The photodecarboxylation of [N-N-bis-4-pyridylmethyOamino acidato]phenanthrolinecobalt(ni) complexes: - Formation and decomposition of metallacyclic species

Article abstract of DOI:10.1039/b003095n

Photolysis of [N,N-bis(2-pyridylmethyl)amino acidato]phenanthrolinecobalt(in) complexes (amino acidato = glycinato (dpg) or alaninato (dpa) or 2-cyclopropylglycinato (dpc)) with UV light led to elimination of carbon dioxide from the amino acidato chelate and the formation of cobalt(ni) complexes which contain a Co-C-N metallacycle. These photolysis products were characterised by NMR and UV-vis spectroscopy and, in the case of dpg, by X-ray crystallography. The eventual decomposition reactions of the photolysis products were monitored by NMR spectroscopy in both D2O and dilute DC1. The decomposition products included bis(2-pyridylmethyl)amine (bpa) and an aldehyde derived from the carbon atom of the metallacycle and its alkyl substituents. A u-peroxo dinuclear cobalt(in) complex, [(bpa)(phen)(Co(O₂)Co(phen)(bpa)]4⁺, was formed from the decomposition products and its structure determined by X-ray crystallography. Experiments with a cyclopropyl derivative demonstrate that any intermediates with a radical centre on what was the u-carbon of the amino acid must be very short-lived. The Royal Society of Chemistry 2000.

Full text of DOI:10.1039/b003095n

View Article Online / Journal Homepage / Table of Contents for this issue
The photodecarboxylation of [N,N-bis(2-pyridylmethyl)amino
acidato]phenanthrolinecobalt(III) complexes: – formation and
decomposition of metallacyclic species†
Richard M. Hartshorn* and Shane G. Telfer
Department of Chemistry, University of Canterbury, Private Bag 4800, Christchurch,
New Zealand. E-mail: r.hartshorn@chem.canterbury.ac.nz
Received 17th April 2000, Accepted 28th June 2000
Published on the Web 27th July 2000
Photolysis of [N,N-bis(2-pyridylmethyl)amino acidato]phenanthrolinecobalt() complexes (amino
acidato = glycinato (dpg) or alaninato (dpa) or 2-cyclopropylglycinato (dpc)) with UV light led to elimination of
carbon dioxide from the amino acidato chelate and the formation of cobalt() complexes which contain a Co–C–N
metallacycle. These photolysis products were characterised by NMR and UV-vis spectroscopy and, in the case of
dpg, by X-ray crystallography. The eventual decomposition reactions of the photolysis products were monitored by
NMR spectroscopy in both D2O and dilute DCl. The decomposition products included bis(2-pyridylmethyl)amine
(bpa) and an aldehyde derived from the carbon atom of the metallacycle and its alkyl substituents. A µ-peroxo
dinuclear cobalt() complex, [(bpa)(phen)(Co(O2)Co(phen)(bpa)]4ϩ, was formed from the decomposition products
and its structure determined by X-ray crystallography. Experiments with a cyclopropyl derivative demonstrate
that any intermediates with a radical centre on what was the α-carbon of the amino acid must be very short-lived.
Introduction
The investigation of photochemical reactions of ligands co-
ordinated to transition metals is a branch of inorganic
chemistry which has not attracted a great deal of attention.1,2
Reactions of this kind have the potential to provide synthetic
avenues to interesting or otherwise inaccessible organic or
co-ordination compounds. The reagent (light) is cheap, readily
available and is compatible with most solvents and
temperatures.
An example of this kind of chemistry can be found in the
formation of a complex containing a three-membered Co–C–N
Scheme 1
metallacycle via irradiation of the bis(2,2Ј-bipyridine)-
glycinatocobalt() ion, [Co(gly)(2,2Ј-bpy)2]2ϩ 1, with UV light.3
ing a cobalt()-bound aminoacyloxy radical intermediate 2.
This complex is proposed to undergo a decarboxylation reac-
tion to give a species which contains a co-ordinated aminoalkyl
radical, 3. The alkyl radical is then thought to react with the
cobalt() centre, yielding the metallacyclic product 4.
The metallacyclic product, [Co(CH2NH2)(2,2Ј-bpy)2]2ϩ 2, is
reasonably stable and has been characterised by X-ray
crystallography.4
A range of similar reactions has been reported, but the stabil-
ity of the resulting cobalt–alkyl products varies consider-
ably.1a,c,3 In particular, photolysis of complexes where the amino
acid is substituted at the α position gives metallacycles which
have been characterised only by UV-vis spectroscopy in frozen
solutions. The decomposition products that are observed from
these photochemical reactions include carbonyl compounds in
which the substituents are derived from the α-substituents of
the original amino acid ligand.5 It has been noted, however, that
the stability of the metallacyclic products is enhanced when the
metallacycle is part of a polydentate ligand and when the
remaining co-ordination sites on the metal ion are occupied by
π-acidic ligands.1a
We have examined the photochemistry of a cyclopropyl-
glycinato complex 1 (R = cyclopropyl) as part of a mechanistic
study of this reaction.5 We found that the cyclopropane ring
was retained in the final product, cyclopropanecarbaldehyde.
This aldehyde was presumed to result from decomposition
of the metallacyclic compound 4. This presumption, together
with the results of a flash photolysis study of this system,6
means that the final step shown in Scheme 1 cannot be rate
determining.
It remained a possibility that the cyclopropanecarbaldehyde
was not formed via compound 4 (R = cyclopropyl), but rather
resulted from some other reaction pathway. The purpose of the
study described in this article was to attempt to stabilise
metallacyclic species derived from α-substituted amino acidate
ligands by incorporating the amino acid (and hence the
metallacycle) into a polydentate ligand framework. The sub-
sequent reactions could then be monitored to establish whether
the observed carbonyl compounds do indeed result from the
decomposition of the metallacycles.
A mechanism for the formation of the metallacyclic com-
pounds has been proposed1a (Scheme 1), wherein absorption of
UV light leads to homolysis of the cobalt–oxygen bond, afford-
† Electronic supplementary information (ESI) available: details of
decomposition experiments and crystallographic experimental. See
http://www.rsc.org/suppdata/dt/b0/b003095n/
DOI: 10.1039/b003095n
J. Chem. Soc., Dalton Trans., 2000, 2801–2808
This journal is © The Royal Society of Chemistry 2000
2801

View Article Online
J = 16.6 Hz, picolyl CH2), 7.94 (t, 2 H), 8.08 (d, 2 H), 8.51 (t, 2
Experimental
Materials
H) and 8.71 (d, 2 H). 13C NMR (D2O): δ 4.47, 5.50, 12.05,
54.90, 70.69 (cpg α-C), 127.15, 127.91, 141.98, 148.13, 154.02,
176.61 (CO2). MS (FAB): m/z 298 ([H2dpc]ϩ).
Reagent grade reagents and solvents were obtained from com-
mercial sources and used without further purification for all
syntheses unless stated. 2-Cyclopropylglycine was prepared by
a modification of a literature method.5,7 All ion-exchange resins
and NMR solvents were purchased from Aldrich Chem. Co.
