Observation of ferromagnetic exchange, spin crossover, reductively induced oxidation, and field-induced slow magnetic relaxation in monomeric cobalt nitroxides

Article

pubs.acs.org/IC

Observation of Ferromagnetic Exchange, Spin Crossover,

Reductively Induced Oxidation, and Field-Induced Slow Magnetic

Relaxation in Monomeric Cobalt Nitroxides

Ian A. Gass,^†Subrata Tewary,^#Ayman Nafady,^∇Nicholas. F. Chilton,^†Christopher J. Gartshore,^†

Mousa Asadi,^†David W. Lupton,^†Boujemaa Moubaraki,^†Alan M. Bond,^†John F. Boas,^‡Si-Xuan Guo,^†

Gopalan Rajaraman,^#and Keith S. Murray*^,†

^†School of Chemistry, Monash University, Clayton, Victoria 3800, Australia

^‡School of Physics, Monash University, Clayton, Victoria 3800, Australia

^§School of Chemistry, University of Melbourne, Victoria 3010, Australia

^#Department of Chemistry, Indian Institute of TechnologyBombay, Powai, Mumbai, India

^∇Department of Chemistry, King Saud University, P.O. Box 2455, Riyadh−11451, Kingdom of Saudi Arabia

S

* Supporting Information

ABSTRACT: The reaction of [Co^II(NO₃)₂]·6H₂O with the

nitroxide radical, 4-dimethyl-2,2-di(2-pyridyl) oxazolidine-N-

oxide (L^•), produces the mononuclear transition-metal

complex [Co^II(L^•)₂](NO₃)₂(1), which has been investigated

using temperature-dependent magnetic susceptibility, electron

paramagnetic resonance (EPR) spectroscopy, electrochemis-

try, density functional theory (DFT) calculations, and variable-

temperature X-ray structure analysis. Magnetic susceptibility

measurements and X-ray diﬀraction (XRD) analysis reveal a

central low-spin octahedral Co²⁺ion with both ligands in the

neutral radical form (L^•) forming a linear L^•···Co(II)···L^•

arrangement. This shows a host of interesting magnetic

3

properties including strong cobalt-radical and radical−radical intramolecular ferromagnetic interactions stabilizing a S = /₂

ground state, a thermally induced spin crossover transition above 200 K and ﬁeld-induced slow magnetic relaxation. This is

supported by variable-temperature EPR spectra, which suggest that 1 has a positive D value and nonzero E values, suggesting the

possibility of a ﬁeld-induced transverse anisotropy barrier. DFT calculations support the parallel alignment of the two radical

π*_NOorbitals with a small orbital overlap leading to radical−radical ferromagnetic interactions while the cobalt-radical interaction

is computed to be strong and ferromagnetic. In the high-spin (HS) case, the DFT calculations predict a weak antiferromagnetic

cobalt-radical interaction, whereas the radical−radical interaction is computed to be large and ferromagnetic. The monocationic

complex [Co^III(L⁻)₂](BPh₄) (2) is formed by a rare, reductively induced oxidation of the Co center and has been fully

characterized by X-ray structure analysis and magnetic measurements revealing a diamagnetic ground state. Electrochemical

studies on 1 and 2 revealed common Co-redox intermediates and the proposed mechanism is compared and contrasted with that

of the Fe analogues.

One-electron redox processes in such ligands invariably leads to

the generation of radical species, which, when coupled with

redox-active metal centers, can show strong magnetic exchange⁶

and electronic delocalization,⁷resulting in ambiguous electronic

structures.

1. INTRODUCTION

Metal complexes involving redox active ligands such as α-

dithiolenes,¹α-diimines,²and 1,2-dioxolenes,³exhibit non-

innocent behavior and have been studied extensively to gain a

fuller understanding of their electronic structure by assignment

of the “correct” oxidation state of the metal, identiﬁcation of the

nature of the ligands involved, and determination of any redox

intermediates via a combination of electrochemical, spectro-

scopic, magnetic, and computational studies. Such ligands are

also biologically relevant, where, for example, 1,2-dioxolenes are

thought to chelate to the nonheme iron center in catechol

dioxygenases⁴and α-dithiolene groups are present in the

cofactor molybdopterin found in a variety of metalloenzymes.⁵

Studies on redox active nitroxide (formally aminoxyl) radical

species coordinated to Cu(II) oﬀered a great model system to

2

examine the direct exchange between the d_{x −y}magnetic orbital

on the Cu(II) and the π* molecular orbital of the nitroxide NO

group. Axial coordination of the nitroxide NO group to Cu(II)

Received: March 5, 2013

A

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

has been reported to induce moderate ferromagnetic

interactions while equatorial coordination results in strong

antiferromagnetic exchange.^6c−eThese interactions have been

rationalized by an orbital overlap argument where signiﬁcant

overlap between the magnetic orbital(s) leads to antiferro-

magnetic exchange whereas an orthogonal arrangement of the

magnetic orbital(s) leads to a ferromagnetic interaction. This

degree of overlap which contributes to the antiferromagnetic

component of the exchange is particularly sensitive to the Euler

angles ψ and θ when the nitroxide is equatorially bound while

in the axially bound case for a given θ angle the largest overlap

is when either ψ is close to 90° and θ is close to 0° or 90° or

when ψ is close to 0° and θ is close to 45°. In each case, the

magnitude of the exchange is related to the Mⁿ⁺−O_nitroxidebond

distances.⁸However Luneau and co-workers have reported

ferromagnetic interactions in equatorially bound Cu(II)

nitroxides,⁹the axially bound Cu(II) nitroxide analogue of

the complexes reported here shows antiferromagnetic inter-

actions,¹⁰and large antiferromagnetic interactions were found

center (J = −315 cm⁻¹, using the spin Hamiltonian −2JS₁·S₂).

Here, we report on the synthesis of the congeners [Co^II(L^•)₂]^-

(NO₃)₂(1) and [Co^III(L⁻)₂](BPh₄) (2) with the diamagnetic

complex 2 formed by a reductively induced oxidation of the

central cobalt ion, where the tetraphenylborate anion acts as a

reducing agent. Complex 1 is a rare example of an octahedral

Co(II) ion undergoing a spin transition and exhibiting

ferromagnetic exchange between the coordinating nitroxide

radicals and the central Co(II) ion in a linear L^•−Co^II−L^•

3

arrangement. This stabilizes an S = /₂ground state, which

shows slow magnetic relaxation (i.e., SMM) eﬀects under an

applied DC ﬁeld. The low-temperature (and low-spin) region

oﬀers a unique model system to look at the direct exchange

2

between the d_zmagnetic orbital on the low-spin Co(II) ion

and the π* molecular orbital of the nitroxide NO group,

providing an interesting comparison with previous studies on

2

Cu(II) nitroxides where the magnetic orbital is usually d_{x −y}. It

should be emphasized that these systems act as one-electron

switches where a one-electron reduction of one of the ligands in

solution results in a change in the iron system from the

in Mn(hfac)₂(proxyl)₂(proxyl = 2,2,5,5-tetramethylpyrrolidyl-

ground state.¹¹Clearly, the

diamagnetic [Fe^II(L^•)₂]²⁺dication to the S = /₂[Fe^III(L⁻)₂]⁺

3

1

1-oxy) stabilizing a S =

/

2

magnitude and type of exchange seen in such d-block metal

nitroxides is therefore very sensitive to the geometrical

arrangement of the bound nitroxide and subsequent overlap

with the magnetic orbital(s) of the metal in question. Apart

from the orientation of the radical coordination, this overlap is

also highly dependent on the metal ion d-electron conﬁg-

urations. Metals exhibiting spin crossover, such as iron or

cobalt, when coupled with coordinated radicals, can be ideal

model systems to study how the metal-radical exchange

interaction changes as the central metal ion undergoes a spin

transition. This oﬀers yet another other degree of freedom to

ﬁne-tune the magnetic coupling. In this regard, we have decided

to investigate octahedral Co(II) nitroxides, which form a linear

L^•−Co^II−L^•exchange coupled “hybrid” system exhibiting a

host of interesting magnetic and electronic properties, including

spin crossover, ferromagnetic exchange, reductively induced

oxidation, and ﬁeld-induced slow magnetic relaxation, i.e.,

single-molecule magnetic behavior.

monocation, whereas in the cobalt system presented here, it

changes from the S = ³/₂[Co^II(L^•)₂]²⁺dication to the

diamagnetic [Co^III(L⁻)₂]⁺monocation.

2. EXPERIMENTAL SECTION

2.1. General. 4,4-Dimethyl-2,2-di(2-pyridyl)oxazolidine N-oxide

(L^•) was synthesized as described previously.¹²All other reagents and

solvents were of reagent-grade and used as-received. Microanalyses

were performed by the Campbell Microanalytical Laboratory,

Chemistry Department, University of Otago, Dunedin, New Zealand.

2.2. Syntheses. [Co^II(L^•)₂](NO₃)₂(1). Ten milligrams (10 mg

(0.037 mmol)) of 4,4-dimethyl-2,2-di(2-pyridyl) oxazolidine N-oxide

(L^•) and 5.4 mg (0.0186 mmol) of Co^II(NO₃)₂·6H₂O were dissolved

in 10 mL of acetonitrile. After 5 min of stirring, the resultant brown

solution was ﬁltered and diﬀused with Et₂O to produce X-ray quality

crystals of 1 after 1 day. Yield: 7 mg (52.3%). Anal. Calcd (%) for 1,

C₃₀H₃₂N₈O₁₀Co: C, 49.8; H, 4.5; N, 15.5. Found: C, 50.0; H, 4.5; N,

15.6. IR (ATR cm⁻¹) 1600m, 1571w, 1465m, 1443m, 1409m, 1334s,

1303s, 1267s, 1155w, 1081m, 1022w, 999w, 928w, 807w, 773w.

[Co^III(L⁻)₂](BPh₄) (2). Ten milligrams (10 mg (0.037 mmol)) of 4,4-

dimethyl-2,2-di(2-pyridyl) oxazolidine N-oxide (L^•), 5.2 mg (0.0185

mmol) of Co^II(SO₄)·7H₂O, and 12.6 mg (0.037 mmol) of Na(BPh₄)

were dissolved in 10 mL of methanol. After 60 min of stirring, the

resultant brown precipitate was dissolved in CH₂Cl₂, ﬁltered, and

diﬀused with Et₂O to produce X-ray quality crystals of 2 after 2 days.

Yield: 5 mg (29.4%). Anal. Calcd (%) for 2, C₅₄H₅₂N₆O₄BCo: C, 70.6;

H, 5.7; N, 9.2. Found: C, 70.7; H, 5.8; N, 8.7. IR (ATR cm⁻¹): 3106s,

2974s, 2938w, 2882w, 1606m, 1578w, 1475m, 1461s, 1440s, 1429s,

1362w, 1269w, 1249m, 1219m, 1194w, 1157m, 1142m, 1075w,

1057w, 1030w, 987m, 954m, 941m, 874w, 851w, 767s, 730s, 701s,

682s, 667m, 653m, 606w.

Coordinated 4,4-dimethyl-2,2-di(2-pyridyl) oxazolidine-N-

oxide (Figure 1) exists in its neutral radical (L^•) or

2.3. Magnetic Susceptibility Measurements. Variable-temper-

ature magnetic susceptibility measurements were performed on a

Quantum Design MPMS 7T SQUID magnetometer from 1.8 K to 400

K were in applied DC ﬁelds ranging from 0 to 50 000 G. The SQUID

magnetometer was calibrated using a standard palladium sample

(Quantum Design) of accurately known magnetization or through the

use of magnetochemical calibrants such as CuSO₄·5H₂O. Micro-

crystalline samples were dispersed in petroleum jelly in order to avoid

torquing of the crystallites. Measurements up to 400 K were

performed in a quartz tube. Alternating current (AC) magnetic

susceptibility measurements were undertaken in oscillating ﬁelds of 3

G, frequencies in the range of 1−1500 Hz, and applied DC ﬁelds of 0,

2000, and 5000 G. The sample mulls were contained in a calibrated

capsule held at the center of a drinking straw that was ﬁxed at the end

of the sample rod.

