytically pure product.
It was important to optimise the synthetic route to the
[Co(aa)(2,2Ј-bpy)2]2ϩ (aa = α-amino acidate) complexes, given
the small amount of cyclopropylglycine (Hcpg) which we
were able to isolate, and its comparatively laborious prepar-
ation. We employed the valinato (val) complex as a model
and found the method of Tatehata10 superior to several
alternatives which appear in the literature.17 Tatehata pre-
pared a series of [Co(S-aa)(phen)2]2ϩ complexes (phen = 1,10-
phenanthroline), from [CoCl2(phen)2]ϩ in basic methanol, and
separated the reaction products by ion-exchange chrom-
atography. The yields following this procedure, substituting
2,2Ј-bpy for phen, were greater than those obtained when
using [Co(CO3)(2,2Ј-bpy)2]ϩ as a precursor, or via the PbO2
oxidation of an aqueous solution containing CoII, an amino
acid, and bipyridine. The formation of side-products,
notably [Co(aa)2(2,2Ј-bpy)]ϩ and [Co(2,2Ј-bpy)3]3ϩ, persisted
however.
Fig. 1 The calculated geometries for the reactants, transition states
and calculated products of the unprotonated (A) and protonated (B)
radicals; bond lengths in Å.
publication of the paper by Smith et al,7 the recommendations
in that report subsequently shaped our approach to the the-
oretical determination of a rate constant for ring opening of
the aminocyclopropylmethyl radical. The calculated geom-
etries for the reactants, products, and transition states of the
protonated and unprotonated radicals, along with selected
bond lengths, are displayed in Fig. 1. The energies, zero-
point corrections, spin distributions, and atomic co-ordinates
for all the radicals and the transition states are given in SUP
57632.
The unprotonated radical is predicted to have C1 symmetry
and the geometry around the spin-bearing carbon is marginally
non-planar (sum of bond angles = 351.9Њ, RMS deviation from
plane = 0.094 Å), like the aminomethyl radical.15 The shortness
of the C4–N1 bond (1.409 Å) and the calculated spin density on
the nitrogen centre are indicative of significant delocalisation of
the unpaired electron. This is a well documented phenomenon
for α-aminoalkyl radicals.16
The atoms which are part of, or attached to, the incipient
double bond are predicted to be close to planar in the calcu-
lated geometry of the transition state for the ring opening of
the aminocyclopropylmethyl radical (RMS deviation from
plane for C1,C3, H3, C4, H4, N1 = 0.042 Å, cf. 0.028 Å for the
product). The constant length of the C4–N1 bond throughout
the reaction is consistent with continued delocalisation of the
nitrogen lone pair.
The geometry of the protonated aminocyclopropylmethyl
radical is broadly similar to that of its unprotonated analogue.
The relative conformation of the amine group remains
unchanged, with a hydrogen atom being present in the position
the nitrogen lone pair occupies in the unprotonated molecule.
Protonation of the lone pair appears to mitigate its ability to
overlap with the radical centre, hence the longer C4–N1 bond
(1.491 Å), and the absence of spin density on N1. The radical
centre, C4, is marginally non-planar (bond angle sum = 354.6Њ,
RMS deviation from plane = 0.076 Å). The location of spin
Complexes [Co(aa)(2,2Ј-bpy)2]2ϩ of chiral amino acids may
exist as two diastereoisomers. A preference for one diastereo-
isomer was noted in the preparation of the alaninato (3:1),
valinato (9:1) and cyclopropylglycinato (5:2) complexes. In the
case of the [Co(val)(2,2Ј-bpy)2]2ϩ complex it was possible to
purify one of the diastereoisomers by fractional crystallisation,
however both isomers persisted to some degree for the [Co(ala)-
(2,2Ј-bpy)2]2ϩ and [Co(cpg)(2,2Ј-bpy)2]2ϩ complexes. Alterna-
tive separation methods were not explored as, for the purposes
of our photolysis experiments, the presence of two diastereo-
isomers was considered to be of little consequence. The acquisi-
tion of COSY spectra proved useful for the assignment of the
1H NMR spectra, especially in the cases where both diastereo-
isomers were present.