Preparation of the [Co(dpx)(phen)]2؉ complexes
General method. Co(NO3)2ؒ6H2O (4.64 g, 16 mmol) and
NaNO2 (4.96 g, 70 mmol) were dissolved in a buffer solution of
glacial acetic acid (3.5 mL) and NaOH (1.2 g) in water (30 mL).
To this was added a solution of HdpxؒHCl (12 mmol), with
NaOH (1.44 g, 36 mmol), in water (10 mL). The solution was
aerated overnight. An orange-brown precipitate of [Co(dpx)-
(NO2)2] deposited, which was filtered off, washed with ice-cold
water, and dried in an oven (50–60 ЊC). These [Co(dpx)(NO2)2]
complexes were characterised by NMR and mass spectrometry
(see below).
[Co(dpx)(NO2)2] (2.5 mmol) was suspended in 0.2 M NaCl
solution (5 mL) with stirring, and a solution of phen (0.49 g, 2.5
mmol) in methanol (5 mL) was added. Stirring was continued
at rt for 3 h. Any remaining solid was removed by filtration and
the orange filtrate was loaded on to SP Sephadex C25 (Hϩ
form, 3 × 20 cm). An intense orange band eluted with 0.25 M
HCl, and the eluate was reduced in volume on a rotary evapor-
ator. Concentrated HClO4 and ethanol were added, and crystal-
line [Co(dpx)(phen)][ClO4]2 formed after standing overnight
in a refrigerator. Recrystallisation from dilute HClO4 gave
analytically pure products. The yield was generally around 30%.
Measurements
1H NMR spectra were recorded on a Varian Unity 300 MHz
spectrometer at 23 ЊC. For those recorded in dimethyl-d6 sulf-
oxide (DMSO-d6) the DMSO-d5 line (δ 2.50) was used as a
reference. Sodium 3-trimethylsilylpropanesulfonate (TMPS, δ 0,
singlet) was added as an internal reference for those recorded in
D2O. The signals are described as singlets (s), doublets (d), trip-
lets (t), quartets (q) or multiplets (AB, ABX or m), and broad
(br) where appropriate. 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-d5 peak (δ 39.5) or, in D2O, to added
1,4-dioxane (δ 67.4). A GBC-920 spectrophotometer was used
to record the UV-visible spectra in water and the data are
reported as λmax (εmax, L molϪ1 cmϪ1). The abbreviation sh
refers to a shoulder. Elemental analyses were performed by the
University of Otago Microanalytical Service. 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.
1
[Co(dpg)(NO2)2]. H NMR (DMSO-d6): δ 3.92 (s, 2 H, gly
CH2), 4.77 and 5.09 (AB, 4 H, picolyl CH2), 7.65 (m, 4 H), 8.15
(t, 2 H) and 8.58 (d, 2 H). 13C NMR (DMSO-d6): δ 66.46 (gly
CH2), 67.34 (picolyl CH2), 123.64, 125.61, 140.42, 151.05,
162.31, 176.92 (CO2). MS (FAB): m/z 315 ([Co(dpg)]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 of N,N-bis(2-pyridylmethyl)amino acid (Hdpx)
ligands
[Co(dpg)(phen)][ClO4]2 5. H NMR (DMSO-d6): δ 4.68 (s,
2 H, gly CH2), 5.32 and 6.19 (AB, 4 H, picolyl CH2, J = 17.5
Hz), 7.16 (d, 2 H, dpg), 7.31 (t, 2 H, dpg), 7.81 (d, 2 H dpg), 7.99
(t, 1 H, phen), 8.08 (t, 2 H, dpg), 8.38 and 8.57 (AB, 2 H, phen),
8.74 (m, 1 H, phen), 8.92 (d, 1 H, phen), 9.42 (d, 1 H, phen),
General method. The synthesis of this series of ligands was
based on a literature method,8 however the work-up procedure
differed from the original account. Following the extraction of
the aqueous reaction mixture with CH2Cl2 the aqueous portion
was acidified and loaded on to a column of Dowex AG50W-X2
ion exchange resin. Elution with 1.5 M HCl gave a yellow band
identified as the monoalkylated amino acid. Elution with 2.5 M
HCl gave the desired product as a pale yellow eluate which
yielded a pale yellow oil when taken to dryness on a rotary
evaporator. Precipitation of the ligand as the hydrochloride salt
was induced by the addition of acetone–PriOH (1:1), and the
off-white solid was filtered off, washed with PriOH and diethyl
ether, and air-dried.
1
9.62 (d, 1 H, phen) and 9.75 (d, 1 H, phen). H NMR (D2O):
δ 4.79 (s, partially obscured by HOD peak, gly CH2), 5.38 and
6.25 (AB, 4 H, J = 17.6, picolyl CH2), 7.20–7.27 (m, 4 H), 7.80
(d, 2 H), 7.95–8.06 (m, 3 H), 8.25 and 8.45 (AB, 2 H, J = 8.8
Hz), 8.79 (m, 1 H), 8.79 (d, 1 H), 9.30 (d, 1 H), 9.43 (d, 1 H) and
9.89 (d, 1 H). 13C NMR (DMSO-d6): δ 67.58 (picolyl CH2),
70.76 (gly CH2), 125.23 (dpg), 127.52 (dpg), 127.71, 128.07,
128.47, 129.22, 130.63, 131.44, 141.14, 141.55 (dpg), 141.98,
145.16, 146.45, 150.99 (dpg), 153.68, 155.11, 163.50 (dpg) and
177.10 (CO2). Calc. for [CoC26H22N5O2][ClO4]2: C, 44.98; H,
3.19; N, 10.09. Found: C, 44.67; H, 3.04; N, 10.09%. UV-vis:
480 nm (220); λmin 420 nm (71). MS (FAB): m/z 594 ([Co(dpg)-
(phen)(ClO4)]ϩ, 3) and 495 ([Co(dpg)(phen)]ϩ, 23%).
N,N-Bis(2-pyridylmethyl)glycine (HdpgؒHCl). 1H NMR
(D2O): δ 3.75 (s, 2 H, gly CH2), 4.51 (s, 4 H, picolyl CH2), 7.99
(t, 2 H), 8.03 (d, 2 H). 8.55 (t, 2 H) and 8.76 (d, 2 H). 13C NMR
(D2O): δ 55.94 (gly α-C), 56.45 (pyridyl CH2), 127.20, 127.86,
142.15, 148.13, 153.33, 175.30 (CO2) MS (FAB): m/z 258
([H2dpg]ϩ).
[Co(dpa)(NO2)2]. 1H NMR (DMSO-d6): δ 1.28 (d, 3 H, CH3,
J = 6.8 Hz), 3.57 (m, obscured by H2O peak), 4.74–4.92 and
5.10–5.16 (picolyl CH2 overlapping AB quartets, total 4 H),
7.60–7.80 (m, 4 H), 8.14 (br s, 2 H), 8.37 (d, 1 H) and 8.75 (d,
1 H). 13C NMR (DMSO-d6): δ 13.56 (CH3), 59.22, 66.70, 67.12,
122.63, 123.89, 125.62, 125.69, 140.47, 140.71, 151.11, 151.26,
162.17, 162.71, 178.05 (CO2). MS (FAB): m/z 329 ([Co(dpa)]ϩ).