Figure 1. Structural formula of 4-dimethyl-2,2-di(2-pyridyl) oxazoli-

dine N-oxide (L^•).

hydroxylamino anionic form (L⁻), as exempliﬁed by our

previous work on iron nitroxide chelates.¹²In that system, we

postulated that the dication [Fe^II(L^•)₂]²⁺undergoes a one-

electron reduction to form the intermediate [Fe^II(L^•)(L⁻)]⁺in

which an intramolecular electron transfer takes place to

generate the monocationic species [Fe^III(L⁻)₂]⁺. In the

[Fe^II(L^•)₂]²⁺dication, we observed a strong radical−radical

antiferromagnetic coupling via the diamagnetic low-spin Fe(II)

B

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

Table 1. Crystallographic Data for 1 (at 123, 273 and 353 K) and 2

1·123 K

1·273 K

1·353 K

2

formula

M_r

C₃₀H₃₂N₈O₁₀Co

723.57

C₃₀H₃₂N₈O₁₀Co

723.57

C₃₀H₃₂N₈O₁₀Co

723.57

C₅₄H₅₂N₆O₄BCo

918.76

crystal system

space group

a (Å)

monoclinic

P2₁/c

monoclinic

P2₁/c

monoclinic

P2₁/c

monoclinic

P2₁/c

9.3825(10)

14.1324(17)

11.7395(15)

90

9.4853(5)

14.0862(6)

12.0890(6)

90

9.4854(13)

14.0059(17)

12.346(2)

90

12.4781(14)

15.8940(18)

23.212(3)

90

b (Å)

c (Å)

α (°)

β (°)

γ (°)

101.421(3)

90

101.495(5)

90

101.807(13)

90

104.920(4)

90

V (Å³)

T (K)

Z

1525.8

1578.4

1605.5(4)

353(2)

4448.3

123(2)

273(2)

2

123(2)

2

4

ρ_calcd(g cm⁻³

)

1.575

1.523

1.497

1.372

a

λ

(Å)

0.71073

3483

0.71073

3623

0.71073

2815

0.71073

10081

number of ind. reﬂns

number of reﬂns with I > 2σ(I)

number of parameters

number of restraints

2499

2827

1693

5821

225

227

599

0

b

ﬁnal R1, wR2 [I > 2σ(I)]

0.0665, 0.1749

0.0945, 0.1925

1.015

0.0444, 0.1143

0.0600, 0.1247

1.036

0.0584, 0.1281

0.1121, 0.1582

1.052

0.0667, 0.1344

0.1243, 0.1605

0.999

b

R1, wR2 all data

goodness of ﬁt, GOF

largest residuals (e Å⁻³

)

0.925, −0.516

0.580, −0.478

0.482, −0.357

0.834, −0.623

a

b

Graphite monochromators. R1 = ∑ ||F_o| − |F_c||/∑ |F_o|, wR2 = {∑[w(F_o∑F_c²)²]/∑[w(F_o)]}^1/2

.

2

2.4. X-ray Crystallography. X-ray crystallographic measurements

Noodleman.¹⁸The hybrid B3LYP¹⁹functional has been employed

throughout our study. B3LYP has a proven track record of obtaining

good numerical estimates of magnetic exchange in variety of transition

metals, lanthanides, and metal-radical complexes.²⁰⁻²²However, there

are also instances where B3LYP has failed to reproduce the correct

ground state.²³Here, we have employed the Ahlrichs triple-ζ valence

(TZV)²⁴basis set for all of the elements, as implemented in Gaussian

09. We have employed the following notation ^multcomplex_{spin state}for

our study to diﬀerentiate the diﬀerent spin ground states arising from

the exchange coupling and the spin state of the metal ions. Here,

“mult” in the superscript denotes the total multiplicity of the

computed state, while “spin state” in the subscript denotes the nature

of the spin state (low-spin (LS) or high-spin (HS)) at the Co(II)

center. The spin-density plots were generated using Molekel version

4.3 with a cutoﬀ of 0.005, and MO diagrams are plotted using

Chemcraft version 1.6.²⁵Some selected low-lying spin conﬁgurations

were also optimized using acetonitrile as a solvent and by employing

the Polarizable Continuum Model (PCM).²⁶Onsager’s SCRF code

extended by Wiberg and co-workers for Gaussian codes was applied to

our system.^27,28To calculate the magnetic susceptibility from the

computed J values, we relied on the MAGPACK²⁹software.

2.7. Electrochemistry. Voltammetric measurements were under-

taken in acetonitrile (0.1 M [Bu₄N](PF₆)) at 293 2 K under a ﬂow

of nitrogen gas or inside a glovebox using a BAS100B computer-

controlled electrochemical workstation and a standard three-electrode

cell. Glassy carbon (Cypress, 1-mm diameter for 1 and 1.5-mm

diameter for 2) macroelectrodes and a carbon ﬁber (7 μm diameter,

ALS, Japan) microelectrode were used as the working electrode,

whereas a platinum mesh and Ag/AgCl electrode was used as the

counter and reference electrodes, respectively. The procedures

employed for polishing the working electrode are described else-

where.³⁰All potentials given in this paper are referred to the

ferrocene/ferrocenium ([FeCp₂]^0/+) reference couple. Mechanistic

aspects of the voltammetric processes were investigated by applying

the appropriate diagnostic criteria.³¹

on 1 and 2 were performed at 123(2) K using a Bruker Smart Apex X8

diﬀractometer with Mo Kα radiation (λ = 0.7107 Å). Single crystals

were mounted on a glass ﬁber using oil. Higher-temperature data

collections for complex 1 were performed at 273 K and 353 K, using

an Oxford Gemini Ultra diﬀractometer with Mo Kα radiation (λ =

0.7107 Å). Crystallographic data and reﬁnement parameters for 1 and

2, given in Table 1, were solved by direct methods (SHELXS-97), and

reﬁned (SHELXL-97) by full least-squares on all F²data.¹³In complex

1 (at all temperatures), the asymmetric unit contains half a monomer

and one nitrate anion with the inversion center located on the Co ion.

In complex 2, the asymmetric unit contains one complete monomer

and one tetraphenylborate anion. For complexes 1 and 2, all non-

hydrogen atoms are reﬁned anisotropically and all H atoms are placed

in calculated positions. Full crystallographic data for 1 and 2 are

available upon request from the Cambridge Crystallographic Data

Centre, 12 Union Road, Cambridge CB2 1EZ, U.K. (http://www.

ccdc.cam.ac.uk/). The CCDC numbers are 925600 (1·123 K), 925601

(1·273 K), 925602 (1·353 K), and 925603 (2).

2.5. EPR Measurements. X-band (9.4−9.7 GHz) EPR spectra

were recorded with a Bruker ESP380E CW/FT spectrometer. Sample

temperatures from room temperature (293 K) to 110 K were obtained

using the standard rectangular TE₁₀₂rectangular cavity, in conjunction

with a Bruker VT 4111 temperature controller and its associated

nitrogen gas ﬂow insert. Spectra in the temperature range from 100 K

down to 2.2 K were obtained using a Bruker ER4118 dielectric

resonator inserted in an Oxford Instruments (Model CF 935 helium

cryostat). Temperatures below 100 K were calibrated against a

germanium thermometer, using a carbon resistor as a transfer

standard. Microwave frequencies were measured with an EIP

Microwave (Model 548A) frequency counter, and the g-factors were

determined by reference to the F⁺line in CaO (g = 2.0001

0.0001).¹⁴Spectral simulations were performed using the Bruker

XSophe−Sophe-XeprView computer simulation software suite.¹⁵

2.6. DFT Calculations. The X-ray crystal structures of 1 were

subject to DFT¹⁶calculations using the Gaussian 09 suite of

programmes¹⁷to calculate the energetics of diﬀerent spin conﬁg-

urations. Exchange coupling constant (J) values was calculated using

the broken symmetry approach, as developed by Ginsberg and

C

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

well as assignment of the neutral radical (L^•) or anionic form

(L⁻) of the ligand.

3. RESULTS AND DISCUSSION

3.1. Synthesis and Structure. [Co^II(L^•)₂](NO₃)₂(1) was

isolated by reacting 4,4-dimethyl-2,2-di(2-pyridyl) oxazolidine

N-oxide (L^•) with [Co^II(NO₃)₂]·6H₂O in a 2:1 ratio in MeCN.

Diﬀusions of Et₂O into the resultant brown solution produced

crystalline 1 in ∼30% yield. Complex 2 was isolated by reacting

L^•with Co^IISO₄·7H₂O and NaBPh₄in MeOH in a 2:1:2 ratio,

and the resulting brown precipitate was dissolved in dichloro-

methane (DCM), followed by Et₂O diﬀusion to yield crystalline

2 in ∼40% yield. A small number of crystals of 2 could be

obtained by Et₂O diﬀusions of the MeOH ﬁltrate. Complexes

1·123 K, 1·273 K, 1·353 K, and 2 crystallize in the monoclinic

space group P2₁/c, with the asymmetric unit of 1 containing

half the [Co^II(L^•)₂]²⁺dication and one nitrate anion, while 2

contains one complete [Co^III(L⁻)₂]⁺monocation and one

tetraphenylborate anion (see Figures 2 and 3). An inversion

Each tridentate ligand coordinates to the Co ion equatorially

via two pyridyl donors and axially via the oxygen, completing

the coordination sphere (see Figure 2). In 1·123 K, this results

in an axially elongated octahedral geometry (cis, 86.99(13)°−

93.01(13)°; trans, all 180°; Co−O 2.113(3)Å, Co−N 1.959(3),

1.986(3)Å). Upon heating complex 1 we see a trend toward a

more perfectly realized octahedral geometry (see Figure 3 and

Table 2) with a marked increase in the Co−N bond lengths

Table 2. Selected Bond Lengths and Angles for 1·123 K,

1·273 K, and 1·353 K

[Co^II(L^•)₂]²⁺

1·123 K

1·273 K

1·353 K

Bond Lengths (Å)

2.113(3)

Co(1)−O(1)

Co(1)−N(1)

Co(1)−N(2)

O(1)−N(3)

N(3)−C(6)

N(3)−C(13)

C(6)−O(2)

O(2)−C(12)

C(12)−C(13)

C(13)−C(14)

C(13)−C(15)

2.117(2)

2.040(2)

2.056(2)

1.285(2)

1.478(3)

1.480(3)

1.397(3)

1.387(4)

1.510(4)

1.509(4)

1.499(4)

2.107(3)

2.056(3)

2.073(4)

1.285(4)

1.481(6)

1.471(5)

1.394(5)

1.369(6)

1.494(8)

1.504(8)

1.496(7)

1.959(3)

1.986(3)

1.294(4)

1.455(5)

1.478(5)

1.416(5)

1.429(6)

1.547(7)

1.515(6)

1.505(6)

Bond Angles (deg)

Co(1)−O(1)−N(3)

115.4(2)

115.13(13)

115.5(2)

Figure 2. Molecular structure of the dication [Co^II(L^•)₂]²⁺in 1·123 K

(left) and the monocation [Co^III(L⁻)₂]⁺in 2 (right). The structure

and atomic labeling in 1·123 K is representative of 1·273 K and 1·353

K. Hydrogen atoms and anions omitted for clarity. Legend: oxygen,

red; nitrogen, dark blue; and cobalt, turquoise.

and an accompanying small decrease in the Co−O bond length

(1·273 K: cis, 85.44(8)°−94.56(8)°; trans, all 180°; Co−O,

2.117(2) Å; Co−N, 2.040(2), 2.056(2) Å; 1·353K,

84.77(14)°−95.23(14)°; trans, all 180°; Co−O, 2.107(3) Å;

Co−N, 2.056(3), 2.073(4) Å). In 2, there is a slight axial

compression (cis, 85.86(11)°−97.81(11)°; trans, 170.62(10)°−

175.91(12)°; Co−O, 1.879(2), 1.892(2) Å; Co−N, 1.916(3),

1.921(3), 1.928(3), 1.930(3) Å) (see Figure S1 in the

Supporting Information and Table 3). The nitroxide N−O

bond lengths are 1.294(4) Å in 1 and 1.401(3) and 1.411(3) Å

in 2. To accommodate the longer nitroxide N−O bond lengths

in 2 the ﬁve membered ring of the ligand is buckled quite

signiﬁcantly (Figure S2 in the Supporting Information). In 1,

center sits on the cobalt in 1 generating the [Co^II(L^•)₂]²⁺

dication. Both the dication [Co^II(L^•)₂]²⁺and monocation

[Co^III(L⁻)₂]⁺are structurally similar but vary in a few

signiﬁcant bond lengths and angles, which, along with

electrochemistry, EPR, and magnetic studies (vide infra), help

us to assign oxidation and spin states of the central Co ion, as

Figure 3. Molecular structure of the dication [Co^II(L^•)₂]²⁺in 1·123 K (left), 1·273 K (middle), and 1·353 K (right), with relevant bond lengths

given in Ångstroms without uncertainties (see Table 2). Hydrogen atoms and anions are omitted for clarity. Legend: oxygen, red; nitrogen, dark

blue; and cobalt, turquoise.