A similar degree of stereoselectivity in the formation of
[Co(S-aa)(phen)2]2ϩ complexes was observed by Tatehata10 who
3568
J. Chem. Soc., Dalton Trans., 1999, 3565–3571

View Article Online
postulated that the energy differences between the diastereo-
isomers was a result of repulsive interactions between the
α-H of the diimine ligand and the α-substituent of the amino
acid. He concluded that the major diastereoisomer had the
Λ-S configuration after analysis of both CD and H NMR
spectra.
Anisotropic magnetic fields in the vicinity of the 2,2Ј-bpy
ligands leads to the observation of significant differences in the
1H NMR chemical shifts for the two diastereoisomers of the
alaninato, valinato, and cyclopropylglycinato complexes. These
differences were consistent with the ∆-R/Λ-S configuration also
being preferred for the complexes we have studied.
Discussion
Ab initio calculations
The geometries, energies, and vibrational frequencies of a react-
ing species and of the transition state along the reaction path-
way are the fundamental properties required to evaluate a rate
constant with transition state theory. In the present study these
values were assessed at the B3LYP/6-31G* level of theory
which has been shown to produce accurate values for a number
of related systems.7
1
Although the B3LYP technique has gained reputation for
underestimating reaction barriers,19 the reverse appears to be
true for cyclopropylmethyl radicals. For example, in the case of
the cyclopropylmethyl radical, the B3LYP/6-31G* level of
theory actually overestimates the reaction barrier by 3.8 kJ
molϪ1. Moreover, for the key reaction in the present study, the
rearrangement of the protonated aminocyclopropylmethyl
radical, a 5 kJ molϪ1 increase in the reaction barrier would
induce a mere sixfold reduction in the rate constant (298 K).
The differences between the rates calculated for the reactions
of the protonated and unprotonated radicals are presumably
related to the degree to which the unpaired electron is delocal-
ised towards or away from the nitrogen substituent, and there-
fore the degree to which it can interact with the cyclopropyl
ring. The equilibria for both the unprotonated and protonated
cases are predicted to lie towards the ring opened forms. We
consider that the protonated case is likely better to model the
cobalt bound system since the absence of a lone pair on the
nitrogen atom drastically reduces the degree to which the spin
density can be delocalised onto the nitrogen atom.
All calculations were based on the presumption that the
reacting species are in the gas phase. This means that favourable
transition-state polarisation effects may be overlooked by the
theoretical procedure, as no account is made for solvation
effects. Experimentally, ring opening of an ester-substituted
cyclopropylmethyl radical was found to accelerate with increas-
ing solvent polarity,20 and it rearranges faster than its unsubsti-
tuted parent. This contrasts with theoretical predictions at the
B3LYP/6-31G* level which estimate a 30-fold drop in the rate
constant upon ester substitution.8c A possible contribution
from an iminium resonance form in the ring opening of the
unprotonated radical may lead to the calculated rate constants
being less reliable than for other cases.
Steady state photochemistry
Steady state photolysis experiments were performed on D2O
solutions of a series of [Co(aa)(2,2Ј-bpy)2]2ϩ complexes and
1H NMR spectroscopy was used to identify the photolysis
products. Acetaldehyde, 2-methylpropanal, and acetone were
identified as photolysis products from the alaninato, valinato,
and 2-aminoisobutyrato complexes respectively. The hydrated
forms of the two aldehydes were also present, as would be ex-
pected in aqueous solution. Free amino acid was always present
in solution following photolysis, and we were also able to
identify [Co(2,2Ј-bpy)3]3ϩ and free 2,2Ј-bipyridine. The relative
concentrations of the products were determined by reference to
an added internal standard (TMPS), and the identities of the
products were confirmed by spiking the solution with authentic
samples.
Addition of DCl to the neutral photolysates resulted in large
increases in the integrals of the free bipyridine and free amino
1
acid signals in the H NMR spectra. It has been noted previ-
ously3 that a significant amount of CoII is produced during
irradiation of [Co(aa)(2,2Ј-bpy)2]2ϩ complexes. The interaction
of this paramagnetic cation with the free bipyridine and free
1
amino acid is likely to reduce the intensity of their H NMR
peaks. However, this association, and the accompanying reduc-
tion of the NMR integrals, will be lessened in the presence of
acid, accounting for the above observation.