N,N-Bis(2-pyridylmethyl)-L-alanine (HdpaؒHCl). -Alanine
was used in the synthesis. 1H NMR (D2O): δ 1.50 (d, 3 H, CH3,
J = 6.8 Hz), 3.83 (q, 1 H, ala α-H), 4.43 (s, 4 H, picolyl CH2),
7.93 (t, 2 H), 8.04 (d, 2 H), 8.50 (t, 2 H) and 8.69 (d, 2 H). 13
C
[Co(dpa)(phen)][ClO4]2ؒH2O 6. 1H NMR (DMSO-d6): δ 1.61
(d, 3 H, CH3, J = 6.8 Hz), 4.40 (m, 1 H, α-H), 5.32 (apparent t,
2 H, picolyl CH2), 6.03–6.11 (m, 2 H, dpa picolyl CH2), 7.13
(m, 2 H, dpa), 7.30 (d, 2 H, dpa), 7.72 (d, 1 H, dpa), 7.93–8.12
(m, 4 H, dpa pyridyl H (3 H), and phen (1 H)), 8.38 (d, 1 H,
phen), 8.57 (d, 1 H, phen), 8.74 (m, 1 H, phen), 8.92 (d, 1 H,
NMR (D2O): δ 14.24 (ala CH3), 54.01 (picolyl CH2), 60.57 (ala
α-C), 127.20, 127.96, 142.08, 148.18, 153.78, 177.31 (CO2). MS
(FAB): m/z 272 ([H2dpa]ϩ).
Cyclopropyl-N,N-bis(2-pyridylmethyl)glycine (HdpcؒHCl). 1H
NMR (D2O): δ 0.46–0.54 (br m, 3 H, cyclopropyl CH2),
0.62–0.68 (br m, 1 H, cyclopropyl CH2), 1.03–1.09 (br m, 1 H,
cyclopropyl CH), 2.92 (d, 1 H, J = 9.7, α-C), 4.57 (AB, 4 H,
1
phen), 9.43 (d, 1 H, phen) and 9.75 (m, 2 H, phen). H NMR
(D2O): δ 1.80 (d, 3 H, CH3, J = 7.8), 4.63 (q, partially obscured
2802
J. Chem. Soc., Dalton Trans., 2000, 2801–2808

View Article Online
by HOD peak), 5.33 and 6.10 (AB, 2 H, J = 16.9), 5.48 and 6.15
(AB, 2 H, J = 18.6 Hz), 7.15–7.28 (m, 3 H), 7.76 (d, 1 H), 7.87
(d, 1 H), 7.97–8.08 (m, 2 H), 8.28 (d, 1 H), 8.46 (d, 1 H), 8.70
(m, 1 H), 8.81 (d, 1 H), 9.32 (d, 1 H), 9.56 (d, 1 H) and 9.92 (d,
1 H). 13C NMR (DMSO-d6): δ 13.60, 59.03, 66.43, 71.19,
124.28, 125.52, 127.58 (2C), 127.76, 128.08, 128.53, 129.19,
130.64, 131.48, 141.14, 141.57, 142.01 (2C), 145.16, 146.50,
151.18 (2C), 153.54, 155.54, 163.16, 163.90 and 178.08 (CO2).
Calc. for [CoC27H24N5O2][ClO4]2ؒH2O: C, 44.66; H, 3.58; N,
9.64. Found: C, 44.56; H, 3.38; N, 9.66%. UV-vis: 481 (187),
262 (31000), 227 (53200) and 204 nm (58800). MS (FAB): m/z
608 ([Co(dpa)(phen)(ClO4)]ϩ, 10) and 509 ([Co(dpa)(phen)]ϩ,
34%).
mercury lamp equipped with a 254 nm transmission Pyrex filter
(Corning, 7–54). The samples were cooled by immersion in a
quartz ice–water bath.
Method 3: UV-vis monitoring of the photolysis reaction.
Aqueous solutions of the [Co(dpx)(phen)][ClO4]2 complexes
5–7 (concentrations in the range 1.0–1.5 mM) were photolysed
in quartz cuvettes. A 200 W high-pressure mercury lamp,
equipped with a Pyrex 254 nm transmission filter (Corning
7–54), was used as the light source. Two water-filled 1 cm quartz
cuvettes in the irradiation path served as IR filters.
Photolysis of [Co(dpg)(phen)]2؉ 5: production of [Co(dgm)-
(phen)]2؉
8
[Co(dpc)(NO2)2]. The crude product was rather impure, as
shown by its H NMR spectrum, but was not further purified
Method 1. The crude red-orange precipitate of complex 8 was
recrystallised from warm, dilute HClO4. Yield: ca. 15% at 70%
conversion. 1H NMR (D2O): δ 4.60 (s, Co–CH2, partially
obscured by HOD peak), 4.80 and 5.59 (AB, the peak at 4.80
was hidden by the HOD signal, J = 17.1 Hz, dgm picolyl CH2),
5.96 (d, 2 H, dgm), 7.41 (d, 2 H, dgm), 7.66 (m, 2 H, dgm),
8.04–8.09 (m, 1 H, phen), 8.37–8.46 (m, 3 H, phen), 8.83 (d,
1 H, phen), 9.06 (d, 1 H, phen), 10.20 (d, 1 H, phen) and 10.40
1
before being used in the reaction to give the [Co(dpc)(phen)]-
[ClO4]2 complex. A 13C NMR spectrum was not acquired due to
the small amount of complex which was isolated and the
1
difficulty in recovering it from DMSO. H NMR (DMSO-d6):
δ 0.37–0.42 (br m, 1 H), 0.60–0.95 (br m, 4 H), 3.02 (d, obscured
by H2O peak, α-H), 4.66 (d, 1 H), 4.97–5.35 (m, 3 H), 7.58–7.83
(m), 8.10–8.16 (m), 8.45 (d) and 8.63 (d); major unassigned
peaks 1.22 (s), 7.50 (m), 8.00 (m), 8.60 (d) and 8.82 (d). MS
(FAB): m/z 355 ([Co(dpc)]ϩ).
1
(d, 1 H, phen). H NMR (DMSO-d6): δ 4.51 (s, 2 H), 4.50 (s,
2 H, CH2), 4.73 and 5.75 (AB, 4 H, J = 16.6 Hz), 5.91 (d, 2 H),
6.71 (t, 2 H), 7.53 (d, 2 H), 7.74 (t, 2 H), 8.20 (m, 1 H, phen),
8.43–8.57 (m, 3 H, all phen), 9.00 (d, 1 H, phen), 9.22 (d, 1 H,
phen), 10.47 (d, 1 H, phen) and 10.51 (d, 1 H, phen). 13C NMR
(DMSO-d6): δ 46.17 (Co–CH2), 60.63, 123.13 (dgm), 124.54
(dgm), 126.96, 127.75, 127.99, 128.25, 130.52, 131.08, 139.17,
139.42 (dgm), 140.12, 145.14, 147.22, 148.95 (dgm), 154.71,
157.89 and 159.99 (dgm). Calc. for [CoC25H22N5][ClO4]2ؒ
H2Oؒ0.25NaClO4: C, 42.96; H, 3.43; N, 10.02. Found: C, 42.98;
H, 3.11; N, 9.74%. UV-vis: 460 (180), 296 (sh, 9000), 274
(25600), 254 (20800) and 222 nm (36000); λmin 422 nm
(142). MS (FAB): m/z 550 ([Co(dgm)(phen)(ClO4)]ϩ, 25) and
([Co(dgm)(phen)]ϩ, 22%).