D

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

temperature structures at 273 K (1·273 K) and 353 K (1·353

K) reveal the gradual disappearance of this pseudo-Jahn−Teller

distortion, such that 1·353 K has a ﬁnal Co−O bond length of

2.107(3) Å and Co−N bond lengths of 2.056(3) Å and

2.073(4) Å (see Figure 3, Table 2). The pseudo-Jahn−Teller

distortion at 123 K is evidenced by the shortening of the four

equatorial Co−N bond lengths (see Figure 3 and Table 2),

which then lengthen as we increase the temperature. The Co−

O bond length is unaﬀected by an increase in temperature, with

the three Co−O bond lengths (2.113(3) Å at 123 K, 2.117(2)

Å at 273 K, and 2.107(3) Å at 353 K) remaining eﬀectively the

same when we consider the standard deviations. Libration

eﬀects³²in high-temperature structural studies leads to bond-

length shortening, which conﬁrms that the overall increase in

bond length seen here represents a signiﬁcant change in the

electronic structure of complex 1. A strong pseudo-Jahn−Teller

Table 3. Selected Bond Lengths and Angles for 2 at 123 K

2, [Co^III(L⁻)₂]⁺

Bond Lengths (Å)

Co(1)−O(1)

Co(1)−O(3)

Co(1)−N(1)

Co(1)−N(2)

Co(1)−N(4)

Co(1)−N(5)

O(1)−N(3)

O(3)−N(6)

N(3)−C(6)

1.879(2)

1.892(2)

1.930(3)

1.928(3)

1.921(3)

1.916(3)

1.411(3)

1.401(3)

1.483(4)

1.487(4)

1.403(4)

1.444(4)

1.519(5)

1.512(5)

1.526(4)

1.472(5)

1.491(5)

1.407(4)

1.448(4)

1.501(5)

1.530(5)

1.515(5)

N(3)−C(13)

C(6)−O(2)

O(2)−C(12)

C(12)−C(13)

C(13)−C(14)

C(13)−C(15)

N(6)−C(21)

N(6)−C(28)

C(21)−O(4)

O(4)−C(27)

C(27)−C(28)

C(28)−C(29)

C(28)−C(30)

1

eﬀect is expected for the low-spin 2E(t_2g⁶e_g) state, in

comparison to the high-spin ⁴T₁(^t_2g⁵e_g²) state for Co(II),

which leads us to propose that the disappearance of such a

distortion as we increase the temperature is consistent with a

gradual spin crossover. The average bond length diﬀerence

around the octahedral Co(II) ion at 123 and 353 K is 0.0424 Å,

which is lower than the range of 0.07−0.11 Å observed for

complete Co(II) spin crossover,³³suggesting that the spin

transition is incomplete at 353 K. This is conﬁrmed by

magnetic susceptibility measurements (vide infra) made up to

400 K.

The overall bond-length shortening and large reduction on

Co−O bond lengths in particular in 2 is consistent with a

diamagnetic low-spin Co³⁺center, where the greater charge

density and unﬁlled antibonding e_gorbital set contributes to the

short distances, as seen previously in the analogous low-spin

complex [Fe^III(L⁻)₂](BPh₄)·0.5H₂O¹²(see Figure S1 in the

Supporting Information and Table 3).

Nitroxide N−O bond lengths are useful in assigning the

ligand to the neutral radical (L^•) or anionic hydroxylamino

form (L⁻). For the conceptually similar ligands Tempo and

Proxyl, typical free ligand nitroxide radical N−O bond lengths

lie between 1.27 and 1.28 Å,^6ccoordinated nitroxide radical N−

O bond lengths lie in the range of 1.261(12)−1.300(3) Å³⁴⁻⁴⁰

and coordinated hydroxylamino anion or hydroxylamine groups

Bond Angles (deg)

Co(1)−O(1)−N(3)

Co(1)−O(3)−N(6)

111.73(18)

111.80(19)

two nitrate anions are present per dication with no signiﬁcant

intermolecular interactions or lattice solvent present and the

shortest distance (O_NO−O_NO) between neighboring nitroxide

N−O groups is 6.586 Å. In 2, one tetraphenylborate anion is

present per cation with, again, no signiﬁcant intermolecular

interactions or lattice solvent present.

The pseudo-Jahn−Teller distortion in 1·123 K contrasts with

the Fe−O distances of 1.876(2) and 1.884(2) Å and the Fe−N

distances in the range of 1.962(2)−1.989(2) Å found in the

analogous low-spin [Fe^II(L^•)](BF₄)₂complex¹²and suggests

we have a typical low-spin d⁷Co ion at 123 K, where the

2

unpaired electron lies in the σ-antibonding orbital d_z. Higher-

Figure 4. (Left) Plot of χ_MT vs T for 1 between 2 K and 300 K in (black circles) and between 5 K and 400 K (red circles, inset) in an applied ﬁeld of

10 000 G. (Right) Plot of M (Nβ) vs ﬁeld (0−50 000 G) at 2 K (top), 3 K, 4 K, 5.5 K, 10 K, and 20 K (bottom). The black solid lines are a guide

only.

E

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

Figure 5. Plot of χ_MT vs T for 1 in the temperature range of 2−300 K (left). The solid red line represents a ﬁt of the experimental data in the

temperature range of 2−119 K to an exchange interaction model shown (right).

have N−O bond lengths in the range of 1.379(5)−1.413(3)

Å.⁴¹⁻⁴³The N−O bond lengths of 1.294(4) Å in 1 suggests

both ligands in 1 are in the neutral radical form (L^•), whereas

the longer N−O bond lengths found in 2 of 1.401(3) Å and

1.411(3) Å suggest both ligands are in the anionic

hydroxylamino form (L⁻). The assignments of a Co(II)

gradual spin-crossover with two neutral radical ligands (L^•) in

1 ([Co^II(L^•)₂](NO₃)₂) and a low-spin d⁶Co(III) ion with two

anionic hydroxylamino ligands (L⁻) in 2 ([Co^III(L⁻)₂](BPh₄))

have been made, not only on the basis of crystallographic data

but also from interpretation of magnetic measurements,

electrochemistry, EPR, and DFT calculations (vide infra).

3.2. Magnetic Studies. Direct current (DC) magnetic

susceptibility measurements were performed on a bulk sample

of crystals of [Co^II(L^•)₂](NO₃)₂(1) in the 300-2 K range

under an applied ﬁeld of 10 000 G (Figure 4, left).

Magnetization measurements on 1 were undertaken between

2 K and 20 K in the applied ﬁeld range of 0−50 000 G (Figure

4, right). Both the 2−300 K magnetic susceptibility and the

magnetization measurements were performed on crystalline

complex 1 in a petroleum jelly mull to minimize any torquing

eﬀects. A second susceptibility measurement was performed

between 5 K and 400 K under an applied ﬁeld of 10 000 G

(Figure 4, left, inset) with polycrystalline sample 1 unre-

strained, in a quartz tube, to conﬁrm the apparent increase in

the χ_MT value as T is increased from ∼225 K. The two

susceptibility measurements have similar, but not identical,

proﬁles, with the only signiﬁcant deviation being at lower

temperatures where the 5−400 K measurement shows a small

maximum at 8.7 K, possibly attributable to torquing eﬀects.

Alternating current (AC) magnetic susceptibility measurements

were performed on crystalline complex 1 between 2 and 20 K,

in applied DC ﬁelds of 0, 2000, and 5000 G and oscillating

frequencies in the range 50−1500 Hz. Direct current (DC)

magnetic susceptibility measurements were also performed on

crystals of [Co^III(L^•)₂](BPh₄) (2) and show that the compound

is diamagnetic, as expected.

then increases rapidly to a broad maximum starting at 40 K

with a χ_MT value of 1.93 cm³mol⁻¹K (Figure 4, left) The χ_MT

value then decreases rapidly upon further cooling, most likely

because of a combination of Zeeman and zero-ﬁeld splitting

eﬀects, reaching a ﬁnal value of 1.37 cm³mol⁻¹K at 2 K. The

maximum χ_MT value of 1.93 cm³mol⁻¹K at 40 K is slightly

3

higher than the expected χ_MT value for a S = /₂ground state

(assuming g = 2.00) of 1.875 cm³mol⁻¹K. This behavior is

1

indicative of ferromagnetic exchange between the S =

/

2

centers on the radical ligands and the low-spin Co(II) S =

3

¹/₂, resulting in a S = /₂ground state.

The χ_MT vs T data in the temperature range of 2−119 K

could be ﬁtted satisfactorily with program PHI,⁴⁴using a 2J

̂

trimer model (Figure 5) with the spin Hamiltonian H =

̂

−2J₁(S₁·S₂+ S₂·S₃) − 2J₂(S₁·S₃) where J₁describes the

interaction between the radicals (S₁and S₃) and the Co(II)

ion (S₂) and J₂the radical−radical interaction via the low-spin

Co(II) ion. The best-ﬁt parameters are as follows: J₁= +63.8

cm⁻¹, J₂= +63.9 cm⁻¹, and g = 2.10.

