Optical investigation of spin-crossover in cobalt(II) bis-terpy complexes

(1)

Article

Reference

Optical investigation of spin-crossover in cobalt(II) bis-terpy complexes

ENACHESCU, Cristian, et al.

Abstract

The spin transition of the [Co(terpy)2]2+ complex (terpy = 2,2':6',2″-terpyridine) is analysed based on experimental data from optical spectroscopy and magnetic susceptibility measurements. The single crystal absorption spectrum of [Co(terpy)2](ClO4)2 shows an asymmetric absorption band at 14 400 cm−1 with an intensity typical for a spin-allowed d–d transition and a temperature behaviour typical for a thermal spin transition. The single crystal absorption spectra of suggest that in this compound, the complex is essentially in the high-spin state at all temperatures. However, the increase in intensity observed in the region of the low-spin MLCT transition with increasing temperature implies an unusual partial thermal population of the low-spin state of up to about 10% at room temperature. Finally, high-spin → low-spin relaxation curves following pulsed laser excitation for [Co(terpy)2](ClO4)2 dispersed in KBr discs, and as a comparison for the closely related [Co(4-terpyridone)2](ClO4)2 spin-crossover compound are given.

ENACHESCU, Cristian, et al . Optical investigation of spin-crossover in cobalt(II) bis-terpy complexes. Inorganica Chimica Acta , 2007, vol. 360, no. 13, p. 3945-3950

DOI : 10.1016/j.ica.2007.06.022

Available at:

http://archive-ouverte.unige.ch/unige:3585

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

Optical investigation of spin-crossover in cobalt(II) bis-terpy complexes

Cristian Enachescu

^a,*

, Itana Krivokapic

^b

, Mohamed Zerara

^b

, Jose Antonio Real

^c

, Nahid Amstutz

^b

, Andreas Hauser

^b,*

aDepartment of Solid State and Theoretical Physics, ‘‘Alexandru Ioan Cuza’’ University, 11 Blvd. Carol, R-700506 Iasi, Romania

bDe´partement de chimie physique, Universite´ de Gene`ve, Baˆtiment de Science 2, 30 quai Ernest-Ansermet, CH-1211 Gene`ve 4, Switzerland

cInstitut de Ciencia Molecular/Departament de Quimica Inorganica, Universitat de Valencia, P.O. Box 22085, 46071 Valencia, Spain Received 12 December 2006; received in revised form 8 June 2007; accepted 17 June 2007

Available online 28 June 2007

Paper presented in the MAGMANet-ECMM, European Conference on Molecular Magnetism, that took place last October 10–15 in Tomar, Portugal.

Abstract

The spin transition of the [Co(terpy)₂]²⁺complex (terpy = 2,2⁰:6⁰,2⁰⁰-terpyridine) is analysed based on experimental data from optical spectroscopy and magnetic susceptibility measurements. The single crystal absorption spectrum of [Co(terpy)2](ClO4)2shows an asymmetric absorption band at 14 400 cm¹with an intensity typical for a spin-allowed d–d transition and a temperature behaviour typical for a thermal spin transition. The single crystal absorption spectra of suggest that in this compound, the complex is essentially in the high- spin state at all temperatures. However, the increase in intensity observed in the region of the low-spin MLCT transition with increasing temperature implies an unusual partial thermal population of the low-spin state of up to about 10% at room temperature. Finally, high- spin!low-spin relaxation curves following pulsed laser excitation for [Co(terpy)2](ClO4)2dispersed in KBr discs, and as a comparison for the closely related [Co(4-terpyridone)₂](ClO₄)₂spin-crossover compound are given.

Keywords: Spin-crossover; Cobalt(II) bis-terpyridine; Absorption spectra; Magnetic susceptibility; High-spin!low-spin relaxation

1. Introduction

Spin-crossover solids [1] are molecular compounds of transition metal ions with electronic configurations d⁴, d⁵, d⁶, and d⁷, switchable between two states with different optical and magnetic properties: the low-spin (LS) state with maximum number of paired and the high-spin (HS) state with maximum number of unpaired d electrons. The occurrence of a thermal spin transition can be readily understood based on simple thermodynamic consider- ations [2]: if the zero-point energy difference between the two states is small and such that the low-spin state is the quantum mechanical ground state, an entropy driven transition from the low-spin state at low temperatures to the

high-spin state at higher temperatures may be observed.