[Co(dpc)(phen)][ClO4]2 7. 1H NMR (DMSO-d6): δ 0.60–0.65
(br m, 1 H, cyclopropyl CH2), 0.82–0.87 (br m, 1 H, cyclopropyl
CH2), 1.02–1.08 (br m, 2 H, cyclopropyl CH and CH2), 1.14–
1.19 (br m, 1 H, cyclopropyl CH2), 3.85 (d, 1 H, α-H, J = 8.8),
5.17 and 6.15 (AB, 2 H, J = 17.6), 5.80 and 6.19 (AB, 2 H,
J = 18.1 Hz), 7.13 (m, 2 H, dpc pyridyl), 7.30 (m, 2 H, dpc
pyridyl), 7.75 (d, 1 H, dpc picolyl), 7.99–8.10 (m, 4 H, dpc
pyridyl (1 H) and phen (3 H)), 8.39 (d, 1 H), 8.57 (d, 2 H), 8.76
(m, 1 H), 8.93 (d, 1 H), 9.44 (d, 1 H) and 9.78–9.82 (m, 2 H). 1H
NMR (D2O): δ 0.68–0.74 (br m, 1 H), 1.03–1.08 (br m, 1 H),
1.19–1.25 (br m, 3 H), 3.92 (d, 1 H, J = 9.3), 5.21 and 6.20 (AB,
2 H, J = 17.6), 6.03 and 6.24 (AB, 2 H, J = 18.6 Hz), 7.15–7.29
(m, 4 H), 7.72 (d, 1 H), 7.86 (d, 1 H), 7.96–8.07 (t, 3 H), 8.26 (d,
1 H), 8.44 (d, 1 H), 8.68 (m, 1 H), 8.80 (d, 1 H), 9.31 (d, 1 H),
9.58 (d, 1 H) and 9.95 (d, 1 H). 13C NMR (DMSO-d6): δ 3.42,
5.28, 11.44, 60.25, 67.38, 79.46, 124.31, 125.54, 127.54 (2C),
127.68, 128.04, 128.50, 129.19, 130.63, 131.43, 141.11, 141.39,
141.94 (2C), 145.16, 146.44, 150.85, 151.08, 153.50, 155.44,
163.41, 163.76 and 176.90 (CO2). UV-vis: 481 nm. MS (FAB):
m/z 634 ([Co(dpc)(phen)(ClO4)]ϩ) and 535 ([Co(dpc)(phen)]ϩ,
9%).
Method 2. Samples of [Co(dpg)(phen)]2ϩ 5 were photolysed
in both DMSO-d6 and D2O. The 1H NMR spectra of the result-
ing solutions were identical to those described above for
chromatographically isolated 8.
Photolysis of [Co(dpa)(phen)]2؉ 6: production of [Co(dam)-
(phen)]2؉
9
1
Method 2. H NMR of [Co(dam)(phen)]2ϩ 9 (D2O, assign-
ments were aided with a COSY spectrum): δ 1.13 (d, 3 H, CH3,
J = 6.4), 4.85 (α-H, obscured by HOD peak), 5.56 (picolyl AB,
other half hidden by HOD peak, 1 H, J = 16.6), 5.72 (picolyl
AB, other half hidden by HOD peak, 1 H, J = 17.6 Hz), 5.87–
5.92 (m, 2 H, dam pyridyl), 6.53–6.65 (m, 2 H, dam pyridyl),
7.36–7.42 (m, 2 H, dam pyridyl), 7.89–7.94 (m, 1 H), 8.00–8.05
(m, 2 H, dam pyridyl and phen), 8.30–8.47 (m, 2 H, phen),
8.81 (d, 1 H, phen), 9.03–9.08 (m, 2 H, phen), 10.17 (d, 1 H,
phen) and 10.44 (d, 1 H, phen). 1H NMR (DMSO-d6): δ 1.02 (d,
3 H, J = 6.3), 4.65–4.79 (m, 3 H, α-H and 2 picolyl CH2), 5.71
(d, 1 H, CH2, J = 17.1 Hz), 5.82–5.92 (m, 3 H, 1 picolyl CH2 and
2 aromatic dam pyridyl), 6.66–6.70 (m, 2 H, dam pyridyl), 7.50–
7.56 (m, 2 H, dam pyridyl), 7.71–7.74 (m, 2 H, dam pyridyl),
8.18–8.21 (m, 1 H, phen), 8.42–8.57 (m, 3 H, phen), 8.99 (d,
Photolysis of the [Co(dpx)(phen)]2؉ complexes 5–7
Method 1: Large scale photolysis and chromatographic isol-
ation of photolysis products. Given the quantity of material
required, this technique was only suitable for complex 5. An
aqueous solution (approximately millimolar concentration) was
photolysed with an immersible Jelight PS-3004-30 mercury
lamp for around 80 min. The solution was kept below 20 ЊC by
flowing ice-cold water through a jacket around the cell. Several
60 mL batches were combined and chromatographed on CM
Sephadex C25 (Naϩ, 3 × 25 cm). Elution with 0.2 M NaClO4
developed two bands; an orange band of the unchanged
starting material and an intense orange-yellow band of the
photolysis product. This second eluate was concentrated on a
rotary evaporator (<30 ЊC) until an orange crystalline solid
deposited. This solid was filtered off, washed with ethanol and
ether, and air dried.
1 H, phen), 9.22 (d, 1 H, phen) and 10.49 (br m, 2 H, phen). 13
C
NMR (DMSO-d6): δ 14.52 (CH3), 52.71 (Co–CH2), 57.65,
61.13, 121.31, 123.21, 124.18, 124.87, 126.89, 127.72, 127.96,
128.25, 130.43, 131.07, 139.08, 139.41, 139.74, 139.89, 140.07,
147.28, 148.58, 148.77, 154.19, 157.55, 159.95 and 161.79.