Since the electron spins are coupled ferromagnetically, a

quartet state lies lowest at energy − J₁− J₂/2, and there are

45

3

doublet states at /₂J₂and 2J₁− J₂/2. This results in a well-

isolated S = ³/₂ground state with the two excited state S = ¹/₂

doublets 191.4 cm⁻¹(3J₁) and 191.6 cm⁻¹(J₁+ 2J₂) higher in

energy. The χ_MT vs T data could also be ﬁtted from 2 K to 205

K (see Figure S3 in the Supporting Information), yielding the

parameters J₁= +58.0 cm⁻¹, J₂= +58.8 cm⁻¹and g = 2.10, also

3

giving an S = /₂ground state, although the ﬁts were of lesser

quality. The excellent ﬁt to the χ_MT vs T data (see Figure 7,

presented later in this paper) below 119 K strongly suggests

that the majority of the Co(II) ions are in the LS state below

this temperature, while the lower-quality higher-temperature

χ_MT vs T ﬁt (see Figure S3 in the Supporting Information)

suggest an increasing population of HS Co(II) ions. These ﬁts

do not take into account the contributions to the magnetic

susceptibility from the introduction of the spin crossover of the

Co(II) ion and possible antiferromagnetic exchange (vide infra)

between the emerging high-spin Co(II) ions and the radicals,

which becomes more prevalent at higher temperatures. The

radical−radical exchange interaction via the low-spin Co(II) ion

For complex 1 measured between 2 K and 300 K, the χ_MT

value of 1.69 cm³mol⁻¹K at 300 K decreases slowly upon

cooling to a χ_MT value of 1.56 cm³mol⁻¹K at 218 K, which

F

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

Figure 6. Plot of M (Nβ) vs ﬁeld (0 − 50,000 G) at (top), 3, 4, 5.5, 10, and 20 K (bottom). The solid red lines represent ﬁts of the experimental data

with the parameters shown and in the text, D being positive (left) and negative (right).

is of a similar magnitude to the radical-Co(II) ion interaction

(∼60 cm⁻¹), which seems counterintuitive, but we have

previously reported a large radical−radical (J₂) antiferromag-

netic exchange (J = −315 cm⁻¹using the spin Hamiltonian

to examine the eﬀect on the χ″_Mvalues and their frequency

dependence with static DC ﬁeld variations. These measure-

ments were undertaken at 2 K in a 3.5 G AC ﬁeld oscillating

with frequencies ranging from 1 Hz to 1500 Hz in applied static

DC ﬁelds of 2000−15 000 G. At 2000 G, the χ″_Mvalue reached

a maximum of 0.22 cm³mol⁻¹at 1500 Hz and 2 K, while at

5000 G, the χ″_Mvalue peaks around a maximum of 0.18 cm³

mol⁻¹at 1400 Hz and 2 K with an obvious downturn at higher

frequencies. To investigate this further, AC susceptibility

measurements were undertaken under static DC ﬁelds of

2000 and 5000 G and plotted as χ″_Mvs T (Figure 7). Upon

cooling, the imaginary component of the susceptibility χ″_M

shows some frequency dependence, ﬁnally reaching a value of

0.24 cm³mol⁻¹at 1500 Hz and 1.8 K in 2000 G. In a 5000 G

static DC ﬁeld, we observe a peak maxima centered at 2 K and

1500 Hz with a χ″_Mvalue of 0.17 cm³mol⁻¹. To investigate this

possible SMM-like behavior, we plotted ln (χ″_M/χ′_M) vs 1/T

for 1 at diﬀering frequencies, and applying linear ﬁts, we have

estimated E_a≈ 10.6 K and τ° ≈ 10⁻⁶s at 2000 G and E_a≈ 7.6

K and τ° ≈ 10⁻⁶s in 5000 G (see Figure S5 in the Supporting

Information). The appearance of such temperature- and

frequency-dependent peaks in the AC susceptibility measure-

ments under an applied DC ﬁeld suggests a negative D-value

whose magnitude is at least 3.7 cm⁻¹, when considering the

relationship concerning the upper limit of the energy barrier for

̂

−2J(S₁·S₂) via the low-spin Fe(II) ion in the analogous

compound [Fe^II(L^•)₂](BF₄)₂.¹²The magnitude and type of

exchange between the radical and metal ion is highly dependent

on geometrical parameters, which aﬀects the degree of overlap

between the metal and ligand-based orbital(s). The interaction

between the radical π* orbital and the Co(II) magnetic orbital,

2

which is expected to be mainly d_zin character, in this case, is

ferromagnetic, which suggests that they are orthogonal in

1

nature. This ﬁt treats the low-spin Co^IIas an isotropic S = /₂

with a g-value of >2, because of close lying excited states. The

possibility of a small residual fraction of the high-spin Co(II)

dication and admixture of closely lying excited states into the

ground state are not explicitly accounted for in the model.

Indeed the M vs H data (Figure 4) do not saturate to the

3

expected M value of 3 Nβ for an isolated S = /₂ground state,

even at 2 K in an applied ﬁeld of 50 000 G. The M vs H data

could be ﬁtted satisfactorily with PHI,⁴⁴using the giant spin

3

approximation with S =

/ (see Figure 6) yielding the

2

parameters g = 2.03 and D = +2.89 cm⁻¹. The corresponding

Hamiltonian is given by eq 1:

2

̂

⃗

H = DS_z+ E(S_x− S_y) + gμ_BB·S

(1)

1

half-integer spins U = (S²− /₄)|D| and is consistent with the

magnitude of ∼2.7 cm⁻¹for D found in the ﬁts of M (Nβ) vs

ﬁeld (see Figure 6). Although we cannot unambiguously

determine the sign of D from ﬁts to the M vs H data, we were

unable to simulate the EPR spectrum at 2.2 K (see Figure 9,

presented later in this work) with D < 0, the simulations

yielding positive D and nonzero E values.

2

where D is the axial anisotropy, μ_Bthe Bohr magneton, S_zthe

easy-axis spin operator, and B the applied ﬁeld. However, these

data could also be ﬁtted with a negative D value, yielding the

parameters g = 2.03 and D = −2.74 cm⁻¹. The combination of

the susceptibility and magnetization ﬁts give us a consistent

3

view of an S = /₂system with a g-value greater than the free-

Similar ﬁeld-induced slow magnetic relaxation eﬀects in

materials with positive D values were observed in the octahedral

mononuclear Co(II) complex, cis-[Co^II(dmphen)₂-

(NCS₂)]·0.25EtOH, where dmphen = 2,9-dimethyl-1,10-

phenanthroline, and these were due to a transverse anisotropy

barrier governed by the E parameter.^46aIn this case, it may be

expected that, for the M_s= ¹/₂, Kramers doublet ground-state

quantum tunneling eﬀects would prevent such a preferential

axis in the xy-plane; however, upon applying a DC ﬁeld, these

eﬀects are quenched. In that work,^46athe reported energy

barrier for the relaxation of the magnetization ΔE = 16.2−18.1

electron value and a nonzero axial zero-ﬁeld splitting parameter

D. The assignment of an S = /₂ground state and nonzero D

3

value is conﬁrmed by EPR measurements (vide infra), which

suggest we also have a nonzero E (rhombic zero-ﬁeld splitting)

parameter that allows mixing of the pure m_Sstates within the S

3

= /₂manifold.

AC susceptibility data were obtained on 1 with measure-

ments undertaken between 1.8 K and 20 K in zero DC ﬁeld

with a 3.5 G AC ﬁeld oscillating with frequencies ranging from

50 Hz to 1500 Hz. In zero ﬁeld, the imaginary component of

the susceptibility χ″_Mis negligible and close to zero, leading us

G

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

ﬁelds. Additional compounds which showed slow magnetic

relaxation eﬀects with D > 0 include the mixed-valence SCM

[Fe^II(ClO₄)₂{Fe^III(bpca)₂}]ClO₄where Hbpca = bis(2-

pyridylcarbonyl)amine^46fand the heterometallic complex

[Fe₃Cr(L)₂(dpm)₆]·Et₂O,^46gwhere H₃L = 2-hydroxymethyl-

2-phenylpropane-1,3-diol and Hdpm = dipivaloylmethane.

It may be expected, then, that if the experimentally derived

energy barrier for the relaxation of the magnetization for

complex 1 is similar to the transverse anisotropy barrier of ΔE

= 2E, this would suggest that the relaxation mechanism involves

a transverse anisotropy barrier; however, if it is similar to the

energy gap between the M_s= ¹/₂and M_s= ³/₂states, this

would imply that the relaxation mechanism follows an Orbach

pathway. Unfortunately, we have no direct experimentally

derived values for the energy barrier for the relaxation of the

magnetization due to the absence of discernible peaks in the

χ″_Mvs T plots (the reported values previously are estimations

only (see Figure S5 in the Supporting Information)).

Therefore, we cannot compare the experimental energy barrier

for the relaxation of the magnetization with the theoretical

transverse anisotropy barrier or indeed the energy gap between

the M_s= ¹/₂and M_s= ³/₂states, so we cannot comment,

with any degree of conﬁdence, on the exact nature of the

relaxation mechanism in complex 1.

The radical-based complex [Co(hfpip)₂(4NOpy)₂] in the

diluted frozen solution reported by Koga et al.^46cis a S = /₂

3

SMM with a U_effvalue of ∼29 K and highlights the potential

that even monometallic radical species, such as those reported

here, have in this regard.

The previous DC susceptibility ﬁts are based on the low-

temperature magnetic data; however, as we increase the

temperature, we see an increase in the χ_MT value of 1.56 cm³

mol⁻¹K at 218 K to a value of 1.69 cm³mol⁻¹K at 300 K (see

Figure 4). A second measurement undertaken between 5 K and

400 K (Figure 4, inset) conﬁrms this increase in the χ_MT values,

ﬁnally reaching a value of 2.13 cm³mol⁻¹K at 400 K, which

suggests the possibility of the Co(II) center undergoing a

gradual spin-crossover transition. In cobalt(II) spin-crossover

d⁷systems, only one electron is transferred to the antibonding

e_gorbital set, which results in a smaller average bond length

change of ∼0.10 Å, compared to ∼0.20 Å seen in typical Fe(II)

d⁶spin transitions. In addition, the majority of the spin-

transition curves for Co(II) compounds are of a gradual nature

Figure 7. Plot of χ_M″ vs T for complex 1 in an applied DC ﬁeld of

2000 G (top) and 5000 G (bottom) at frequencies shown.

cm⁻¹matched the theoretical transverse barrier for rotation in

the xy plane of ΔE = 2E = 16.8 cm⁻¹given the reported

experimental E value of 8.4 cm⁻¹. In addition, a high-ﬁeld EPR

spectroscopic study on model compounds based on the single-

46b

chain magnet (SCM) (DMF)₄MnReCl₄(CN)₂

found that

the slow magnetic relaxation eﬀects in (DMF)₄MnReCl₄(CN)₂

were due to the transverse E term associated with the Re^IVion,

where the quantum tunneling eﬀects were suppressed, in that

case, by the intrachain spin correlations. In both cases

mentioned above, slow magnetic relaxation eﬀects were

observed with the presence of an easy-plane anisotropy (D >

0) and a signiﬁcant E component. However, in a pseudo-

tetrahedral Co(II) complex [(3G)CoCl](CF₃SO₃),^46dwhere

3G = 1,1,1-tris[2N-(1,1,3,3-tetramethylguanidino)methyl]-

ethane, AC data showed ﬁeld-induced slow magnetic relaxation

with positive D and E values; however, in this case, it was

suggested that the relaxation mechanism followed an Orbach

pathway through the excited state M_s= ³/₂levels after a ﬁeld-

induced bottleneck of the direct relaxation between the ground

M_s= ¹/₂states. This was rationalized by consideration of the

experimental energy barrier for the relaxation of the magnet-

ization ΔE = 24 cm⁻¹, which closely matched the separation

between the M_s= ¹/₂and M_s= ³/₂states and not the

theoretical transverse barrier. A similar Orbach mechanism was

also reported for the trinuclear Cu(II) complex [Bu₄N₂]-

[Cu₃(μ-Cl)₂(μ-pz)₃Cl₃],^46ewhere pz = the pyrazolato anion,

which showed slow magnetic relaxation eﬀects under applied

and complementary techniques such as Mossbauer spectrosco-

̈

py, used for Fe(II), are unavailable for Co(II) systems. In the

next section, we see that EPR spectroscopy is useful in

identifying the doublet state (^{2E(t 6}e_g¹)) of low-spin Co(II)

2g

giving valuable information on the local symmetry, anisotropy,

and splitting of the ground state. Spin−lattice relaxation eﬀects

broaden the resonances from high-spin Co(II) to such an

extent that they are usually only observable at very low

temperatures. Temperature-dependent magnetic susceptibility,

however, remains the main technique for studying Co(II) spin

crossover, but even this remains diﬃcult with ligand-ﬁeld

strength, spin−orbit coupling eﬀects, and doublet and quartet

levels mixing in the spin-crossover region making an

unambiguous magnetic analysis very diﬃcult.⁴⁷For example,

the 400 K χ_MT value of 2.13 cm³mol⁻¹K in complex 1 could

represent anywhere between a 60% and 80% incomplete

gradual spin crossover. Fortunately, in this case, the low-spin

octahedral Co(II) ion exhibits a strong pseudo-Jahn−Teller

distortion, which we have monitored by means of variable-

temperature crystallographic measurements (see Figure 3,

H

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

Table 2). At 123 K, we have an unambiguous low-spin Co(II)

ion, exhibiting a pseudo-Jahn−Teller distortion, which

disappears with an increase in temperature up to 353 K. The

disappearance of this distortion suggests complex 1 undergoes

an incomplete, gradual spin transition up to 400 K, which is

consistent with the magnetic susceptibility measurements. In

summary, compound 1 is a rare, perhaps unique, example of a

mixed radical−d-block complex displaying magnetic exchange,

spin-crossover features, and slow magnetic relaxation (SMM)

eﬀects.

