Molecular and electronic structure of four- and five-coordinate cobalt complexes containing two o-phenylenediamine- or two o-aminophenol-type ligands at various oxidation levels: An experimental, density functional, and correlated ab initio study

Molecular and Electronic Structure of Four- and Five-Coordinate Cobalt

Complexes Containing Two o-Phenylenediamine- or Two o-Aminophenol-

Type Ligands at Various Oxidation Levels: An Experimental, Density

Functional, and Correlated ab initio Study**

Eckhard Bill, Eberhard Bothe, Phalguni Chaudhuri, Krzysztof Chlopek, Diran Herebian,

Swarnalatha Kokatam, Kallol Ray, Thomas Weyhermüller, Frank Neese,* and

Karl Wieghardt*^[a]

Abstract: The bidentate ligands N-

phenyl-o-phenylenediamine, H₂(²L_N^IP),

or its analogue 2-(2-trifluoromethyl)-

anilino-4,6-di-tert-butylphenol, (⁴L_O^IP),

react with [Co^II(CH₃CO₂)₂]·4H₂O and

triethylamine in acetonitrile in the

presence of air yielding the square-

reduced form of 4, namely [Co(Cp)₂]

[Co(⁴L_O)₂] (5)has been generated

ture magnetic susceptibility measure-

ments have been used to experimental-

ly determine the electronic structures

of these complexes. Density functional

theoretical (DFT)and correlated ab

initio calculation have been performed

on the neutral and monoanionic spe-

cies [Co(¹L_N)₂]^0,ꢀin order to under-

stand the structural and spectroscopic

properties of complexes. It is shown

that the corresponding nickel com-

plexes 8 and 9 contain a low-spin nick-

el(ii)ion regardless of the oxidation

level of the ligand, whereas for the cor-

responding cobalt complexes the situa-

tion is more complicated. Spectroscopic

oxidation states describing a d⁶(Co^III)

or d⁷(Co^II)electron configuration

cannot be unambiguously assigned.

through the reduction of

4

with

[Co(Cp)₂]. Complexes 1, 2, 2a, 3, 4, 5,

7, 8, and 9 have been characterized by

X-ray crystallography (100 K). The li-

gands are non-innocent and may exist

as catecholate-like dianions (²L_N^IP)^2ꢀ,

(⁴L^I_O^P)^2ꢀor p-radical semiquinonate

monoanions (²L^I_N^SQ)C^ꢀ, (⁴L^I_O^SQ)C^ꢀor as

planar,

four-coordinate

species

4

[Co(²L_N)₂] (1)and [Co( L_O)₂] (4)with

an S=¹= ground state. The correspond-

2

ing nickel complexes [Ni(⁴L_O)₂] (8)and

neutral

benzoquinones

(²L^I_N^BQ)⁰,

its

cobaltocene

reduced

form

(⁴L^I_O^BQ)⁰; the spectroscopic oxidation

states of the central metal ions vary ac-

cordingly. Electronic absorption, mag-

netic circular dichroism, and EPR spec-

troscopy, as well as variable tempera-

[Co^III(Cp)₂][Ni(⁴L_O)₂] (9)have also

been synthesized. The five-coordinate

species [Co(²L_N)₂(tBu-py)] (2) (S=¹= )

2

and its one-electron oxidized forms

[Co(²L_N)₂(tBu-py)](O₂CCH₃) (2a)or

[Co(²L_N)₂I] (3)with diamagnetic

ground states (S=0)have been pre-

pared, as has the species [Co-

(⁴L_O)₂(CH₂CN)] (7). The one-electron

Keywords: ab initio calculations ·

cobalt · density functional calcula-

tions · nickel · radical ions

Introduction

[a] Dr. E. Bill, Dr. E. Bothe, Prof. Dr. P. Chaudhuri, K. Chlopek,

Dr. D. Herebian, S. Kokatam, K. Ray, Dr. T. Weyhermüller,

Dr. F. Neese, Prof. Dr. K. Wieghardt

In 1966 Balch and Holm^[1]reported that the reaction of o-

phenylenediamine with CoCl₂·6H₂O (or [Co(CH₃CO₂)₂]·

4H₂O^[2])in the ratio 2:1 in an aqueous ammonia affords, in

Max-Planck-Institut für Bioanorganische Chemie

Stiftstrasse 34–36, 45470 Mülheim an der Ruhr (Germany)

Fax : (+49)208-306-3952

the presence of air,

a

deep violet precipitate of

[Co{C₆H₄(NH)₂}₂], which we will abbreviate here as

[Co(¹L_N)₂] according to Scheme 1. The room-temperature

crystal structure of this complex was reported by Peng

et al.^[3]in 1985 (see Figure 1). The neutral, mononuclear

complex is square planar; it is paramagnetic and possesses

E-mail: wieghardt@mpi-muelheim.mpg.de

[**] In this paper we use the following ligand abbreviations: the aromatic

dianion, a diiminophenolate(2ꢀ) , [CH₄(NH)₂]^2ꢀis (¹L_N^IP)^2ꢀ; its p radi-

6

cal monoanion,

(¹L^I_N^SQ

a

diiminosemiquinonate(1ꢀ)

,

[CH₄(NH)₂]C^ꢀis

6

)

^1ꢀ, and the neutral, diamagnetic diiminobenzoquinone form

[C₆H₄(NH)₂]⁰is (¹L^I_N^BQ)⁰. In cases in which the oxidation level of the

ligand is not established or of relevance but its chemical composition

is [C₆H₄(NH)₂] we will use the abbreviation (¹L_N). The o-aminophe-

nolato complexes (^XL_O)are analoguously abbreviated. See also

Scheme 1.

an S=¹= ground state as was established by its X-band EPR

2

spectrum and magnetochemistry.^[1]

In the above reports^[1–3]the authors have suggested that

both organic ligands are identical and that they are mono-

204

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

DOI: 10.1002/chem.200400850

Chem. Eur. J. 2005, 11, 204 – 224

FULL PAPER

highly colored, isostructural but diamagnetic complexes of

Ni^II, Pd^II, and Pt^II, namely [M(C₆H₄(NH)₂)₂], had also been

II

prepared,^[1]and that 2)square planar complexes of Co with

redox innocent ligands such as acetylacetonate(1ꢀ),

(acac)^[1–4]or (salen)^[2–5]are known; they also possess an

S=¹= ground state.^[6]

2

Even more persuasive was the observation that it is possi-

ble to oxidize [Co(C₆H₄(NH)₂)₂] with air or iodine in the

presence of coordinating (solvent)molecules (or anions)

such as pyridine (py)or phosphines or simple halide anions

(Cl^ꢀ, Br^ꢀ, I^ꢀ)affording diamagnetic, five-coordinate spe-

cies:^[1–3][Co^III(C₆H₄(NH)₂)₂X] (X=Cl;^[2,3]I,^[1,2]SCN^ꢀ,^[2]

[Co^III(C₆H₄(NH)₂)₂(PPh₃)]PF₆,^[2,7]and [Co^III(C₆H₄(NH)₂)₂-

(py)]Cl.^[8]These have in part been characterized by X-ray

crystallography (Figure 1)at ambient temperature. The pro-

posed bonding picture requires then the presence of two p

radical monoanions, (¹L^I_N^SQ)C^ꢀand an additional apical anion,

X, in cases in which five-coordinate neutral species are gen-

erated or a neutral apical ligand when monocationic species

are formed. In both instances the central cobalt(iii)ion pos-

sesses an S_Co=0 ground state. Thus, on going from the four-

coordinate to these five-coordinate species an oxidation of

the central cobalt(ii)to cobalt( iii)is believed to occur.

Scheme 1.

It has been observed that in these diamagnetic square-

base pyramidal complexes of low-spin cobalt(iii)the mono-

anionic ligands [C₆H₄(NH)₂]C^ꢀexhibit geometrical features

that are readily ascribed to o-diiminosemiquinonate(1ꢀ) p

radicals:^[9]the average C N bond length is in the range

ꢀ

1.34ꢁ0.01 irrespective of the nature of the fifth apical

ligand. Furthermore, the six-membered rings display typical

ꢀ

quinoid type distortions with two alternating short C C dis-

tances and four longer ones.

It is therefore quite surprising that the geometrical fea-

tures of the two ligands apparently differ slightly in the

ꢀ

four- and five-coordinate species (Figure 1). The C N bonds

are longer in the four than in the five-coordinate species

(Figure 1). We note that the quality of the reported struc-

ture determinations at room temperature is not satisfactory.

Figure 1. Schematic structures with bond lengths in of square-planar

[Co(¹L_N)₂]⁰from reference [3] (top)and square-base pyramidal

[Co^III(¹L)₂Cl] from reference [3] (bottom). The experimental errors from

the room-temperature crystal structures are ~ꢁ0.03 (ꢂ3s)for the

former and ꢁ0.015 for the latter.

ꢀ

The average experimental error of a given C C or C N

bond length is ~ꢁ0.03 (ꢂ3s), which does not allow to

safely assign an oxidation level of the ligands in the four-co-

ordinate species. Thus, the previous authors have not been

able to distinguish experimentally between the two different

electronic structures A and B shown in Scheme 2 for the

four-coordinate species. The difference between A and B is

that A has two ligand p radicals and a central Co^IIion

anionic p radicals (S_rad=¹= )of the diiminobenzosemiquino-

2

nate(1ꢀ)type and that, therefore, the central cobalt ion pos-

sesses a +II oxidation state (low spin d⁷, S_Co=¹= ). In this

(S_Co=¹= ), whereas B may be described as a species contain-

2

model the spins of the ligand p radicals are assumed to be

strongly intramolecularly antiferromagnetically coupled.^[3]

We note that Balch and Holm suggested a different

model. Using a valence-bond picture, they suggested a cen-

ing a dianionic ligand, (¹L^I_N^P)^2ꢀ, and a single p radical mono-

anion, (¹L_N^ISQ)^1ꢀ, as well as a Co^IIIion in a square-planar

ligand field (S_Co=1). The ligand mixed valency in B may

then be delocalized (class III)ensuring the observed struc-

tural equivalency of the two ligands on the time scale of an

X-ray diffraction experiment. Model B would explain the

tral low-spin cobalt(ii)ion ( S=¹= )coordinated to two closed

2

shell [C₆H₄(NH)₂]^nꢀligands, which they formulated as two

resonance hybrids between an aromatic dianion (¹L_N^IP)^2ꢀand

longer C N distances as a consequence of the presence of

ꢀ

its

neutral

diiminobenzoquinone

form

(¹L^I_N^BQ)⁰:

an aromatic dianion and a monoanionic p radical. In fact,

[Co^II(¹L^I_N^BQ)(¹L_N^IP)]Q[Co^II(¹L^I_N^P)(¹L^I_N^BQ)]. These conclusions ap-

one would expect the C C and C N distances to be the

arithmetic average of the corresponding distances in the

ꢀ

peared to be bolstered by the fact that 1)the corresponding

Chem. Eur. J. 2005, 11, 204 – 224

www.chemeurj.org

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

205

K. Wieghardt, F. Neese et al.

Scheme 2.

1ꢀ

mono- and dianion, (¹L^I_N^SQ

)

and (¹L^I_N^P)^2ꢀ, as shown in

Scheme 3. These data are quite close to the observed values

of [Co(¹L_N)₂] from reference [3] (Figure 1)and seem to indi-

Figure 2. Schematic structures of diamagnetic five-coordinate monocation

of cis-[Co^III(²L)₂(py)]CH₃CO₂·H₂O^[10](top)and of trans-[Co^III(²L)₂I] (3)

of this work (bottom); the experimental error (3s)is ꢁ0.01 for the

ꢀ

C C, C N distances of the former and the latter species.

been possible to prepare

a salt of the monoanion

[Co(⁴L_O)₂]^ꢀ, namely [Co(Cp)₂]⁺[Co(⁴L_O)₂]^ꢀ(5), in which

Cp^ꢀis the cyclopentadienyl anion and [Co(Cp)₂]⁺is the co-

baltocenium cation. Both 4 and 5 have been structurally

characterized. Oxidation of 4 with air in acetonitrile affords

[Co(⁴L_O)₂(CH₂CN)] (7). The corresponding neutral species

[11]

Scheme 3.

[Co(⁴L_O)₂X] (X=Cl, I)have been described previously.

These five-coordinate complexes possess a diamagnetic cen-

4

cate

[Co^III(¹L_N^IP)(¹L_N^ISQ)] with charge delocalization over both li-

gands. The cobalt(iii)ion should possess an _Co=1 ground

a

charge distribution of this species as in

tral cobalt(iii)ion and two ( L^I_O^SQ)C^ꢀradicals that are antifer-

romagnetically coupled yielding the observed S=0 ground

S

state. Octahedral [Co^III(³L_O^ISQ)₃] (6)possesses an

S=²=

3

state as is readily deduced from ligand-field theoretical con-

siderations (d⁶in a square planar field). Intramolecular anti-

ferromagnetic coupling between the spins of the Co^IIIion

ground state due to the presence of three orthogonally coor-

dinated (L^I_O^SQ)C^ꢀp-radical monoanions.^[15]Interestingly, it has

also been possible to synthesize the nickel(ii)-containing

complexes [Ni^II(⁴L_O^ISQ)₂] (8)and its monoanionic analogue

[Co(Cp)₂][Ni^II(⁴L^I_O^P)(⁴L^I_O^SQ) ] 9(). As we will show here, these

analogues of 4 and 5 possess different ligand oxidation

levels and, concomitantly, differing spectroscopic oxidation

states of the metal ions (possibly Co^IIIin 4 and 5, but cer-

tainly Ni^IIin 8 and 9).

and one ligand radical yields then the observed S_t=¹=

2

ground state.

In order to clarify this ambiguity we have synthesized a

series of three cobalt complexes with the ligand N-phenyl-o-

phenylenediamine, H₂(²L_N^IP). In addition, we have performed

DFT and correlated ab initio theoretical calculations on neu-

tral [Co(¹L_N)₂] and its monoanion. The structure of the

monocationic complex in cis-[Co^III(²L^I_N^SQ)₂(py)]CH₃CO₂·H₂O

has been reported by Peng et al.^[10]and is shown in Figure 2.

We have structurally characterized the four-coordinate spe-

cies [Co(²L_N)₂] (1), its neutral five coordinate pyridine deriv-

ative trans-[Co(²L_N)₂(tBu-py)]⁰(2), and its one-electron oxi-

dized species cis-[Co(²L_N)₂(tBu-py)]CH₃CO₂(2a)and also

the iodo complex trans-[Co(²L_N)₂I] (3).

Throughout this paper we carefully differentiate between

formal^[17]and spectroscopic (or physical)^[18]oxidation states

as C. K. Jçrgensen proposed in 1968. A formal oxidation

state of a metal ion in a given coordination compound is a

non-measurable, physically meaningless, usually integral

number that is derived by heterolytic removal of all ligands

in their closed-shell electron configuration, whereas a spec-

troscopic oxidation state is derived from the actual (often

n

Finally, we have used the bulky ligand 2-(2-trifluoro-

methyl)anilino-4,6-di-tert-butylphenol,^[11–16]H₂(⁴L_O^IP), shown

in Scheme 1 and prepared the neutral four coordinate spe-

cies [Co(⁴L_O)₂] (4), which is the exact analogue of 1. It has

observable)d electron configuration of the metal ion in a

given complex. For example, the spectroscopic oxidation

state of the metal ion of a metalloprotein containing a single

iron ion may be +III, since its high- or low-spin d⁵electron

206

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemeurj.org

Chem. Eur. J. 2005, 11, 204 – 224

Cobalt Coordination Complexes

FULL PAPER

tate formed that was collected by filtration. Recrystallization from a

CH₂Cl₂/CH₃OH (1:1)mixture afforded single crystals suitable for X-ray

crystallography. Yield: 0.81 g (54%); EI MS: m/z (%): 785 (100) [M⁺];

elemental analysis calcd (%)for C ₄₂H₄₈N₂O₂F₆Co: C 64.2, H 6.1, N 3.6,

Co 7.5; found: C 64.0, H 5.9, N 3.4, Co 7.6.