1
Measurement of H NMR signal integrals in the acidified
samples against an internal standard allowed us to account
for all of the starting material following irradiation of the
[Co(aib)(2,2Ј-bpy)2]2ϩ complex. We found that, for a 50 mM
solution, the ratio of free aib:acetone in the photolysate was
approximately 3:1. The ratio increases with increasing concen-
tration of starting material. Free bipyridine and [Co(2,2Ј-
bpy)3]3ϩ were usually present in approximately equal amounts,
although the concentration [Co(2,2Ј-bpy)3]3ϩ increased grad-
ually over time.
Our best prediction for the rate constant for ring opening of
the protonated aminocyclopropylmethyl radical is in the range
of 108–109 sϪ1. It is this reaction that we expect to be the closer
model for our cobalt containing system.
Provided the structural, steric, and electronic differences
will not have major implications for the rate constant for ring
opening, the best estimate for the rate of rearrangement of the
co-ordinated aminocyclopropylmethyl radical has an upper
bound at 109 sϪ1 at room temperature and a reasonable lower
bound could be set at 107 sϪ1. Furthermore, the rate constant
for the reverse reaction should be around two orders of magni-
tude smaller than that of the forward reaction.
Unfortunately, full material balances could not be achieved
for the other complexes. We ascribe this to a combination of (i)
the volatility of the aldehydes, (ii) persistent interactions of the
free amino acids with the paramagnetic cobalt() cation, (iii)
poor signal to noise ratio in the NMR spectra at high concen-
trations of DCl and (iv) possible aldol polymerisation of the
aldehydes.
Given that formation of the metallacycle from the radical
and cobalt() is proposed to occur with a rate constant (kc) of
103 sϪ1, it should be several orders of magnitude too slow to
compete with the rearrangement reaction. Thus, the currently
accepted mechanism would predict that exclusively ring-opened
products will be observed following the photolysis of [Co(cpg)-
1
The results from H NMR analysis of irradiated solutions
of [Co(cpg)(2,2Ј-bpy)2]2ϩ were entirely analogous to those
obtained for the other amino acid complexes. The products
identified were free amino acid, [Co(2,2Ј-bpy)3]3ϩ, free 2,2Ј-
bipyridine, and cyclopropanecarbaldehyde. A COSY spectrum
was obtained in order to confirm the identification of the latter.
Also, the aldehyde was extracted into CDCl3 so that the 1H and
13C NMR spectra could be compared to those previously pub-
lished for this compound.18 The coupling pattern shown by the
COSY spectrum was fully consistent with that expected with
the methine proton, displaying a correlation with the cyclo-
propyl ring protons and the aldehyde proton. The spectra run
in chloroform were nearly identical to those published for
authentic samples.
(bpy)2]2ϩ
.
Steady state photolysis products
The use of 1H NMR spectroscopy has allowed us to show
that carbonyl compounds, derived from the C–N fragment
of the amino acid, are produced in the photolysis of [Co(aa)-
(2,2Ј-bpy)2]2ϩ complexes. It has been noted by Povlovskii
and Poznyak3 that substituents on either the α-carbon or the
J. Chem. Soc., Dalton Trans., 1999, 3565–3571
3569

View Article Online
nitrogen atom will destabilise the cobalt()–alkyl photoprod-
uct, although this is the first report detailing the nature of the
decomposition products. Similar behaviour has been reported
for the complexes [Co(edta)]Ϫ 21 (edta = ethylenedinitrilotetra-
acetate), [Co(gly)3] and [Co(ala)3],22 for which aldehydes were
detected as photooxidation products of the aminocarboxylato
ligands.
Poznyak and co-workers have characterised, by electronic
spectroscopy, the primary photoproducts of the [Co(val)(2,2Ј-
bpy)2]2ϩ,3 [Co(val)(en)2]2ϩ 2a and [Co(ala)(trien)]2ϩ 2a complexes,
following UV irradiation at low temperature (trien = triethylene-
tetramine). The UV-Vis spectrum of the thermally unstable
photoproduct of [Co(val)(2,2Ј-bpy)2]2ϩ closely resembled that
of the crystallographically characterised [Co(CH2NH2)(2,2Ј-
bpy)2]2ϩ complex ion and, on this basis, the authors concluded
that photolysis of the valinato complex also yielded a Co–C–N
metallacycle.
aminocyclopropyl radical co-ordinated to cobalt() could
reasonably be predicted to have a rate constant for ring opening
in the range 107–109 Ϫ1, with an equilibrium constant of
s
approximately 200, favouring the ring opened form.