∼2200 G. Although the g ≈ 2 resonance had a similar peak-to-

peak derivative width at 110 K, since, at 295 K, there appeared

to be rather more intensity in the wings at the lower

temperature (Figure 8). As shown in Figure 8, further

reductions in temperature resulted in a narrowing and an

increase in the intensity of the feature at g ≈ 3.7, relative to that

at g ≈ 2, and a rather less-symmetric appearance of the latter. At

temperatures below ∼20 K, the g ≈ 3.7 resonance was of

similar intensity to the g ≈ 2 resonance and began to show a

resolved feature on the low-ﬁeld side at with a peak at ∼1550

G. The shoulder feature on the g ≈ 3.7 resonance and the

change in shape of that at g ≈ 2.0 are most clearly seen at 2.2 K,

as shown in Figure 8. It can also be seen that, as well as

becoming less symmetric, the g ≈ 2 resonance had broadened

to a width of ∼150 G and shifted to a g-value of 1.99.

3.3. EPR Spectroscopy of [Co^II(L^•)₂](NO₃)₂(1). At 295 K,

the spectrum of 1 over the ﬁeld range from 50 G to 6000 G

only exhibited an intense symmetric resonance of width ∼80 G

and centered on g ≈ 2.003, as shown in Figure 8. The spectrum

remained unchanged until the temperature was reduced to

∼120 K, when a broad but rather less-intense resonance could

be discerned at g ≈ 3.7, which extended from 1200 G to above

The ﬁt of the 2.2 K spectrum was approached using the

SOPHE program¹⁵and two diﬀerent models, as shown in

Figure 9. The ﬁrst, shown in Figure 9a, assumed that the

Figure 9. Comparison of simulated spectra (a, b) as described in the

text with experimental spectrum (c) of 1 recorded at 2.2 K (see Figure

10). The spin Hamiltonian and linewidth parameters for the simulated

spectrum are listed in Table 4. Spectrum (a) is for “high spin Co (II)”

1

Sz =

/ doublet and Gaussian line shape; spectrum (b) is for

2

3

SOPHE with S = /₂and Lorentzian line shape.

resonances were due to M_s= 1 transitions within a Kramers

1

doublet with eﬀective spin S = /₂. The second assumed a

system with spin S = ³/₂, where the zero-ﬁeld splitting

parameters D and E were two of the variables (Figure 9b). The

parameters giving the best ﬁts to the experimental spectrum are

listed in Table 4.

We note that the narrow resonances at ∼1750 G and 4000 G

are due to cavity artifacts and the “hump” at ∼3400 G is most

likely due to a radical impurity.

1

The simulation of Figure 9a, assuming a spin S = /₂, and

gave eﬀective g-values of g_x= 4.50, g_y= 3.71, and g_z= 1.955,

which are clearly not appropriate for S = ¹/₂for either a radical

or for low-spin Co(II). They are also not consistent with

Figure 8. EPR spectra of polycrystalline 1 at the temperatures

indicated. Spectrometer settings: upper spectra (295 and 110 K),

microwave frequency = 9.434 GHz; microwave power = 5.26 mW;

receiver gain = 1.0 × 10⁵; 100 kHz modulation amplitude = 0.2 mT;

ﬁeld scan range/time = 600 mT/84 s; time constant = 41 ms. Lower

spectra (80 K, 20 K, 6.8 K, 2.2 K): microwave frequencies = 9.704

GHz (80 K, 20 K, 6.8 K), 9.691 GHz (2.2 K); microwave power = 1.05

mW (80 K), 10.5 μW (20 K, 6.8 K, 2.2 K); receiver gain = 1.0 × 10⁴

(80 K), 1.0 × 10⁵(20 K), 2.5 × 10⁴(6.8 K, 2.2 K); 100 kHz

modulation amplitude = 0.1 mT; ﬁeld scan range/time = 600 mT/84

s; time constant = 41 ms.

1

expectation for transitions within the low-lying M_S=

/

2

Kramers doublet of high-spin Co(II), for the following reasons:

(i) the average g-value here is ∼3.3, whereas, for most

Co(II) high-spin systems, it is ∼4.3,⁴⁸and

(ii) the hyperﬁne interactions obtained through the simu-

lation appear to be rather too small for a high-spin

Co(II) system.⁴⁹

We also note here that a zero-ﬁeld splitting of ≥50 cm⁻¹is

considered appropriate for six-coordinated high spin Co(II),

I

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

low-spin Co(II) than those listed in Table 1 and are consistent

with previous studies on a series of spin-crossover Co(II) alkyl

chain terpyridine-based complexes.⁵³

Table 4. Spin Hamiltonian and Linewidth Parameters Used

To Simulate the Spectra Shown in Figure 9

a

1

3

“S = /₂”

SOPHE,¹⁵S = /₂

The characteristics of the spectra observed at temperatures

between 2.2 K and 295 K depend on the spin-relaxation

behavior of the individual components and the trimeric spin

system as a whole. For low-spin Co(II) ions, spin−lattice

relaxation eﬀects generally result in line broadening above

∼100 K, and these would appear to be the cause of the

disappearance of the broad resonances at ∼1700 G above 110

K. As discussed by Gatteschi,^6dspin relaxation at higher

temperatures can occur via the doublet states which become

populated at these temperatures. The spin-crossover transition

also introduces a gradually increasing population of high-spin

Co(II) above 120 K. The short spin−lattice relaxation time of

the Co(II) ion means that any resonances associated with it will

be so broad that they would be unobservable at higher

temperatures. The seemingly isotropic resonances at g ≈ 2.003

observed in the high-temperature region (above 200 K) would

appear to result from the averaging out of the anisotropic g- and

hyperﬁne interactions of the nitroxide radicals, as a

consequence of these spin-relaxation pathways.

g_x

4.50

3.71

1.955

80

2.090

2.050

1.990

45

g_y

g_z

A_x

A_y

60

25

A_z

12

D

7.0

E/D

σ_x

0.064

60

0.075

20

σ_y

50

20

σ_z

25

20

δg_x/g_x

δg_y/g_y

δg_z/g_z

0.08

0.02

0

−0.08

0.025

0

a

Hyperﬁne parameters, and linewidths are in units of 10⁻⁴cm⁻¹, while

the zero-ﬁeld splitting parameter D is in cm⁻¹. Uncertainties for the S

3

= /₂SOPHE simulation spin Hamiltonian parameters are estimated

as follows: g-values, 0.005; A-values (× 10⁻⁴cm⁻¹) A_x, A_y5, A_z

2; D 1.0 cm⁻¹; E/D 0.005. The ratio E/D for the “S = /₂” case

1

3.4. Electrochemistry. Unlike the iron(II) complex,

[Fe^II(L^•)₂](BF₄)₂,¹²EPR measurements on the conceptually

and structurally related congener [Co^II(L^•)₂](NO₃)₂, 1 did not

show any contribution arising from the mixed-valent

[Co^III(L^•)(L⁻)]²⁺species, which can be formed, presumably,

via an intramolecular metal-to-ligand charge-transfer reaction.

Thus, in solution [Co^II(L^•)₂]²⁺(1) is likely to be the

thermodynamically favored form, as is its oxidized form

[Co^III(L⁻)₂]⁺(2). Clearly, electrochemical studies on 1 and 2

may provide details of the relationship between 1 and 2 in a

redox sense, as was the case with the iron analogues where the

reaction scheme was deduced to be as depicted in Scheme 1.¹²

was calculated using the g-values listed here and the equations of

Pilbrow.⁵⁴

and this is rather larger than that found here (vide infra).⁵⁰The

most viable interpretation of the EPR spectrum at 2.2 K is that

suggested by a consideration of the X-ray crystallographic and

magnetic susceptibility measurements, namely, that it arises

from an approximately linear trimeric spin system composed of

a low-spin (S = ¹/₂) Co(II) ion bracketed by two radicals, each

1

with spin S = /₂. As described above, for the present linear

trimeric system, where J₁and J₂are both strong and

ferromagnetic (J₁= 63.8 cm⁻¹and J₂= 63.9 cm⁻¹), a quartet

state (S = ³/₂) lies lowest in energy and is well-isolated from the

two S = ¹/₂doublets, which lie 3J₁and J₁+ 2J₂higher in energy,

relative to the ground state. Thus, the resonances observed at

2.2 K arise from transitions within the quartet state, and the

simulations shown in Figure 9b, proceed on this basis, giving

the parameters listed in Table 4. In order to arrive at the

simulated spectrum shown in Figure 9b using SOPHE, it was

necessary to assume that the x- and y-axes of the zero-ﬁeld

splitting tensor were interchanged, relative to those of the g-

and A matrices.

Scheme 1. Square Scheme Depicting the Diﬀerent Species

Involved in the Redox Reaction of [Fe^II(L^•)₂](BF₄)₂.¹²

The voltammetry (Figure 10) of [Co^III(L⁻)₂](BPh₄), (2), in

acetonitrile (0.1 M [Bu₄N](PF₆)) at a 1.5 mm glassy carbon

(GC) electrode under transient conditions (Figures 10a and c,

scan rates = 20−200 mV s⁻¹) and near-steady-state conditions

(Figure 10b, scan rate = 1 mV s⁻¹), resembles that reported for

the [Fe^III(L⁻)₂](BPh₄) congener, but exhibits some additional

complexity.¹²Thus, four processes, labeled I, II, III, and IV in

Figure 10 are found in the potential range of +1.0 V to −1.7 V

vs [FeCp₂]^0/+in this solvent. Other irreversible processes were

detected outside this potential range (as also applies to 1), but

are not relevant to the theme of interest in this paper and,

therefore, are not discussed. The transient cyclic voltammo-

grams associated with the designated processes I, II, III, and IV

in Figure 10a reveals that chemically irreversible process IV,

which corresponds to the oxidation of [BPh₄]⁻,¹²leads to

diﬃculties in the analysis of process I, which occurs at a slightly

more positive potential. Chemically reversible process II is

diﬀusion-controlled, as evidenced by the linear dependence of

both reduction and oxidation peak currents on the square root

The spectrum observed at temperatures in the liquid helium

region arises from transitions between the levels of the Sz =

¹/₂Kramers doublet of the quartet state, which, since D is

positive, lies lowest and is the only doublet with a signiﬁcant

3

spin population at 2.2 K. A negative D results in the Sz = /₂

doublet having a signiﬁcant population at 2.2 K and exhibiting

resonances at g_z≈ 6 (∼1100 G) and at magnetic ﬁelds

approaching those for which g ≈ 0. These resonances are not

observed. The eﬀective g-values of the lowest doublet arising

from the quartet state of a S = ³/₂spin system derived from the

1

coupling of three S = /₂entities can be derived from the

expressions of Banci et al.⁵¹for a linear, exchange-coupled

trimer. Using the g-values for nitroxide radicals, estimated from

the spectrum of a nitroxide radical in the analogous system

[Fe^II(L^•)₂](BF₄)₂,¹²which are similar to those given by

Dzuba,⁵²we estimate the “real” g-values for the uncoupled

low-spin Co (II) ion as g_x= 2.25, g_y= 2.14, and g_z= 1.97. These

g-values are rather more in accordance with the expectation for

J

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

The steady-state E_1/2value for oxidation process II is −0.03 V,

as is the midpoint potential (E_m) derived from cyclic

voltammetry from the average of the oxidation and reduction

peak potentials. By analogy, like the iron congener, this is

assumed to occur as in square scheme 2, where K_1eqand K_2eq

Scheme 2. Square Scheme Depicting the Diﬀerent Species

Involved in the First Oxidation Step (Process II) of 2

are equilibrium constants. Further evidence for this reaction

will be provided below when the voltammetry of 1 is presented.