Photolysis of [Co(dpc)(phen)]2؉ 7: production of [Co(dcm)-
(phen)]2؉ 10
Method 2. 1H NMR of [Co(dcm)(phen)]2ϩ 10 (D2O): δ
Ϫ0.030 to 0.05 (br m, 1 H, cyclopropyl CH), 0.85 (m, cyclo-
Method 2: NMR scale photolysis. Milligram amounts of the
[Co(dpx)(phen)][ClO4]2 complexes 5–7 were dissolved in around
150 µL of deuteriated solvent (D2O or DMSO-d6) and placed in
3 mm NMR tubes. They were irradiated with a high-pressure
J. Chem. Soc., Dalton Trans., 2000, 2801–2808
2803

View Article Online
propyl CH2), 1.09–1.13 (m, 2 H, cyclopropyl CH2), 1.27–1.31
(m, 1 H, cyclopropyl CH2), 4.23 (d, 1 H, J = 9.3, α-H), 4.73 and
5.55 (AB, 2 H, J = 16.8), 4.94 and 5.72 (AB, 2 H, J = 17.1 Hz),
5.90–5.96 (m, 2 H, dcm pyridyl), 6.52–6.63 (m, 2 H, dcm
pyridyl), 7.34–7.42 (m, 2 H, dcm pyridyl), 7.59–7.68 (m, 2 H,
dcm pyridyl), 8.00–8.05 (m, 1 H), 8.30–8.47 (m, 3 H), 8.81 (d,
1 H), 9.05 (d, 1 H), 10.15 (d, 1 H) and 10.50 (d, 1 H). 1H NMR
(DMSO-d6): δ Ϫ0.27 to Ϫ0.22 (br m, 1 H, CH), 0.72–0.77,
0.95–1.00, 1.18–1.23, 1.38–1.43 (all br m, total 4 H, cyclopropyl
CH2), 4.21 (d, 1 H, J = 9.0, α-H), 4.70 and 5.69 (AB, 2 H,
J = 17.1 Hz), 4.74 and 5.90 (AB, downfield doublet obscured by
dcm aromatic protons, upfield peak overlapping with other dcm
CH2 peak), 5.90 (m, 2 H, dcm pyridyl), 6.63–6.72 (m, 2 H, dcm
pyridyl H), 7.42–7.57 (m, 3 H, dcm pyridyl (2 H) and phen
(1 H)), 7.66–7.77 (m, 2 H, dcm pyridyl), 8.17–8.21 (m, 1 H,
phen), 8.46–8.57 (m, 2 H), 8.99 (d, 1 H), 9.22 (d, 1 H), 10.39 (d,
1 H) and 10.47 (d, 1 H). 13C NMR (DMSO-d6): δ 5.75 (2C,
cyclopropyl CH2), 10.24 (cyclopropyl CH), 58.10 (Co–C), 61.09
and 61.21 (dcm picolyl CH2), 121.4, 123.08, 123.16, 123.59,
124.13, 124.90, 126.87, 127.90, 128.22, 130.38, 137.31, 139.01,
139.23, 139.76, 140.05, 147.31, 148.65, 148.95, 154.03, 157.46,
159.93 and 161.83. Some peaks were assigned to thermal
decomposition products (δ 6.50, 50.00).
Fig. 1 Crystal structure of complex 5. All hydrogen atoms have been
omitted from the diagram. Selected bond lengths (Å) and angles (Њ):
Co–O(1) 1.880(2); Co–N(1) 1.930(3); Co–N(2) 1.933(3); Co–N(3)
1.921(3); Co–N(4) 1.919(3); Co–N(5) 2.010(3); O(1)–Co–N(1)
87.57(11), N(3)–Co–N(1) 84.73(13); N(3)–Co–N(2) 168.71(12); N(1)–
Co–N(2) 84.03(13); N(4)–Co–N(5) 83.01(11); N(1)–Co–N(5)
100.80(11).
Decomposition of the photolysis products 8–10
In neutral solutions. The NMR-scale photolysates were
allowed to stand for prolonged periods at room temperature
and the appearance of decomposition products was monitored
by NMR spectroscopy. The data from the spectra are given in
the supplementary material.
Of complex 8. This complex has a half-life of around 1 week
in D2O at rt. Bis(2-pyridylmethyl)amine (bpa) and phen were
detected.
could be used was a consequence of crystal damage during the
data collection.
CCDC reference number 186/2068.
See http://www.rsc.org/suppdata/dt/b0/b003095n/ for crystal-
lographic files in .cif format.
Of complex 9. Acetaldehyde, [(bpa)(phen)Co(O2)Co(phen)-
(bpa)]4ϩ (11). [Co(phen)3]3ϩ, bpa, and phen were detected.
Of complex 10. This complex has a half-life of around 8–9
hours in D2O at room temperature. Cyclopropanecarbaldehyde
and 11 were detected.
Results
Syntheses
The nucleophilic displacement of chloride from two equivalents
of 2-chloromethylpyridine produces N,N-bis(2-pyridylmethyl)-
glycine in good yield, as reported previously.8 The preparation
of [Co(dpg)(NO2)2] was based on a published synthesis of
[Co(NO2)3(tacn)]9 (tacn = 1,4,7-triazacyclononane) and as the
desired complex, [Co(dpg)(NO2)2], is also a neutral species it
too has the attractive property of precipitating out of the aque-
ous reaction mixture. The Hdpa and Hdpc ligands, and their
nitro complexes, were prepared in a similar manner.
In acidic solution. Of complex 8. The peaks due to complex 8
disappeared over a period of weeks; Bpa and phen were
detected.
Of complex 9. The peaks due to complex 9 disappeared with
a half-life of around 24 hours. Acetaldehyde, bpa and phen
were detected.
1
The [Co(dpg)(NO2)2] complex was characterised by H and
X-Ray crystallography
13C NMR spectroscopy. It has the pyridyl donor groups
arranged in a trans fashion (Cs symmetry). This geometry is
conserved during the formation of the [Co(dpg)(phen)]2ϩ com-
plex 5, and the substituted derivatives (Scheme 2), as indicated
by examination of 1H and 13C NMR spectra.
Experimental details are given in the supplementary inform-
ation. Crystallographic data are given in Table 1.
[Co(dpg)(phen)][ClO4]2 5. Crystals suitable for the diffraction
study were obtained by the slow cooling of a hot solution of the
complex in dilute HClO4. All hydrogen atoms were placed at
calculated positions and refined with a riding model.
[Co(dgm)(phen)][ClO4]2ؒH2O 8. X-Ray quality crystals were
produced by a recrystallisation from water–ethanol (50:50).
Two independent molecules were present in the asymmetric
unit. The hydrogen atoms bound to the carbon atom of
the metallacycle (C(13)) were found on a difference electron
density map and refined isotropically. Other hydrogen atoms
were placed in calculated positions and refined with a riding
model.
[(bpa)(phen)CoO2Co(phen)(bpa)][ClO4]4 11. Small orange-
brown crystals formed in an NMR tube during the thermal
decomposition of complex 10 at room temperature. All hydro-
gen atoms were included in calculated positions and refined
using a riding model. The limited number of reflections that
Scheme 2
The solid state structure of [Co(dpg)(phen)][ClO4]2 5 was
determined by X-ray crystallography (Fig. 1). The geometry
2804
J. Chem. Soc., Dalton Trans., 2000, 2801–2808

View Article Online
Table 1 Crystallographic data for complexes 5, 8 and 11
Formula
M
Crystal system
Space group
a/Å
b/Å
c/Å
C26H22Cl2CoN5O10
C25H24Cl2CoN5O9
666.31
Orthorhombic
Pbcm
18.158(5)
8.978(2)
34.173(9)
C48H42Cl4Co2N10O18
1306.58
Monoclinic
C2/m
21.652(10)
11.320(5)
12.352(6)
114.142(6)
2763(2)
2
694.32
Monoclinic
P21/c
11.620(8)
24.487(13)
10.296(6)
114.54(3)
2665(3)
4
β/Њ
V/Å3
5571(2)
8
Z
T/K
293(2)
0.687
26963
3873
0.0351
0.0888
159(2)
0.870
42616
11940
0.0431
0.1006
168(2)
0.875
6574
2287
0.0497
0.1017
µ/mmϪ1
Reflections
Independent
R [I > 2σ(I)]
RЈ [I > 2σ(I)]
around the cobalt centre is approximately octahedral. There is a
marked difference in the Co–Nphen bond lengths (2.01 Å trans to
the carboxylate group, and 1.92 Å trans to the tertiary nitrogen
donor) which is presumably indicative of the greater trans
influence of the more basic amine donor. The five-membered
picolylamine chelate adopts the common envelope motif, with
the picolyl methylene unit slightly out of the plane of the other
four atoms. The metal–ligand bond lengths are marginally
shorter than those found in a related tetranuclear cobalt()
complex of N-(2-pyridylmethyl)glycinate.10
The assignment of the stereochemistry of the N,N-bis(2-
pyridylmethyl)alaninato (6) and cyclopropyl-N,N-bis-(2-
pyridylmethyl)glycinato (7) complexes from NMR data was less
straightforward than for the parent complex, as the alkyl sub-
stituent breaks the Cs symmetry. However, the chemical shifts
of the protons of the two pyridine rings and those of the picolyl
methylene groups in both complexes are similar to those
observed for 5. This implies that 6 and 7 also have a trans-
pyridyl geometry. Furthermore, the UV-vis spectra of these
complexes are very similar to that of 5.