[(Cp)₂Co^III][Co(⁴L_O)₂]·2CH₃CN (5): [Co(Cp)₂] (0.19 g; 1.0 mmol)was

added to a deaerated solution of 4 (0.79 g; 1.0 mmol)in CH ₂Cl₂(15 mL)

under an Ar blanketing atmosphere. After stirring for 1 h a purple pre-

cipitate was filtered off and recrystallized from a CH₃CN/diethyl ether

mixture (1:1). Single crystals of 5 suitable for X-ray crystallography were

obtained in this fashion. Yield: 0.32 g (30%); electrospray MS (CH₂Cl₂)

(pos. ion mode): m/z (%): 189 (100) [CpCo⁺]; MS (neg. ion mode): m/z:

785 [Co(⁴L_O)₂^ꢀ]; elemental analysis calcd (%)for C ₅₆H₆₄N₄F₆O₂Co₂: C

63.69, H 6.10, N 5.30, Co 11.17; found: C 63.5, H 6.0, N 4.9, Co 10.9.

configuration has been determined spectroscopically as such

without even knowing what the ligands are (the formal oxi-

dation state cannot be determined at this point). If redox-

active, non-innocent^[18]ligands are involved the formal and

spectroscopic oxidation states cannot be identical.^[12]

Experimental Section

The ligand H₂(²L_N^IP)is commercially available (Aldrich).

2-(2-Trifluoromethyl)anilino-4,6-di-tert-butylphenol, H₂(⁴L_O^IP): 3,5-Di-tert-

butylcatechol (11.1 g; 50 mmol)and 2-trifluoromethylaniline (50 mmol)

were dissolved in n-heptane (60 mL)containing NEt (0.5 mL). The so-

[Co^III(³L^I_O^SQ)₃] (6): This complex has been prepared as described in refer-

3

lution was heated to reflux for 3 h in the presence of air. The volume of

the colored reaction solution was reduced to ~30 mL by rotary evapora-

tion of the solvent, whereupon a white precipitate formed that was col-

lected by filtration and washed with a small amount of n-hexane. Yield:

7.6 g (43%); EI MS: m/z (%): 365 (100) [M⁺], 350 (28)[ M⁺ꢀCH₃]; ele-

mental analysis calcd (%)for C ₂₁H₂₆NOF₃: C 69.10, H 7.18, N 3.83;

found: C 69.2, H 7.3, N 3.7.

ence [15].

[Co^III(⁴L^I_O^SQ)₂(CH₂CN)] (7): Solid Co(ClO₄)₂·6H₂O (0.5 mmol)was added

to

a deaerated solution of NEt₃(0.2 mL)and the ligand H

₂(³L^IP)

(1.5 mmol)in CH ₃CN (30 mL)under an argon atmosphere. Then the so-

lution was heated to reflux of 30 min. On cooling in the presence of air,

X-ray quality black crystals of 7 precipitated. Yield: 0.25 g (60%); IR

(KBr): n˜ =2193 cm^ꢀ1(CꢂN); ESI MS (pos. ion; CH₂Cl₂): m/z (%): 785

(100)[ M⁺ꢀCH₂CN]; elemental analysis calcd (%)for C ₄₄H₅₀F₆N₃O₂Co:

C 64.00, H 6.10, N 5.09, Co 7.14; found: C 64.0, H 6.0, N 5.1, Co 7.3.

[Ni(⁴L^I_O^SQ)₂] (8): This diamagnetic complex has been prepared as de-

scribed for complexes 4a or 4b in reference [12] by using the ligand

H₂(⁴L^I_O^P). The complex has been characterized by single-crystal X-ray

crystallography (see below)for the purpose of comparison of the ligand

geometrical features with those of 4 of this work. EI MS: m/z: 784 [M⁺

ꢀH]; elemental analysis calcd (%)for C ₄₂H₄₈N₂O₂F₆Ni: C 64.33, H 6.17,

N 3.57, Ni 7.39; found: C 64.4, H 6.2, N 3.5, Ni 7.3.

trans-[Co(²L_N)₂] (1): [Co(CH₃CO₂)₂]·4H₂O (0.25 g; 1.0 mmol)and NEt

3

(1 mL)was added to a solution of the ligand H ₂(²L_N^IP)(0.37 g; 2.0 mmol)

in CH₃CN (10 mL). The mixture was stirred at 208C in the presence of

air for 2 h. A deep violet microcrystalline precipitate formed that was

collected by filtration, washed twice with cold CH₃CN (5 mL)and dried

in vacuo. X-ray quality crystals were grown by slow evaporation of the

solvent of a solution of 1 in toluene under strictly anaerobic conditions.

Yield: 0.34 g (81%); EI MS: m/z: 423 [M⁺]; IR (KBr): n˜ =3335,

3347 cm^ꢀ1(NH); elemental analysis calcd (%) for C ₂₄H₂₀N₄Co: C 68.09,

H 4.76, N 13.23; found: C 68.2, H 4.9, N, 13.2.

[Co(Cp)₂][Ni(⁴L_O)₂] (9): Cobaltocene (0.19 g; 1.0 mmol)was added to a

degassed solution of 8 (0.79 g; 1.0 mmol)in CH ₂Cl₂(15 mL)under an

argon blanketing atmosphere. After stirring for 3 h at 208C a green pre-

cipitate was obtained by filtration. Yield: 0.67 g (69%); ESI MS (CH₂Cl₂,

pos. ion): m/z (%): 189.2 (100) [CoCp₂⁺]; MS (neg. ion): m/z: 784.6

trans-[Co(²L_N)₂(tBu-py)] (2): A fiftyfold excess of 4-tert-butylpyridine

(28.4 mmol)was added to a stirred dark blue solution of 1 (0.24 g;

0.57 mmol)in dry toluene (7 mL)under an Ar blanketing atmosphere.

After 24 h of stirring at 208C the solution was filtered. The resulting so-

lution was allowed to stand under Ar for four months at 208C. Dark

blue-black crystals formed that were collected by filtration. Yield: 31 mg

(10%); elemental analysis calcd (%) for C ₃₃H₃₃N₅Co: C 70.96, H 5.95, N

12.54, Co 10.55; found: C 70.8, H 5.9, N 12.5, Co 10.3.

[Ni(⁴L_O)₂^ꢀ]; elemental analysis calcd (%)for

64.20, H 6.0, N 2.88, Ni 6.0, Co 6.0; found: C 64.0, H 5.9, N 3.0, Ni 6.0,

Co 6.1.

C ₅₂H₅₈N₂O₂F₆CoNi: C

X-ray crystallographic data collection and refinement of the structures:

A dark red single crystal of 1, a dark brown crystal of 5, and black crys-

tals of 2, 2a, 3, 4, 8, 8a, 8b, 8c, and 9 were coated with perfluoropolyeth-

er and mounted in the nitrogen cold stream of a Nonius Kappa-CCD dif-

fractometer equipped with a Mo-target rotating-anode X-ray source and

cis-[Co(²L_N)₂(tBu-py)]CH₃CO₂(2a): The compound has been prepared

according to the published procedure for [Co(²L_N)₂(py)]CH₃CO₂·H₂O.^[10]

Single crystals of 2a suitable for X-ray analysis were grown from a so-

lution of 2a in 4-tert-butylpyridine into which diethyl ether was allowed

to diffuse. ESI MS (pos. ion, CH₂Cl₂): m/z: 558 [M⁺]; IR (KBr): n˜ =

a

graphite monochromator (Mo_Ka, l=0.71073 ). All measurements

3316 cm^ꢀ1(N H); ¹H NMR (300 K, CD₂Cl₂, 400 MHz): d=1.27 (s, 9H;

ꢀ

were recorded at a temperature of 100(2)K. A dark red crystal of 7 was

treated the same way, but was mounted on a Siemens SMART diffrac-

tometer system equipped with a Cu fine focus tube (Cu_Ka, l=1.54178 ).

Final cell constants were obtained from least-squares fits of subsets of

several thousand strong reflections. Crystal faces of 1, 3, 4, 5, 8, 8a, 8b,

and 8c were determined and the corresponding intensity data were cor-

rected for absorption using the Gaussian-type routine embedded in

XPREP.^[19]Data set of 7 was corrected for absorption using the program

SADABS.^[20]The Siemens SHELXTL^[19]software package was used for

solution and artwork of the structure, SHELXL97^[21]was used for the re-

finement. The structures were readily solved by direct and Patterson

methods and subsequent difference Fourier techniques. All non-hydrogen

atoms were refined anisotropically. Hydrogen atoms attached to carbon

atoms were placed at calculated positions and refined as riding atoms

with isotropic displacement parameters. Crystallographic data of the

compounds are listed in Table 1. Further details are available from the

Cambridge Crystallographic Data Centre. The crystals of 8a, 8b, and 8c

represent polymorphs of 8 that have been obtained on various occasions.

These results are given in the supplementary crystallographic data only.

They will not be discussed in the text.

tert-butyl), 1.62 (s, 3H; CH₃CO₂), 7.34–7.66 (m, 22H), 10.53 ppm (brs,

ꢀ

2H; (N H···O)); elemental analysis calcd (%) for C₃₅H₃₆N₅O₂Co: C

68.06, H 5.87, N 11.34, Co 9.54; found: C 67.9, H 5.9, N 11.3, Co 9.6.

trans-[Co(²L_N)₂I] (3): This complex was prepared from a mixture of

aqueous NH₃(25%; 0.5 mL), the ligand H₂(²L_N^IP)(0.18 g; 2.0 mmol)and

CH₃CN (10 mL)to which a solution of CoI (0.16 g; 0.5 mmol)dissolved

2

in H₂O (5 mL)was added. The resulting solution was stirred in the pres-

ence of air for 3 h at 708C. A dark blue precipitate formed that was col-

lected by filtration. Single crystals suitable for X-ray crystallography

were grown from a solution of 3 in THF into which diethyl ether was al-

lowed to slowly diffuse. Yield: 0.37 g (67%); MS (EI): m/z: 550 [M⁺]; IR

(KBr): n˜ =3331, 3316 cm^ꢀ1(NH); ¹H NMR (300 K, CD₂Cl₂, 250 MHz):

d=7.10–7.24 (m, 4H), 7.43–7.53 (m, 2H), 7.60–7.64 (m, 2H), 8.11–8.13

(m, 1H), 10.01 ppm (brs, 1H); ¹³C(¹H)NMR (63 MHz, CD ₂Cl₂, 300 K):

d=119.64, 121.62, 124.60, 125.58, 125.62, 127.29, 127.72, 129.48, 130.19,

151.47, 165.28, 166.19 ppm; elemental analysis calcd (%)for

C₂₄H₂₀N₄CoI: C 52.38, H 3.66, N 10.18, Co 10.71, I 23.06; found: C 52.46,

H 3.58, N 10.12, Co 10.55, I 23.19.

[Co(⁴L_O)₂] (4): The ligand H₂(⁴L_O^IP)(2.19 g; 6 mmol)and Co(ClO ₄)₂·6H₂O

(0.73 g; 2.0 mmol)were added to a deaerated solution of methanol

(50 mL)and NEt ₃(0.8 mL). The solution was heated to reflux for 1 h and

then stirred at 208C in the presence of air for 2 h. A deep blue precipi-

CCDC-243458–243468 contains the supplementary crystallographic data

for complexes 1, 2, 2a, 3, 4, 5, 7, 8, 8a–c, whereas CCDC-243843 contains

the material for 9. These data can be obtained free of charge via

Chem. Eur. J. 2005, 11, 204 – 224

www.chemeurj.org

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

207

K. Wieghardt, F. Neese et al.

Table 1. Crystallographic data for 1, 2, 2a·H₂O, 3 ·THF, 4, 5·2CH₃CN, 7, 8, and 9·CH₃CN

1

2

2a

3

formula

M_r

space group

a []

b []

c []

a [8]

C₂₄H₂₀CoN₄

423.37

P1 (No. 2)

9.6250(6)10.207(2)15.1278(8) 8.8007(3)

10.3004(6)10.862(2) 13.0595(6) 8.8335(3)

11.5432(8)13.285(3) 15.6260(8) 32.1615(12)

74.17(1)78.19(2)90

75.86(1)79.88(2)91.61(1) 90

62.59(1)74.70(2)90

967.91(11)1379.0(5) 3085.9(3) 2500.3(2)

2

C₃₃H₃₃CoN₅

558.57

P1 (No. 2)

C₃₅H₃₈CoN₅O₃

635.63

P2₁/n (No. 14)

C₂₈H₂₈CoIN₄O

622.37

P2₁2₁2₁(No. 19)

¯

90

b [8]

g [8]

V [³]

Z

2

4

1_calcd[gcm^ꢀ3

2q_max

]

1.453

56.6

1.345

50.00

1.368

55.00

1.653

60.94

reflns collected

unique reflns

20688

4796

8481

4827

51821

7071

33248

7522

observed reflns [I>2s(I)]

3878

4402

5420

6982

parameters/restraints

271/0

9.04

0.0349/1.034

0.0745

+0.43/ꢀ0.34

341/7

6.54

0.1003/1.065

0.2457

+5.07/ꢀ0.92

415/1

6.00

0.0555/1.057

0.1149

+0.76/ꢀ0.44

322/1

19.50

0.0374/1.078

0.0784

+1.32/ꢀ0.68

ꢀ1

m(Ka)[cm

]

R1^[a]/GoF^[b]

wR2^[c][I>2s(I)]

ꢀ3

residual density [e

]

4

5

7

8

9

formula

M_r

C₄₂H₄₈CoF₆N₂O₂

785.75

P1 (No. 2)

17.0501(8)12.3198(6) 20.068(2) 9.7701(6) 10.2271(4)

17.3146(6)13.5850(6) 12.8531(121)3.8343(8)20.4853(10)

18.0229(8)17.6577(9) 33.765(3) 15.2068(102)3.6043(14)

64.51(1)93.31(1) 90

85.16(1)105.23(1)106.45(19)4.75(1)90

60.78(1)110.58(1)90

4137.9(13)2632.4(2) 8352.7(14)2048.3(2) 4945.2(4)

4

1.261

55.0

56392

18976

14329

982/0

C₅₆H₆₄Co₂F₆N₄O₂

1056.97

P1 (No. 2)

C₄₄H₅₀CoF₆N₃O₂

825.80

C2/c (No. 15)

C₄₂H₄₈F₆N₂NiO₂

785.53

P2₁/c (No. 14)

C₅₄H₆₁CoF₆N₃NiO₂

1015.70

P2₁2₁2₁(No. 19)

¯

space group

a []

b []

c []

a [8]

90

b [8]

g [8]

90

V [³]

Z

2

8

2

4

1_calcd[gcm^ꢀ3

2q_max

]

1.333

55.0

1.313

1.274

61.94

23258

6485

5488

247/0

5.37

0.0383/1.067

0.0892

+0.44/ꢀ0.37

1.364

50.0

62008

8669

6489

125.74

20233

6408

5718

523/1

reflns collected

unique reflns

observed reflns [I>2s(I)]

35495

11968

10003

676/60

6.96

0.0658/1.083

0.1796

+1.37/ꢀ1.32

parameters/restraints

617/0

7.82

ꢀ1

m(Ka)[cm

]

4.77

37.84

R1^[a]/GoF^[b]

0.0561/1.030

0.1388

+1.65/ꢀ0.54

0.0760/1.214

0.1470

+0.57/ꢀ0.48

0.0491/1.035

0.0718

+0.42/ꢀ0.31

wR2^[c][I>2s(I)]

ꢀ3

residual density [e

]

[a] Observation criterion: I>2s(I). R1=ꢀjjF_ojꢀjF_cjj/ꢀjF_oj. [b] Goodness of fit (GoF)=[ꢀ{w(F²_oꢀF²_c)²}/(nꢀp)]^1/2

.