The cyclopropylmethyl radical clock, and many substituted
analogues, have found wide application in the investigation of
reaction mechanisms involving radical intermediates.24 It is
generally accepted that the appearance of ring-opened products
is direct evidence that a discrete free radical is formed on the
α-carbon, with the ratio of ring-opened to ring-intact products
providing an indication of the relative rates of the intra-
molecular rearrangement and the reaction of interest. However,
if only products containing a cyclopropane ring are isolated,
then several possibilities exist. The first is that there is no radical
intermediate involved in the reaction. The second is that there is
a radical intermediate, but that its lifetime is so short that the
ring-opening reaction cannot compete with the reaction that is
being probed. The final possibility is that reaction of the ring-
opened radical is so slow that, in spite of the equilibrium con-
stant favouring the ring-opened radical, all the products result
from reaction of the ring-closed form. It would be remarkable
if reaction of a reactive species like the ring-opened radical
could not compete with a reaction which is proposed to proceed
with a rate constant of 103 sϪ1. This is particularly so when the
equilibrium gives the ring-opened radical a significant concen-
tration advantage (by a factor of 200) over the ring-closed
radical. We therefore feel able to discount this third possibility.
According to the published mechanism for formation of the
metallacyclic complexes, photoinduced homolysis of the Co–O
bond will lead to decarboxylation and the generation of a
cobalt()-bound aminocyclopropylmethyl radical. This radical
can act as a radical clock, whereby ring opening will compete
with recombination with the metal centre. The relative rates of
these two competing processes, kc versus kr, will determine the
ratio of ring-opened to ring-intact products.
Given that irradiation of the complexes of α-substituted
amino acids does indeed lead to the formation of metallacyclic
complexes, the carbonyl compounds which we observe are
likely the result of the thermal decomposition of these
metallacycles. A possible pathway is shown in Scheme 2, where
The cyclopropane ring was found to survive the photolysis
of [Co(cpg)(2,2Ј-bpy)2]2ϩ and, as expected on the basis of our
investigation of analogous complexes, it was found as cyclo-
propanecarbaldehyde. No ring-opened products were detected
by 1H NMR spectroscopy.
The appearance of an intact cyclopropane ring in the
photolysis products demonstrates that, if the co-ordinated rad-
ical is formed, reaction to give a cobalt()-alkyl complex dom-
inates the potential ring-opening reaction. Therefore, provided
any further reaction of the ring-opened radical species is not
very slow, the rate of this Co–C bond formation step (kc) must
be significantly faster than 107 sϪ1. This means that the co-
ordinated aminocyclopropylmethyl radical, if formed at all, has
a lifetime much shorter than 10Ϫ7 s. This lifetime is four orders
of magnitude shorter than that inferred by the assignment of
the rate determining step in the original kinetic study.
Scheme 2
the Co–C–N metallacycles 2a–2e, which can also be considered
as π-bound cobalt()–methyliminium complexes,4 lose the C–N
fragment as an iminium ion, leaving the cobalt centre in the ϩ1
oxidation state. Iminium ions are prone to hydrolysis and, in
D2O, will readily convert into the corresponding carbonyl
compound.
The cobalt() species could then undergo redox reactions,
either with starting material or with water, eventually to give the
observed cobalt() compounds. Redox reactions with starting
material would also account for the presence of free amino acid
following photolysis, as the resulting cobalt() complexes
would be substitutionally labile and could catalyse ligand
exchange reactions of further molecules of starting material.
The concentration dependence of the aib : acetone ratio is con-
sistent with these proposals, since the rates of the second order
redox reactions that could lead to production of the free amino
acid would have a greater dependence on the concentration of
the complex ions than those of reactions leading to the carb-
onyl compounds, which would be expected to be first order in
complex concentration. Amino acid products would therefore
be favoured at higher concentrations.
Conclusion
We have shown that the lifetime of any cobalt()-bound ami-
noalkyl radical that may be formed in the UV photolysis of
complexes of the type [Co(aa)(2,2Ј-bpy)2]2ϩ is less than 10Ϫ7 s.