Consequently, process II is ascribed mainly to the oxidation of

one of the monoanionic forms of the oxazolidine−N-oxide

ligands to yield [Co^III(L^•)(L⁻)]²⁺. This species then presum-

ably undergoes rapid intramolecular ligand-to-metal charge

transfer to form the thermodynamically favored [Co^II(L^•)₂]²⁺

species (1), which contains the original radical form of the

ligand. The cobalt(II) species, along with the equilibrium

concentration of [Co^III(L^•)(L⁻)]²⁺, is further oxidized via

process I (E_1/2= 0.61 V), as in eq 3, to generate the tricationic

cobalt(III) species, [Co^III(L^•)₂]³⁺.

−e⁻

•

[Co^II(L )₂]²⁺HoooI [Co^III(L )₂]³⁺

(3)

The voltammetry of complex 1 as a nitrate salt in acetonitrile

(0.1 M [Bu₄N](PF₆)) is displayed in Figure S6 in the

Supporting Information. Some of the characteristics predicted

on the basis of the voltammetry of 2 and the relationships of

their structures is evident, but also some additional features are

observed. Since 1 is a nitrate salt, process IV arising from

oxidation of [BPh₄]⁻present as the anion for 2, is of course

absent. This enables process I to be more readily studied in the

case of 1 than 2. Cyclic voltammetry at a GC macrodisk

electrode now clearly reveals a well-deﬁned chemically

reversible, diﬀusion-controlled, one-electron oxidation step

with E_mat +0.60 V (see Figures S6a and S6b in the Supporting

Information), which corresponds to the reaction in eq 3, for

both 1 and 2. Process II for 1 also is well-deﬁned, with E_m=

−0.04 V, but the peak current is smaller than that for process I.

Importantly, the steady-state voltammogram at a carbon ﬁber

microelectrode in Figure S6c in the Supporting Information

conﬁrms that process II, which is derived from 1, is now a

reduction rather than oxidation step, as found for 2. This is as

expected on the basis of square scheme 3, which is the

Figure 10. (a) Transient cyclic and (b) linear sweep near-steady-state

voltammograms obtained with a 1.5-mm-diameter glassy carbon disk

electrode, using scan rates of 100 and 1 mV s⁻¹, respectively, for ∼1

mM 2 in CH₃CN (0.1 M [Bu₄N](PF₆)) at 298 K. (c) Cyclic

voltammograms as a function of scan rate.

of the scan rate, whereas process III, while diﬀusion-controlled

at the reduction peak potential, is chemically irreversible.

Analysis of the relative magnitudes and signs of limiting

currents and wave shapes associated with the slow-scan-rate

near-steady-state current−potential voltammetric response at a

slow scan rate of 1 mV s⁻¹at a glassy carbon macrodisk

electrode in Figure 10b reveals that 2 gives rise to two one-

electron oxidation processes I and II (current positive in Figure

10b) and one reduction process (current negative in Figure

10b), in addition to multielectron process IV associated with

oxidation of the [BPh₄]⁻anion. Process III has a halfwave

potential of E_1/2= −1.22 V and is assigned to a metal-based

Co(III)/Co(II) reduction step, as proposed in eq 2 to give the

corresponding neutral [Co^II(L⁻)₂] product, which, unlike the

iron analogue, is unstable on the voltammetric time scale.

Scheme 3. Square Scheme Depicting the Diﬀerent Species

Involved in the First Reduction Step (Process II) of 1

+e⁻

fast

−

[Co^III(L )₂]⁺HoooI [Co^II(L )₂] ⎯⎯→ products

(2)

K

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

Table 5. Reversible Potentials (V vs [FeCp₂]^0/+) for

Designated Processes I, II, and III Exhibited by the Fe and

Co Complexes

reduction counterpart of reaction scheme 2. Figure S6b in the

Supporting Information reveals that the potential region, which

is slightly more negative than process II in 1, is now more

complicated than when starting with 2, because of the presence

of processes V and VI. The steady-state response for process V

has a nonsigmoidal peak shape, which suggests that

precipitation or adsorption of a product has occurred so that

this process corresponds to a surface conﬁned rather than

solution-phase reaction. Indeed, if the initial potential is held in

the region of processes V and VI, then severe electrode fouling

and blocking of processes II and I occurs. That is, the

voltammetry observed on scanning in the positive potential

direction is now poorly deﬁned. Also noteworthy for 1 is that

process II, in combination with process V (see Figures S6b and

S6c in the Supporting Information), has a current magnitude

corresponding to that for process I and, hence, represents the

total of a one-electron reduction in a coulometric sense. It is

likely that the nitrate anion is not innocent in the voltammetry

of 1. The analogue of irreversible process III reported for 2 is

located at approximately −1.24 V (cyclic voltammetry scan rate

100 mV s⁻¹) in the case of 1. This process, since it is derived in

part from the reduction of products formed in processes V and

VI, can be designated as process VII (not shown). In summary,

the introduction of surface-conﬁned electrochemistry in 1

modiﬁes process II and, hence, all processes at more-negative

potentials in a manner that does not apply to 2. Nevertheless,

the electrochemical data, despite being far more complex than

that for the corresponding iron compound, where processes I,

II, and III are all fully reversible,¹²does explain why the

chemical oxidation of 1 leads to structurally rearranged 2 and

vice versa, given that square reaction schemes are still operative.

Taken together, the voltammetric data for the two cobalt

complexes, 1 and 2, along with their iron analogues, clearly

establish that the four members of this family essentially share

inter-related voltammetric features, i.e., exhibit three one-

electron processes, I, II, and III, with a square scheme

providing chemical rearrangement with equilibrium constants

K_1eqand K_2eqthat couple two-electron transfer processes. As

seen for the iron system, the redox chemistry of the cobalt

compounds 1 and 2 underscores the fact that the oxazolidine−

N-oxide ligand is a noninnocent ligand and is the main cause of

the redox-induced structural changes associated with square

reaction schemes 2 and 3. Very signiﬁcant in this regard is the

chemical synthesis of the monocations [Co^III(L⁻)₂]⁺and

[Fe^III(L⁻)₂]⁺achieved from their parent [Co^II(L^•)₂]²⁺and

[Fe^II(L^•)₂]²⁺complex via a fast, reductively induced inter-

molecular metal-to-ligand charge transfer that takes place upon

one-electron reduction of the ligand. Table 5 provides a

comparison of the reversible potentials for processes I, II, and

III for the cobalt and iron congeners, and it shows that the

diﬀerences are not very large.

I

II

III

ref

[Fe^II(L^•)₂]²⁺

0.68

0.60

0.62

−0.10

−0.04

−0.03

−1.15

−1.24

−1.22

12

[Fe^III(L⁻)₂]⁺

12

a

c

[Co^II(L^•)₂]²⁺(1)

[Co^III(L⁻)₂]⁺(2)

this work

ab

,

a

An irreversible process is involved, so the peak potential value is

b

given. Overlap with [BPh₄]⁻oxidation process introduces signiﬁcant

uncertainty. An irreversible process is involved, so the E_1/2value is

c

given.

3

very strong radical−radical interaction ensures an S =

ground state (⁴1_HS).

/

2

Here, we have considered both the HS and LS Co(II) cases

and extracted the exchange coupling for each to rationalize the

observed magnetic behavior. The exchange interaction

computed for the HS Co(II) case will be especially useful,

because there are no experimental data on the type, sign, and

magnitude of any possible exchange coupling due to diﬃculties

in extracting such parameters in a system that has simultaneous

spin-crossover, exchange, and temperature-dependent HS

Co(II) behavior.

3.5.1. Structure and Energetics. Structure optimizations

were performed for conﬁgurations ⁴1_LSand ⁴1_HS(see Table S1

in the Supporting Information for the computed spin-density

values), and the computed structural parameters for conﬁg-

4

urations 1_LSand 1_HSare given in Figure 11. Although some

structural parameters are overestimated, compared to the

experimental structural parameters,⁵⁵it is apparent from the

4

given selected structural parameters that structure 1_LSis a

better match to the experimental structure determined at 123 K

4

(see Figure 11) while 1_HScorresponds to the structure

determined at 353 K (see Figure 3 for experimental bond

4

lengths). Also, 1_LSis expected to have a pseudo-Jahn−Teller

distortion and the computed structure nicely reproduces this

distortion along the O−Co−O vector, as seen in the X-ray

4

structure. For conﬁguration 1_HS, computations reveal larger

metal−ligand bond lengths as expected for high-spin Co(II)

complexes.

For comparison, computation has also been performed on

the sextet state (⁶1_HS) corresponding to ferromagnetic

exchange between both the radical ligands and HS Co(II).

4

6

Calculations reveal that 1_LSis the ground state with 1_HSand

⁴1_HSlying 1.1 and 5.6 kJ/mol higher in energy, respectively.

Our energetics conﬁrm that ⁴1_LSis the ground state and this is

3

consistent with the experimental observation of an S = /₂

ground state at low temperature, resulting from a strong

ferromagnetic interaction between the LS Co(II) ion and the

radical ligand spins.

3.5.2. Magnetic Exchange. The exchange parameters have

been computed on the X-ray structure assuming (a) Co(II) in

the LS state and (b) Co(II) in the HS state. Two diﬀerent

exchange interactions have been assumed based on the spin

3.5. Theoretical Studies. Experimental measurements

3

suggest that the ground state for complex 1 is S = /₂; this

can arise from several diﬀerent spin-exchange coupled

scenarios. The ﬁrst scenario considered involves two radical

ligands coupled ferromagnetically to the LS Co(II) ion and a

simultaneous ferromagnetic superexchange between the two

radical ligands mediated via the LS Co(II) ion and corresponds

to the experimental observations (⁴1_LS). The second scenario

involves a HS Co(II) case, where the two radical ligands are

coupled ferromagnetically to each other with a weak

antiferromagnetic coupling with the central Co(II) ion. A

̂

Hamiltonian H = −2J₁(S₁·S₂+ S₂·S₃) − 2J₂(S₁·S₃) (where S₁

and S₃are for the radicals and S₂is for the Co^IIion). For

conﬁguration 1_LS, the parameter set J₁= +67.3 cm⁻¹and J₂=

4

+15.1 cm⁻¹has been obtained. This gives rise to an S = /₂

3

ground state with two S = /₂levels at 97.5 cm⁻¹and 201.9

1

cm⁻¹, respectively. Although the estimate of J₁matches with the

experimental (susceptibility) data, the magnitude of J₂is

L

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

Figure 11. B3LYP optimized structures for conﬁgurations (a) ⁴1_LSand (b) ⁴1_HS, along with selected structural parameters. Bond lengths are given in

Ångstroms and bond angles are given in degrees.

computed to be smaller than J₁and, here, our data deviates

from the experimental estimates. For HS Co(II), the exchange

interactions between the metal and the radical centers are

antiferromagnetic (AF) and weak in nature, with J₁= −5 cm⁻¹,

whereas the radical−radical interaction is computed to be large

and ferromagnetic, with J₂= +109.7 cm⁻¹. This leads to an

eﬀective ground state of S = ¹/₂. With these estimates, the ﬁrst

excited S = ³/₂₅state lies 15.0 cm⁻¹above the ground state. The

remaining S = /₂and S = ³/₂states lie 40 cm⁻¹and 244 cm⁻¹

above the ground state, respectively. We have previously

reported a strong AF radical−radical super exchange in the

[Fe^II(L^•)₂](BF₄)₂complex,¹²so the strong ferromagnetic

interaction calculated is not without precedent and implies a

large orbital overlap between the magnetic orbitals in the HS

Co(II) ion and the π*_NOorbital of the ligand. Computation of

the J parameters on the optimized geometries also yield a

similar trend −albeit despite the fact that the magnitudes of the

J parameters are diminished (see Table S2 in the Supporting

Information).