[c] wR2=[ꢀ{w(F_o2ꢀF²_c)²}/

ꢀ{w(F²_o)²}]^1/2in which w=1/s²(F_o²)+(aP)²+bP and P=(F_o²+2F²_c)/3.

www.ccdc.cam.ac.uk/conts/retrieving.html (or from the Cambridge Crys-

tallographic Data Centre, 12 Union Road, Cambridge, CB2 1EZ, UK;

fax (+44)223–336–033; or deposit@ccdc.cam.ac.uk).

netic circular dichroism (MCD)spectra were performed by using the

PeakFit program and C/D ratios were calculated by using Equation (1).

R

Physical measurements: Electronic absorption spectra of complexes and

spectra from the spectroelectrochemical measurements were recorded on

HP 8452 A diode array spectrophotometer (range: 200–1700 nm). Cyclic

voltammograms and coulometric electrochemical experiments were per-

formed with an EG&G potentiostat/galvanostat. Temperature-dependent

(2–298 K)magnetization data were recorded with a SQUID magnetome-

ter (MPMS Quantum Design)in an external magnetic field of 1 T. The

experimental magnetic susceptibility data were corrected for underlying

diamagnetism by use of tabulated Pascalꢁs constants. X-band EPR spec-

tra were recorded with a Bruker ESP 300 spectrometer. Magnetic circu-

lar dichroism spectra were obtained on a home built instrument consist-

ing of a JASCO J-715 spectropolarimeter and an Oxford Instruments

SPECTROMAG magnetocryostat (generating magnetic fields of up to

11 T). Spectra were taken for samples dissolved in butyronitrile, which

gave high-quality glasses suitable for optical spectroscopy at cryogenic

temperatures. Simultaneous gaussian resolution of absorption and mag-

C

D

k_BT ^DeðnÞdn

n

¼

ð1Þ

R

^eðnÞdn

m_BB

n

Since we have determined the Gaussian fit from a spectrum recorded at

1.8 K and 5 T, the C/D ratio defined above was multiplied by a factor of

2.22 (determined from the appropriate S=¹= simulation)in order to ac-

2

count for partial saturation of the signal at low temperatures and high

magnetic fields.

Calculations: All calculations reported in this paper were done with the

program package ORCA.^[22]All geometry optimizations were carried out

at the BP86 level^{[23, 24]}of DFT. This functional has proved in many appli-

cations its ability to reliably predict structures of transition-metal com-

plexes. The all-electron Gaussian basis sets used were those reported by

the Ahlrichs group.^[25]Accurate triple-z valence basis sets with one set of

polarization functions on the metal and nitrogen atoms were used

(TZV(P)).^[25b]The carbon and hydrogen atoms were described by a

208

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemeurj.org

Chem. Eur. J. 2005, 11, 204 – 224

Cobalt Coordination Complexes

FULL PAPER

slightly smaller polarized split-valence SV(P)basis sets that is of double-

z quality in the valence region and contains a polarizing set of d-func-

tions on the non-hydrogen atoms.^[25a]The auxillary basis sets used to fit

the electron density were taken from the Turbomole library^[26]and were

chosen to match the orbital basis. The SCF calculations were always of

the spin-polarized type and were tightly converged (10^ꢀ7Eh in energy,

10^ꢀ6Eh in the density change and 10^ꢀ6in maximum element of the

DIIS^[27]error vector). Single-point calculations with the B3LYP function-

al^[23,28,29]were carried out at the optimized geometries. Ab initio calcula-

tions of the optical spectra were also undertaken with the ORCA pro-

gram and utilized the recently developed spectroscopy oriented configu-

[30]

ration interaction (SORCI)formalism.

In these calculations the TZVP

basis set was used on the central cobalt atom and the SV(P)basis set on

all other atoms. The selection threshold T_selwas 10^ꢀ6Eh. All other pa-

rameters were default values described in reference [30]. The lowest thir-

teen roots were calculated for the neutral species and the lowest fourteen

roots for the anion. The calculations were started with spin-restricted or-

bitals from BP86 DFT calculations. For the anion it turns out that the

highest occupied MOs are unbound and much too diffuse. Consequently,

we started the SORCI calculations for the anion from the orbitals of the

neutral species obtained at the optimized geometry of the anion. The ref-

erence space were CAS(11,8)for the neutral comlex and CAS(12,8)for

the anion. As described below this encompasses all important valence or-

bitals of both ligand and metal parentage.

Figure 3. Temperature dependence of the magnetic moment of solid 5

(top)and 2 (bottom). The solid lines represent best fits with parameters

for 5: S=1; jDj=57 cm^ꢀ1; g=2.25, and for 2: S=¹= , g_iso=2.09, c_TIP

=

2

25010^ꢀ6emu; q=ꢀ3.0 K.

Results

dark blue precipitate of [Co(²L_N)₂I] (3)was obtained. This

material is also diamagnetic as was judged from its normal

¹H NMR and ¹³C NMR spectra (Experimental Section). The

Syntheses and characterization: The reaction of the ligand

N-phenyl-o-phenylenediamine (H₂(²L_N^IP)) and [Co^II(CH₃-

CO₂)₂]·4H₂O in the ratio 2:1 in acetonitrile in the presence

of the base triethylamine and air affords deep violet micro-

crystals of trans-[Co(²L_N)₂] (1)in excellent yield. From varia-

ble-temperature measurements of the molar magnetic sus-

ꢀ

infrared spectrum (KBr disk)shows two n(N H)modes at

3331 and 3316 cm^ꢀ1.

By using the ligand 2-(2-trifluoromethyl)anilino-4,6-di-

tert-butylphenol, H₂(⁴L_O^IP), as starting material for its reac-

ceptibility it was established that 1 possesses an S=¹=

ground state (m_eff(20–298 K)=1.9 m_B). Complex 1 displays

two n(N H)stretching frequencies at 3335 and 3347 cm

(KBr disk)in the infrared region.

From an anaerobic solution of 1 in toluene, to which an

excess of 4-tert-butylpyridine had been added under an

argon blanketing atmosphere, dark blue-black crystals of the

neutral adduct trans-[Co(²L_N)₂(tBu-py)] (2)were obtained.

tion with Co(ClO₄)₂·6H₂O (2:1)in methanol and NEt in

2

3

the presence of air deep blue crystals of [Co(⁴L_O)₂] (4)were

obtained in good yield. In the temperature range 80–298 K,

complex 4 displays a temperature-independent magnetic

moment of 2.35 m_B(g=2.43, c_TIP=0.1410^ꢀ3emu)which is

ꢀ1

ꢀ

indicative of its S=¹= ground state.

2

When a deaerated solution of 4 in CH₂Cl₂reacted with

one equivalent of the reductant cobaltocene, [Co(Cp)₂],

under argon a purple precipitate of the salt [Co^III(Cp)₂]

[Co(⁴L_O)₂] (5)was obtained. The bis(acetonitrile)salt

5·2CH₃CN crystallized from CH₃CN. The crystal structure

(see below)shows that the diamagnetic cobalticenium

cation, a [Co(⁴L_O)₂]^ꢀion, and two uncoordinated acetonitrile

molecules of crystallization are present. Figure 3 displays

the temperature dependence of the effective magnetic

moment of 5. A satisfactory model for the data has been ob-

tained by using the following fit parameters: S=1, D=

57 cm^ꢀ1, and g=2.26. A very similar large zero-field splitting

parameter (D=32 cm^ꢀ1)has been reported for a bis(benzo-

dithiolato)cobalt(iii)monoanion, which also has an S=1

ground state.^[31]These results imply that the charge distribu-

tion in the monoanion is best described as [Co^III(⁴L_O^IP)₂]^ꢀ,

whereby two closed-shell dianions, (⁴L_O^IP)^2ꢀ, are bound in a

square-planar fashion to a cobalt(iii)ion affording an S_Co=1

ground state. A number of square-planar cobalt(iii)com-

plexes with an S=1 ground state containing innocent N,O-

donor ligands have been described in the literature.^[32]These

ꢀ

In the infrared spectrum two n(N H)modes are observed

at 3335 and 3346 cm^ꢀ1. The temperature dependence of the

magnetic moment of 2 is shown in Figure 3. A model com-

prising temperature-independent paramagnetism of 250

10^ꢀ6emu and a Weiss constant q of ꢀ3 K indicating weak in-

termolecular antiferromagnetic coupling, a g_isovalue of 2.09,

and an S=¹= ground state was found to yield a satisfactory

fit of the data (m_eff(70–298 K): 1.85 m_B).

2

When a mixture of the ligand H₂(²L_N^IP)and [Co ^II(CH₃-

CO₂)₂] (2:1)was stirred in the presence of air in 4- tert-butyl-

pyridine deep blue, diamagnetic crystals of cis-[Co(²L_N)₂-

(tBu-py)]CH₃CO₂·H₂O (2a)were obtained in good yield. The

material is formally the one-electron oxidation product of 2.

Previously, the complex cis-[Co^III(²L^I_N^SQ)₂(py)]CH₃CO₂·H₂O

was structurally characterized at ambient temperature^[10]

(Figure 2).

When a solution of CoI₂and the ligand H₂(²L_N^IP)(1:2)in

CH₃CN, to which a solution of aqueous ammonia (25%)

had been added, was stirred at 708C in the presence of air a

Chem. Eur. J. 2005, 11, 204 – 224

www.chemeurj.org

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

209

K. Wieghardt, F. Neese et al.

complexes also display remarkably large zero-field splittings

(jDj>50 cm^ꢀ1). This is in stark contrast to the previously

reported octahedral analogue [Co^III(³L_O^ISQ)₃] (6), which pos-

tion of their respective oxidation level, namely (²L_N^ISQ)C^ꢀand

0

(²L^I_N^BQ

)

in [Pd^II(bpy)(²L^I_N^SQ)]PF₆(S=¹= )^[13]and [Ru^II-

2

(bpy)₂(²L^I_N^BQ)](PF₆)₂,^[10]respectively, and (³L_O^IP)^2ꢀin

sess an S=²= ground state. This is typical for the presence

[Pd^II(bpy)(³L^I_O^P)], (³L_O^ISQ)C^ꢀin [Pd^II(bpy)(³L_O^ISQ)]PF₆,^[14], and

[14]

3

0

[16]

(³L^IBQ

)

in [Ni^II(tren)(³L_O^IBQ)](PF₆)_2.

Clearly, the C N,

ꢀ

of three orthogonally coordinated o-iminobenzosemiquino-

nate(1ꢀ)radicals and a diamagnetic central cobalt( iii)

ion.^[15]

ꢀ^O

ꢀ

C O, and C C distances vary with the ligand oxidation

level in a predictable manner and, conversely, the measured

distances in a given complex should allow the experimental

determination of the oxidation level of such a ligand in a

given complex.^[9]Selected bond lengths of complexes are

summarized in Table 2.

Interestingly, when the above synthesis of 4 was carried

out in acetonitrile instead of methanol in the presence of

air, a different crystalline, deep blue-violet (black)product

was obtained, namely [Co^III(⁴L_O^ISQ)₂(CH₂CN)] (7). The com-

plex is diamagnetic (S=0). The compound contains a

carbon-coordinated ion (CH₂CN)^ꢀ. Some cobalt complexes

containing such a carbon-coordinated CH₂CN^ꢀion have

been described.^[33]

The structures of neutral, mononuclear [Co(²L_N)₂] (1),

[Co(⁴L_O)₂] (4), and [Ni(⁴L_O)₂] (8)are square planar as

ꢀ

shown in Figure 5. Average C N and C C bond lengths are

given in Figure 6. Note that at this stage we do not assign

oxidation states to the central cobalt ion or the ligands in

We have also prepared the deep blue, neutral, diamagnet-

ic nickel(ii)complex [Ni ^II(⁴L_O^ISQ)₂] (8)in a completely analo-

gous fashion as described for [Ni^II(³L_O^ISQ)₂] in reference [12].

The structure of 8 has been determined by X-ray crystallog-

raphy at 100 K (see below)in order to compare the dimen-

sions of the coordinated ligands with those in 4. It has also

been possible to reduce 8 by cobaltocene affording the

green, paramagnetic salt [Co(Cp)₂][Ni(⁴L_O)₂] (9). The anion

ꢀ

complexes 1 and 4, because the experimentally observed C

ꢀ

O, C N, and C C distances do not closely resemble any pat-

tern in Figure 4. On the other hand, it is remarkable that

the ligand oxidation level of the square-planar nickel com-

plex 8 (Figure 5)is in excellent agreement with the notion

that two monoanionic p-radical ligands (⁴L_O^ISQ)C^ꢀare present

and, therefore, the nickel ion possesses an experimentally

determined spectroscopic oxidation state of +II (d⁸). Very

similar dimensions of the p radical ligands have been report-

ed in reference [12] for [M(³L_O^ISQ)₂] (M=Ni^II, Pd^II, Pt^II)com-

plexes. Thus, the charge distribution in 8 is best described as

[Ni^II(³L_O^ISQ)₂]. This is not the case for the square-planar

cobalt complexes 1, 4, and [Co(¹L_N)₂] (Figure 1)for which

of complex 9 possesses an S=¹= ground state (m_eff(70–

2

298 K)=1.8 m_B)and displays a charge distribution as in

[Ni^II(⁴L^I_O^SQ)(⁴L^I_O^P)]^ꢀQ[Ni^II(⁴L_O^IP)(⁴L^I_O^SQ)]^ꢀ.

ꢀ

Crystal structures: Figure 4 shows the C N, C O, and C C

bond lengths in N,N- and N,O-coordinated ligands as a func-

ꢀ

the observed C N, C O, and C C distances are in better

agreement with values of the arithmetic mean between

those of an aromatic dianion and the corresponding p-radi-

cal monoanion (Scheme 3). Since the two ligands in the neu-

tral molecules are crystallographically identical and no indi-

cation for static disorder has been detected, it appears that

the unpaired electron is delocalized over both ligands.

We

propose

the

following

charge

distributions:

[Co^III(¹L^I_N^SQ)(¹L^I_N^P)]for [Co(¹L_N)₂], [Co^III(²L_N^ISQ)(²L^I_N^P)] for 1,

and [Co^III(⁴L_O^ISQ)( L_O)] for 4. The observed C O, C N, and

4

IP

ꢀ

C C distances in these compounds closely resemble those

which were calculated from the arithmetic mean of one

mono- and one dianion.