The large majority of known spin-crossover complexes are based on a Fe(II) as central ion, but other transition metal ions also give rise to spin-crossover complexes, such as Co(II), Co(III), Mn(II), Mn(III), Cr(II) or Fe(III). To date, only a few detailed studies discuss the class of Co(II) spin-crossover complexes [3].

The [Co(terpy)]²⁺ complex (terpy = 2,2⁰:6⁰,2⁰⁰-terpyridine) is one of the ﬁrst and most important cobalt(II) spin-crossover complexes discovered to date [4,5], for which the transition occurs from the²E low-spin manifold to the⁴T1high-spin manifold. It has a two tridentate coordination motif. In the low-spin state, the equatorial metal to ligand bond length is 0.17 A˚ longer than the axial bond length in the low-spin state [6]. This results in a stabilisation of the low-spin state as a consequence of the Jahn–Teller eﬀect [7]. The approximate symmetry of the complex isD2dwith an additional Jahn–Teller distortion[7]

doi:10.1016/j.ica.2007.06.022

* Corresponding authors. Tel.: +0040 232201175; fax: +0040 232201205.

E-mail addresses:[email protected](C. Enachescu),andreas.

[email protected](A. Hauser).

www.elsevier.com/locate/ica

(3)

resulting in a symmetry lowering toC2v. However, this distortion is small, and for the sake of simplicity will be neglected in the ensuing discussion. The transition to the high-spin state is accompanied by a substantial change in metal–ligand bond lengths: the axial bond length increases by 0.21 A˚ , the equatorial one by 0.08A˚[6]. As a result, the distortion of the coordination octahedron is much less pro- nounced in the high-spin state.

Previous investigations on [Co(terpy)]²⁺showed that its spin-crossover behaviour depends crucially on the anion and on solvent content. The anhydrous [Co(terpy)2] (ClO4)2, for instance, is fully in the low-spin state at low temperatures and shows a transition to the high-spin state with a thermal population80% at room temperature[4,5]. The various hydrated forms, on the other hand, show diﬀerent degrees of partial spin transitions with remnant high-spin fractions at low temperature, or temperature independent populations of the two spin states. In this paper, we present experimental results derived mainly from optical absorption measurements on the [Co(terpy)]²⁺complex with two diﬀer- ent anions, CLO₄and PF₆, respectively, in the absence of any solvent molecules in the crystal lattices.

2. Experimental

Single crystals of [Co(terpy)2](ClO4)2 and [Co(terpy)2]- (PF6)2 were grown from ethanol/acetonitrile solutions according to literature methods[4]. Red, transparent crystals of square pyramidal shape were obtained, which are dichroic in the case of [Co(terpy)2](PF6)2. As the presence of water molecules in the unit cell may inﬂuence the geom- etry and the local electronic structure of the title complexes [4,5], the synthesis was performed in water free solvents in order to synthesise the compounds without any lattice water. Elemental analyses conﬁrmed the composition without any lattice solvent molecules. The PF₆ salts of Co²⁺

and Zn²⁺ are isomorphous and crystallise in the space groupP421c, with the molecular S4axis parallel to the crystal c-axis[8].

The [Co(terpy)2](ClO4)2solution for spectroscopic measurements was prepared in a mixture of proprionitrile/

butyronitrile (4:5), and corresponding absorption spectra were recorded on a Cary50 spectrometer.

For optical measurements single crystals with a thickness varying from 90 to 140lm were mounted so as to cover entirely a small aperture in a copper sample holder.

For low temperature measurements, the sample holder was inserted into a closed cycle cryostat capable of achiev- ing temperatures down to 10 K, with the sample sitting in 1 bar of He exchange gas for eﬃcient cooling. Complete absorption spectra of crystals, that is, from 6000 to 27 000 cm¹, were recorded on an FT-IR spectrometer (Bruker IFS66/S). The source light was provided either by a tungsten halogen or a xenon lamp, and the signal was measured using a Ge (6000–12 000 cm¹), a Si (9000- 21 000 cm¹) or a GaP (18 000–27 500 cm¹) diode detectors.