The magnetic exchange of LS Co(II) with the radicals, and

the radical−radical super exchange via the LS Co(II), are

4

computed to be ferromagnetic for 1_LSwhile, for 1_HS, the

computed HS Co(II)−radical exchange is AF and the radical−

radical super exchange via the HS Co(II) is strongly

ferromagnetic. To probe the origin of this interaction further,

we have analyzed the electronic structure of these two species.

The magnetic orbitals of ⁴1_LSand ⁴1_HSare shown in Figure 12,

along with their computed energies. Our calculations on the

⁴1_LSconﬁguration clearly reveal that the unpaired electron in

2

Co(II) is located in the d_zorbital (see Figure 13), while the

radical unpaired electrons are located in the π*_NOorbitals,

which are perpendicular to the ﬁve-membered ring of the

2

radical ligand. As evident from Figure 12a, the d_zorbital of

Figure 12. Eigenvalue plots of the LS Co(II) and radical ligands along

with the DFT computed energies for the (a) 1_LSand (b) 1_HS

conﬁgurations.

4

Co(II) is orthogonal to the π*_NOorbital(s) of the nitroxide

ligand, yielding a ferromagnetic coupling between the radical

and the LS Co(II) center. The two radical π*_NOorbitals are

aligned parallel to each other with a very small overlap expected

between them. This leads to ferromagnetic coupling between

radical centers. These qualitative arguments are also supported

by overlap integrals computed between the MOs (see see Table

S3 in the Supporting Information).

4

The computed orbital diagram for the 1_HSconﬁguration is

shown in Figure 12b. For the HS Co(II) ion the following

e l e c t r o n i c c o n ﬁ g u r a t i o n w a s d e t e r m i n e d :

(d_xy)²(d_xz)²(d_yz)¹(d_z²)¹(d_{x −y}²)¹. As evidenced by Figure 12b,

2

the unpaired electron in the t_2gset is in the d_yzorbital, which

M

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

interacts with the symmetry compatible π*_NOorbitals. The

energy spacing between these two magnetic orbitals is large,

leading to a small overlap between the d_yzand π*_NOorbitals,

resulting in a weak AF interaction between the HS Co(II) ion

and the radical centers. On the other hand, the π*_NOorbitals,

which lie parallel to each other, interact rather strongly. A weak

overlap between the π*_NOorbital and the d_yzorbital of the

Co(II) ion promotes a polarization mechanism that leads to

strong ferromagnetic coupling between two radical centers. A

comparison between the radical −LS Co(II) and radical −HS

Co(II) interactions are shown in Scheme 4.

was estimated. The estimated spin density on the oxygen atoms

are much lower than that of the 1_LSconﬁguration, indicating

that, along the O−Co−O direction, there is some spin

polarization, as suggested earlier with the HS case. (See

Table S1 in the Supporting Information for computed spin-

density values.)

To validate our computed J values, we have calculated the

molar magnetic susceptibility using the HS and LS Co(II) DFT

computed values. Since spin crossover has been observed for

complex 1, the molar magnetic susceptibility of the complex at

a particular temperature would have contributions from both

LS and HS states of the metal ion, depending on their relative

abundance (mole fraction) at that temperature. The HS mole

fraction (x) can be calculated using the following equation⁵⁶at

a temperature T (given in Kelvin) and, thereby, the mole

fraction for LS would be of (1 − x).

4

Scheme 4. Illustrative Molecular Orbital (MO) Diagram

Depicting the Signiﬁcant Interactions between Radical and

Co(II) Center When Considering Low-Spin (LS) (Left) and

High-Spin (HS) (Right) States of Co(II) in 1

χT − (χT)LS

x =

(χT)HS − (χT)LS

Taking the DFT estimated values and assuming a g-value of

2.0 for both spin states, the above equations yield a

susceptibility curve that is strikingly similar to that of the

experiment (see Figure 14). Particularly, the shape of the curve

4

Computed spin density plots for 1_LSand 1_HSare shown in

Figure 13. For ⁴1_LS, the d_z²orbital shape of Co(II) is very clear

Figure 14. Plot of χ_MT vs T for 1 between 2 K and 300 K in (black

circles). The blue circles are DFT calculated points (see text for

details).

is very well reproduced. Given the fact that there are numerous

parameters involved, we have not attempted to ﬁt the

experimental data, but this simulation adds conﬁdence to the

computed results and suggests how the coupling between the

radical centers can be tuned by the nature of the spin state at

the metal center.

4

Figure 13. DFT computed spin density for the 1_LS(left) and 1_HS

(right) conﬁgurations. Blue represents spin up, and red represents spin

down.

4. CONCLUSIONS

The two related complexes of the nitroxide chelating ligand,

[Co^II(L^•)₂](NO₃)₂(1) and [Co^III(L⁻)₂](BPh₄) (2), have been

synthesized and structurally characterized. The Co ions in 1

and 2 display an octahedral geometry formed by equatorial

coordination of four nitrogens and axial coordination from two

oxygens of the radical ligand, forming a linear

from the spin density plot and the radical spin densities are

equally distributed to O and N atoms, with nitrogen possessing

slightly larger values (0.483 on oxygen vs 0.524 on nitrogen).

The spin density on the LS Co(II) ion is 1.012, suggesting that

spin delocalization is not prominent. For the ⁴1_HSstate, a cubic

shape was detected with a spin density of 2.74, revealing

signiﬁcant spin delocalization. On the radical centers, a spin

density of −0.378 on the oxygen and −0.552 on the nitrogen

L

^•/−···Co^II/III···L^•/−arrangement. The cations in 1 and 2 are

structurally similar but vary in regard to a few signiﬁcant bond

lengths and angles. Complex 1, [Co^II(L^•)₂](NO₃)₂, contains

N

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

the neutral radical form, L^•, with the cobalt in the low-spin

Co(II) state at low temperatures and undergoes a thermally

induced gradual spin crossover to the high-spin Co(II) state.

Complex 2, [Co^III(L⁻)₂](BPh₄), contains the reduced form, L⁻,

of the ligand with the cobalt in the low-spin Co(III) state

throughout the temperature range studied. The existence of the

cobalt spin/oxidation states and neutral or anionic form of the

ligand has been characterized by structural, solid-state magnetic

susceptibility, and EPR spectroscopic studies and conﬁrmed by

DFT calculations. Electrochemical studies support the existence

of 1 and 2 in solution and the interplay between the redox

states of 1 and 2 are fully described with the use of square

schemes. Complex 1 shows strong radical-cobalt and radical−

Basuli, F.; Pierpont, C. G. Inorg. Chem. 2002, 41, 5810. (e) Haga, M.;

Dodsworth, E. S.; Lever, A. B. P. Inorg. Chem. 1986, 25, 447.

(f) Kalinina, D.; Dares, C.; Kaluarachchi, H.; Potvin, P. G.; Lever, A. B.

P. Inorg. Chem. 2008, 47, 10110.

(4) Kaim, W.; Schwederski, B. Coord. Chem. Rev. 2010, 254, 1580.

(b) Kaim, W. Eur. J. Inorg. Chem. 2012, 343.

(5) Burgmayer, S. J. N.; Pearsall, D. L.; Blaney, S. M.; Moore, E. M.;

Sauk-Schubert, C. J. Biol. Inorg. Chem. 2004, 9, 59.

(6) (a) Okazawa, A.; Hashizume, D.; Ishida, T. J. Am. Chem. Soc.

2010, 132, 11516. (b) Vostrikova, K. E. Coord. Chem. Rev. 2008, 252,

1409. (c) Caneschi, A.; Gatteschi, D.; Rey, P. Prog. Inorg. Chem. 1991,

39, 331. (d) Bencini, A.; Gatteschi, D. EPR of Exchange Coupled

Systems; Springer−Verlag: Berlin, Heidelberg, 1990. (e) Caneschi, A.;

Gatteschi, D.; Sessoli, R. Acc. Chem. Res. 1989, 22, 392.

(7) Kaim, W. Inorg. Chem. 2011, 50, 9752.

radical ferromagnetic interactions below 200 K stabilizing a S =

3

(8) Caneschi, A.; Gatteschi, D.; Grand, A.; Laugier, J.; Pardi, L.; Rey,

P. Inorg. Chem. 1988, 27, 1031.

/

ground state with ac susceptibility measurements

2

conﬁrming ﬁeld-induced slow magnetic relaxation eﬀects.

Complex 2 is formed by a “reductively induced oxidation” of

the central cobalt ion by use of the tetraphenylborate anion and

it is diamagnetic over the temperature range studied.

The most important and signiﬁcant feature of complex 1 is

that it is a rare, if not unique, example of a hybrid molecular

magnetic material that displays the multiple functions of spin-

crossover, intramolecular ferromagnetic exchange, single-

molecule magnetism, and metal−ligand redox processes. Such

materials are of much current interest in developing nanoma-

terials therefrom and exploring both fundamental and applied

properties. The redox-active features of such metal-radical

compounds can also lead to applications in organic reactivity

and catalysis.⁵⁷

(9) Luneau, D.; Rey, P.; Laugier, J.; Fries, P.; Caneschi, A.; Gatteschi,

D.; Sessoli, R. J. Am. Chem. Soc. 1991, 113, 1245.

(10) Ito, A.; Nakano, Y.; Urabe, M.; Tanaka, K.; Shiro, M. Eur. J.

Inorg. Chem. 2006, 3359.

(11) Dickman, M. H.; Porter, L. C.; Doedens, R. J. Inorg. Chem.

1986, 25, 2595.

(12) Gass, I. A.; Gartshore, C. J.; Lupton, D. W.; Moubaraki, B.;

Nafady, A.; Bond, A. M.; Boas, J. F.; Cashion, J. D.; Milsmann, C.;

Wieghardt, K.; Murray, K. S. Inorg. Chem. 2011, 50, 3052.

(13) (a) Sheldrick, G. M. SHELXL-97. In Program for Reﬁnement of

Crystal Structures, University of Gottingen, Germany, 1997. (b) Spek,

̈

A. L. Acta Crystallogr., Sect. A: Found. Crystallogr. 1990, 46, C34.

(14) Wertz, J. E.; Orton, J. W.; Auzins, P. Faraday Discuss. 1961, 31,

140.

(15) (a) Hanson, G. R.; Gates, K. E.; Noble, C. J.; Griffin, M.;

Benson, S. J. Inorg. Biochem. 2004, 98, 903.

(16) Introduction to Computational Chemistry, 2nd Ed.; Jensen, F.;

John Wiley & Sons, Ltd.: Chichester, U.K., 2007.

ASSOCIATED CONTENT

* Supporting Information

Crystallographic data as .cif ﬁles. Figures S1 and S2 are

molecular structure pictures; Figures S3−S5 are magnetism

plots; Figure S6 shows electrochemical voltammograms; Tables

S1−S5 give computational (DFT) data. This material is

available free of charge via the Internet at http://pubs.acs.org.

■

S

(17) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;

Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci,

B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H.