Complex 5 contains the cobaltocenium cation and the

[Co(⁴L_O)₂]^ꢀion shown in Figure 7. The overall geometrical

features of the monoanion in crystals of 5 are similar to

those of the neutral species in crystals of 4; both are square

ꢀ

planar species. As shown in Figure 7 the C N, and C O

bond lengths in the monoanion of 5 are slightly longer than

ꢀ

those in the neutral species in 4 (Figure 6). The C C bond

lengths of the aminophenol ring in 5 and 4 differ also signifi-

ꢀ

cantly. In complex 5, six C C bonds are equidistant within

experimental error at 1.40ꢁ0.01 ; this clearly indicates the

presence of two aromatic, closed-shell, dianionic (⁴L_O^IP)^2ꢀli-

ꢀ

gands. In 4 these six C C distances show quinoid-type dis-

ꢀ

tortions with two alternating shorter C C bonds. Without

any ambiguity the oxidation level of the ligands in 5 are

ꢀ

Figure 4. Average C O, C N, and C C distances in in complexes con-

taining (²L_N)^nꢀor (³L_O)^nꢀligands (n=2ꢀ,1ꢀ,0).

therefore aromatic dianions; this renders the oxidation state

210

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemeurj.org

Chem. Eur. J. 2005, 11, 204 – 224

Cobalt Coordination Complexes

Table 2. Selected bond lengths [].

FULL PAPER

Complex 1^[a]

ꢀ

Co N1

1.826(2)C2

1.855(2)C3

1.354(2)C4

1.411(2)C5

1.424(2)

C3

ꢀ

C4

1.379(3)C6

1.406(3)N7

1.379(3)C8

1.408(2)C8

N7

ꢀ

C8

1.376(2)C9

1.431(2)C10

1.392(3)C11

1.396(2)C12

C10

1.391(3)

1.390(3)

1.387(3)

1.393(3)

ꢀ

Co N7

C11

C12

ꢀ

C13

ꢀ

N1 C1

C5

ꢀ

C1 C2

C6

C9

C13

ꢀ

C1 C6

Complex 2

ꢀ

Co N41

2.076(4)C1

1.864(3)C2

1.898(4)C3

1.891(4)C4

1.866(4)C5

1.339(6)C1

C2

ꢀ

C3

1.415(7)C6

1.370(9)N7

1.395(10)N27

1.377(9)C25

1.411(8)C24

1.434(7)C24

N7

1.357(6)C23

1.424(6)C21

1.358(6)C21

1.402(7)C21

1.385(8)

C22

1.365(7)

1.410(7)

1.437(7)

1.344(6)

ꢀ

C22

Co N1

C8

C26

ꢀ

Co N7

C4

C26

N21

ꢀ

Co N27

C5

C26

C25

C23

ꢀ

Co N21

C6

ꢀ

N1 C1

C6

1.398(8)

Complex 2a

ꢀ

Co N1

1.845(2)C1

1.853(2)C1

1.934(2)C2

1.940(2)C3

1.989(2)C4

1.323(3)C5

C2

ꢀ

C6

1.429(4)C6

1.443(4)N7

1.357(4)N21

1.434(4)C21

1.364(4)C21

1.421(4)C22

N7

1.345(3)C23

1.447(3)C24

1.318(4)C25

1.431(4)C26

1.452(4)N27

1.367(4)

C24

1.420(4)

1.370(4)

1.428(4)

1.342(4)

1.442(4)

ꢀ

C25

Co N21

C8

C21

ꢀ

Co N7

C3

C26

ꢀ

N27

Co N27

C4

C22

C26

C23

ꢀ

C28

Co N40

C5

ꢀ

N1 C1

C6

Complex 3

ꢀ

Co I

2.5745(4)C2

1.856(2)C3

1.860(2)C4

1.882(3)C5

1.887(2)C6

1.328(4)N7

1.420(4)C8

1.433(4)

C3

ꢀ

C4

1.362(4)C8

1.418(4)C9

1.365(4)C10

1.417(4)C11

1.347(4)C12

1.434(4)N21

1.381(5)C21

C9

1.388(5)C21

1.382(5)C22

1.385(6)C23

1.379(6)C24

1.397(5)C25

1.329(4)C26

1.422(4)N27

C26

1.440(4)

1.362(4)

1.421(5)

1.370(4)

1.414(4)

1.338(4)

1.432(4)

ꢀ

C23

Co N1

C10

C11

ꢀ

Co N21

C5

C24

ꢀ

C25

Co N7

C6

C12

C13

C21

C22

ꢀ

Co N27

N7

ꢀ

C8

C26

N27

ꢀ

N C1

ꢀ

C13

ꢀ

C28

C1 C2

ꢀ

C1 C6

Complex 4^[a]

ꢀ

Co O1

1.822(2)C1

1.823(2)C2

1.832(2)C3

1.840(2)C4

1.329(3)C5

1.415(4)C6

C6

ꢀ

C3

1.424(4)N7

1.384(4)O26

1.420(4)C26

1.382(4)C26

1.402(4)C27

1.373(3)C28

C8

1.434(3)C29

1.327(3)C30

1.419(4)C31

1.420(4)N32

1.387(4)

C30

1.378(4)

1.411(4)

1.374(3)

1.432(3)

ꢀ

C31

Co O26

C26

C27

ꢀ

Co N7

C4

N32

C33

ꢀ

Co N32

C5

C31

C28

C29

ꢀ

O1 C1

C6

ꢀ

C1 C2

N7

1.420(4)

Complex 5^[a]

ꢀ

Co O1

1.830(2)C1

1.837(3)C1

1.339(3)C2

C6

ꢀ

C2

1.402(4)C3

1.404(4)C4

1.402(4)C5

C4

ꢀ

C5

1.397(4)C6

1.394(4)N7

1.393(4)

N7

ꢀ

C8

1.388(4)

1.413(4)

ꢀ

Co N7

ꢀ

C3

ꢀ

C6

O1 C1

Complex 7^[b]

ꢀ

Co O1

1.845(3)C2

1.854(3)C3

1.855(3)C4

1.859(3)C5

2.004(5)C6

1.305(5)N7

1.419(6)C8

1.428(6)C8

C3

ꢀ

C4

1.367(6)C9

1.431(6)C10

1.364(6)C11

1.410(6)C12

1.358(5)O31

1.441(5)C31

1.384(6)C31

1.394(6)C32

C10

1.385(6)C33

1.383(7)C34

1.382(7)C36

1.381(6)N37

1.305(5)C60

1.429(6)C61

1.430(6)

C34

1.430(6)

1.368(6)

1.346(5)

1.422(5)

1.446(7)

1.136(6)

ꢀ

C35

Co N7

C11

C12

ꢀ

Co O31

C5

N37

ꢀ

C38

Co N37

C6

C13

C31

C36

C32

ꢀ

C61

Co C60

N7

ꢀ

C8

ꢀ

N62

O1 C1

ꢀ

C13

ꢀ

C1 C6

ꢀ

C1 C2

C9

C33

1.372(6)

Complex 8

ꢀ

Ni O1

ꢀ

Ni N8

1.8365(9)C2

1.844(1)C2

1.315(2)C3

C3

ꢀ

C7

1.425(2)C4

1.432(2)C5

1.385(2)C6

C5

ꢀ

C6

1.430(2)C7

1.376(2)N8

1.418(2)

N8

ꢀ

C9

1.355(2)

1.428(2)

ꢀ

C4

ꢀ

C7

O1 C2

Complex 9

ꢀ

Ni N38

ꢀ

Ni N8

1.831(3)C2

1.843(3)C3

1.844(2)C4

1.846(3)C5

1.329(5)C6

1.407(6)C7

C3

ꢀ

C4

1.428(5)N8

1.384(6)O31

1.407(6)C32

1.386(6)C32

1.410(6)C33

1.378(5)C34

C9

1.421(5)C35

1.345(5)C36

1.403(6)C37

1.422(6)N38

1.411(6)

C36

1.401(6)

1.394(6)

1.382(5)

1.429(5)

ꢀ

C37

C32

C33

ꢀ

Ni O1

C5

N38

C39

ꢀ

Ni O31

C6

C37

C34

C35

ꢀ

O1 C2

C7

ꢀ

C2 C7

N8

1.386(6)

[a] Data for one crystallographically independent molecule is given only. [b] Angle: Co-C60-C61: 112.7(3)8.

of the central cobalt ion +III. It is interesting to note that

the ligand dimensions in 5 are identical to those observed

for diamagnetic, square-planar [Pd^II(bpy)(³L_O^IP)]⁰.^[14]The

charge distribution in the monoanion of 5 is best formulated

as [Co^III(⁴L_O^IP)₂]^ꢀ. It is also worth noting that the average

ꢀ

Co N, and Co O bond lengths in 4 and 5 are 1.835, 1.822

and 1.837, 1.830 , respectively, and are identical within ex-

perimental error of ꢁ0.006 (3s). The reduction of the

Chem. Eur. J. 2005, 11, 204 – 224

www.chemeurj.org

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

211

K. Wieghardt, F. Neese et al.

ꢀ

Figure 6. Average C O, C N, and C C distances in of the neutral mol-

ecules in 1 (in red)and in [Ni( ²L)₂] reference [9a] (in black). Values for

complex 4 (middle)and 8 (bottom)are also given. The average experi-

mental error (3s)is ~ꢁ0.01 .

In each case the cobalt ion is in a square-base pyramidal

environment composed of two bidentate N,N- or N,O-coor-

dinated organic ligands (L_N)^nꢀor (L_O)^nꢀin basal positions

and a single halide ion or pyridine as the fifth ligand in the

apical position. As mentioned earlier, diamagnetic com-

plexes of this type have been structurally characterized pre-

viously: [Co^III(¹L_N^ISQ)₂Cl],^[3]cis-[Co^III(²L_N^ISQ)₂(py)]CH₃CO₂

·H₂O,^[11][Co^III(¹L^I_N^SQ)₂(PPh₃)]PF₆,^[7][Co^III(¹L_N^ISQ)₂(py)]Cl,^[8]

and [Co^III(³L^I_O^SQ)₂X] (X=Cl, I).^[10]In all of these cases the

ꢀ

C N, C O, and C C bond lengths unambiguously indicate

that the bidentate ligands possess benzosemiquino-

a

nate(1ꢀ) p-radical oxidation level and, consequently, the

central cobalt ion invariably possess a +III (d⁶, low spin,

S_Co=0)spectroscopic oxidation state.

The structure of the neutral, paramagnetic species trans-

[Co(²L_N)₂(tBu-py)] (2)is therefore of great interest, because

a different charge distribution must prevail, while the coor-

dination polyhedron remains square-base pyramidal. The

charge distribution is either trans-[Co^III(²L_N^ISQ)(²L_N^IP)(tBu-py)]

containing a diamagnetic Co^IIIion (d⁶, low spin), a paramag-

netic benzosemiquinonate(1ꢀ) p-radical (²L_N^ISQ)C^ꢀ, and a dia-

Figure 5. Structure of one of the crystallographically independent mole-

4

cules of [Co(²L_N)₂]⁰in crystals of 1 (top)and of [Co( L_O)₂] in crystals of 4

(middle), and of [Ni^II(⁴L)₂] in crystals of 8 (bottom).

magnetic (²L_N^IP)^2ꢀion, or, alternatively, trans-[Co^II(²L^I_N^SQ

)

2

7

neutral species in 4 yielding 5 is therefore unlikely to be a

metal-centered process involving Co^III+e^ꢀ!Co^II.

(tBu-py)] containing a low-spin cobalt(ii)ion (d , S_Co=¹= )

2

and two antiferromagnetically coupled benzosemiquinonate

ꢀ

In the following we describe the structures of the five-co-

ordinate species 2, 2a, 3, and 7 which are shown in Fig-

p radicals (²L_N^ISQ)C . The C N and C C bond lengths in 2

(Figure 8)indicate that the former description is more ap-

propriate than the latter, because the ligand dimensions re-

semble those of an averaged mono- and dianion, (²L_N^ISQ)C^ꢀ

and (²L^I_N^P)^2ꢀ. In contrast, these bonds in diamagnetic cis-

ꢀ

ures 8–10. Average C N, C O, and C C bond lengths of the

ligands are given in Figure 8 for 2 and 2a, Figure 2 for 3,

and Figure 10 for 7.

212

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemeurj.org

Chem. Eur. J. 2005, 11, 204 – 224

Cobalt Coordination Complexes

FULL PAPER

ꢀ

Figure 8. Average C N, C O, and C C distances [] in a diamagnetic

neutral molecule of [Co^III(⁴L)₂X] (X=Cl, I)from reference [10] and

[Co(²L_N)₂(tBu-py)] in crystals of 2 (middle)and cis-[Co(²L_N)₂(tBu-py)]⁺

in 2a (bottom).

two (⁴L^I_O^SQ)C^ꢀp radicals. As shown in Figure 6 these bond

lengths differ slightly from those in the cobalt complex 4;

this indicates that the respective ligand oxidation levels in

both species differ (and, concomitantly, the metal oxidation

states)and we assign them as [Ni ^II(⁴L_O^ISQ)₂] in 8 and

[Co^III(⁴L^I_O^SQ)(L_O^IP) ] in4.

Figure 7. Structures of the monoanions [Co^III(⁴L)₂]^ꢀin crystals of 5 (top)

ꢀ

Similarly, the C O, C N, and C C bond lengths in the

structures of the monoanions in crystals of 5 and 9 shown in

Figure 7 differ slightly. Both are square-planar species, but

the dimensions of the ligands indicate the presence of two

o-iminophenolate(2ꢀ)dianions in 5 as shown above, where-

and [Ni^II(⁴L)(⁴L)]^ꢀin crystals of 9 (bottom). The error of the C N, C O,

ꢀ

and C C bond lengths is ꢁ0.01 (3s)for 5 and ꢁ0.015 for 9.

[Co^II(²L^I_N^SQ)₂(py)]^+[11]are in excellent agreement with the

presence of two (²L_N^ISQ)C^ꢀp radicals. They are the same as in

[Co^III(²L^I_N^SQ)₂I] (3)(Figure 2).

ꢀ

as in 9 the C O, C N, and C C distances are more in agree-

ment with the average between one o-iminobenzosemiquin-

onate(1ꢀ)and an o-iminophenolate(2ꢀ)and a central nick-

The structure of the neutral complex in crystals of 7 clear-

ly indicates the presence of two ligand p radicals (⁴L_O^ISQ)C^ꢀ

and the C-coordinated monoanion of acetonitrile,

(CH₂CN)^ꢀ, in the apical position (Figures 9 and 10). The

geometrical features of the benzosemiquinonate(1ꢀ)radi-

cals are, within experimental error, identical with those re-

ported previously for [Co^III(³L_O^ISQ)X] (X=Cl, I).^[10]A few

complexes containing C-coordinated (CH₂CN)^ꢀions have

el(ii)ion: [Ni (⁴L_O^ISQ)(⁴L^I_O^P)]^ꢀ.

II

ꢀ

It is also interesting to note that the corresponding M O

ꢀ

and M N bond lengths of the neutral complexes 4 and 8 are

slightly shorter in the cobalt complex 4. The same is true for

the monoanions in 5 and 9, in which these bonds are also

shorter in the cobalt complex 5. If both metal ions possess

the same oxidation state +II, we would expect that the low-

II

been structurally characterized.^[33]The C N, C O, and C C

spin Co X bond lengths to be longer than those of the cor-

ꢀ

1ꢀ

II

bond lengths of the (⁴L^I_O^SQ

)

ligands found here in five-coor-

responding Ni X bonds.