Relaxation curves were recorded on [Co(terpy)2](ClO4)2

dispersed in KBr by monitoring the transient bleaching of the MLCT band of the low-spin species at 505 nm, following pulsed excitation with the second harmonic of a Q-switched Nd:YAG laser at 532 nm, with a repetition rate of 20 Hz. The light provided by a 50 W tungsten halogen lamp was used as probe beam. After passing through the sample, the probe beam was dispersed in a 1/4 m mono- chromator (Spex 280M), then ampliﬁed with a photo-mul- tiplier (R928). Transient absorption curves were recorded using a digital oscilloscope (Tektronix TDS540B).

3. Results and discussion 3.1. The thermal spin transition

The temperature dependence of the solution spectrum of [Co(terpy)₂](ClO₄)₂ in proprionitrile/butyronitrile (4:5) at room temperature is presented inFig. 1a. Absorption max-

Fig. 1. (a) Temperature dependence of the electronic absorption spectrum for a 10³M solution of [Co(terpy)2](ClO4)2in proprionitril/butyronitril (4:5). (b) Temperature dependent absorption spectra of single crystals of [Co(terpy)2](ClO4)2between 10 and 300 K.

3946 C. Enachescu et al. / Inorganica Chimica Acta 360 (2007) 3945–3950

(4)

ima are found at 15 000 cm¹ (e= 86 l mol¹cm¹), 18 200 cm¹(e= 405 l mol¹cm¹), 19 800 cm¹(e= 1100 l mol¹ cm¹), and 22 600 cm¹ (e= 1285 l mol¹cm¹).

The three more intense peaks have alternatively been assigned either to metal to ligand charge-transfer (MLCT) transitions[7]or to d–d transitions[5,7]. The broad band between 13 500 and 15 500 cm¹ has unanimously been attributed to the transitions from the ²A1 component of the low-spin²E ground state in compressedD2dsymmetry to low-symmetry and spin-orbit levels arising from the²T1

and²T2ligand-ﬁeld states of octahedral parentage[5], using the energy level diagram of Liehr[9].

The maximum absorbances at 18 200, 19 800 and 22 600 cm¹ corresponding to MLCT transitions of the low-spin species decrease in intensity when the temperature increases, in accordance with a classical thermal spin transition. Likewise, the d–d band of the low-spin species between 13 500 and 15 500 cm¹decreases with increasing temperature. Indeed, previous studies using magnetic measurements and the Evans technique[10]have shown that in solution at room temperature about 30% of the Co²⁺complexes are in the low-spin state[7,11].

Fig. 1b presents the temperature dependence of the absorption spectrum of a single crystal of [Co(terpy)2]- (ClO4)2of 100lm thickness. As before, the absorption band between 13 500 and 15 500 cm¹with e55 l mol¹cm¹ decreases in intensity with increasing temperature as a result of the thermal spin transition. In order to obtain the MLCT bands at higher energies havingevalues of up to 2500 l mol¹cm¹, a crystal of a thickness not exceeding 2lm would have to be used. This is not feasible.

Fig. 2 depicts the absorption spectra of [Co(terpy)2]- (PF₆)₂ at diﬀerent temperatures for two polarisation modes, that is parallel and perpendicular to the tetragonal axis. The low temperature absorption spectra, as well as the temperature dependence of the intensity are surprisingly diﬀerent from those of the perchlorate salt. In the region of the MLCT transitions of the low-spin species, the peaks that appear at the same energies as for the perchlorate salt increase in intensity for increasing temperatures. Likewise, the absorption band between 13 500 and 15 500 cm¹corresponding to the low-spin ²A1!²T1, ²T2 d–d transitions increases in intensity with increasing temperature, with vir- tually zero intensity at 10 K. This suggests that [Co(terpy)2]- (PF6)2has a high-spin ground state and shows an unusual partial population of the low-spin state at elevated temperatures. This is supported by the observation of additional absorption bands, centred at, respectively, 9200 and 11 600 cm¹, which can be assigned to the transitions from the ⁴E ground state component of the tetragonally split

4T1ðt⁵_2ge²_gÞ manifold to the ⁴E and ⁴B2 components of the

4T2ðt⁴_2ge³_gÞ state. They are presented more detailed in the insets ofFig. 3for the two polarisation modes.