P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.;

Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima,

T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A.;

Peralta, Jr. J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.;

Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.;

Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi,

J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J.

B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R.

E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.;

Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador,

AUTHOR INFORMATION

Corresponding Author

*E-mail: keith.murray@monash.edu.

Notes

■

The authors declare no competing ﬁnancial interest.

̈

P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.; Ortiz, J.

V.; Cioslowski, J.; Fox, D. J. Gaussian 09, Revision A.01, Wallingford,

CT, 2009.

(18) Noodleman, L.; Peng, C. Y.; Case, D. A.; Mouesca, J. M. Coord.

Chem. Rev. 1995, 144, 199.

(19) (a) Becke, A. D. J. Chem. Phys. 1986, 84, 4524. (b) Perdew, J. P.

Phys. Rev. B. 1986, 33, 8822.

(20) Berg, N.; Rajeshkumar, T.; Taylor, S. M.; Brechin, E. K.;

Rajaraman, G.; Jones, L. F. Chem.Eur. J. 2012, 18, 5906.

(21) (a) Rajeshkumar, T.; Singh, S. K.; Rajaraman, G. Polyhedron

2013, 52, 1299. (b) Singh, S. K.; Neeraj, K. T.; Rajaraman, G. Dalton

Trans 2011, 40, 10897. (c) Rajeshkumar, T.; Rajaraman, G. Chem.

Commun. 2012, 48, 7856. (d) Singh, S. K.; Rajaraman, G. Dalton

Trans. 2013, 42, 3612.

ACKNOWLEDGMENTS

■

K.S.M. acknowledges the support of an Australian Research

Council Discovery grant and, with G.R., the support of an

Australia-India Strategic Research Fund grant.

REFERENCES

■

(1) See articles in: Dithiolene Chemistry: Synthesis, Properties and

Applications; Karlin, K. D., Stiefel, E. I., Eds.; Progress in Inorganic

Chemistry, Vol. 52; Wiley: Hoboken, NJ, 2003.

(2) (a) Kaim, W.; Reinhardt, R.; Sieger, M. Inorg. Chem. 1994, 33,

4453. (b) Zal

Chem. 2003, 42, 5185. (c) Sieger, M.; Kaim, W.; Stufkens, D. J.;

Snoeck, T. L.; Stoll, H.; Zalis, M. Dalton Trans. 2004, 3815. (d) Ray,

K.; Petrenko, T.; Wieghardt, K.; Neese, F. Dalton Trans. 2007, 1552.

́ ̌

is, S.; Sieger, M.; Greulich, S.; Stoll, H.; Kaim, W. Inorg.

(22) Cremades, E.; Gomez-Coca, S.; Aravena, D.; Alvarez, S.; Ruiz, E.

J. Am. Chem. Soc. 2012, 134, 10532.

́

̌

(23) (a) Tewary, S.; Gass, I. A.; Murray, K. S.; Rajaraman, G. Eur. J.

Inorg. Chem. 2013, 1024. (b) Paulsen, H.; Duelund, L.; Winkler, H.;

Toftlund, H.; Trautwein, A. X. Inorg. Chem. 2001, 40, 2201.

(c) Paulsen, H.; Paulsen, H.; Zimmermann, A.; Averseng, F.;

Gerdan, M.; Winkler, H.; Toftlund, H.; Trautwein, A. X. Monatsh.

(e) Khusniyarov, M. M.; Weyhermuller, T.; Bill, E.; Wieghardt, K. J.

Am. Chem. Soc. 2009, 131, 1208.

(3) (a) Pierpont, C. G.; Lange, C. W. Prog. Inorg. Chem. 1994, 41,

331. (b) Pierpont, C. G. Coord. Chem. Rev. 2001, 95, 216. (c) Griffith,

W. P. Trans. Met. Chem. 1993, 18, 250. (d) Bhattacharya, S.; Gupta, P.;

̈

O

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

Chem. 2003, 134, 295. (d) Kripovic, I., K.; Zerara, M.; Daku, M. L.;

Vargas, A.; Enachescu, C.; Ambrus, C.; Tregenna-. Piggott, P.;

Amstutz, N.; Krausz, E.; Hauser, A. Coord. Chem. Rev. 2007, 251, 364.

(24) (a) Schafer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97,

2571. (b) Schafer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100,

5829.

(25) Molekel, Advanced Interactive 3D-Graphics for Molecular

Sciences; available via the Internet at http:\\www.cscs.ch/molkel/.

Version 1.6(build 338), Programming Zhurko, G.A.; available via the

Internet at www.chemcraftprog.com.

(49) Kremer, S.; Henke, W.; Reinen, D. Inorg. Chem. 1982, 21, 3013.

Pilbrow, J. R. Transition Ion Electron Paramagnetic Resonance;

Clarendon Press: Oxford, U.K., 1990; p 148.

(50) Makinen, M. W.; Kuo, L. C.; Yim, M. B.; Wells, G. B.;

Fukuyama, J. M.; Kim, J. E. J. Am. Chem. Soc. 1985, 107, 5245.

(51) Banci, L.; Bencini, A.; Gatteschi, D. Inorg. Chem 1983, 22, 2681.

(52) Dzuba, S. A. Phys. Lett. A 1996, 213, 77. A simulation of the

nitroxide radical spectrum published in ref gave g_x= 2.009, g_y= 2.006,

g_z= 2.002. (From J. F. Boas, unpublished data.)

(53) Nielsen, P.; Toftlund, H.; Bond, A. D.; Boas, J. F.; Pilbrow, J. R.;

Hanson, G. R.; Noble, C.; Riley, M. J.; Neville, S. M.; Moubaraki, B.;

Murray, K. S. Inorg. Chem. 2009, 48, 7033.

(54) Pilbrow, J. R. J. Magn. Reson. 1978, 31, 479.

(55) Vargas, A.; Zerara, M.; Krausz, E.; Hauser, A.; Daku, L. M. L. J.

(26) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005, 105,

2999−3093.

(27) Wong, M. A.; Frisch, M. J.; Wiberg, K. B. J. Am. Chem. Soc.

1991, 113, 4776.

Chem. Theory Comput. 2006, 2, 1342−1359.

(56) Kahn, O. Molecular Magnetism; VCH Publishers: New York,

1993. (ISBN: 1-56081-566-3.)

(57) Darmon, J. M.; Stieber, S. Ch. E.; Sylvester, K. T.; Fernandez, I.;

́

Lobkovsky, E.; Semproni, S. P.; Bill, E.; Wieghardt, K.; DeBeer, S.;

(28) Wong, M. A.; Frisch, M. J.; Wiberg, K. B. J. Am. Chem. Soc.

1992, 114, 523.

(29) (a) Borras

Tsukerblat, B. S. Inorg. Chem. 1999, 38, 6081. (b) Borras

J.; Clemente-Juan, J. M.; Coronado, E.; Tsukerblat, B. S. J. Comput.

Chem. 2001, 22, 985.

́

-Almenar, J. J.; Clemente-Juan, J. M.; Coronado, E.;

-Almenar, J.

́

Chirik, P. J. J. Am. Chem. Soc. 2012, 134, 1725.

(30) Nafady, A.; O’Mullane, A. P.; Bond, A. M.; Neufeld, A. K. Chem.

Mater. 2006, 18, 4375.

(31) Geiger, W. E. In Laboratory Techniques in Electrochemistry, 2nd

Edition; Kissinger, P. T., Heineman, W. R., Eds.; Marcel Dekker: New

York, 1996; Chapter 23.

(32) Schomaker, V.; Trueblood, K. N. Acta Crystallogr., Sect. B: Struct.

Crystallogr. Cryst. Chem. 1968, 24, 63.

(33) Goodwin, H. A. Top. Curr. Chem. 2004, 234, 23.

(34) Fujino, M.; Hasegawa, S.; Akutsu, H.; Yamada, J.; Nakatsuji, S.

Polyhedron 2007, 26, 1989.

(35) Porter, L. C.; Doedens, R. J. Inorg. Chem. 1985, 24, 1007.

(36) Porter, L. C.; Dickman, M. H.; Doedens, R. J. Inorg. Chem.

1986, 25, 678.

(37) Matsunaga, P. T.; McCall, D. T.; Carducci, M. D.; Doedens, R.

J. Inorg. Chem. 1990, 29, 1655.

(38) Porter, L. C.; Dickman, M. H.; Doedens, R. J. Inorg. Chem.

1988, 27, 1548.

(39) Kinoshita, H.; Akutsu, H.; Yamada, J.-I.; Nakatsuji, S. Inorg.

Chim. Acta 2008, 361, 4159.

(40) Dickman, M. H. Acta Crystallogr., Sect. C: Cryst. Struct. Commun.

1997, 53, 1192.

(41) Ahlers, C.; Dickman, M. H. Inorg. Chem. 1998, 37, 6337.

(42) Jaitner, P.; Huber, W. J. Organomet. Chem. 1983, 259, C1.

(43) Jaitner, P.; Huber, W.; Gieren, A.; Betz.. J. Organomet. Chem.

1986, 311, 379.

(44) Chilton, N. F.; Anderson, R. P.; Turner, L. D; Soncini, A.;

Murray, K. S. J. Comput. Chem. 2013, 34, 1164−1175.

(45) See, for example: Griffith, J. S. Struct. Bonding (Berlin) 1972, 10,

87. Jotham, R. W.; Kettle, S. F. A. Inorg. Chim. Acta 1970, 4, 145.

(46) (a) Vallejo, J.; Castro, I.; Ruiz-Garcia, R.; Cano, J.; Julve, M.;

Lloret, F.; De Munno, G.; Wernsdorfer, W.; Pardo, E. J. Am. Chem.

Soc. 2012, 134, 15704. (b) Feng, X.; Liu, J.; Harris, D.; Hill, S.; Long, J.

R. J. Am. Chem. Soc. 2012, 134, 7521. (c) Karasawa, S.; Yoshihara, D.;

Watanabe, N.; Nakano, M.; Koga, N. Dalton Trans. 2008, 1418.

(d) Zadrozny, J. M.; Liu, J.; Piro, N. A.; Chang, C. J.; Hill, S.; Long, J.

R. Chem. Commun. 2012, 48, 3927. (e) Sanakis, Y.; Pissas, M.;

Krzystek, J.; Telser, J.; Raptis, R. G. Chem. Phys. Lett. 2010, 493, 185.

(f) Kajiwara, T.; Nakano, M.; Kaneko, Y.; Takaishi, S.; Ito, T.;

Yamashita, M.; Igashira-Kamiyama, A.; Nojiri, H.; Ono, Y.; Kojima, N.

J. Am. Chem. Soc. 2005, 127, 10150. (g) Tancini, E.; Rodriguez-

Douton, M. J.; Sorace, L.; Barra, A.-L.; Sessoli, R.; Cornia, A. Chem.

Eur. J. 2010, 16, 10482.

(47) Krivokapic, I.; Zerara, M.; Daku, M. L.; Vargas, A.; Enachescu,

C.; Ambrus, C.; Tregenna-Piggott, P.; Amstutz, N.; Krausz, E.; Hauser,

A. Coord. Chem. Rev. 2007, 251, 364.

(48) Abragam, A.; Bleaney, B. Electron Paramagnetic Resonance of

Transition Ions; Clarendon Press: Oxford, U.K., 1970; pp 447 and

751−752.

P

dx.doi.org/10.1021/ic400565h | Inorg. Chem. XXXX, XXX, XXX−XXX

Article Doi

Observation of ferromagnetic exchange, spin crossover, reductively induced oxidation, and field-induced slow magnetic relaxation in monomeric cobalt nitroxides

DOI: 10.1021/ic400565h

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ic400565h

Full text of DOI:10.1021/ic400565h

Products guided by the article

R&D Labs maybe for 1443533-86-5

Relevant to this article

Hot Product