ꢀ

dinate 7 are also identical to those reported in octahedral

[Co^III(³L^I_O^SQ)₃] with an S=²= ground state.^[15]

Electro- and spectroelectrochemistry: Cyclic voltammo-

grams (CV)of complexes 1, 4, and 8 were recorded in

CH₂Cl₂containing 0.10m [nBu₄N]PF₆as a supporting elec-

trolyte at a glassy carbon working electrode and an Ag/

3

The nickel complex 8 contains square-planar, diamagnetic,

neutral [Ni^II(⁴L_O^ISQ)₂] molecules (Figures 5 and 6). The C O,

ꢀ

C N, and C C bond lengths clearly show the presence of

Chem. Eur. J. 2005, 11, 204 – 224

www.chemeurj.org

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

213

K. Wieghardt, F. Neese et al.

Figure 9. Structures of the ion pair in crystals of 2a and of the neutral complexes in crystals of 2, 3, and 7.

Table 3. Redox potentials of complexes versus Fc⁺/Fc.^[a]

E¹₁[V] E²₁[V]

E³₁[V]

=

2

=

2

=

2

+1Ð0

0Ðꢀ1

ꢀ1Ðꢀ2

ꢀ1.98

[Co(¹L_N)₂]^[b]

1

4

8

ꢀ0.32

ꢀ0.95

ꢀ1.03

ꢀ0.85

ꢀ1.01

ꢀ0.50

ꢀ0.20

+0.04(2e)

ꢀ1.74

[a] Conditions: glassy carbon working electrode, Ag/AgNO₃reference

electrode; 208C; CH₂Cl₂containing 0.10m [(nBu)₄N]PF₆supporting elec-

trolyte; [complex]~10^ꢀ3m: scan rate 100 mVs^ꢀ1. [b] Reference [1].

Figure 10. Bond lengths [] of the neutral complex [Co^III(⁴L)₂(CH₂CN)]

in crystals of 7. Experimental error: ꢁ0.02 (3s).

electron oxidation at E¹₁=ꢀ0.50 V and a reversible one-

=

2

AgNO₃reference electrode. Ferrocene was used as an inter-

nal standard; all potentials are referenced versus the ferro-

cenium/ferrocene couple (Fc⁺/Fc). Table 3 summarizes the

results.

electron reduction at E²₁=ꢀ1.03 V. Similar waves have been

=

2

reorted by Balch and Holm^[1]for [Co(¹L_N)₂]; they report a

second one-electron reduction at E³₁=ꢀ1.98 V for the

=

2

mono-/dianion couple.

The CV of 1 displays two reversible one-electron transfer

waves in the potential range ꢀ1.5 to 0.0 V; a reversible one-

The CV of 4 is very similar; again a reversible one-elec-

tron oxidation (E¹₁=ꢀ0.20 V)and a reversible one-electron

=

2

214

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemeurj.org

Chem. Eur. J. 2005, 11, 204 – 224

Cobalt Coordination Complexes

FULL PAPER

reduction (E²₁=ꢀ0.85 V)are observed. Since we have struc-

Table 4. Electronic spectra of complexes.

=

2

l_max[nm] (10⁴e [m^ꢀ1cm^ꢀ1])

turally characterized the neutral species 4 and its monoan-

ion in 5 and shown that both contain a cobalt(iii)ion, we

assign the charge distribution as shown in Equations (2)and

(3).

Solvent

1

CH₃CN

CH₃CN^a)

CH₂Cl₂

360sh(0.85), 625(3.0), 816(1.5), 1230(0.40)

445(0.9), 530(0.9), 740(1.0), 870(0.80)

350sh(2.6), 580(1.2), 800(1.5)

324sh(0.8), 628(2.5), 819(1.3), 1230(0.3)

280(1.9), 463(0.4), 594(1.6), 786(1.3)

269(3.2), 397(0.8), 436(0.85), 514(1.2),

636(2.6), 786(1.9)

1^ꢀ

1⁺

2

2a

3

ꢀe

½Coð L_NÞ₂

½Co^IIIð L^ISQÞð L^IPÞ

½Co^IIIð L^IPÞ

2

0

2

ꢀ

^þG_þe

H

G_þe

H

ð2Þ

ð3Þ

2

N

ꢀe

^þG_þe

H

G_þe

H

4

0

4

½Co^IIIð L^ISQÞð L^IPÞ

½Co^IIIð L^IPÞ

4

CH₂Cl₂

₂Cl₂

CH₂Cl₂

300sh(2.0), 676(1.6), 916(2.5), 1600(0.25)

400sh(0.6), 535(0.35), 720sh(0.33), 827(0.7)

300sh(1.28), 350sh(0.5), 620(0.7), 960(1.1)

280(0.3), 440sh, 505(0.9), 629(0.34),

831(0.35)

½Coð L_OÞ₂

2

O

[a]

4^ꢀ(5)CH

[a]

4⁺

7

The oxidized and reduced species of 1 and 4 are stable in

solution on the timescale of a coulometric experiment.

Therefore, it has been possible to record the electronic spec-

tra of the monocations of the neutral species 1 and 4, and of

their corresponding monoanions. The results are shown in

Figure 11 and summarized in Table 4. From the crystal struc-

CH₃CN

8

9

CH₂Cl₂

350(3.0), 600(0.2), 900(1.6)

382sh, 608(0.1), 885(0.3), 1340(1.0)

337(0.5), 420(0.3), 588(1.9), 763(1.2),

1135(0.4)

260(1.1), 410(0.2), 490(0.4), 607(1.6),

751(0.8)

[Co(¹L_N)₂]^[b]DMF,

DMSO

[Co(¹L_N)₂I]^[c]DMF

[a] Electrochemically generated in solutions containing 0.10m

[(nBu)₄N]PF₆supporting electrolyte; [b] Reference [1]. [c] Ref-

erence [15].

ꢀe

½Ni^IIð L^I_O^SQÞ

½Ni^IIð L^ISQÞð L^I_O^PÞ

½Ni^IIð L^IPÞ

4

2ꢀ

^ꢀG_þe

H

ð4Þ

G_þe

H

2

O

Controlled-potential coulometry at +0.6 V shows that 8

can be oxidized in a nearly reversible one-step, two-electron

oxidation yielding the [Ni^II(⁴L_O^IBQ)₂]²⁺ion.

X-band EPR spectra: The X-band EPR spectra of four-co-

ordinate 1 at 10 K in CH₂Cl₂/toluene (1:1 v/v)and of five-

coordinate 2 in 4-tert-butylpyridine/toluene at 10 K shown in

Figure 12 confirm the S=¹= ground state of both species.

2

The spectrum of 1 has been successfully simulated by

using the following parameters: g_x=1.9906, g_y=2.0508, g_z=

Figure 11. Electronic spectra of 1 and its electrochemically generated

monoanion and monocation (top), respectively, in CH₃CN solution

(0.10m [(nBu)₄N]PF₆)and, correspondingly, those of 4, 4^ꢀ, and 4⁺

(bottom).

ture of 5, which contains the square-planar, paramagnetic

ion [Co^III(⁴L_O^IP)₂]^ꢀ, we conclude that the electrochemically

generated ion [Co^III(²L_N^IP)₂]^ꢀis also square-planar and para-

magnetic (S=1). The four-coordinate monocations have not

yet been isolated as salts or structurally characterized; only

the five-coordinate species 2, 2a, 3, and 7 are synthetically

accessible.

The CV of

8 is very similar to that reported for

[Ni^II(³L^I_O^SQ)₂].^[12]Two successive one-electron reductions of 8

in CH₂Cl₂(0.10m [(n-Bu)₄N]PF₆)are observed at ꢀ1.01 and

ꢀ1.74 V. These processes have been assigned as ligand-cen-

tered processes as in Equation (4).

Figure 12. X-band EPR spectra of 1 (top)and 2 (bottom). Conditions for

1: CH₃CN at 30 K; frequency 9.6351 GHz; power 505.3 mW; modulation

amplitude 10 G. Conditions for 2: 4-tert-butylpyridine/toluene mixture

(1:3)at 10 K; frequency 9.6345 GHz; power 20 mW; modulation ampli-

tude 10 G.

Chem. Eur. J. 2005, 11, 204 – 224

www.chemeurj.org

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

215

K. Wieghardt, F. Neese et al.

2.8100 (g_iso=2.3140)with ⁵⁹Co hyperfine coupling constants

A_xx=24.0, A_yy=0, A_zz=42 G (line width W_x=20 G, W_y=

37 G, W_z=38 G). The spectrum indicates that the unpaired

electron resides in a metal d orbital; it resembles closely

those reported for many square-planar Co^II(low spin d⁷)

complexes.^[6]On the other hand, a cobalt(iii)ion in a

square-planar ligand field possesses an S_Co=1 local spin

state;^[31,32]strong antiferromagnetic coupling then yields an

a single very intense band at 900 nm (e=4.610⁴m^ꢀ1cm^ꢀ1)

and 1340 nm (1.010⁴m^ꢀ1cm^ꢀ1), respectively.^[12]For compar-

ison, the corresponding complexes [Ni^II(²L_N^ISQ)₂] and

[Ni^II(²L^I_N^SQ)(²L_N^IP)]^ꢀdisplay these bands at 839 (4.0

10⁴m^ꢀ1cm^ꢀ1)and at 1119 nm (1.710 ⁴m^ꢀ1cm^ꢀ1),^[9]respec-

tively.

For the neutral species [Ni^II(¹L_N^ISQ)₂], it has been establish-

ed^[9]that the upper valence region contains four doubly oc-

cupied MOs that are predominantly centered on the central

Ni^IIion (d⁸). The LUMO+1 orbital is dominated by the Ni

d_x₂_ꢀy₂MO, which is strongly s antibonding with the ligands.

The HOMO–LUMO transition 1b_1u!2b_2grepresents the in-

tense LLCT band. No other LMCT bands are observed in

the visible region. In the corresponding monoanion the

former LUMO (2b_2g)becomes the SOMO, which has ~15%

Ni 3d_xzcharacter; the transition 1b_1u!2b_2gis again electric

dipole and spin-allowed and represents a ligand-to-ligand in-

tervalence charge-transfer band (LLIVCT).

S=¹= ground state again with an unpaired electron in the

2

same metal d orbital as above. Thus, it is not straightfor-

wardly possible to discern between A and B in Scheme 2 by

EPR spectroscopy.

The rhombic signal of five-coordinate 2 has been satisfac-

torily modeled by using the following parameters: g_x=

1.9694, g_y=2.1275, g_z=2.3100 (g_iso=2.1402)with ⁵⁹Co hyper-

fine coupling constants A_xx=84.8, A_yy=15.1, A_zz=0 G

(W_x=34.7, W_y=76.0, W_z=190 G).

The rhombic X-band EPR spectrum of 4 in frozen CH₂Cl₂

solution at 90 K is very similar to that of 1: g_x=1.97, g_y=

Figure 13 exhibits the spectra of 2, 2a, and 3; the similari-

ty of these is quite remarkable. Complexes 1 (Figure 11)and

2 exhibit each a LLCT band at 1230 nm with an intensity of

2.03, g_z=3.11 (g_iso=2.43). The S=¹= signal displays no re-

2

solvable ⁵⁹Co hyperfine splitting.

The X-band EPR spectrum of 9 in frozen CH₂Cl₂at 90 K

exhibits a rhombic signal (g_x=2.075, g_y=2.014, g_z=2.046

(g_iso=2.045)), which resembles closely those reported for

many square-planar bis(dioxolene)nickel(ii)monoanions

with an S=¹= ground state.^[9]These spectra have recently

2

been interpreted^[9]in terms of a central, diamagnetic nick-

8

el(ii)ion (d ), a (L_O^IP)^2ꢀion and a p-radical ion (L^I_O^SQ)C^ꢀ, in

which the unpaired electron is delocalized over both ligands.

The SOMO of these compounds, b_2gunder D_2hsymmetry, is

basically the antisymmetric combination of the SOMO of

the free semiquinonate(1ꢀ)ligand; it transforms “gerade”

under inversion and, therefore, mixes with the out-of-plane

d_xzorbital of the nickel(ii)and thereby acquires some metal

d character (~15%), which in turn gives rise to a sizeable

2

nickel hyperfine coupling. Since the ground state B_2greadi-

Figure 13. Electronic spectra of 2, 2a, and 3 in CH₂Cl₂.

ly mixes with relatively low-lying d–d excited states, it has a

reasonably large orbital angular momentum that manifests

itself in a relatively large g anisotropy.^[9,14]The charge

distribution in the monoanion 9 is therefore correctly

~310³m^ꢀ1cm^ꢀ1. Thus, increasing the coordination number

from four to five on going from 1 to 2 does not influence

the electronic spectrum greatly. In fact, the spectral differen-

ces between 1 and 2 in solution are so small that the solid-

state spectra of both species were also recorded and con-

firmed their similarity. In contrast, the spectra of five-coor-

dinate 2a and its formally one-electron oxidized analogue 3

(Figure 13)and, similarly, that of 7 do not display an intense

absorption maximum >900 nm. Instead, two intense

maxima at ~800 and ~600 nm are observed that are as-

signed LMCT bands.

described

by

the

two

resonance

structures:

[Ni^II(⁴L^I_O^P)(⁴L_O^ISQ)]^ꢀQ[Ni^II(⁴L_O^ISQ)(⁴L^I_O^P)]^ꢀ.

Electronic spectra: The electronic spectra of complexes

have been recorded in the range 280–2000 nm in CH₂Cl₂or

CH₃CN. In some cases the species were generated electro-

chemically by means of controlled potential coulometry in

solutions containing 0.1m [N(n-Bu)₄]PF₆as a supporting

electrolyte. The results are summarized in Table 4.

Complexes 1, 2, 2a, 3, 4, 5, 7, 8, and 9 are all highly col-

ored and display a varying number of intense absorption

maxima (e>510³m^ꢀ1cm^ꢀ1)in the visible and near infrared

regions. These absorption maxima do not represent d–d

transitions; they are spin-allowed ligand-to-metal (LMCT)

or ligand-to-ligand charge-transfer bands (LLCT).

Interestingly, the spectrum of the electrochemically gener-

ated four-coordinate monoanion of 1 in CH₃CN exhibits

four intense (~310³m^ꢀ1cm^ꢀ1)LMCT bands at 445, 530,

740 and 870 nm.

The spectrum of square-planar 4 shown in Figure 11 is

very similar to that of 1:

a band at 1600 nm (2.5

It is instructive to first discuss the spectra of neutral

nickel complex 8 and its monoanion 9, both of which exhibit

10³m^ꢀ1cm^ꢀ1), two intense LMCT maxima at 916 and

676 nm, and a shoulder at ~300nm are observed. The spec-

216

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemeurj.org

Chem. Eur. J. 2005, 11, 204 – 224

Cobalt Coordination Complexes

FULL PAPER

trum of the monoanion in 5 is interesting; it displays two in-

tense maxima at 827 and 535 nm and two shoulders at 720

and ~390 nm.

the 2b_3gorbitals remain singly occupied yielding a spin-trip-

let ground state for the monoanion (as is observed). The

energy gap between these two SOMOs is found to be too

small to stabilize the alternative spin-singlet state as had

been suggested for some bis(benzodithiolato)cobalt(iii)com-

plexes.^[34]The first spin-triplet state is found to be

~10 kcalmol^ꢀ1(0.43 eV)lower in energy than the corre-

sponding spin singlet state in the B3LYP calculations.