To summarise these results,Fig. 3presents a comparison between representative single crystal absorption spectra for [Co(terpy)2](PF6)2and [Co(terpy)2](ClO4)2at 10 K

and at 300 K (top) and the dependence of the intensity on the temperature measured at 18 200 cm¹ (bottom). The average intensity for the dichroic crystals of [Co(terpy)2]- (PF6)2 at 18 200 cm¹ was obtained using the relation:

e= (ei+ 2e^)/3. At room temperature this value is 120 l mol¹cm¹, that is, about three times lower that the corresponding value for the absorption spectrum of [Co(terpy)2](ClO4)2in solution. Thus, we can deduce that the low-spin fraction for [Co(terpy)₂](PF₆)₂at room temperature must be around 10%.

The magnetic susceptibility measurements, presented in Fig. 4, conﬁrm the conclusions obtained form the analysis of the optical spectra. The curve corresponding to [Co- (terpy)2](ClO4)2 shows the typical behaviour for a cobal- t(II) spin-crossover complex, with vT close to 0.4 cm³ mol¹K at lower temperature (spin-only value for the low-spin state) and around 2.4 cm³mol¹K at room temperature, indicating a high-spin fraction of around 80%

[4,5]. The curve for [Co(terpy)2](PF6)2is characteristic for a high-spin compound at low temperatures. An unexpected

Fig. 2. Temperature dependence of single crystal absorption spectra for [Co(terpy)2](PF6)2inE_^(top) andEi(bottom) polarisations with respect to the C4-axis of the crystal.

(5)

decrease in vT for [Co(terpy)2](PF6)2 at higher temperatures conﬁrms that for this compound the low-spin state may become partially populated at higher temperatures.

This unusual partial thermal population of the low-spin state for a system with a high-spin ground state as observed for [Co(terpy)2](PF6)2 can be explained by the compara- tively small differences in the electronic properties of the two spin states of cobalt(II). In contrast to iron(II) spin- crossover or high-spin complexes, for which the high-spin state is highly favoured by the entropic term that leads to a negligible low-spin population at high temperature, for cobalt(II) complexes the ratios of the electronic degenera- cies and the vibrational densities of state are smaller. The latter is due to the smaller average bond-length difference between the two states of 0.1 A˚ [12,13] in comparison to the typical value of 0.2 A˚ for iron(II) complexes [14,15]. For this reason, the cobalt(II) complexes are more sensitive to the local environment, that is, anion or solvent effects, and thus in the two compounds the [Co(terpy)2]²⁺

complex shows opposite behaviours.

A more detailed understanding of the electronic structure of the two spin states is provided by the powder EPR spectra (X-band) of the dilute mixed systems [Zn_1xCox(terpy)2](PF6)2 and [Fe_1xCox (terpy)2](PF6)2

(x= 2–4%) at 4.2 K. The values obtained for theg values for the high-spin state of the former and for the low-spin state of the latter are in line with a tetragonally compressed octahedron reﬂecting the shorter axial Co–N bond lengths as compared to the equatorial bond lengths[16].

3.2. The HS!LS relaxation

The HS!LS relaxation in cobalt(II) spin-crossover compounds has been the object of only very few papers.

Previous studies, using the ‘‘laser temperature jump’’ technique, have been performed and a relaxation time of around 30 ns was suggested as upper limit at room temperature[17,18]. Ultrasonic relaxation measurements, realised by Beattie[17]at room temperature in solution conﬁrmed that the HS!LS relaxation is very fast with a value of around a few nanoseconds.

As stated previously, the intense absorption of cobalt(II) complexes at 532 nm, where the pulsed laser irradiation was performed, would imply the use of very thin crystals for transient transmission measurements. In order to avoid this problem, relaxation measurements were performed on [Co(terpy)₂](ClO₄)₂dispersed in KBr discs.

First of all, we veriﬁed whether the thermal behaviour of the compound dispersed in KBr is compatible with the one of a single crystal. The absorption spectra for [Co(terpy)2]- (ClO4)2in a KBr disc as function of temperature are presented inFig. 5. Apart from the sloping baseline due to dif- fuse scattering, they are very similar to those obtained for a crystal. The transition curve seems slightly more gradual in the case of the KBr disc (see inset ofFig. 5), which is a well- known eﬀect for spin-crossover systems due to grinding [19].