Photochemistry of [Co(dpx)(phen)]2؉ complexes 5–7
Fig. 2 Crystal structure of one of the two cations of complex 8 found
in the asymmetric unit. Most hydrogen atoms have been omitted from
the diagram. Selected bond lengths (Å) and angles (Њ): Co–N(13)
1.898(4); Co–C(13) 1.924(6); Co–N(15) 1.931(4); Co–N(20) 1.940(4);
Co–N(1) 1.970(4); C(13)–N(13) 1.417(7); Co–N(10) 2.044(4); N(13)–
Co–C(13) 43.5(2); N(13)–Co–N(15) 84.31(18); N(13)–Co–N(20)
84.12(17); N(13)–Co–N(1) 157.79(18); N(13)–Co–N(10) 119.13(17);
N(1)–Co–N(10) 83.07(17).
The UV photolysis reaction of complex 5 was examined by
three different methods: (i) large-scale photolysis with separ-
ation of the product by ion-exchange chromatography; (ii) H
1
NMR spectroscopy; and (iii) UV-vis spectroscopy.
Photolysis of an aqueous solution of complex 5 with broad
spectrum light from an immersible mercury lamp induced a
change from orange-red to yellow. Chromatography of this
photolysate produced two closely spaced bands. The band
which eluted first was orange-red, and identified as unchanged
5. The similar chromatographic behavior of the second band
implied that it also had a 2ϩ charge. This eluate was noticeably
more yellow than the starting material and, upon concentration
of the solution, orange crystals that contained some sodium
perchlorate deposited. Recrystallisation furnished crystals suit-
able for X-ray crystallography. The solid state structure of the
photolysis product 8 is shown in Fig. 2.
The X-ray diffraction study revealed that carbon dioxide had
been eliminated from the amino acidato chelate to leave a three-
membered Co–C–N ring (Scheme 3). The bite angle of the new
chelate is 43.5Њ, with bond lengths Co–C 1.93 Å and Co–N 1.90
Å. Both hydrogen atoms attached to the co-ordinated carbon
atom (C13) were found on a electron density difference map
during the refinement. The C13–N13 bond length (1.42 Å) is of
particular interest (see Discussion). The Co–Npy bond lengths
fall between 1.92 and 1.94 Å and the chelate bite angles are all
clustered around 84–85Њ. The picolylamine chelate rings retain
their envelope configurations in the photolysis product.
Scheme 3
photoelimination of CO2. This has major ramifications for the
appearance of the H NMR spectra of 8, as the pyridine rings
1
have been rotated around their Co–N axis, moving the ortho
protons closer to the π electrons of the phen ligand. These
protons now resonate at δ 5.96, which is an unusually low value
for aromatic protons. Another notable shift occurs for two of
the phen resonances (probably the two ortho protons) which
move downfield, to beyond 10 ppm. The methylene groups
Comparison of the solid state structures of complexes 5 and
8 indicates that there are significant changes in the orientation
of the pyridylmethyl arms of the tetradentate ligand upon
J. Chem. Soc., Dalton Trans., 2000, 2801–2808
2805

View Article Online
attached to the pyridyl arms remain as an AB system, one half
of which was found under the HOD peak through collection of
a COSY spectrum.
1H NMR spectra of complex 8 obtained directly after
photolysis of a D2O or DMSO-d6 sample of 5 were identical to
spectra of the photolysis product which had been isolated fol-
lowing column chromatography. The results of the NMR scale
photolyses indicated that the photochemical conversion is very
clean. A 13C NMR spectrum in DMSO-d6 showed that, as ex-
pected, the signal for the carboxylate carbon had disappeared.
The photolysis of an aqueous solution of complex 5 was also
monitored by UV-vis spectroscopy. The 480 nm ligand field
band shifts to shorter wavelengths upon irradiation of the
complex is consistent with the replacement of a carboxylate
donor by a strongly basic carbon donor. There is a significant
rise in absorption in the 400–425 nm region. The final spectrum
was identical to that obtained from a solution of an isolated
sample of 8. These spectral changes are similar to those noted
by Poznyak and Pavlovskii during the photolysis of [Co(gly)-
(phen)2]2ϩ 11
.
Fig. 3 A portion of the visible spectrum of [Co(dpc)(phen)]2ϩ 7 and its
photolysis product 10, in aqueous solution.
The photolysis of a D2O solution of the alaninato analogue 6
was monitored by 1H NMR spectroscopy. Many of the spectral
changes observed during the photolysis of complex 5 were
repeated in this instance. For example, two of the phen reson-
ances were observed downfield of 10 ppm, and a multiplet
appeared at δ 5.95. In addition, the doublet assigned to the
methyl group was seen to move from δ 1.80 to 1.13. Examin-
ation of molecular models reveals that, upon formation of 9,
the methyl group would be expected to move into a shielded
1
region over a pyridine ring, and its H NMR signal should
therefore be observed at higher field.
The photolysis of the dpa complex 6 in aqueous solution was
also monitored by UV-vis spectroscopy. The ligand field band,
initially centred at 481 nm, shifted to 462 nm, and absorption in
the 400–430 nm region rose significantly. These changes are
consistent with the formation of 9 during the photolysis.
Unfortunately complex 9 could not be isolated as a crystal-
line solid. This has prevented a full characterisation of the
photolysis product as neither elemental analysis nor X-ray
crystallography was possible.
Upon irradiation of the cyclopropyl derivative 7 with 254 nm
light a clean reaction takes place to form complex 10. The
1
changes that occur in the H NMR spectrum are similar to
those seen in the photolyses of 5 and 6. For instance, two of the
phen signals shift downfield to δ 10.50 and 10.15, and two
protons which can be assigned to the pyridyl rings move upfield
to around δ 5.95. Thus it appears likely that a similar metalla-
cycle has been formed. A broadened set of signals centered at
around δ 1 is seen for 7, which is characteristic of a cyclopropyl
group. The cyclopropyl methine proton of the photolysis prod-
uct resonates at δ 0. Examination of a molecular model reveals
that this proton is likely to lie directly over a pyridine ring.
Similar spectral changes were observed when the photolysis was
carried out in DMSO-d6.
The UV-vis spectra of complexes 7 and 10 are shown in Fig.
3. The shift of the d–d absorption peak to shorter wavelengths
(462 nm), accompanied by a rise in absorption in the 400–430
nm region, is also consistent with 10 being the expected
metallacyclic photolysis product.