On the basis of our results and those of the flash photolysis
study,5 either the currently accepted mechanism for the photo-
decarboxylation of cobalt()-amino acid complexes is incor-
rect or, at least, the rate determining step has been incorrectly
assigned. If the latter is true then one of the other steps, Co–O
bond homolysis or decarboxylation,25 must be significantly
slower than has been observed previously. We are continuing
our investigations of these reactions with the aim of obtaining
information which may give us further insights into the mech-
anism of metallacycle formation. In particular, we will have to
consider the possibilities that the reaction does not proceed by a
radical pathway, or that a radical intermediate is involved, but
reacts rapidly (e.g. by β scission of the radical to generate a
The use of a cyclopropylglycinato complex as a radical clock
The cyclopropylmethyl radical is a well calibrated radical clock
which ring-opens to the 1-butenyl radical at 1.0 × 108 sϪ1 (298
K),23 and the foregoing discussion led to the conclusion that the
3570
J. Chem. Soc., Dalton Trans., 1999, 3565–3571

View Article Online
10 A. Tatehata, Inorg. Chem., 1982, F21, 2497.
11 M. P. Hancock, J. Josephsen and C. E. Schäffer, Acta Chem. Scand.,
1976, 30, 79.
cobalt() species and iminium ion) to give other intermediates
which can then give rise to metallacycles and their eventual
decomposition products.
12 M. J. Frisch, G. W. Trucks, H. B. Schlegel, P. M. W. Gill, B. G.
Johnson, M. A. Robb, J. R. Cheeseman, T. Keith, G. R. Petersson,
J. A. Montgomery, K. Raghvachari, M. A. Al-Laham, V. G.
Zakrzewski, J. V. Ortiz, J. V. Foresman, J. Cioslowski, B. B. Stefanov,
A. Nanayakkara, M. Challacombe, C. Y. Peng, P. Y. Alaya,
W. Chen, M. W. Wong, J. L. Andres, E. S. Replogle, R. Gomperts,
R. L. Martin, D. J. Fox, J. S. Binkley, D. J. Defrees, J. Baker,
J. P. Stewart, M. Head-Gordon, C. Gonzales and J. A. Pople,
GAUSSIAN 94, revision B.1, Gaussian Inc., Pittsburgh, 1995.
13 A. P. Scott and L. Radom, J. Phys. Chem., 1996, 100, 16502.
14 J. P. A. Heuts, R. G. Gilbert and L. Radom, Macromolecules, 1995,
28, 8771.
Acknowledgements
We acknowledge Dr Mark Rees for his help with the ab initio
calculations. The New Zealand Lottery Science Grants Board
provided funding towards this project, and the University of
Canterbury provided a Doctoral Scholarship to S. G. T.
References
15 D. A. Armstrong, A. Rauk and D. Yu, J. Am. Chem. Soc., 1993,
115, 666.
1 For reviews of early work see: (a) V. I. Pavlovskii and A. L. Poznyak,
Angew. Chem., Int. Ed. Engl., 1988, 27, 789; (b) J. Sykora and
J. Sima, Coord. Chem. Rev., 1990, 107, 1; (c) V. I. Pavlovskii and
A. L. Poznyak, Z. Chem., 1989, 29, 6.
16 S. F. Nelson, in Free Radicals (Vol. II), ed. J. Kochi, Wiley-
Interscience, New York, 1973; Y. L. Chow, W. C. Danen, F. S.
Nelsen and D. H. Rosenblatt, Chem. Rev., 1978, 78, 243.
17 V. I. Pavlovskii and A. L. Poznyak, Russ J. Inorg. Chem., 1980, 25,
422; D. R. Boone, R. N. Kust and M. B. Monroe, Polyhedron, 1989,
3, 49; N. Maki, Bull. Chem. Soc. Jpn., 1969, 42, 2275.
18 C. J Pouchert, The Aldrich Library of NMR Spectra, 2nd edn,
Aldrich Chemical Co. Inc., Milwaukee, WI, 1983; A. K. Khusid, G.
V. Kryshtal, V. A. Dombrovsky, V. F. Kucherov, L. A. Yanovskaya,
V. I. Kadentsev and O. S Chizhov, Tetrahedron, 1977, 33, 77.
19 R. L. Bell, D. L. Taveras, T. N. Truong and J. Simons, Int. J.
Quantum. Chem., 1997, 63, 861; M. N. Glukhovtsev, R. D. Bach,
A. Pross and L. Radom, Chem. Phys. Lett., 1996, 260, 550.
20 M. Newcomb, J. H. Horner and C. J. Emanuel, J. Am. Chem. Soc.,
1997, 119, 7147.