The 2b_2gorbital is an out-of-plane orbital with almost

equal contributions from the Co

Calculations

Bonding schemes: As shown in Table 5 the optimized calcu-

lated structures of [M(¹L_N)₂]^0,1ꢀ(M=Co, Ni)are in excellent

agreement with the experimental findings. The slight overes-

3d_xzand the ligand b_2gfragment

Table 5. Calculated and experimental (in parentheses)bond lengths [] in [Co( ¹L_N)₂], [Co(¹L_N)₂]^ꢀ, and

[Ni(¹L_N)₂], [Ni(¹L)₂]^ꢀ.

orbitals (Table 6). The 2b_3gorbi-

tal, on the other hand, is pre-

dominantly Co 3d_yzin charac-

ter; it undergoes little mixing

with the low-lying ligand b_3g

fragment orbital. The remaining

two Co 3d_x₂_ꢀy₂(1a_g)and 3d

z²

[Co¹L_N)₂]^[a]

[Co(¹L_N)₂]^ꢀ[b]

[Ni(¹L)₂]^[c]

[Ni(¹L_N)₂]^ꢀ[c]

(2a_g)orbitals do not strongly in-

teract with ligand orbitals and

are placed at comparatively

lower energies. The d-orbital

splitting in [Co(¹L_N)₂]^1ꢀis thus

ꢀ

M N

1.841 (1.841(2))

1.367 (1.365(2))

1.446 (1.424(2))

1.417 (1.408(2))

1.399 (1.379(3))

1.419 (1.406(3))

1.866 (1.833(3))

1.377 (1.369(4))

1.438 (1.417(4))

1.411 (1.391(4))

1.415 (1.398(4))

1.406 (1.401(4))

1.841 (1.822)

1.350 (1.350)

1.446 (1.429)

1.416 (1.425)

1.386 (1.383)

1.421 (1.423)

1.857 (1.843(3))

1.361 (1.378(5))

1.427 (1.407(6))

1.410 (1.410(6))

1.403 (1.389(6))

1.407 (1.405(6))

ꢀ

C N

ꢀ

C1 C2

ꢀ

C2 C3

ꢀ

C3 C4

ꢀ

3d_x₂_ꢀy₂

,

3d_z₂<3d_xz, 3d_yz<3d_xy,

C4 C5

[a] From this work for 1. [b] From this work for 5. [c] From reference [9a,b] for [Ni(¹L)₂] in which the ligand

which is nicely compatible with

has two tertiary butyl groups at C3 and C5.

ligand-field theory. Analysis of

ꢀ

timation of the Co N and Ni N distances is typical of pres-

ent day DFT functionals. However, the metrical parameters

for the organic ligands are very accurately reproduced by

the calculations with the typical error in computed bond

lengths not exceeding ~0.02 .

For the MO description of the complexes [Co(¹L_N)₂]^0,1ꢀ

within the D_2hpoint group we choose the z axis to be along

the normal of the planar complexes, the x axis along the

long axis, and the y axis along the short axis of the com-

plexes as shown in Figure 14. The qualitative bonding

scheme derived from the spin-unrestricted B3LYP DFT cal-

culations are also shown in Figure 14, in which the spin-up

and spin-down Kohn–Sham MOs are plotted in order of in-

creasing energy. Since the MO scheme qualitatively does

not change on going from the neutral [Co(¹L_N)₂] species to

its one-electron reduced monoanionic form, it is only given

for [Co(¹L_N)₂]^1ꢀ. The compositions of selected orbitals are

summarized in Table 6. We will argue below that the de-

scription of neutral [Co(¹L_N)₂]⁰complex in terms of a single

configuration is slightly oversimplified, but for all practical

purposes of this section the single-determinant description

is adequate. The ground-state electronic configuration

of [Co(¹L_N)₂]^1ꢀ, as shown in Figure 14, is (1a_g)²(2a_g)²-

(1b_3g)²(1a_u)²(1b_2g)²(1b_1u)²(2b_2g)¹(2b_3g)¹(1b_1g)⁰.

Figure 14. Kohn–Sham MO orbitals and energy scheme of 1 from an spin

unrestricted B3LYP DFT calculation.

the optical and magnetic data will show that the 3d_z₂and the

3d_yzorbitals lie very close to each other in energy.

One-electron oxidation of the monoanion yielding the

neutral complex generates an additional hole in the out-of-

plane 2b_2gorbital, which becomes the LUMO of the

[Co(¹L_N)₂] species. The calculated ground-state electronic

The 1b_1gLUMO of the monoanion represents the anti-

bonding combination of the Co 3d_xyand the ligand s orbi-

tals. Due to the ligand geometry the overlap between these

two orbitals is very favorable and provides an efficient path-

way for the ligand-to-metal electron donation. The 2b_2gand

Chem. Eur. J. 2005, 11, 204 – 224

www.chemeurj.org

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

217

K. Wieghardt, F. Neese et al.

Table 6. Percentage composition of the selected orbitals of [Co(¹L_N)₂] and [Co(¹L_N)₂]^ꢀ.

which brings the 3d orbitals

much lower in energy relative

to the ligand orbitals. The

upper valence region of the two

nickel complexes is therefore

composed of four doubly occu-

pied MOs that are predomi-

nantly centered on nickel (d⁸,

Ni^II).

MO

M(3d_yz)M(3d

_xz)M(3d

_xy)N(2p

_z)N(2p

_x,y)C(2p

_z)C(2p

)

x,y

[Co(¹L_N)₂]^ꢀ

[Co(¹L_N)₂]

2b_2g

2b_3g

1b_1g

47.1

35.2

28

10

24

12

75

2

3

61

24

22

2b_2g

2b_3g

1b_1g

34.2

12

28

14

73

In contrast, for [Co(¹L_N)₂]^1ꢀ

the DFT calculations allow us

configuration for this neutral species (S=¹= )is then

2

(1a_g)²(2a_g)²(1b_3g)²(1a_u)²(1b_2g)²(1b_1u)²(2b_3g)¹(2b_2g)⁰(1b_1g)⁰.

to envisage two extreme descriptions for its electronic struc-

ture (Figure 15): 1) If the 2b_2gorbital has predominantly

This oxidation is accompanied by a 12% decrease in the

Co 3d_xzcharacter of the 2b_2gorbital. However, the composi-

tions of the SOMO (2b_3g)and of the LUMO +1 (1b_1g)do

not change significantly upon oxidation. The unpaired elec-

tron of the neutral species is located in a predominantly

metal based 2b_3gorbital in agreement with the observed

large anisotropy of the EPR g tensor in 1 (Figure 12).

It is now interesting to compare the electronic structures

of 1 and of its monoanion with those reported for the corre-

sponding two nickel complexes, namely [Ni(¹L_N)₂] and

[Ni(¹L_N)₂]^1ꢀ, which have in part been studied previously.^[9]

The frontier orbitals calculated here for the cobalt com-

plexes are very similar to those reported for the nickel spe-

cies with the notable exception that the 2b_2gand 2b_3gorbi-

tals are predominantly ligand-based orbitals in the nickel

cases. As can be verified from the data in Table 6, these or-

bitals are significantly more metal-based in the correspond-

ing two cobalt species. This is consistent with the considera-

bly higher effective nuclear charge of nickel(ii)relative to

Co^II. This has been previously shown by X-ray absorption

spectra of divalent metal chlorides.^[35]

Thus it appears to be justified by the DFT calculations to

assign a +II spectroscopic oxidation state to the central

nickel ions in both complexes, namely in [Ni(¹L_N)₂] and

[Ni(¹L_N)₂]^ꢀ. Thus, for neutral, square-planar [Ni^II(¹L_N^ISQ)₂] a

detailed model study as well as high-level electronic struc-

ture calculations^[9]have shown that the complex is best de-

scribed as a species that contains two strongly interacting

ligand radicals, (L^I_N^SQ)C^ꢀcoordinated to a diamagnetic central

Figure 15. MO schemes of [Co(¹L_N)₂]^1ꢀand of [Ni^II(¹L_N^ISQ)(¹L^I_N^P)]^1ꢀ

.

metal character (Co), then the anionic complex could be de-

scribed as a species that contains a d⁶configuration with an

intermediate spin (S=1)cobalt( iii)ion; and 2)if, on the

other hand, the 2b_2gorbital possesses pure ligand character,

a Co^IIlow-spin d⁷configuration with a ferromagnetically

coupled (¹L_N^ISQ)C^ꢀligand p radical would prevail. The actual

DFT calculations on [Co(¹L_N)₂]^1ꢀreveal that the 2b_2gorbital

has almost equal metal and ligand contributions.

8

nickel(ii)ion (d ). The ligands have been shown to interact

with each other through an efficient superexchange mecha-

nism mediated by a back-bonding interaction to the central

metal ion yielding a singlet diradical.

From the corresponding DFT calculations for the mono-

anion [Ni(¹L_N)₂]^1ꢀ, an unambiguous interpretation of the va-

lence state has also been found.^[9]The ground-state configu-

ration is in this case (1a_g)²(2a_g)²(1b_3g)²(1a_u)²(1b_2g)²(2b_3g)²-

(1b_1u)²(2b_2g)¹(1b_1g)⁰. The 2b_2glevel is again ligand-centered

with some 26% Ni 3d_xzcharacter.

The redox-active 2b_2gorbital (LUMO of the neutral spe-

cies, but SOMO in the anion)is predominantly ligand-based

and is associated with a small p interaction with the low-

lying Ni 3d_xzorbital. As pointed out above, the predomi-

nance of the ligand character in the 2b_2glevel arises from

the higher effective nuclear charge of Ni (relative to Co)

It is therefore not possible to determine an unequivocal

dⁿ(n=6 or 7)electron configuration at the metal ion and

assign unambiguously a spectroscopic oxidation state of the

central cobalt ion. We note that the closed-shell-like ligand

configuration (⁴L_O^IP)^2ꢀas determined by X-ray crystallogra-

phy for 5 does point to a more pronounced Co^III(d⁶, S=1)

spectroscopic oxidation state than

mixed-valent configuration.

a

[Co^II(⁴L_O^ISQ)(⁴L_O^IP)]^ꢀ

Similarly, for the neutral complex [Co(¹L_N)₂] the metal

contribution in the 2b_2gorbital (the LUMO)is high (35%),

and the description of the electronic structure of this neutral

species as containing a low-spin Co^IIion coordinated to two

antiferromagnetically coupled ligand p radicals, (¹L_N^ISQ)^1ꢀ, is

218

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemeurj.org

Chem. Eur. J. 2005, 11, 204 – 224

Cobalt Coordination Complexes

FULL PAPER

ambiguous. However, such a description is in somewhat

better agreement with the multiconfigurational ab initio cal-

culations presented below. We stress that the question of the

spectroscopic (physical)oxidation state of the central cobalt

ion may again not be resolvable for the neutral [Co(¹L_N)₂]

species in an unambiguous fashion.

Excited state calculations: The optical electronic spectra of

the corresponding [M(¹L_N)₂]ⁿcomplexes of Ni, Pd, and Pt

[9]

(n=0,1ꢀ)have been analyzed in some detail previously. It

was shown that the very intense absorption in the near-IR

region (e>10⁴Lmol^ꢀ1cm^ꢀ1)of the spectra corresponds to a

ligand-to-ligand charge transition (LLCT)that contains con-

siderable information on the diradical character of the neu-

tral species and the ground-state ligand–ligand exchange

8

coupling. In the monoanionic (and monocationic)metal d

species, this intense absorption band corresponds to an inter-

valence transition (IVCT)of the type [LML C]^{1ꢀ or 1+}

Q[LCML]^{1ꢀ or 1+}

, in which LC represents the semiquino-

nate(1ꢀ) p-radical form of the ligand and L denotes the ar-

omatic dianionic form in the monoanion or the neutral qui-

nonane form in the monocation. The position, shape, and in-

tensity of these observed IVCT were clearly indicative of

class III mixed-valence systems with complete electron de-

localization of the unpaired electron over both ligands.^[9]

In the present case of the corresponding cobalt complexes

these optical absorption spectra (Figure 11)have a consider-

ably more complicated appearance and, consequently, their

analysis is more involved.

Figure 16. Deconvoluted electronic absorption spectrum [Co(¹L_N)₂]⁰(top)

and its magnetic circular dichroism (MCD)spectrum (bottom)recorded

at 1.8 K and 5.0 T.

The room-temperature absorption and low-temperature

MCD (1.8 K, 5 T)spectra of 1 have been subjected to Gaus-

sian deconvolution in the range 25000–5000 cm^ꢀ1in order to

detect the individual transitions (Figure 16). As is commonly

observed, the bands show a slight shift on going from ambi-

ent to liquid-helium temperature and, therefore, equal band

positions have not been enforced in the fit. The minimum

number of Gaussian bands was used such that it is consistent

with the experimentally observed peaks and shoulders yield-

ing a total of 11 detectable transitions for neutral 1 below

25000 cm^ꢀ1(bands 3–13 in Figure 16; see also Table 7).

The absorption spectrum is dominated by three reasona-

bly intense transitions (bands 3, 5, and 8). A weaker band in

between bands 3 and 5 and two weaker bands between

5 and 8 show up, and apparently there is a whole series of

transitions to higher energy than band 8 that are not re-

solved in the absorption spectra, but lead to weak MCD fea-

tures of variable sign. The temperature dependence of the

MCD spectrum (not shown)indicates that all observed fea-

tures arise from a C-term mechanism that is expected to

dominate for paramagnetic molecules studied at low temper-

ature. However, the C/D ratios of all bands (with the excep-

tion of band 4)are ꢃ0.01, indicating that the MCD spec-

trum is at least not dominated by transitions of d–d charac-

ter, which would require C/D ratios ꢄ0.03. In particular

bands 3 and 5, which are quite prominent in the absorption

spectrum, have very small MCD intensities and the domi-

nant absorption band 8 still only has a C/D ratio of 0.011.

These findings are consistent with the interpretation of the

spectrum resulting from correlated ab initio calculations de-

scribed below.

Of particular interest is the transition 1b_1u!2b_2g, which

corresponds to the LLCT transition of A_usymmetry and is

allowed in x polarization (along the long axis of the com-

plex). In the case of the neutral cobalt complex this transi-

tion leads from a doubly occupied MO to a virtual MO

(Figure 14). Thus, in the excited state there are three un-

paired electrons that can spin couple to give two doublet

and one quartet states. The absorption intensity for the

quartet state will be very low since it is spin forbidden. For-

mally, the first doublet arises from the straightforward

1b_1u!2b_2gsinglet excitation, while the second excitation in-

volves the 1b_1u!2b_2gtriplet excitation with a concomitant

spin flip of the unpaired electron in the 2b_3gSOMO. Thus,

the second doublet is referred to as a “trip-doublet” in Gou-

termanꢁs nomenclature^[36]and formally corresponds to a

double excitation. However, the two doublets interact

through the Hamiltonian operator and can, therefore, mix

and acquire absorption intensity. From the SORCI calcula-

tions there is little doubt that bands 3 and 8 must be as-

signed to the two components of the 1b_1u!2b_2gtransition,

with the lower energy transition corresponding to the trip-

doublet component. These transitions are calculated at 6175

and 12390 cm^ꢀ1and observed at 8224 and 15927 cm^ꢀ1, re-

spectively, which is considered to be good agreement. The

calculated oscillator strengths of 0.006 and 0.359 are also in

Chem. Eur. J. 2005, 11, 204 – 224

www.chemeurj.org

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

219

K. Wieghardt, F. Neese et al.

Table 7. Detailed analysis of the optical transitions for 1 following Gaussian deconvolution of the experimental data shown in Figure 14 combined with

the results of spectroscopy-oriented configuration interaction calculations.