Scheme 4
thermal decomposition of 5, although this complex had a
relatively long half-life in D2O (<1 week).
If the D2O solutions of the (decomposing) complexes 9 or 10
were allowed to stand at room temperature for prolonged
periods a new grouping of eight aromatic protons appeared in
1
the H NMR spectra. Clearly the growth of these signals was
due to the formation of a new compound. We were rather
fortunate in that crystals of it formed in a NMR sample. X-Ray
crystallography revealed the structure shown in Fig. 4.
The complex is a peroxo-bridged dinuclear complex, [(bpa)-
(phen)CoO2Co(phen)(bpa)]4ϩ 11. The geometry around each
cobalt is identical, with the peroxo ligand occupying a co-
ordination site trans to the secondary nitrogen of the facially
co-ordinated bpa ligand. This gives a highly symmetrical struc-
ture (the atoms which are labeled in Fig. 4 comprise the asym-
metric unit). The remainder of the complex is generated by a
mirror plane which contains the two cobalt ions and the peroxo
ligand (bisecting the phen ligands), and a centre of inversion,
which is located at the midpoint of the O–O bond. The geom-
Thermal decomposition of the metallacycles 8–10
The complexes 9 and 10 decomposed on standing to give alde-
hydes (acetaldehyde and cyclopropanecarbaldehyde respect-
ively) in both D2O and DMSO-d6 (Scheme 4). These complexes
have half-lives at room temperature in D2O of 5 days and 7–9
hours respectively. Extraction of the resulting D2O solutions
with CDCl3 followed by NMR spectroscopy established the
presence of free phen and bis(2-pyridylmethyl)amine (bpa) in
both cases. These products were also observed following the
2806
J. Chem. Soc., Dalton Trans., 2000, 2801–2808

View Article Online
bond (1.50 Å) and the C–N bond length found in pyridine15
(1.34 Å). Thus we may conclude that the C–N bond order is
about 1.25.
Four other complexes that contain a Co–C–N chelate ring
have been characterised by X-ray crystallography. Two of these
were synthesized by photochemical decarboxylation of an
amino acidato chelate,2g,16g whilst the other two17,18 resulted
from intramolecular reactions of macrocyclic complexes. The
range of C–N bond lengths in these molecules (1.35–1.45 Å)
brackets that revealed in our structure.
The downfield chemical shift changes that we observed for
the phen resonances are also consistent with the idea that the
C–N bond of the metallacycle has some π character, with the
ortho protons in particular being influenced by the resulting
magnetic anisotropy.
Taking all these data together, we believe that the structure of
these complexes may best be represented as a combination of
the resonance forms shown in Scheme 4. The complex can be
viewed as a combination of a distorted octahedral cobalt()
complex and a trigonal bipyramidal cobalt() complex (in which
the C(13)–N(13) fragment is a π-bound η2 imine ligand). We
believe that this imine character in the metallacyclic fragment
of these molecules may be significant in the context of the
decomposition reactions that we observed.
The metallacyclic complexes 8–10 decompose with time, with
those derived from alanine and cyclopropylglycine giving the
related carbonyl compounds, acetaldehyde and cyclopropane-
carbaldehyde respectively. These same products were observed
following photolysis of the related bidentate amino acid com-
plexes 1.5 This study establishes a direct link between the
decomposition of the metallacyclic complexes and the form-
ation of the carbonyl compounds. This supports the contention
that the unsubstituted amino acid complexes do indeed react
via metallacyclic species.
Fig. 4 Crystal structure of the complex ion 14. All hydrogen atoms
have been omitted from the diagram. Selected bond lengths (Å) and
angles (Њ): Co–O(1) 1.833(5); Co–N(1) 1.925(4); Co–N(2) 2.004(6); Co–
N(3) 1.943(4); N(1)–Co–N(2) 84.20(18); N(3)Ј–Co–N(3) 84.2(3);
N(1)Ј–Co–N(1) 88.9(2); O(1)Ј–O(1)–Co 113.0(4).
etry around the cobalt atom is approximately octahedral, with
the largest deviations appearing in the chelate bite angles. The
O(1)Ј–O(1)-Co bond angle (113.0Њ), and the O(1)–O(1)Ј bond
length (1.436 Å) are similar to those seen for other dinuclear,
monobridged peroxo complexes of Co().12,13 The relatively
long Co–N(2) bond (2.004 Å) is in keeping with other µ-peroxo
dimers which have similar donor sets.14
The effect of acid conditions on the decomposition was
probed by monitoring DCl solutions ([DCl] ≈ 0.5 M) of com-
plexes 8 and 9 by H NMR spectroscopy. In the case of 9 the
methyl group of the amino acid fragment was incorporated into
acetaldehyde, however the fate of the remainder of the complex
was initially less clear due to the crowding in the aromatic
region of the spectra. However, free bpa and phen were found
for both complexes following extraction of alkaline solutions
with chloroform.
Formation of the amine fragment of the metallacycle was
also observed, both directly and through isolation of a dimeric
µ-peroxo complex. The most common method of synthesis of
µ-peroxo complexes is the air oxidation of neutral solutions of
CoII and the required ligands.13 Thus, the discovery of the dimer
11 in the decomposing photolysates is not surprising given that
bpa, phen and (probably) CoII are all by-products from the
decomposition reactions of 9 and 10. Of the nine possible geo-
metrical isomers for the dimeric complex 11, the observed iso-
mer is one with the same conformation around each cobalt
centre, and with the peroxo ligand trans to the secondary nitro-
gen of the bpa ligand. This trans configuration may be favoured
because the Co–peroxo ligand bonding interaction, i.e. the fill-
ing of the O2 π* orbitals with electron density from the metal d
orbitals, is maximised.13
1
Discussion
The preparation of complexes 5–7 was quite straightforward.
The observed lability of the nitro ligands is rather remarkable.
Nitro complexes are common intermediates in cobalt() chem-
istry. Typically they are treated with acid (HCl) solutions to
generate intermediate (chloro) complexes which readily under-
go substitution reactions. For the [Co(dpx)(NO2)2] complexes,
reaction via this two step route is also possible but results in a
diminished yield. It is possible that a trace cobalt() impurity
may have been catalysing the direct reactions of the nitro com-
plexes with phen, however we did not pursue evidence for this
hypothesis.
Irradiation of complexes 5–7 resulted in decarboxylation and
formation of complexes that contain Co–C–N metallacycles.
The metallacyclic complex 8 that results from photolysis of the
5 has fully been characterised. Characterisation of the analo-
gous complexes 9 and 10 was based on their NMR and UV-vis
spectra, and the structural assignments are supported by
the similarities of their spectra to those observed for the fully
characterised complex.
Cobalt–carbon bond cleavage can occur in three ways during
the decomposition of the metallacycle. They differ according to
how the bonding electrons are shared between the alkyl frag-
ment and the cobalt centre;19 eqns. (1)–(3). Reaction via path (1)
CoIII–R → CoIII ϩ RϪ
(1)
(2)
(3)
III
II
ؒ
Co –R → Co ϩ R
CoIII–R → CoI ϩ Rϩ
would lead to an α-amino carbanion, which would be expected
to acquire a deuteron from the solvent (D2O or DMSO-d6) gen-
erating a substituted methylamine. Alternatively, a cobalt()
complex and an α-amino radical may be formed via heterolytic
cleavage of the Co–C bond (path 2). This radical could abstract
a hydrogen atom from a species in solution or from the solvent,
generating an amine, or react with another odd-electron species
(e.g. self-termination, or reaction with O2) to give a diamine or
a carbonyl compound (see below). The third possibility is the
production of a cobalt() complex and a carbocation (path 3).