21 A. L. Poznyak and V. E. Stelmashok, Inorg. Chim. Acta, 1984, 83,
L59.
22 V. Balzani, V. Carassiti, L. Moggi and N. Sabbatini, Inorg. Chem.,
1965, 4, 1247.
23 M. Newcomb and A. G. Glenn, J. Am. Chem. Soc., 1989, 111, 275;
A. Effio, D. Griller, K. U. Ingold, A. L. J. Beckwith and A. K.
Serelis, J. Am. Chem. Soc., 1980, 102, 1736.
2 For more recent examples see: (a) V. V. Pansevich and A. L.
Poznyak, Russ. J. Inorg. Chem., 1991, 36, 1425; (b) A. L. Poznyak
and V. E. Stel’mashok, Metallorg. Khim., 1992, 5, 611; (c) A. L.
Poznyak and L. V. Stopolyanskaya, Russ. J. Inorg. Chem., 1995,
40, 1080; (d) T. Ama, H. Kawaguchi, K. Okamoto, T. Yasui,
T. Yonemura and M. Yoshida, Bull. Chem. Soc. Jpn., 1995, 68, 874;
(e) T. Yonemura, H. Kawaguchi, K. Shibuya, K. Murakami,
T. Yasui, K. Okamoto and J. Hidaki, Bull. Chem. Soc. Jpn., 1995,
68, 2287; ( f ) M. Kojima, F. M. D. Akhter, K. Nakajima and Y.
Yoshikawa, Bull. Chem. Soc. Jpn., 1996, 69, 2889; (g) D. M. Tonei,
L. J. Baker, P. J. Brothers, G. R. Clark and D. C. Ware, Chem.
Commun., 1998, 2953; (h) T. Yonemura, K. Shibuya, T. Ama, H.
Kawaguchi, K. Okamoto, J. Hidaka and T. Yasui, Inorg. Chem.,
1999, 38, 3244.
3 V. I. Pavlovskii and A. L. Poznyak, Z. Anorg. Allg. Chem., 1982, 485,
225.
4 E. B. Chuklanova, V. I. Pavlovskii, T. V. Polynova, M. A. Porai-
Koshits and A. L. Poznyak, Monatsh. Chem., 1982, 113.
5 E. Natarajan and P. Natarajan, Inorg. Chem., 1992, 31, 1215.
6 D. Meyerstein and H. Schwarz, J. Chem. Soc., Faraday Trans. 1,
1988, 2933; H. Cohen, D. Meyerstein and Y. Sorek, J. Chem. Soc.,
Faraday Trans. 1, 1989, 1169; J. F. Endicott and T. S. Roche, J. Am.
Chem. Soc., 1972, 94, 8622.
7 D. M. Smith, A. Nicolaides, B. T. Golding and L. Radom, J. Am.
Chem. Soc., 1998, 120, 10223.
8 (a) F. N. Martinez, H. B. Schlegel and M. Newcomb, J. Org. Chem.,
1996, 61, 8547; (b) F. N. Martinez, H. B. Schlegel and M. Newcomb,
J. Org. Chem., 1998, 63, 3618; (c) J. H. Horner, N. Tanaka and
M. Newcomb, J. Am. Chem. Soc., 1998, 120, 10379.
24 For general reviews see: D. Griller and K. U. Ingold, Acc. Chem.
Res., 1980, 13, 317; C. J. Suckling, Angew. Chem., Int. Ed. Engl.,
1988, 27, 537; M. Newcomb, Tetrahedron, 1993, 49, 1151; D. C.
Nonhebel, Chem. Soc. Rev., 1993, 347. For recent examples see: A.
L. J. Beckwith and V. W. Bowry, J. Am. Chem. Soc., 1994, 116, 2710;
M. He and P. Dowd, J. Am. Chem. Soc., 1996, 118, 711; M. A.
Filipkowski, C. Ha, J. H. Horner, M. Newcomb and S. U. Park,
J. Am. Chem. Soc., 1995, 117, 3674.
25 D. Budac and P. Wan, J. Photochem. Photobiol. A, 1992, 67, 135.
9 P. J. Lawson, M. J. McCarthy and A. M. Sargeson, J. Am. Chem.
Soc., 1982, 104, 1734.
Paper 9/04451E
J. Chem. Soc., Dalton Trans., 1999, 3565–3571
3571