Band

Energy [cm^ꢀ1

]

Method^[c]

Exptl

C₀/D₀

Oscillator strength

Assignment

¯

exptl

calcd

exptl

calcd

1

2

3

4

–

1616

3789

6175

7983

8552

abs

–

0.000

0.001

0.006

0.000

²A_g(2a_g(d_z₂)!2b_3g(d_yz))

–

8224

9698

²B_1u(1b_1u(p*)!2b_3g(d_yz))

0.026

0.012

²A_u(1b_1u(p*)!2b_2g(p*))

²B_3u(1b_1u(p*),2a_g(d_z₂)!2b_3g(d_yz),2b_2g(p*))

²B_2g(1b_2g(d_xz)!2b_3g(d_yz))

11384

12245

MCD

abs

ꢀ0.0240

5

11779

0.113

0.008

²A_u(1b_1u(p*),1,2a_g(d_x₂_ꢀy₂,d_z₂)!2b_3g(d_yz),1b_1g(d_xy)) +

²A_u(1b_1u(p*), 1,2a_g(d_x₂_ꢀy₂,d_z₂),1b_2g(d_xz)!

2b_3g(d_yz),1b_1g(d_xy),2b_2g(p*))

12788

13112

14501

14210

15031

15927

15600

17164

16654

18704

17533

20305

19290

21746

20564

23263

23376

MCD

abs

MCD

abs

MCD

abs

MCD

abs

MCD

abs

MCD

abs

MCD

abs

MCD

abs

ꢀ0.0004

0.0082

6

12262

0.038

0.032

0.200

0.100

0.060

0.032

0.029

0.021

0.000

²A_g(1a_g(d_x₂_ꢀy₂)!2b_3g(d_yz))

[a]

7

0.0075

8

12390

17482

18148

20033

21339

0.359

²A_u(1b_1u(p*)!2b_2g(p*))

0.0110

[b]

9

–

ꢀ0.0040

ꢀ0.0130

ꢀ0.0036

0.0100

[b]

10

11

12

13

[a]

MCD

ꢀ0.0042

[a] No transitions calculated. [b] Not analyzed. [c] abs=electronic absorption spectra, MCD=magnetic circular dichroism.

reasonable agreement with the experimentally determined

values of 0.026 and 0.20, respectively, but it appears that the

calculations underestimate the mixing of the two compo-

nents somewhat thus leading to too low an intensity for the

lower and a too large intensity for the upper state. The C/D

ratio for the trip-doublet component is very low, as might be

expected from its relatively large double excitation charac-

ter, but the C/D ratio of band 8 approaches ~0.01. This

shows that there must be considerable metal character in-

volved in the acceptor orbital, since a pure, unipolarized

LLCT transition should show a fairly weak MCD response.

In the SORCI calculations four states are calculated be-

tween bands 3 and 8; these results agree nicely with the four

bands resolved in the simultaneous fit of the MCD and ab-

sorption spectra. The first of the intermediate states is domi-

tor strength of 0.0079. We assign this transition to the rea-

sonably intense band 5 observed at 12249 cm^ꢀ1. Its calculat-

2

ed symmetry is A_u, and it therefore can obtain considerable

intensity from mixing with the intense LLCT band, which

has the same symmetry. The origin of the experimentally ob-

served intensity must arise from such mixing mechanisms,

since the calculated origin of this band unexpectedly reveals

that it is mainly composed of double (~50% 1b_1u(p*),

1,2a_g(d_x₂_ꢀy₂,d_z₂)!2b_3g(d_yz),1b_1g(d_xy)) and even triple excitations

(~25%

1b_1u(p*),1,2a_g(d_x₂_ꢀy₂,d_z₂),1b_2g(d_xz)!2b_3g(d_yz),1b_1g-

(d_xy),2b_2g(p*)). The interpretation of band 5 results from the

ab initio calculations; this could have hardly been anticipat-

ed in advance and demonstrates the importance of an un-

biased multireference approach. In the present case the

SORCI calculations include up to and including pentuple

excitations from the leading ground state configuration and

these high excitations are also required in order to properly

capture the differential dynamic correlation effects.^[37]It is

evident from these considerations that any single reference

treatment of the excited states of [Co(¹L_N)₂]⁰will be inflicted

with large difficulties and is likely to lead to unbalanced re-

sults.

nated by the double excitation B_3u(1b_1u(p*), 2a_g(d_z²)!

2

2b_3g(d_yz) , 2b (p*)at 7983 cm ^ꢀ1, but with small absorption in-

2g

tensity owing to its two-electron nature. This double excita-

tion is so exceptionally low-lying because the 2a_g(d_z₂)!

2b_3g(d_yz)transition is very low in energy (vide infra); this in

turn leads to considerable complexity in the calculated excit-

ed-state structure. However, this is characteristic of the co-

ordinated radical species under investigation. The second

transition calculated in the region of band 4 is the

²B_2g(1b_2g(d_xz)!2b_3g(d_yz)) electric-dipole-forbidden “d–d”

transition at 8552 cm^ꢀ1. Owing to the relatively large C/D

ratio, this transition is likely to make the dominant contribu-

tion to band 4. The next intermediate transition is of compli-

cated origin and is calculated at 11779 cm^ꢀ1with an oscilla-

The final transition calculated below the intense band 8 is

the electric-dipole-forbidden ²A_g(1a_g(d_x₂_ꢀy₂)!2b_3g(d_yz)) “d–

d” transition. It is calculated at 12262 cm^ꢀ1and occurs in

the region of bands 6 and 7 observed at 13112 and

14210 cm^ꢀ1, respectively. However, the case for two bands

in this region is not unambiguous. The calculated oscillator

strength is zero, but the transition may borrow intensity

220

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemeurj.org

Chem. Eur. J. 2005, 11, 204 – 224

Cobalt Coordination Complexes

FULL PAPER

from the near-lying, intense, allowed transitions by low-sym-

metry distortions and vibronic coupling effects. Beyond the

intense band 8 the density of states increases and we have

calculated additional states at 17482, 18148, 20033 and

21339 cm^ꢀ1consistent with the appearance of bands 9–13,

which occur in these regions. All of these transitions are of

complicated nature and involve low-lying double excitations

that were not further analyzed. Their MCD features are

very weak.

possible for strong intersite interactions. The remaining tran-

sitions are of LMCT and d–d origin. It is particularly inter-

esting that the calculations predict a very low-lying excited

d–d state below 2000 cm^ꢀ1which is responsible for the very

large g_maxshift observed in the EPR experiments.

The [Co(¹L_N)₂]^ꢀcomplex proved to be too unstable for

the MCD measurements. Hence, we will skip the assignment

of the transitions for the anion, since in the absence of the

MCD it will not be possible to prove the validity of our as-

signments.

The origin of the MCD intensities is subject to debate. In

the D_2hpoint group all transitions are unipolarized and,

thus, spin-orbit coupling (SOC)to other states with non-col-

linear transition moments is required in order to obtain non-

zero MCD intensity.^[38]Since the SOC only couples states of

like parity, it is either the SOC of the ground state with

other “gerade” states or the coupling of electric dipole al-

lowed states with other “ungerade” states that gives rise to

the observed signs. Since in our interpretation the main in-

tensity in the absorption spectrum arises from x-polarized

Comparison of SORCI and DFT ground-state description:

We will now compare the calculated ground-state electron

and spin distribution between the DFT and SORCI calcula-

tions. The first meeting point between DFT and correlated

ab initio theory is the one-electron density. The comparison

in terms of a Lçwdin population analysis is meaningful,

since both methods were carried out with the same basis set

and geometry (Table 8).

2

²B_3g! A_utransitions, the nature of the coupling states is not

immediately evident. For the excited-state SOC mechanism,

which gives rise to a sum rule,^[38]the prime candidates are y-

Table 8. Comparison of the charge and spin populations at the cobalt

ions resulting from a Lçwdin analysis of the one-electron density of the

ground state from ab initio SORCI and B3LYP density functional (in pa-

rentheses)calculations.

2

polarized B_1ustates of which we calculate only one at very

low energy (vide infra)and with a low transition moment. It

is, however, conceivable that other excited states of this

symmetry that exist outside the studied spectral window sig-

nificantly contribute to the observed MCD intensities. That

such mechanisms are operative, at least in addition to the

ground-state SOC mechanism, is witnessed by the fact that

the spectrum is not strongly dominated by bands of only

one sign, which is often the case if the ground-state SOC

dominates the MCD response.

Electron d

Electron 4s

Spin d

Spin s

[Co(¹L_N)₂]

7.65 (7.64)0.43 (0.51) 0.97 (1.14)0.07 (0.00)

7.42 (7.60)0.54 (0.53) 1.56 (1.47)0.00 (0.00)

[Co(¹L_N)₂]^1ꢀ

For the neutral species, the analysis shows a d population

between seven and eight electrons and a spin density of es-

sentially one unpaired electron at the central cobalt atom.

Both values would be consistent with a central, low-spin

Co^IIion and this would also be in accord with the interpreta-

tion of the optical spectrum, which shows that essentially all

intensity arises from the LLCT transition of the coordinat-

ing ligand in the diradical form. Thus, the ground state

In the SORCI calculations, two additional very low-lying

states were found that are outside the detection range of the

absorption and MCD experiments. The first state corre-

2

sponds to the A_g(2a_g(d_z₂)!2b_3g(d_yz)) parity forbidden d–d

transition, which is calculated to occur as low as 1616 cm^ꢀ1.

Since this state spin-orbit couples to the ground state

through the x component of the SOC operator, its very low

energy nicely explains the very large g_maxshift observed for

[Co(¹L_N)₂] in the EPR experiments. A more detailed analy-

sis of the ground-state magnetic properties of cobalt com-

plexes within this family will be presented elsewhere. The

second low-lying excited state corresponds to the

²B_1u(1b_1u(p*)!2b_3g(d_yz)) electric-dipole-allowed LMCT

transition, which is calculated at 3789 cm^ꢀ1but with a fairly

low oscillator strength of 9.310^ꢀ4.

wavefunction would contain three open-shell S=¹= frag-

2

ments that are coupled to a total spin of S_t=¹= . In fact the

2

SORCI many-electron ground-state wavefunction contains

~7% of the double excitation (1b_1u)²!(2b_2g)², which de-

scribes the diradical character of the ligand and rules out

the alternative of ferromagnetic coupling between the li-

gands and antiferromagnetic coupling to the central metal.

It is clear that such a complicated behavior can only be

crudely modelled with DFT methods. In fact the DFT value

2

ˆ

The SORCI ab initio calculations lead to rather good

agreement with the experimentally observed absorption

spectrum of [Co(¹L_N)₂]⁰(the errors in the calculated transi-

tion energies do not exceed 0.2–0.3 eV). The calculations

also reveal quite a high degree of electronic structure com-

plexity, which we believe to require a multireference ap-

proach such as the SORCI method to be properly dealt

with. The two intense peaks 3 and 8 are interpreted as the

two components of a multiplet-split LLCT transition with

the splitting approaching 1 eV; this type of splitting is only

of 1.14 unpaired electrons on the central cobalt and hS i=

0.83 show a partial broken symmetry character in a “desper-

ate” attempt of the variational principle to minimize the

energy towards the truly multiconfigurational ground state;

this is, however, impossible within the restrictions imposed

by a single determinantal wavefunction.^[39]In addition, the

natural orbitals of the SORCI calculation show a relatively

pure d_xz-based b_2gorbital in contrast to the strong mixing

observed in the DFT calculations and which made the oxi-

dation state assignment ambiguous. Thus, from the point of

Chem. Eur. J. 2005, 11, 204 – 224

www.chemeurj.org

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

221

K. Wieghardt, F. Neese et al.

view of the SORCI calculations, the assignment of the spec-

troscopic oxidation state as Co^IIis the preferred one.

in four occupied metal d orbitals, the neutral, mono- and

dianions and their electronic structures may be adequately

represented by the following resonance structures:

Interestingly, for the anionic species the d-orbital popula-

tion at the central cobalt atom decreases, while the spin pop-

ulation increases to more than 1.5 unpaired electrons. Both

observations would seem to indicate that the central cobalt

is oxidized to a Co^IIIion with S=1 upon reduction of the

complex, while the ligands would exist in the dianionic

redox innocent form. This is also in accord with the inter-

pretation of the structural data discussed above. However,

both the d population (which is somewhat higher than

expected for Co^III)and the spin population (which is

lower than expected for Co^III)show that this description

is oversimplified. Thus the actual electronic structure

may be best described by the resonance forms

[Co^III(L^2ꢀ)(L^2ꢀ)]^ꢀQ[Co^II(LC^ꢀ)(L^2ꢀ)]^ꢀQ[Co^II(L^2ꢀ)(LC^ꢀ)]^ꢀ, but

with a somewhat larger weight for the first structure. The

DFT results are also in accord with this interpretation, since

Very similar behavior is deduced for the series

[Ni(⁴L_O)₂]^nꢀ(n=0,1,2). The electronic absorption spectra of

the neutral and monoanionic species of nickel are dominat-

ed by a single intense band that is of ligand-to-ligand charge

transfer character in the former and of intervalence charge

transfer character in the latter. This band is absent in the di-

anionic form because the 2b_2gorbital is now filled (HOMO).

The four-coordinate cobalt complexes are more difficult

to understand, even qualitatively, because the out-of-plane

orbital 2b_2ghas almost equal contributions from the Co 3d_xz

and the b_2gligand fragment orbitals. It is therefore not possi-

ble to assign straightforwardly a spectroscopic oxidation

state for the cobalt ion. On the other hand, it is clear that

the one-electron oxidation of the monoanion leads to a de-

crease of the metal character of the 2b_2gorbital. Taking the

following resonance structures into account one can describe

the electronic structures as follows:

2

ˆ

the hS i value of 2.01 indicates little spin contamination or

broken symmetry character and the calculated populations

are in reasonable accord with the SORCI calculations, al-

though the loss of metal d-based charge is less pronounced

in the DFT case.

We conclude that multiconfigurational ab initio methods

in form of the SORCI procedure successfully account for

the experimental observations despite the fact that the elec-

tronic structure of the molecules investigated is very compli-

cated. This work is perhaps among the first attempts to sig-

nificantly go beyond the restrictions of DFT in the modeling

of transition-metal radical systems and demonstrates that

such approaches, despite their somewhat higher computa-

tional cost, have become increasingly feasible and, therefore,

add a powerful new tool in the theoretical arsenal available

to treat such systems. This is particularly true for the inter-

pretation of the fascinating optical properties of such sys-

tems that are even more complicated than the already com-

plicated ground states, and any method based on a single

Slater determinant reference will have great difficulties to

arrive at a successful interpretation.

Discussion

In the Introduction we pointed out that small structural dif-

The resonance structures that formally contain Co^IIIions

have significant weights which nicely explains the observed

ligand bond lengths in four-coordinate complexes 1, 4, and

5.

On first sight, it is somewhat counterintuitive that the ad-

dition of a fifth ligand, such as pyridine, to the four-coordi-

nate species 1 to yield complex 2 does not change the elec-

tronic spectra (Figures 11 and 13), but alters the EPR spec-

tra dramatically (Figure 12). However, these results are

readily understood from the calculations presented above.