Fig. 3. (Top) comparison between the absorption spectra for [Co(terpy)2]- (PF6)2and [Co(terpy)2](ClO4)2at 10 K and at 300 K; (bottom) temperature dependence of the absorption intensity at 18 200 cm¹for crystalline [Co(terpy)2](PF6)2(average of the two polarisations) and for [Co(terpy)2]- (ClO4)2in solution.

Fig. 4. The productvTas function of temperature for [Co(terpy)2](ClO4)2

(open circles) and for [Co(terpy)2](PF6)2 (close circles). All curves are corrected for the diamagnetic contribution.

(6)

Fig. 6a shows representative transient absorption curves at 10 and 50 K following pulsed laser excitation at 532 nm, that is, into the MLCT band of the low-spin species. The

bleaching at 500 nm indicates a transient depopulation of the low-spin state. In the inhomogeneous matrix of the KBr disc, the relaxation curves, indicative of the

Fig. 5. Temperature dependence of [Co(terpy)2](ClO4)2absorption spectra in KBr discs. Inset: Comparison between band intensity temperature variations for [Co(terpy)2](ClO4)2in crystals and in KBr discs.

Fig. 6. (a) Representative relaxation curves for [Co(terpy)2](ClO4)2 in a KBr disc. (b) Relaxation curve of [Co(4-terpyridone)2](ClO4)2measured on a 8lm thick crystal at 12 K (excitation at 532 nm, probe beam at 500 nm). Inset: relaxation curve for [Co(4-terpyridone)2](ClO4)2at 10 K in a KBr disc.

Fig. 7. (a) Temperature dependence of the [Co(4-terpyridone)2](ClO4)2

absorption spectrum of a single crystal; (b) temperature dependence of the [Co(4-terpyridone)2](ClO4)2absorption spectrum dispersed in KBr discs.

Inset: relative intensity of the MLCT absorption band of the low-spin species as a function of temperature for [Co(4-terpyridone)2](ClO4)2in a single crystal (circles) and dispersed in KBr (triangles).

Fig. 8. Logarithmic plot of relaxation rateskHLvs. 1/Tfor the HS!LS relaxation for [Co(4-terpyridone)2](ClO4)2 in KBr discs (circles) and crystals (squares).

(7)

high-spin!low-spin relaxation, are not mono-exponential due to inhomogeneous broadening. In order to estimate an average relaxation rate constant, either a biexponential ﬁt or an integral method can be used[20], both giving very similar results. At temperatures above 50 K, the relaxation time is less than 10ls; it increases up to 400ls at 10 K.

For comparison, a similar study was performed on the spin-crossover compound [Co(4-terpyridone)2](ClO4)2[21].

This compound shows all the absorption bands present in [Co(terpy)2](ClO4)2, as one can see inFig. 7, and the temperature dependence of the spectra for a singe crystal and dispersed in KBr discs are fully compatible. The relaxation curve shown inFig. 6b, was measured at 12 K on a very thin crystal[22]. This curve is perfectly mono-exponential, and the ﬁt gives a value around 50ls for the decay.

In analogy to [Co(terpy)2](ClO4)2, further relaxation measurements were performed on [Co(4-terpyridone)₂](ClO₄)₂ dispersed in KBr. As for [Co(terpy)₂] (ClO4)2, the thermal transition curve is slightly more gradual than for the crystal according to the inset of Fig. 7b, which displays the absorption intensity of the MLCT band of the low-spin species as a function of temperature.

The relaxation curve of [Co(4-terpyridone)2](ClO4)2dis- persed in KBr at 10 K is shown inFig. 7b. As in the case of [Co(terpy)2](ClO4)2in KBr discs, the relaxation curve is not mono-exponential and the same procedure as above was used to extract an average relaxation rate constant. The relaxation rate constants obtained in this way are compatible both to those determined for [Co(terpy)2](ClO4)2and to [Co(4-terpyridone)2](ClO4)2in the form of a single crystal. All rate constants are displayed inFig. 8as ln(kHL) ver- sus 1/T. As expected, at the lowest temperatures the relaxation proceeds entirely via a temperature independent tunnelling mechanism, at higher temperatures it is ther- mally activated [23]. In contrast to iron(II) spin-crossover compounds, for which the low-temperature tunnelling rate constants are of the order of 10⁶–10²s¹[24], and to iro- n(III) spin-crossover compounds with values in the range of 10¹–10²s¹ [20,25], the high-spin!low-spin relaxation rate constant of 10⁴s¹ for the compounds of this study is substantially larger. This is due to the shorter difference in metal–ligand bond lengths between the two spin states for cobalt(II) complexes of 0.1 A˚ , as compared to the values of 0.15 and 0.2 A˚ for iron(III) and iron(II), respectively.