The structural and spectroscopic data for complex 8 raise a
question about the nature of the bonding in the metallacycle. In
particular, the C–N bond length (1.42 Å) is rather short, which
may imply that the bond order that is greater than one. This
bond length is midway between that expected for a C–N single
J. Chem. Soc., Dalton Trans., 2000, 2801–2808
2807

View Article Online
A carbocation with an adjacent amine group is a canonical
form of an imine, and hydrolysis of this imine would give an
amine and a carbonyl compound. These are the products that
we observed in both neutral and acidic solution, and this
decomposition pathway would seem to be a reasonable way for
them to be produced. It should be noted however that a one
electron oxidation of the aminoalkyl radical that results from
the second pathway would also generate the imine and could
therefore also explain the products we observe. Aminoalkyl
radicals are known to be strong reductants.20 Indeed, if the
oxidant were the cobalt() ion, the combination of the second
pathway and electron transfer from the radical to cobalt()
would be the equivalent of the third path.
We favour the third decomposition path, with production of
cobalt() and an imine, since there is evidence for π character in
the C–N bond of the metallacyclic complex. Cobalt() would
likely react further to give cobalt() species in aqueous solution.
Metallacycle decomposition could be envisaged as occurring
through structures which had increasing contributions from the
imine canonical form (Scheme 4). This mechanism can also
account for the observation of formaldehyde in the photolysis
of related complexes.21
Decomposition pathway (1) has been observed previously for
a cyclam-based Co–C–N metallacycle in dilute HCl whereby
the metallacycle is (presumably) opened by electrophilic attack
of a proton at the co-ordinating carbon.2g A chloro ligand filled
the vacant co-ordination site with no apparent change of the
cobalt oxidation state. If complexes 8–10 had followed this
pathway in acidic solution, deuteriated bis(2-pyridylmethyl)-
methylamine, free or complexed, would have been observed.
The difference in the decomposition pathways of this com-
plex and our pyridine-based systems can perhaps be ascribed to
the nature of the bonding in the complexes and the auxiliary
ligands. The cyclam-based complex was found to have a C–N
bond length of 1.447 Å, longer than the 1.417 Å found in 8.
This may imply that there is somewhat more π character in the
C–N bond in the complex we have isolated than in the cyclam
complex. This would render the carbon atom less susceptible to
electrophilic attack (since the electrons will be shared with the
nitrogen atom). Furthermore, the π-acidic co-ligands on this
complex would favour the cobalt() canonical form to a greater
degree, and would stabilise any cobalt() products that might be
formed. The strongly σ-donating co-ordination sphere of the
cyclam complex would stabilise cobalt() species and this in
turn would favour the decomposition pathway that generates
the carbanion.
This study has opened the door to the synthesis of a range of
other complexes by similar photochemical methods. Further-
more, the chemistry of such complexes is as yet unexplored and
may itself lead to novel and interesting compounds along with
shedding light on the nature of the bonding in the Co–C–N
metallacycle.
Acknowledgements
S. G. T. acknowledges the University of Canterbury for the
provision of a doctoral scholarship. We are also indebted to
the members of the X-ray crystallography laboratory. We par-
ticularly thank Dr P. J. Steel for his help with the structure
determination of the dinuclear complex.
References
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.
2 For 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.
Stelímashok, 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;
(i) R. M. Hartshorn and S. G. Telfer, J. Chem. Soc., Dalton Trans.,
1999, 3217.
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 R. M. Hartshorn and S. G. Telfer, J. Chem. Soc., Dalton Trans.,
1999, 3565.
6 E. Natarajan and P. Natarajan, Inorg. Chem., 1992, 31, 1215.
7 P. J. Lawson, M. J. McCarthy and A. M. Sargeson, J. Am. Chem.
Soc., 1982, 104, 1734.
8 H. Iikura and T. Nagata, Inorg. Chem., 1998, 37, 4702.
9 M. S. Okamoto and E. K. Barefield, Inorg. Chim. Acta, 1976, 17, 91.
10 T. Ama, K. Okamoto, T. Yonemura, H. Kawaguchi, A. Takeuchi
and T. Yasui, Chem. Lett., 1997, 1189.
11 A. L. Poznyak and V. I. Pavlovskii, Russ. J. Inorg. Chem., 1981, 26,
292.
12 E. C. Niederhoffer, J. H. Timmons and A. E. Martell, Chem. Rev.,
1984, 84, 137.
13 D. A. Buckingham and C. R. Clark, in Comprehensive Coordination
Chemistry, ed. G. Wilkinson, Pergamon Press, Oxford, 1987, vol. 4,
ch. 47, pp. 775–789.
14 J. H. Timmons, R. H. Niswander, A. Clearfield and A. E. Martell,
Inorg. Chem., 1979, 18, 1042; Inorg. Chem., 1979, 18, 2977.
15 A. Albert, Heterocyclic Chemistry, 2nd edn., Athlone Press,
London, 1968.
16 E. B. Chuklanova, T. N. Polynova, M. A. Porai-Koshits, A. L.
Poznyak and V. I. Pavlovskii, Koord. Khim., 1986, 14, 103.
17 C.-K. Poon, W.-K. Wan and S. S. T. Liao, J. Chem. Soc., Dalton
Trans., 1997, 1247.
18 S. M. Polson, L. Hansen and L. G. Marzilli, J. Am. Chem. Soc.,
1996, 118, 4804.
Conclusion
A series of cobalt() complexes which contain amino acidato
chelates which are incorporated into tetradentate frameworks
have been synthesized and characterised. UV photolysis of
these complexes leads to elimination of CO2, as observed for
related bidentate amino acidato complexes. However, in this
case the resulting Co–C–N three-membered chelates are
relatively stable, even with alkyl groups on the co-ordinated
carbon atom, and can be characterised by conventional
methods.
Studies of a cyclopropyl containing substrate have shown
that the cyclopropane ring is retained in both the metallacyclic
product of photodecarboxylation and its decomposition prod-
ucts. This demonstrates that any intermediates in the formation
and decomposition of the metallacyclic compound that have a
radical centre on what was the α-carbon of the amino acid must
be very short-lived.
19 P. J. Toscano, L. G. Marzilli in Prog. Inorg. Chem.; S. J. Lippard
(Ed); Interscience, New York, 1984.
20 D. A. Armstrong, A. Rauk and D. Yu, J. Am. Chem. Soc., 1993, 115,
660.
21 V. Balzani, V. Cassarati, L. Moggi and F. Scandola, Inorg. Chem.,
1964, 4, 1247; A. L. Poznyak and V. E. Stelmashok, Inorg. Chim.
Acta, 1984, 83, L59; C. H. Langford and G. W. Quance, Can. J.
Chem., 1977, 55, 133.
2808
J. Chem. Soc., Dalton Trans., 2000, 2801–2808