Thus, an addition of a fifth ligand does not alter the elec-

tronic configuration at the central metal ion and conse-

quently the number and nature of allowed electronic transi-

tions remains conserved. In particular, it is the equatorial li-

ꢀ

ferences in the ligand C N and C C distances between the

corresponding four-coordinate cobalt and nickel complexes

may exist and might point to differing ligand oxidation

levels (and concomitantly central metal ions)in the two

neutral complexes and the corresponding one-electron re-

duced monoanions. This has been verified by low-tempera-

ture (100 K)crystallography of neutral 4 and 8 on the one

hand and of 5 and 9 on the other. By using DFT and corre-

lated ab initio calculations these differences were traced to

energetic differences of the 2b_2gand 2b_3gorbitals relative to

the metal d orbitals. As shown in Figure 15 in the nickel

cases both orbitals are predominantly ligand-centered and

the central nickel ion has a d⁸electron configuration of Ni^II

222

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemeurj.org

Chem. Eur. J. 2005, 11, 204 – 224

Cobalt Coordination Complexes

FULL PAPER

[41]

gands that give rise to the observed absorption intensities

and these are little affected by the presence of an additional

axial ligand in the present case. That the electronic configu-

ration at the central cobalt stays intact is witnessed by the

presence of the trip-doublet feature^[36]at ~1200 nm, which

can only arise if there are three unpaired electrons in the ex-

cited state.

in fact [Ru^III(acac)₂(²L_N^ISQ)];

[Ru^IICl₂(PhNH₂)₂(²L^I_N^BQ) ] is

[Ru^IIICl₂(PhNH₂)₂(²L^I_N^SQ)], and [Ru^IICl₂(bpy)(²L_N^IBQ

)

]

is

[Ru^IIICl₂(bpy)(²L^I_N^SQ)].

lated as [Os^IIBr₂(²L_N^IBQ)₂] should most probably be under-

stood as mixed-valent ligand class III species

[Os^IIIBr₂(²L_N^ISQ)(²L_N^IBQ)].

Interestingly, the complex formu-

[42]

a

In [Ru^II(bipy)₂(²L^I_N^BQ)](PF₆)₂the

[43]

metal and ligand oxidation states have been correctly as-

The EPR parameters are dominated by a different mecha-

nism. Here the spin-orbit coupling of electric-dipole-forbid-

den states with the ground state determines the g shift (devi-

ation from the free electron g value; g_e=2.0023). In 1 the

state that gives the dominant contribution to the g_zvalue is

in fact very low in energy (<2000 cm^ꢀ1in our calculations)

and involves a transition from the out-of-plane d_z₂orbital

into the d_yz-based SOMO. It is intuitively clear that the ad-

dition of a fifth ligand must indeed have a major effect on

the position of this low-lying state. Thus, already a shift in

the transition energy of a few thousand wavenumbers is

able to explain the enormous reduction of the g_zshift by

almost a factor of three from 0.81 in 1 to a value of 0.31 in

2. That subtle variations have a large influence on g_zis also

consistent with the large change of g_z(Dg_z~0.3)in going

from 1 (g_z=2.81)to the very similar species 4 (g_z=3.11).

Clearly, given the limited value of the one-electron SOC

constant on Co (~600 cm^ꢀ1) g_zvalues larger than 2.5 are

only possible in the presence of near orbital degeneracy.

The monocationic form of 1⁺has not been structurally

characterized, but it should possess an electronic structure

such as [Co^III(¹L^I_N^SQ)₂]⁺or [Co^II(¹L_N^ISQ)(¹L_N^IBQ)]⁺with a singlet

or triplet electron configuration (1a_g)²(2a_g)²(1b_3g)²(1b_2g)²-

(1b_1u)¹(2b_3g)¹(2b_2g)⁰or (1a_g)²(2a_g)²(1b_3g)²(1b_2g)²(1b_1u)²(2b_3g)⁰-

(2b_2g)⁰. Upon five-coordination as in 2a, 3, and 7, the 1b_2g

orbital becomes predominantly metal-centered (d_xz). This

leads to a closed-shell electron configuration (1a_g)²(2a_g)²-

(1b_3g)²(1b_2g)²(1b_1u)²(2b_3g)⁰(2b_2g)⁰(1b_1g)⁰with a singlet ground

signed.^[10]

Acknowledgement

We thank the Max-Planck Society and the Fonds der Chemischen Indus-

trie for financial support of this work.

[1] A.L. Balch, R. H. Holm, J. Am. Chem. Soc. 1966, 88, 5201.

[2] L. F. Warren, Inorg. Chem. 1977, 16, 2814.

[3] S.-M. Peng, C.-T. Chen, D.-S. Liaw, C.-I. Chen, Y. Wang, Inorg.

Chim. Acta 1985, 101, L31.

[4] F. Cariati, F. Morazzoni, C. Busetto, G. Del Piero, A. Zazzetta, J.

Chem. Soc. Dalton Trans. 1976, 342.

[5] W. P. Schaefer, R. E. Marsh, Acta Crystallogr. Sect. B 1969, 25, 1675.

[6] C. Daul, C. W. Schlꢂpfer, A. v. Zelewski, Struct. Bonding 1979, 36,

12.

[7] S. Nemeth, L. I. Simandi, G. Argay, A. Kꢃlman, Inorg. Chim. Acta

1989, 166, 31.

[8] P.-H. Cheng, H.-Y. Cheng, C.-C. Lin, S.-M. Peng, Inorg. Chim. Acta

1990, 169, 19.

[9] a)D. Herebian, E. Bothe, F. Neese, T. Weyhermüller, K. Wieghardt,

J. Am. Chem. Soc. 2003, 125, 9116; b)D. Herebian, K. E. Wieghardt,

F. Neese, J. Am. Chem. Soc. 2003, 125, 10997.

[10] H.-Y. Cheng, C.-C. Lin, B.-C. Tzeng, S.-M. Peng, J. Chin. Chem. Soc.

1994, 41, 775.

[11] D. Herebian, P. Ghosh, H. Chun, E. Bothe, T. Weyhermüller, K.

Wieghardt, Eur. J. Inorg. Chem. 2002, 1957.

[12] P. Chaudhuri, C. N. Verani, E. Bill, E. Bothe, T. Weyhermüller, K.

Wieghardt, J. Am. Chem. Soc. 2001, 123, 2213.

[13] P. Ghosh, A. Begum, D. Herebian, E. Bothe, K. Hildenbrand, T.

Weyhermüller, K. Wieghardt, Angew. Chem. 2003, 115, 581; Angew.

Chem. Int. Ed. 2003, 42, 563.

[14] X. Sun, H. Chun, K. Hildenbrand, E. Bothe, T. Weyhermüller, F.

Neese, K. Wieghardt, Inorg. Chem. 2002, 41, 4295.

[15] C. N. Verani, S. Gallert, E. Bill, T. Weyhermüller, K. Wieghardt, P.

Chaudhuri, Chem. Commun. 1999, 1747.

[16] K. S. Min, T. Weyhermüller, K. Wieghardt, Dalton Trans. 2003, 1126.

[17] L. S. Hegedus, in Transition Metals in the Synthesis of Complex Or-

ganic Molecules, University Science Books, Mill Valley, California

1994, p. 3.

[18] C. K. Jçrgensen, in Oxidation Numbers and Oxidation States, Spring-

er, Berlin, Heidelberg, New York, 1969.

[19] SHELXTL V.5, Siemens Analytical X-Ray Instruments, Inc., Madi-

son, Wisconsin (USA), 1994.

[20] SADABS, G. M. Sheldrick, University of Gçttingen (Germany),

1994.

6

state (as is observed)for a d (Co^III)configuration. This

would explain the structural ligand dimensions in 2a, 3, and

7 as two radical ions (L_N^ISQ)C^ꢀand their similar electronic

spectra. It is interesting that similar electronic effects have

been observed for three complexes containing the redox-in-

nocent ligand o-phenylenebis(biuretate)(4ꢀ)(bbphen:) the

four coordinate species [Co^III(bbphen)]^ꢀpossesses an S=1

ground state, whereas the five- and six-coordinate species

[Co^III(bbphen)(CN)]^ꢀand [Co^III(bbphen)(CN)₂]^3ꢀare both

diamagnetic (S=0).^[32f]

It is here and previously clearly established that in com-

plexes containing only one dioxolene-type ligand, namely

(^XL_N)^0,1ꢀ,2ꢀor (^XL_O)^0,1ꢀ,2ꢀ, the respective oxidation level is

defined by high-quality X-ray crystallography, because their

[21] SHELXL97, G. M. Sheldrick, University of Gçttingen (Germany),

1997.

[22] F. Neese, Orca, an ab initio, DFT, and Semiempirical Electronic

Structure Package, Version 2.2, Revision 74, Max-Planck-Institut für

Bioanorganische Chemie, Mülheim an der Ruhr (Germany), June

2003.

[23] A. D. Becke, J. Chem. Phys. 1986, 84, 4524.

[24] J. P. Perdew, Phys. Rev. B 1986, 33, 8522.

[25] a)A. Schꢂfer, H. Horn, R. Ahlrichs, J. Chem. Phys. 1992, 97, 2571;

b)A. Schꢂfer, C. Huber, R. Ahlrichs, J. Chem. Phys. 1994, 100, 5289.

ꢀ

C O, C N and C C bond lengths are characteristic for a

given oxidation state (see Figure 4). Therefore, we feel that

in a series of crystallographically characterized ruthenium

and osmium complexes reported by Goswami et al.^[40–43]the

assigned ligand and metal oxidation states as diimineruthe-

nium(ii)or diimineosmium( ii)species are incorrect. The

ligand dimensions clearly point to the presence of o-diimi-

nosemiquinonato(1ꢀ)radicals: Thus [Ru ^II(acac)₂(²L_N^IBQ) ] is

Chem. Eur. J. 2005, 11, 204 – 224

www.chemeurj.org

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

223

K. Wieghardt, F. Neese et al.

[26] Basis sets were obtained from the ftp server of the quantum chemis-

try group at the university of Karlsruhe (Germany)under http://

www.chemie.uni-karlsruhe.de/PC/TheoChem/.

[27] a)P. Pulay, Chem. Phys. Lett. 1980, 73, 393; b)P. Pulay, J. Comput.

Chem. 1992, 13, 556.

[28] C. Lee, W. Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785.

[29] A. D. Becke, J. Chem. Phys. 1993, 98, 5648.

[30] F. Neese, J. Chem. Phys. 2003, 119, 9428.

[36] a)M. G. Cory, M. C. Zerner, Chem. Rev. 1991, 91, 813; b)M. Gou-

terman, J. Mol. Spectrosc. 1961, 6, 138.

[37] Differential dynamic correlation effects are represented by those

parts of the dynamic correlation that change between different elec-

tronic states and are therefore essential for the calculation of accu-

rate transition energies. As shown in the pioneering work of Mal-

rieu, Caballol and co-workers (J. Miralles, J. P. Daudey, R. Caballol,

Chem. Phys. Lett. 1992, 198, 555; J. Miralles, O. Castell, R. Caballol,

J. P. Malrieu, Chem. Phys. Lett. 1993, 172, 33)a substantial part of

the total dynamic correlation, namely the part which is brought in

by the inactive double excitations, is expected to be nearly state in-

dependent. It is therefore the differential dynamic correlation that is

the focus of the SORCI method, which may be viewed as an approx-

imation to the iterative difference dedicated configuration interac-

tion methodology (V. M. Garcia, O. Castell, R. Caballol, J. P. Mal-

rieu, Chem. Phys. Lett. 1995, 238, 222).

[31] P. J. van der Put, A. A. Schilperoord, Inorg. Chem. 1974, 13, 2476.

[32] a)S. W. Gordon-Wylie, E.L. Bominaar, T. J. Collins, J. M. Workman,

B.L. Claus, R. E. Patterson, S. E. Williams, B. J. Conklin, G. T. Yee,

S. T. Weintraub, Chem. Eur. J. 1995, 1, 528; b)S. W. Gordon-Wylie,

B.L. Claus, C. P. Horwitz, Y. Leychkis, J. M. Workman, A. J. Marzec,

G. R. Clark, C. E. F. Rickard, B. J. Conklin, S. Sellers, G. T. Yee, T. J.

Collins, Chem. Eur. J. 1998, 4, 2173; c)J. J. Bour, P. T. Beurskens,

J. J. Steggarda, J. Chem. Soc. Chem. Commun. 1972, 221; d)T. J.

Collins, R. D. Powell, C. Slebodnick, E. S. Uffelman, J. Am. Chem.

Soc. 1991, 113, 8419; e)J. C. Brewer, T. J. Collins, M. R. Smith, B. D.

Santarsiero, J. Am. Chem. Soc. 1988, 110, 423; f)T. Yagi, H. Hanai,

T. Komorita, T. Suzuki, S. Kaizaki, J. Chem. Soc. Dalton Trans. 2002,

1126.

[33] a)N. Bresciani-Pahor, L. Randaccio, W. Summers, P. J. Toscano,

Inorg. Chim. Acta 1983, 68, 69; b)M. F. Summers, L. G. Marzili,

N. B. Bresciani-Pahor, L. Randaccio, J. Am. Chem. Soc. 1984, 106,

4478; c)S. J. Moore, R. J. Lachicotte, S. T. Sullivan, L. G. Marzili,

Inorg. Chem. 1999, 38, 383; d)N. B. Pahor, S. Geremia, C. Lopez, L.

Randaccio, E. Zangrando, Inorg. Chem. 1990, 29, 1043.

[34] C. R. Ollis, D. Y. Jeter, W. E. Hatfield, J. Am. Chem. Soc. 1971, 93,

547.

[38] F. Neese, E. I. Solomon, Inorg. Chem. 1999, 38, 1847.

[39] F. Neese, J. Phys. Chem. Solids 2004, 65, 781.

[40] C. Das, S. Goswami, Comments Inorg. Chem. 2003, 24, 137.

[41] a)K. N. Mitra, P. Majumdar, S.-M. Peng, A. Castineiras, S. Goswa-

mi, Chem. Commun. 1997, 1267; b)K. N. Mitra, S. Choudhury, A.

Castineiras, S. Goswami, J. Chem. Soc. Dalton Trans. 1998, 2901.

[42] C. Das, K. K. Kamar, A. K. Ghosh, P. Majumdar, C.-H. Hung, S.

Goswami, New J. Chem. 2002, 26, 1409.

[43] a)K. N. Mitra, S. Goswami, Chem. Commun. 1997, 49; b)K. N.

Mitra, S. Goswami, Inorg. Chem. 1997, 36, 1322.

[35] S. E. Shadle, B. Hedman, K. O. Hodgson, E. I. Solomon, J. Am.

Chem. Soc. 1995, 117, 2259.

Received: August 17, 2004

Published online: November 17, 2004

224

ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemeurj.org

Chem. Eur. J. 2005, 11, 204 – 224

Article Doi

Molecular and electronic structure of four- and five-coordinate cobalt complexes containing two o-phenylenediamine- or two o-aminophenol-type ligands at various oxidation levels: An experimental, density functional, and correlated ab initio study

DOI: 10.1002/chem.200400850

Source and publish data:

Authors:

Article abstract of DOI:10.1002/chem.200400850

Full text of DOI:10.1002/chem.200400850

Products guided by the article

R&D Labs maybe for 845726-51-4

Relevant to this article

Hot Product