4. Conclusions

In this paper, we presented a detailed optical investigation of the Co^II spin-crossover compounds [Co(terpy)2]

(ClO4)2and [Co(terpy)2](PF6)2, and we analysed their different thermal behaviour. In addition, we presented relaxation curves for the high-spin!low-spin relaxation following pulsed laser excitation in [Co(terpy)2](ClO4)2, and we correlate the data obtained with data for the related compound [Co(4-terpyridone)2](ClO4)2.

Acknowledgements

Acknowledgements are due to the MAGMANet Net- work of Excellence of the European Union (Contract:

NMP3-CT-2005-515767-2) and to the Swiss National Sci- ence Foundation for ﬁnancial support. C.E. thanks CNC- SIS Romania for a CEEX research grant. J.A.R. thanks the Spanish Ministerio de Educacion y Ciencia (MEC) for the Project CTQ 2004-03456/BQU.

References

[1] P. Gu¨tlich, H.A. Goodwin, Topics Curr. Chem. 233 (2004) 1.

[2] C.P. Slichter, H.G. Drickamer, J. Chem. Phys. 56 (1972) 2142.

[3] H.A. Goodwin, Topics Curr. Chem. 234 (2004) 23.

[4] R. Hogg, R.G. Wilkins, J. Chem. Soc (1962) 341.

[5] J.S. Judge, W.A. Baker, Inorg. Chim. Acta 1 (1967) 68.

[6] B.N. Figgis, E.S. Kucharski, A.H. White, Aust. J. Chem. 36 (1983) 1537.

[7] S. Kremer, W. Henke, D. Reinen, Inorg. Chem. 21 (1982) 3013.

[8] P. Pattison, EPFL Lausanne, unpublished results.

[9] A.D. Liehr, J. Phys. Chem. 67 (1963) 1314.

[10] D.F. Evans, J. Chem. Soc. Chem. Commun. (1959) 2003.

[11] W. Henke, S. Kremer, Inorg. Chim. Acta 65 (1982) L115.

[12] P. Thuery, J. Zarembowitch, A. Michailowicz, O. Kahn, Inorg.

Chem. 26 (1987) 851.

[13] R. Sieber, S. Decurtins, H. Stoeckli Evans, C. Wilson, D. Yuﬁt, J.A.K. Howard, S.C. Capelli, A. Hauser, Chem. Eur. J. 6 (2000) 361.

[14] J.F. Letard, P. Guionneau, L. Rabardel, J.A.K. Howard, A.E. Goeta, D. Chasseau, O. Kahn, Inorg. Chem. 37 (1998) 4432.

[15] D. Chernyshov, M. Hostettler, K.W. Tornroos, H.B. Bu¨rgi, Angew.

Chem. Int. Ed. 42 (2003) 3825.

[16] I. Krivokapic, M. Zerara, M. Lawson Daku, A. Vargas, C.

Enachescu, C. Ambrus, P. Tregenna-Piggott, N. Amstutz, E. Krausz, A. Hauser, Coord. Chem. Rev. 251 (2007) 364.

[17] J.K. Beattie, Adv. Inorg. Chem. 27 (1988) 1.

[18] M.R. Grace, T.W. Swaddle, Inorg. Chem. 32 (1993) 5597.

[19] E.W. Muller, H. Spiering, P. Gutlich, Inorg. Chem. 23 (1984) 119.

[20] C. Enachescu, A. Hauser, J.J. Girerd, M.L. Boillot, ChemPhysChem 7 (2006) 1127.

[21] A.B. Gaspar, C.M. Munoz, V. Niel, J.A. Real, Inorg. Chem. 40 (2001) 9.

[22] M. Zerara, Ph.D. Thesis, Universite de Geneve, 2003.

[23] A. Hauser, Chem. Phys. Lett. 124 (1986) 543.

[24] A. Hauser, Topics Curr. Chem. 234 (2004) 155.

[25] S. Schenker, A. Hauser, R.M. Dyson, Inorg. Chem. 35 (1996) 4676.